Downloaded from https://academic.oup.

com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023

Grain Boundary Character Distribution of
Nanocrystalline Cu Thin Films Using Stereological
Analysis of Transmission Electron Microscope
Orientation Maps
AD Darbal, KJ Ganesh, X Liu, S-B Lee, J Ledonne, T Sun, B Yao, AP Warren,
GS Rohrer, AD Rollett, PJ Ferreira, KR Coffey, K Barmak
Microsc. Microanal. 19, 111–119, 2013
doi:10.1017/S1431927612014055 Microscopy AND

Microanalysis
© MICROSCOPY SOCIETY OF AMERICA 2013

Grain Boundary Character Distribution of Nanocrystalline

Cu Thin Films Using Stereological Analysis of
Transmission Electron Microscope Orientation Maps

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
A.D. Darbal,1,† K.J. Ganesh,2,‡ X. Liu,1 S.-B. Lee,1 J. Ledonne, 1 T. Sun,2,# B. Yao,2,§ A.P. Warren,3
G.S. Rohrer,1 A.D. Rollett,1 P.J. Ferreira,2 K.R. Coffey,3 and K. Barmak 1, *
1
Materials Research Science and Engineering Center, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213,
USA
2
Materials Science and Engineering Program, The University of Texas at Austin, 1 University Station, Austin, TX 78712, USA
3
Advanced Materials Processing and Analysis Center, University of Central Florida, 4000 Central Florida Boulevard, Orlando,
FL 32816, USA

Abstract: Stereological analysis has been coupled with transmission electron microscope ~TEM! orientation
mapping to investigate the grain boundary character distribution in nanocrystalline copper thin films. The use
of the nanosized ~,5 nm! beam in the TEM for collecting spot diffraction patterns renders an order of
magnitude improvement in spatial resolution compared to the analysis of electron backscatter diffraction
patterns in the scanning electron microscope. Electron beam precession is used to reduce dynamical effects and
increase the reliability of orientation solutions. The misorientation distribution function shows a strong
misorientation texture with a peak at 608/@111#, corresponding to the S3 misorientation. The grain boundary
plane distribution shows $111% as the most frequently occurring plane, indicating a significant population of
coherent twin boundaries. This study demonstrates the use of nanoscale orientation mapping in the TEM to
quantify the five-parameter grain boundary distribution in nanocrystalline materials.
Key words: stereology, orientation mapping, grain boundary character distribution, precession microscopy

I NTR ODUCTION is due to the fact that the spatial resolution of SEM-EBSD
orientation imaging is limited to tens of nanometers ~35 6 5
Crystal orientation imaging using automated analysis of elec- in the x direction and 90 6 15 in the y direction!, as deter-
tron backscattered diffraction ~EBSD! patterns in the scan- mined by Zaefferer ~2007!, which is not adequate to map the
ning electron microscope ~SEM! is well established and shapes of grain boundaries with lengths smaller than a few
widely used ~Adams et al., 1993; Dingley, 2004!. With auto- hundred nanometers. This provides the motivation for em-
mated serial sectioning using focused ion beam milling, one ploying the transmission electron microscope ~TEM! for
can obtain three-dimensional ~3D! orientation data from large-scale orientation imaging and grain boundary charac-
which it is possible to extract the complete 3D network of terization. The use of the TEM provides an order of magni-
grain boundaries ~Rohrer et al., 2010b!. Alternately, stereolog- tude improvement in the spatial resolution, thereby making
ical analysis of orientation maps from planar sections can be it possible to study polycrystalline materials not possible by
employed to obtain the average properties of the grain SEM-EBSD analysis ~Dingley, 2006; Ganesh et al., 2010, 2012;
boundary network ~Saylor et al., 2004b!. Methods such as Liu et al., 2011!. The objective of this work is to demonstrate
these have been useful in investigating the crystallography of the quantitative investigation of grain boundary character
grain boundaries and the impact of the distribution of grain distribution in thin Cu films by combining stereological analy-
boundary types on the properties of polycrystalline materi- sis and the use of a ,5 nm probe ~Ganesh et al., 2010!, with
als ~Rohrer et al., 2004a!. However, the use of SEM-EBSD the recently developed TEM orientation imaging technique
orientation imaging is not reliable for the purpose of investi- known as ASTAR TM ~Rauch & Duft, 2005; Rauch & Dupuy,
gating grain boundaries in polycrystalline materials where 2005; Rauch & Veron, 2005; Rouvimov et al., 2008!.
the average grain size is in the 100 nm range. This limitation ASTAR TM was developed recently for automated orien-
tation determination by fast acquisition of spot diffraction
Received April 24, 2012; accepted September 9, 2012 patterns in the TEM. In ASTAR TM, a dedicated hardware
*Corresponding author. E-mail: [email protected] unit is used for precession and automated scanning of a
†
A.D. Darbal, formerly at Carnegie Mellon University, is now at NanoMEGAS USA, nanosized quasi-parallel electron beam probe. A high speed
Tempe, AZ, USA.
‡
K.J. Ganesh, formerly at the University of Texas at Austin, is now at Intel Corpora- external optical camera is then used for rapid acquisition of
tion, Hilsboro, OR, USA. spot diffraction patterns. The acquired spot patterns are
#
T. Sun, formerly of the University of Central Florida, is now at Integrated System indexed automatically using a template matching algorithm.
Ltd., Wanchai, Hong Kong.
§
B. Yao, formerly of the University of Central Florida, is now at the Pacific Northwest Significant improvement in the reliability of the orientation
National Laboratory. maps is achieved with electron beam precession ~Rouvimov
112 A.D. Darbal et al.

