Professional Documents
Culture Documents
Materials Characterization
journal homepage: www.elsevier.com/locate/matchar
A R T I C L E I N F O A B S T R A C T
Keywords: The line intercept method is well developed for estimating the dislocation density in metals, but its application is
Dislocation density limited by the intersection counting, which is deeply depending on the TEM image quality. Manual interscetion
Invisibility criterion counting is quite time and labor consuming. If most of the dislocations has good contrast and homogenous
Two-beam condition
background in an image, it is possible to perform automatic counting by the computer program. In this work a
Zone axis STEM image
series of imaging conditions were investigated on a slightly deformed IF steel thin foil. Though the invisibility
criterion remains applicable for both the conventional transmission electron microscopy (CTEM) and the scan
ning transmission electron microscopy (STEM), the STEM images revealed more dislocations and had more
homogenous background than the CTEM images. The STEM bright field (BF), annular dark field (ADF) and
annular bright field (ABF) images at a low index zone axis contained the most dislocations visible and were
suitable for automatic intersection counting of dislocations.
1. Introduction the two-beam conditions is likely to reveal all the dislocations. However,
the number of two-beam conditions is limited by the sample tilting an
The line intercept method is widely applied in the estimation of the gels and image distortion. Moreover, the uniformity of the image
dislocation density [1–4]. By measuring the foil thickness and the background is ruined by the bend contours and the second phase
intersection count between a given grid and the dislocations on a TEM contrast in CTEM. Therefore, the intersection counting on the CTEM
image, the dislocation density ρ in the current grain can be achieved by images can only be conducted manually. It is quite time and labor
Eq. (1). consuming, and its accuracy is closely dependent on the imaging
condition.
2n
ρ= (1) In the scanning transmission electron microscopy (STEM) mode, due
lt
to the convergent illumination of the electron beam on the sample, a
where n is the intersection count, l is the total length of the grid, and t is STEM image is equivalent to an averaged image of a series of CTEM
the thickness of the foil on this TEM image. This method is effective images acquired with the sample tilted in a range of directions. The
when the dislocation density in the metal is less than 1014 cm− 2. The dislocation contrast is enhanced and the elastic strain contrast in the
thickness t can be accurately measured by the trace analysis [5], and background is reduced by the averaging the intensities from all the di
Kossel-Möllenstedt patterns in convergent beam electron diffraction rections. As a result, the dislocations are much easier to distinguish in
(CBED) method [6]. The difficulty is the measurement of the intersec STEM images [7–10] than in CTEM images. Therefore, it is possible to
tion counting. perform the automatic counting of the intersection on the STEM images,
In the conventional transmission electron microscopy (CTEM) mode, which will greatly enhance the much wider application of measuring the
the electron beam is parallelly illuminating on the sample, and good dislocation density through TEM. Besides, the convergent beam has
contrast of the defects can be achieved at two-beam conditions. Since the higher brightness and can be applied on thicker foil, which is benefit to
extinction of the dislocation is inevitable, a series of images at the same the thickness measurement by CBED method [6].
position with different two-beam conditions must be acquired in order However, in literatures very little work has been found on STEM
to get more dislocations visible. The superposition of the images at all application in measuring the dislocation density. In present work, a
* Corresponding author.
E-mail address: mengyang@shougang.com.cn (Y. Meng).
https://doi.org/10.1016/j.matchar.2021.111065
Received 10 November 2020; Received in revised form 24 March 2021; Accepted 25 March 2021
Available online 26 March 2021
1044-5803/© 2021 Elsevier Inc. All rights reserved.
Y. Meng et al. Materials Characterization 175 (2021) 111065
Fig. 1. The full image of the grain: (a) CTEM BF image and (b) STEM BF image with camera length of 40 cm.
2. Experiment
2
Y. Meng et al. Materials Characterization 175 (2021) 111065
Fig. 3. CTEM observation at two-beam conditions: SAD pattern (a) and BF image (b) with the strongest reflector g1 = (011); SAD pattern (c) and bBF image (d) with
the strongest reflector is g2 = (110); SAD pattern (e) and BFB image (f) with the strongest reflector g3 = (101).