et al., 2008!. The use of precession reduces the dynamical estimate of the relative areas of grain boundary planes with
effects and increases the number of spots in the diffraction all possible orientations ~Saylor et al., 2004a!.
pattern ~Vincent & Midgley, 1994; Oleynikov et al., 2007; To emphasize, the important requirement that needs to
Portillo et al., 2010!. The use of rapidly acquired spot be taken into account for the stereological analysis of the
patterns and the robust template matching algorithm make grain boundary character distribution is the need for a
ASTAR TM highly suitable for obtaining large datasets of sufficient number of grain boundary traces within the plane
crystal orientations and extracting grain boundary section. The number of traces needed depends on the
distributions. resolution of the five parameters and the crystal symmetry.

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
In the macroscopic description of grain boundaries, For a cubic system with the five parameters resolved with a
which forms the basis of the stereological analysis used resolution of 108, approximately 50,000 grain boundary
here, a general grain boundary is characterized by five traces are sufficient ~Saylor et al., 2004a!. Given the advance-
parameters: three parameters to specify the lattice misori- ments in automation of modern SEM-EBSD systems, this
entation Dg between the adjoining crystals meeting at the requirement is not difficult to satisfy. However, automated
grain boundary and two parameters to specify the inclina- orientation mapping in the TEM is a relatively new develop-
tion of the grain boundary plane normal n. The distribu- ment, and acquisition of orientation data at the scale re-
tion l~Dg, n!, known as the grain boundary character quired for this analysis is challenging. Also, it should be
distribution, gives the relative area of the boundaries with noted that stereological analysis assumes that sample has
misorientation Dg and boundary normal n. The grain random orientation texture, and if there is a misorientation
boundary character distribution is measured in units of texture, then the grain boundary character distribution is
multiples of a random distribution ~MRD!, where values accurate only for peaks in the misorientation texture, i.e.,
greater than 1 indicate that the grain boundary type occurs where there is sufficient information to carry out the analysis.
more frequently than expected in a random distribution. In the next section, the samples studied and details of
Apart from the distribution of boundaries in the five- the orientation imaging experiments are discussed. A de-
dimensional space, it is also useful to consider the distribu- tailed discussion on calibrating the frame of reference for
tion of internal grain surfaces l~n!, which gives the the orientation maps has been included. It may be noted
distribution of grain boundary planes in the crystal frame that so far the use of ASTAR TM has been limited to the
of reference, averaged over all misorientations ~Rohrer study of local misorientation texture. Therefore, not much
et al., 2004b!. l~n! is also measured in units of MRD. attention has been given to understanding the rotations
Orientation maps from a planar section can provide between the image and the diffraction pattern. However,
four of the five parameters: the three misorientation param- when one is concerned with absolute orientation, it is
eters and one of the two parameters describing the inclina- important to determine the inertial sample frame of refer-
tion of the boundary plane ~Saylor et al., 2004a!. It is not ence with respect to which the orientation is specified. This
possible to determine the exact inclination of the grain calibration is crucial for obtaining reliable results from
boundary plane by only considering orientation data from a stereological analysis, and the emphasis placed on this as-
plane section. However, the set of possible boundary planes pect here is motivated by the fact that reliable information
are in the zone of the grain boundary trace. If a large on interface character can only be obtained if absolute
number of indistinguishable misorientations are consid- orientations are measured correctly.
ered, then the true boundary planes should make a rela-
tively larger fraction in the set of possible boundary planes,
and an average estimate of l~n! for this misorientation can
M ATERIALS AND M ETHODS
be made. This is the basis for the stereological analysis used Two SiO2 encapsulated Cu thin films with Cu layer thick-
to make a statistical estimate of the grain boundary charac- nesses 36.9 and 46.4 nm were used in this study. Details of
ter distribution. Stereological analysis has been employed the thin film deposition and characterization are available in
successfully for extracting grain boundary character distri- Sun et al. ~2008, 2010!. Briefly, the two Cu films studied were
bution with the SEM-EBSD orientation maps ~Saylor et al., sputter deposited onto Si ~100! substrates having a 150 nm
2004a!. In this work, a similar analysis of orientation maps thick layer of thermally grown SiO2 and cooled to ⫺408C by
is made in nanocrystalline Cu films using the TEM. In a contact with a liquid nitrogen cooled Cu plate. The Cu layer
recent study, grain boundaries in nanoscale Cu intercon- was direct current sputtered from 99.9999% pure Cu targets
nects were investigated using orientation mapping from and was encapsulated between an underlayer of 20 nm thick
ASTAR TM ~Ganesh et al., 2012!. However, the study of radio frequency sputter deposited SiO2 and an overlayer of
Ganesh et al. ~2012! focused on copper lines, instead of 20 nm sputter deposited SiO2 to form a SiO2 ~20 nm
copper films for which the texture is significantly different, sputtered!/Cu/SiO2 ~20 nm sputtered!/SiO2 ~150 nm
and it was based on a grain boundary trace analysis of thermal!/Si structure. The Cu thin films were annealed at
orientation maps that simply classifies S3 grain boundaries 6008C for 30 min in Ar ⫹ 3% H2 . Plan-view TEM samples
as incoherent @not terminated by ~111! planes# or as possi- were prepared using a back-etching technique by thinning
bly being coherent twins. By contrast, the current study initially with HF ⫹ HNO3 solution and subsequently a di-
analyzes all of the boundaries and provides a stereological luted HF solution ~Yao et al., 2006!. The thickness of the Cu
 Stereological Analysis of TEM Orientation Maps 113

layer was measured using X-ray reflectivity at the Stanford sideration is different from the orientation of at least four
Synchrotron Radiation Lightsource ~SSRL; Sun et al., 2010!. nearest neighbors, then the orientation of the point is changed
The thicknesses of the two films are used as labels for the to the orientation of the four nearest neighbors similar in
samples studied for the remainder of this work. The average orientation. Small changes in orientation solutions for points
grain sizes for the 36.9 and 46.4 nm thick films are reported within a grain were commonly observed due to local bend-
to be 81.4 6 4.5 and 112.6 6 7.7 nm ~Sun et al., 2010!. The ing of the thin film. To eliminate these orientation fluctua-
diameter of the equivalent circle with an area equal to the tions, an average orientation was assigned to all points within
average grain area was used as the measure for the aforemen- a grain. After sampling numerous misorientation profiles

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
tioned grain size. Hollow cone dark-field images taken at within grains, it was observed that the maximum misorienta-
multiple sample tilts in the TEM were used to calculate the tion within a grain was 108. Therefore, for the purpose of
grain size ~Yao et al., 2010!. Twin boundaries were excluded assigning a single average grain orientation, two neighboring
in this grain size calculation. points with disorientation less than 108 were considered to
The orientation imaging microscopy of the two SiO2- be part of the same grain. For the specific sequence of clean
encapsulated Cu films studied was performed on a JEOL up steps used in this study, the maximum fraction of points
2010F ~JEOL Ltd., Tokyo, Japan! field emission gun TEM/ changed due to data clean up was approximately 5%. Grain
scanning TEM ~STEM! at an operating voltage of 200 kV. boundary traces were then extracted with a tolerance of 2
The lens configurations to obtain a probe size less than 5 nm pixels from the cleaned orientation data using the method by
were similar to those employed by Ganesh et al. ~2010!. Spot Wright and Larsen ~2002!. Forty-eight thousand and 57,000
diffraction patterns were collected by scanning the beam boundary traces were extracted for the 36.9 and 46.4 nm
using a step size of 8 nm over an area containing 180 ⫻ 180 thick films, respectively. A representative color-coded inverse
steps. Consistent with the probe size, the step size was cho- pole figure map from the 46.4 nm thick Cu film is shown in
sen to optimize the sampling of orientation data and the Figure 1~left! with the corresponding color code also shown
acquisition time. Fourteen and 20 scan areas were collected in Figure 1~right!. As a useful comparison, the corresponding
for the 36.9 and 46.4 nm thick films, respectively. A preces- correlation map is shown in Figure 2. The correlation index
sion angle of 0.48 was used while acquiring all the spot dif- for a given pixel in the orientation map gives the degree of
fraction patterns. The intensity of the forward beam was correlation between the observed diffraction and the precal-
used to reconstruct a virtual bright-field image in a manner culated template ~Rauch & Veron, 2005!. The image in Fig-
similar to bright-field imaging in STEM mode ~Rauch & ure 2 is shaded such that the bright regions correspond to a
Dupuy, 2005!. Therefore, every pixel in the virtual bright- high correlation index. Grain boundary traces extracted from
field image could be associated with a diffraction pattern. the orientation map shown in Figure 1~left! are shown in
The diffraction patterns were collected using the external Figure 3. The choice of parameterization for the description
optical charge-coupled device camera at 25 frames per sec- of misorientation and grain boundary normal for stereologi-
ond at a reduced binning of 150 ⫻150 pixels. For automated cal analysis is consistent with that used by Saylor et al. ~2004a!.
indexing, 1,326 templates were generated so that the disori- The misorientation space was parameterized using the Euler
entation ~minimum misorientation! between two successive angles ~f1 , F, f 2 ! ~Bunge convention! with 108 discretiza-
templates is approximately 18 ~Rauch & Dupuy, 2005!. tion ~Saylor et al., 2004a!. The grain boundary plane normal
The orientation images obtained from ASTAR TM were was parameterized using the two spherical angles u and w to