(exp.) and the upper limits were calculated (cal.) from the inner and beam condition, but no more than in Fig. 4. By comparison with the
outer diameters, and all of them were listed in Table 1. The collection CTEM BF image at the same two-beam condition, the dislocations were
angle of the bright field detector was just a little smaller than the lower marked by white arrows with black border if they were still invisible and
limits of collection angle of the annular detector. marked by white arrow without border if they were visible, as shown in
The STEM BF images were also acquired at the same two-beam Fig. 5(b), (d) and (f). No more extinction was found in any STEM BF
conditions with the cameral length of 40 cm and the range of the image.
collection angle of the BF detector was less than 17.5 mrad according to The dislocation contrast was determined by the invisibility criterion
Table 1. For the ferrite, 2θ110 = 12.4 mrad, so the transmission disk and g⋅b = 0 in the CTEM mode with parallel beam condition. No more
strong reflector {110} disk were collected. There are more dislocations extinction in STEM BF image than in CTEM BF image indicated that the
in each STEM BF image than in each CTEM BF image at the same two- invisibility criterion in the CTEM mode remained applicable in the STEM
3
Y. Meng et al. Materials Characterization 175 (2021) 111065
transmission disk and all the {110} diffraction disks when the camera
length was 40 cm, as the white circle indicated on the CBED pattern at
[111]bcc zone axis in Fig. 6(a), and the STEM BF image was shown in
Fig. 6(b). The STEM annular dark field (ADF) image in Fig. 6(c) was
acquired with the camera length of 20 cm, where the range of the
collection angle was 35–93.3 mrad, and the diffraction disks with index
higher than {123} can be collected (2θ123 = 32.8 mrad). The ADF image
had an inverted contrast to the STEM BF image (Fig. 6(b)). The STEM
annular brigth field (ABF) image in Fig. 6(d) was acquired with the
camera length of 150 cm, where the range of the collection angle was
4.7–12.4 mrad and the outer part of the transmission disk and half of
{110} diffraction disks were collected. The STEM ABF image had a
consistent contrast with the STEM BF image (Fig. 6(b)).
According to the diffraction mechanism, at a two-beam conditon,
Fig. 4. The superposition of the three two-beam conditions: g1 = (011), g2 =
(110) and g3 = (101). Table 1
Collection angle of the annular detector.
mode. Due to the convergent beam condition in STEM mode, the Camera length (cm) Range of the collection angle (mrad)
contrast of the extinction dislocations was enhanced, therefore, only a
8 87.5(exp.) ~ 233.3(cal.)
part of the extinctions still kept invisible, and the others had a fuzzy but 10 70.0(exp.) ~ 186.7(cal.)
recognizable contrast. 12 58.3(exp.) ~ 155.6(cal.)
20 35.0(exp.) ~ 93.3(cal.)
40 17.5(exp.) ~ 46.7(cal.)
3.2. STEM dislocation images at [111]bcc zone axis 80 8.8(exp.) ~ 23.3(cal.)
120 5.8(exp.) ~ 15.6(cal.)
150 4.7(exp.) ~ 12.4(cal.)
In order to get good dislocation contrast at [111]bcc zone axis, the
strong reflectors should be selected. The BF detector can cover the
Fig. 5. Comparison of the dislocation images at two-beam conditions: CTEM (a) STEM (b) BF images with the strongest reflector of g1 = (011); CTEM (c) STEM (d) BF
images with the strongest reflector of g2 = (110); CTEM (e) STEM (f) BF images with the strongest reflector of g3 = (101). By the comparison with Fig. 4 the white
arrows with black border indicated the invisible extinctions and the white arrows without border indicated the extinctions with weak contras in STEM images.
4
Y. Meng et al. Materials Characterization 175 (2021) 111065
there was only one strong reflection, and the dislocations satisfied the image as shown in Fig. 7.
invisibility criteria g⋅b = 0 and were invisible without auxiliary contrast.