then exported to TexSEM Laboratories Orientation Imaging specify the inclination of the boundary normal in the crystal
~TSL OIM TM ; EDAX, Mahwah, NJ, USA! analysis software frame of reference. For the given choice of parameterization,
for further data processing. The first step in data processing the resolution of the discretized five-parameter space is ap-
was to use a cleanup procedure to correct spurious points in proximately 108.
the orientation map due to incorrect indexing. Diffraction
patterns may be incorrectly indexed in cases where the spot Frame of Reference Calibration
intensities are too low or the number of spots is too few. As mentioned in the Introduction, it is important to cali-
Grain dilation with a minimum grain size of 4 pixels fol- brate the fixed sample reference frame with respect to which
lowed by neighbor orientation correlation ~level 4! in the the crystal orientations are specified. This involves two as-
TSL OIM TM software was used as the data cleanup proce- pects: ~a! calibration of the rotation between the image and
dure. The grain dilation operation was used to ensure that the diffraction pattern, and ~b! knowledge of the sample
the orientation of the smallest grain is sampled using at least reference frame with respect to the image. The scanning
4 pixels. This is essential to assign reliable orientations for direction is chosen such that the virtual bright-field image
small grains. The neighbor orientation correlation proce- coincides with the TEM bright-field image. In other words,
dure identifies more likely orientations for points in the ori- this ensures that no additional rotation is introduced be-
entation map that possess a reliability index less than 0.12 tween the image and the diffraction pattern due to the orien-
~Rauch & Dupuy, 2005!. For such points, the orientation is tation of the scanning direction. Furthermore, it is necessary
compared with the orientation of the nearest neighbors. If to calibrate the image rotation with respect to the diffraction
the orientation of at least four of the nearest neighbors is pattern for the microscope used for mapping ~Williams &
similar to each other and the orientation of the point in con- Carter, 2009!. This standard calibration was performed using
114 A.D. Darbal et al.

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
Figure 1. Representative color-coded inverse pole figure map from the 46.4 nm thick Cu film. The color coding for the
orientations is shown in the legend.

Figure 2. Correlation index map corresponding to the orientation Figure 3. Grain boundary traces extracted from the orientation
map shown in Figure 1. map shown in Figure 1.

a MoO3 crystal, and it was found that a rotation of 1868 in normal direction ~ND! ~Dingley, 2004!. Figure 4 shows that
the counter-clockwise direction of the diffraction pattern is there is a net 908 rotation between the reference systems.
needed to make it coincide with the diffraction pattern. The 111 pole figure from a point in the orientation map
The set of three Bunge Euler angles for each pixel in the from a Cu thin film in the frame of reference used in
orientation map was imported into the TSL OIM TM data ASTAR TM is shown in Figure 4a. The 111 pole figure from
analysis software for subsequent analysis. It is important to the same point plotted using the TSL OIM TM data analysis
note that the definition of the Euler reference frame for software is shown in Figure 4b. The difference in the two
TSL OIM TM data analysis software is not consistent with pole figures is due to the fact that definitions of the refer-
ASTAR TM as seen in Figure 4. According to the notation ence systems for TSL OIM TM data analysis and ASTAR TM
used in TSL OIM TM data analysis, the x axis is denoted as are different. This needs to be taken into account when the
the rolling direction ~RD!, the y axis is denoted as the data from ASTAR TM are exported to TSL OIM TM data
transverse direction ~TD!, and the z axis is denoted as the analysis. Therefore, the orientation data were rotated by
 Stereological Analysis of TEM Orientation Maps 115

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
Figure 5. Inverse pole plots from the combined orientation data
for the 36.9 and 46.4 nm thick Cu films. Intensity is given in units
of MRD. A weak $111% fiber texture may be noted.