Whereas at a zone axis there were multiple reflectors, and the invisibility
criterion could not be simultaneously satisfied, only if the Burgers vector 3.3. STEM dislocation images at high index zone axes
of the dislocation was parallel to the zone axis, i.e. [111]bcc for the
present grain. For most dislocations, whose Burger vectors were not High index zone axes, [331]bcc, [231]bcc, and a zone between
parallel to the zone axis, the auxiliary were not zero and they should be [355]bcc and [122]bcc, were achieved by tilting 17.5◦ , 12.2◦ and 16.2◦
visible. In the STEM mode, the image at the zone axis, no matter the BF from [111]bcc, respectively. The corresponding diffraction patterns were
or DF, was the average contrast of all the reflectors, and was equal to shown in Fig. 8(a), (c) and (e), and the dislocation images from the same
overlap the images at all the two-beam conditions around the zone. So area were shown in Fig. 8(b), (d) and (f). Since the magnetic sample was
the comparison between the STEM BF image at zone [111]bcc (Fig. 6(b)) tilted by large angles, the TEM alignment was seriously disturbed, which
and Fig. 4 proved that all of the visible dislocations in CTEM BF images made it hard to eliminate the strong astigmatism of the object lens and
under 3 two-beam conditions could be found in the single STEM BF the quality of STEM BF and ABF images were poor. But it was still
possible to distinguish some dislocations. All the STEM images at high
Fig. 6. The CBED pattern and STEM images at zone [111]bcc, (a) the CBED pattern, the white circle indicates the edge of the bright field detector, (b) the BF image,
(c) the ADF image and (d) the ABF image.
Fig. 7. Comparison of dislocations in the same area within the white frame, (a)the overlapped CTEM image of 3 two-beam conditions around [111]bcc zone axis and
(b) the STEM bright field image at [111]bcc zone axis.
5
Y. Meng et al. Materials Characterization 175 (2021) 111065
The CTEM central DF image (Fig. 9(a)) and weak beam dark field
(WBDF) image (Fig. 9(b)) were obtained at the 1st two-beam condition
with the reflector of g1 = (011). As marked in Fig. 5(a) there were 7
extinction dislocations in the CTEM BF image. All of them were invisible
in the central DF image and 5 out of them were still invisible and the
other 2 were recognizable in the WBDF image, and no more extinction
was found. The invisible ones were marked by the white arrow with
black border and the recognizable ones were marked by white arrows
without border in Fig. 9(a) and (b). The dislocation contrast in the CTEM
DF images was weaker than that in the BF image and uniformity of the
background was worse.
In the STEM mode, the ADF images and the central DF images were
acquired at the 1st two-beam condition with the reflector of g1 = (011).
The ADF images (Fig. 10(a) and 10(b)) were acquired by the annular
detector with camera length of 40 cm and 20 cm, respectively. The
corresponding collection angle were 17.5–46.7mrad and 35–93.3 mrad
according to Table 1. The central DF images were acquired by the BF
detector with shifting the strong diffraction disk g(011) (Fig. 10(c)) and
the weak diffraction disk g(011) (Fig. 10(d)) to the center, respectively.
Fig. 10(c) had nearly consistent contrast with the STEM BF image in
Fig. 5(b): the contrast of the dislocations was darker than the back
ground. It was because the transmission disk was inevitably overlapped
with the disk g(011) and the BF detector collected the whole disk g(011)
and part of the transmission disk. The 7 extinction dislocations in Fig. 5
(a) were examined in every image: only the 6th dislocation was recog
nizable in Fig. 10(a), and 4 out of them were recognizable in Fig. 10(b);
and all of them were invisible in Fig. 10(c), and only the 5th dislocation
was invisible in Fig. 10(d). No more extinction was found. The contrast
of the 7 dislocations in all the images at the 1st two-beam condition was
listed in detail in Table 2.
All the images at the 1st two-beam condition, including Fig. 5(a) and
(b), Fig. 9 and Fig. 10, were compared with the images at the low index
zone axis in Fig. 6. Though the contrast of dislocations varied with the
imaging condition, no more dislocations were found and the images at
low index zone axis showed the most dislocations.