Figure 4. An example of a 111 pole figure in ~a! the ASTAR TM

800 MRD at the ~111! position for the 36.9 nm and 46.4 nm
frame of reference and ~b! the TSL OIM TM frame of reference. For
the latter, x is designated the rolling direction ~RD! and y the
thick Cu films, respectively. This peak corresponds to a
transverse direction ~TD!. The direction normal to the plane of coherent twin boundary ~the coherent S3!. The distribution
the page is z or ND. The orientation data in TSL OIM TM data of grain boundary plane normals averaged over all grain
analysis software should be given a ⫹908 counter-clockwise active boundary misorientations is shown in Figure 8. The distri-
rotation about the normal direction ~z! to account for the differ- bution of the boundary plane normals is relatively isotropic
ence in the ASTAR TM and TSL OIM TM frames of reference. with a clear preference for the $111% planes.

⫹908 in the counter-clockwise direction about the ND

using TSL OIM TM data analysis software. It may be noted D ISCUSSION
that a 908 rotation can also be applied to the data using Earlier experimental and theoretical studies of the distribu-
ASTAR TM before exporting it. tion of grain boundaries show that grain boundaries with
low energies have high populations ~Saylor et al., 2004b;
Rohrer et al., 2004a, 2004b!. As mentioned earlier, the peak
R ESULTS in the grain boundary character distribution in Figure 7
The orientation distribution ~OD! from the combined data corresponds to a coherent twin boundary. It is well known
taken from the 36.9 and 46.4 nm thick Cu films is shown in that the energy of a coherent twin boundary is significantly
Figure 5 as an inverse pole plot. The OD for both the Cu lower compared to other boundaries and that coherent twin
films analyzed was similar, and hence only combined data boundaries are observed frequently in Cu ~Randle et al.,
are shown to have improved counting statistics. The OD 2008; Lu et al., 2009!. Therefore, the peak at the ~111! pole is
shows a weak orientation texture with the frequency of consistent with these observations. It is also interesting to
@111# poles perpendicular to the sample surface as 2.1 compare the distribution of grain boundary planes aver-
MRD. The presence of this relatively low level of orientation aged over all misorientations, which shows that $111% planes
bias does not have a significant impact on the grain bound- occur most frequently. This is consistent with the calculated
ary character distribution determined from stereological surface energy using the broken bond model for face-
analysis ~Saylor et al., 2004b!. centered cubic ~fcc! crystals ~Mackenzie et al., 1962; Rohrer
The misorientation distribution ~MD! is shown in Fig- et al., 2004a, 2004b!, according to which the $111% planes in
ure 6. The axis-angle convention is used to denote the the fcc crystal have the lowest surface energy. Studies of the
misorientation. Figure 6 shows a prominent peak at 608/ grain boundary distribution and grain boundary energy
@111# misorientation with a frequency of 19.2 MRD. This anisotropy in fcc Al and Ni using experimental measure-
misorientation is commonly referred to as the S3 misorien- ments and calculated energies show similar trends between
tation in coincident site lattice convention ~Bollmann, 1970; the surface energy and the frequency of grain boundary
Randle, 1996!. The boundary plane distribution l~n/Dg! planes ~Li et al., 2009; Rohrer et al., 2010a; Holm et al.,
for the S3 misorientation is shown as a stereographic pro- 2011!. A similar trend between the frequent occurrence of
jection in Figure 7; l~n/Dg! shows a peak of 500 MRD and low energy planes along the interfaces between adjoining
116 A.D. Darbal et al.

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
Figure 6. Misorientation distribution from the combined orientation data for the 36.9 and 46.4 nm thick Cu films with
intensity in units of MRD. A strong peak at the S3 misorientation can be observed.

crystals has been observed in other systems such as MgO, the $111% is only a necessary, but not a sufficient, condition
SrTiO3 , MgAl2O4 , and TiO2 ~Saylor et al., 2004c and refer- for a twin boundary to be coherent. This is because the
ences therein!. trace of the boundary plane matches the trace of all the
As mentioned earlier, the misorientation distribution planes whose normals are perpendicular to the boundary
in Figure 6 shows a high density of S3 or twin boundaries. line segment. Therefore, when a boundary is classified as a
Given their frequent occurrence, it is interesting to further coherent boundary, it is ambiguous, and the true fraction of
investigate whether the twin boundaries are coherent or coherent twin boundaries is always less than or equal to the
incoherent. Twin boundaries are considered coherent when estimate from trace analysis. Figure 9 shows the grain bound-
the boundary plane is $111% and are considered incoherent ary traces extracted using the orientation map for the
otherwise. It is well known that the properties of twin 46.4 nm thick Cu film with the coherent and the incoherent
boundaries significantly depend on whether they are coher- twin boundaries identified using trace analysis. It was found
ent or incoherent ~Olmsted et al., 2009!. This is of particu- that 45% of the S3 boundaries are coherent twins. This is
lar interest in the study of classical resistivity size effect in an interesting observation, as earlier investigations of nano-
Cu resistivity increase due to incoherent twin boundaries is crystalline Cu assumed all the twin boundaries to be coher-
significantly higher than that of coherent twin boundaries ent due to the lack of a suitable technique to investigate
~Lu et al., 2004; Feldman et al., 2010!. The grain boundary grain boundaries at this scale ~Lu et al., 2004; Sun et al.,
character distribution in Figure 7 shows a significant occur- 2010!. Clearly, this assumption is not valid and further
rence of boundary planes that are not exactly $111%, thus investigation in this regard may be necessary. It is interest-
indicating the presence of incoherent boundaries. ing to compare the percentage of coherent S3 boundaries
To gain a more quantitative measure of the fraction of observed in this study ~45%! with earlier investigations of
incoherent boundaries, trace analysis was used. In trace twin boundaries in Cu interconnect lines using trace analy-
analysis, to determine whether a twin boundary is coherent, sis of EBSD orientation maps ~Wright & Larsen, 2002!. It
the trace of the boundary plane is matched with the trace of was found that 62% of the S3 boundaries in these Cu
the $111% ~Wright & Larsen, 2002!. If the two traces are interconnect were coherent. The grain size of Cu studied by
within a predefined tolerance, then the twin boundary is Wright and Larsen was in the 2–3 mm range, thus making it
considered to be coherent. However, it should be noted that amenable for EBSD measurements. In this regard, it is also
the matching between the trace of the boundary plane and interesting to discuss the study on five parameter grain
 Stereological Analysis of TEM Orientation Maps 117

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
Figure 7. The distribution of grain boundary planes on a @001#
stereographic projection for the S3, i.e., 608/@111#, misorientation
for ~a! 36.9 nm and ~b! 46.4 nm thick Cu films. The intensity is
expressed in units of MRD. For both samples a peak is found at
the @111# pole, which corresponds to a coherent S3 twin boundary.

boundary characterization on grain boundary engineered

~GBE! bulk Cu ~Randle et al., 2008! with grain sizes in the
tens of micrometer range. They observed that the fraction
of S3 boundaries increased in GBE Cu while the fractional Figure 8. The distribution of grain boundary planes in the crystal
area of $111% remained constant; therefore, a high fraction frame of reference l~n! averaged over all misorientations for
of incoherent twin boundaries had formed during the ~a! 36.9 nm and ~b! 46.4 nm thick Cu films. Intensity of the
thermomechanical treatments associated with GBE. Such distribution is expressed in units of MRD.
differences in the fraction of coherent twin boundaries
among bulk, micrometric, and nanometric Cu samples are providing large orientation datasets so that statistical meth-
likely a consequence of the differences in processing meth- ods such as the stereological analysis used in this study can
ods, i.e., bulk thermomechanical, electroplating, and sputter be used. For this purpose, we have employed a stereological
deposition used for the three types of samples, respectively. analysis on large orientation datasets acquired using the TEM.
This means that the quantitative investigation of microstruc-
ture in polycrystalline materials is no longer limited by the
S UMMARY resolution of SEM-EBSD techniques. However, due care must
The use of TEM-based orientation imaging for quantitative be taken to understand the effects of lens rotation and the
investigation of grain boundary characterization in nanocrys- frame of reference used to represent the orientation.
talline Cu thin films has been demonstrated. It has been The nanocrystalline Cu films studied in this work show
shown here that orientation imaging in TEM is capable of weak ^111& fiber texture but a strong S3, i.e., @111#/608,
118 A.D. Darbal et al.

Feldman, B., Park, S., Haverty, M., Shankar, S. & Dunham,

S.T. ~2010!. Simulation of grain boundary effects on electronic
transport in metals, and detailed causes of scattering. Phys
Staus Solidi B 247, 1791–1796.
Ganesh, K.J., Darbal, A., Rajasekhara, S., Rohrer, G.S., Bar-
mak, K. & Ferreira, P.J. ~2012!. Effect of downscaling copper
interconnects on the microstructure revealed by high resolu-
tion TEM-orientation-mapping. Nanotechnology 23 135702.
Ganesh, K.J., Kawasaki, M., Zhou, J.P. & Ferreira, P.J. ~2010!.

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
D-STEM: A parallel electron diffraction technique applied to
nanomaterials. Microsc Microanal 16, 614–621.
Holm, E.A., Rohrer, G.S., Foiles, S.M., Rollett, A.D., Miller,
H.M. & Olmsted, D.L. ~2011!. Validating computed grain
boundary energies in fcc metals using the grain boundary
character distribution. Acta Mater 59, 5250–5256.
Li, J., Dillon, S.J. & Rohrer, G.S. ~2009!. Relative grain boundary
area and energy distributions in nickel. Acta Mater 57, 4304–4311.
Liu, H.H., Schmidt, S., Poulsen, H.F., Godfrey, A., Liu, Z.Q.,
Sharon, J.A. & Huang, X. ~2011!. Three-dimensional orienta-
tion mapping in the transmission electron microscope. Science
332, 833–834.
Lu, K., Lu, L. & Suresh, S. ~2009!. Strengthening materials by
engineering coherent internal boundaries at the nanoscale.
Figure 9. Grain boundary traces extracted from the orientation Science 324, 349–352.
map for the 46.4 nm thick Cu film with coherent and incoherent Lu, L., Shen, Y., Chen, X., Qian, L. & Lu, K. ~2004!. Ultrahigh
twin boundaries identified using trace analysis. Thick red lines strength and high electrical conductivity in copper. Science 304,
correspond to incoherent twin boundaries and thick black lines 422–426.
correspond to coherent twin boundaries. Thin black lines corre- Mackenzie, J.K., Moore, A.J.W. & Nichols, J.F. ~1962!. Bonds
spond to all other boundaries. broken at atomically flat crystal surfaces—I: Face-centred and
body-centred cubic crystals. J Phys Chem Solids 23, 185–193.
misorientation texture. The distribution of grain bound- Oleynikov, P., Hovmuller, S. & Zou, X.D. ~2007!. Precession
ary planes for this misorientation shows a prominent peak electron diffraction: Observed and calculated intensities. Ultra-
at @111#, which shows a high density of coherent twin microscopy 107, 523–533.
boundaries. However, it was also found that a significant Olmsted, D.L., Foiles, S.M. & Holm, E.A. ~2009!. Survey of
fraction of the S3 boundaries are not coherent twin computed grain boundary properties in face-centered cubic
boundaries. metals: I. Grain boundary energy. Acta Mater 57, 3694–3703.
Portillo, J., Rauch, E.F., Nicolopoulos, S., Gemmi, M. & Bul-
treys, D. ~2010!. Precession electron diffraction assisted orien-
A CKNOWLEDGMENTS tation mapping in the transmission electron microscope. Mater
Sci Forum 644, 1–7.
Financial support of the SRC, Task 1292.008 and 2121.001,
Randle, V. ~1996!. The Role of Coincident Site Lattice in Grain
and of the MRSEC program of the National Science Foun- Boundary Engineering. London: Cambridge University Press.
dation under DMR-0520425 is gratefully acknowledged. Randle, V., Rohrer, G.S., Miller, H.M., Coleman, M. & Owen,
G.S.R. acknowledges financial support from the ONR- G.T. (2008). Five-parameter grain boundary distribution of
MURI under the grant no. N00014-11-1-0678. Portions of commercially grain boundary engineered nickel and copper.
this research were carried out at the SSRL, a national user Acta Mater 56, 2363–2373.
facility operated by Stanford University on behalf of the Rauch, E.F. & Duft, A. ~2005!. Orientation maps derived from
U.S. Department of Energy, Office of Basic Energy Sciences. TEM diffraction patterns collected with an external CCD cam-
E. Rauch is thanked for helpful discussions. era. Mater Sci Forum 495–497, 197–202.
Rauch, E.F. & Dupuy, L. ~2005!. Rapid diffraction patterns identi-
fication through template matching. Arch Metall Mater 50,
R EFER ENCES 87–89.
Adams, B.L., Wright, S.I. & Kunze, K. ~1993!. Orientation imag- Rauch, E.F. & Veron, M. ~2005!. Coupled microstructural obser-
ing: The emergence of a new microscopy. Metal Trans A 24, vations and local texture measurements with an automated
819–831. crystallographic orientation mapping tool attached to a tem.
Bollmann, W. ~1970!. Crystal Defects and Crystalline Interfaces. Materialwiss Werkst 36, 552–556.
New York: Springer Verlag. Rohrer, G.S., El Dasher, B.S., Miller, H.M., Rollett, A.D. &
Dingley, D. ~2004!. Progressive steps in the development of elec- Saylor, D.M. ~2004a!. Distribution of grain boundary planes
tron backscatter diffraction and orientation imaging micros- at coincident site lattice misorientations. Mat Res Soc Symp
copy. J Microsc 213, 214–224. Proc 819, N7.2.
Dingley, D.J. ~2006!. Orientation imaging microscopy for the Rohrer, G.S., Holm, E.A., Rollett, A.D., Foiles, S.M., Li, J.
transmission electron microscope. Microchimica Acta 155, 19–29. & Olmsted, D.L. ~2010a!. Comparing calculated and mea-
 Stereological Analysis of TEM Orientation Maps 119

sured grain boundary energies in nickel. Acta Mater 58, Sun, T., Yao, B., Warren, A.P., Kumar, V., Roberts, S., Barmak,
5063–5069. K. & Coffey, K.R. ~2008!. Classical size effect in oxide-
Rohrer, G.S., Li, J., Lee, S., Rollett, A.D., Groeber, M. & Uchic, encapsulated Cu thin films: Impact of grain boundaries versus
M.D. ~2010b!. Deriving grain boundary character distributions surfaces on resistivity. J Vac Sci Technol A 26, 605–609.
and relative grain boundary energies from three-dimensional Vincent, R. & Midgley, P.A. ~1994!. Double conical beam-
EBSD data. Mater Sci Technol 26, 661–669. rocking system for measurement of integrated electron diffrac-
Rohrer, G.S., Saylor, D.M., Dasher, B.E., Adams, B.L., Rollett, tion intensities. Ultramicroscopy 53, 271–282.
A.D. & Wynblatt, P. ~2004b!. The distribution of internal Williams, D.B. & Carter, C.B. ~2009!. Transmission Electron
interfaces in polycrystals. Z Metallk 95, 197–214. Microscopy: A Textbook for Materials Science, pp. 167–168. New

Downloaded from https://academic.oup.com/mam/article/19/1/111/6932393 by National Science & Technology Library user on 12 July 2023
Rouvimov, S., Moeck, P., Rauch, E.F., Maniette, Y. & Bultreys, York: Springer.
D. ~2008!. Crystallographic characterization of polycrystalline Wright, S.I. & Larsen, R.J. ~2002!. Extracting twins from orien-
materials: High resolution automated crystallite orientation. tation imaging microscopy scan data. J Microsc 205, 245–252.
Microsc Microanal 14, 768–769. Yao, B., Petrova, R.V., Vanfleet, R.R. & Coffey, K.R. ~2006!. A
Saylor, D.M., Dasher, B.S., Adams, B.L. & Rohrer, G.S. ~2004a!. modified back-etch method for preparation of plan-view high-
Measuring the five-parameter grain-boundary distribution from resolution transmission electron microscopy samples. J Electron
observations of planar sections. Metal Mater Trans A 35, Microsc 57, 47–52.
1981–1989. Yao, B., Sun, T., Warren, A., Heinrich, H., Barmak, K. &
Saylor, D.M., Dasher, B.S.E., Rollett, A.D. & Rohrer, G.S. Coffey, K.R. ~2010!. High contrast hollow-cone dark field
~2004b!. Distribution of grain boundaries in aluminium as a transmission electron microscopy for nanocrystalline grain size
function of five macroscopic parameters. Acta Mater 52, quantification. Micron 41, 177–182.
3649–3655. Zaefferer, S. ~2007!. On the formation mechanisms, spatial reso-
Saylor, D.M., El, D.B., Pang, Y., Miller, H.M., Wynblatt, P., lution and intensity of backscatter Kikuchi patterns. Ultramicros-
Rollett, A.D. & Rohrer, G.S. ~2004c!. Habits of grains in copy 107, 254–266.
dense polycrystalline solids. J Am Ceram Soc 87, 724–726.
Sun, T., Yao, B., Warren, A.P., Barmak, K., Toney, M.F., Peale,
R.E. & Coffey, K.R. ~2010!. Surface and grain-boundary scat-
tering in nanometric Cu films. Phys Rev B 81.