According to the above comparison, both the STEM ADF and ABF
images had uniform background and could be used to count in
tersections by digital image processing technology. And the images at
low index zone axis [111]bcc had the most recognizable were suitable for
the measurement of dislocation density. A program had been developed
to automatically count the intersection. In the program, the digital im
ages of ADF and ABF (Fig. 6(c) and (d)) were processed to be black-white
(BW) images and masked by a 20 × 20 grid as shown in Fig. 11. The
intersections between the grid and the dislocations were then counted
Fig. 8. STEM images at high index zone axes: SAD pattern (a) and ABF image
automatically. The total counts in each image were 442 and 449,
(b) at [331]bcc; SAD pattern (c) and BF image (d) at [231]bcc; (e) SAD pattern
respectively. The average counts were 445. In order to verify the validity
and ABF image (f) between [355]bcc and [122]bcc; STEM BF image (g) at
of the automatic counting by program, the manual countingon the ABF
[111]bcc. The arrows with letter a indicated the dislocations invisible in Fig. 8
image was also conducted and the result was 454, which is in good
(b), the arrows with letter b indicated the dislocations invisible in Fig. 8(d), and consistency with the automatic counting result. The nl value was 3.89/
the arrows with letter c indicated the dislocations invisible in Fig. 8(f). μm. The foil thickness was determined to be 201 μm according to the
CBED pattern at a two-beam condition [6] in current grain in Fig. 12.
The dislocation density in this grain was estimated to be 3.9 × 1012
cm− 2.
index axes were compared to the STEM BF image at [111]bcc zone axis in
Fig. 8(g), and all the invisible dislocations were marked by arrows, and
no more dislocation was found. Therefore, the number of visible dislo
cations at [111]bcc zone was larger than that at high index zone axes.
6
Y. Meng et al. Materials Characterization 175 (2021) 111065
Fig. 9. The CTEM DF images at the g1 = (011)two-beam condition, (a)the central DF image and (b) the WBDF image.
Fig. 10. The DF images at the two-beam condition with the reflector of g1 = (011): (a) the ADF image with the collection angle 17.5– 46.7 mrad; (b) the ADF image
with the collection angle 35– 93.3 mrad; (c) the central DF image with strong disk g(011) in the center; (d) the central DF image with weak disk g(011)in the center.
Table 2
The contrast of the 7 extinct dislocations at the two-beam conditions with reflector of g1 = (011).
CTEM STEM
BF DF WBDF BF ADF (17.5–46.7mrad) ADF (35–93.3 mrad) DF,strong g(011) DF, weak g(011)
1 × × × × × o × o
2 × × × o × o × o
3 × × o o × o × o
4 × × × o × o × o
5 × × × × × × × ×
6 × × o o o o × o
7 × × × × × × × o
×: Invisible.
o: Recognizable.
7
Y. Meng et al. Materials Characterization 175 (2021) 111065
Fig. 11. The automatic intersection counting on the STEM images: (a) the BW image of Fig. 6(c); (b) the contrast enhanced image with a 20 × 20 grid on Fig. 6(c); (c)
the BW image of Fig. 6(d); (d) the contrast enhanced image with a 20 × 20 grid on Fig. 6(d).
None.
Acknowledgements
References
Fig. 12. CBED pattern at a two-beam condition for thickness determination.
[1] J.G. Rider, C.T.B. Foxon, An experimental determination of electrical resistivity of
dislocations in aluminium, Philos. Mag. 13 (122) (1966) 289–303.
4. Conclusions [2] B.R. Watts, Calculation of electrical resistivity produced by dislocations in various
metals, J. Phys. F Met. Phys. 18 (6) (1988) 1197–1209.
[3] R.K. Ham, The determination of dislocation densities in thin films, Philos. Mag. 6
The different sample conditions and imaging methods for the dislo
(69) (1961) 1183–1184.
cation density estimation were investigated in present work on a slightly [4] J.E. Bailey, P.B. Hirsch, The dislocation distribution, flow stress, and stored energy
deformed IF steel thin foil. A software was developed to automatically in cold-worked polycrystalline silver, Philos. Mag. 5 (53) (1960) 485–497.
[5] P.B. Hirsch, A. Howie, R.B. Nicholson, D.W. Pashley, M.J. Whelan, Electron
count the intersection in the TEM dislocation images. The following can
Microscopy of Thin Crystals, Butterworths, London, 1965.
be concluded. [6] D.B. Williams, C.B. Carter, Transmission Election Microscopy. A Textbook for
Materials Science II Diffraction, Springer, New York, 1996, pp. 321–323.
1. The STEM images revealed more dislocations than the CTEM images [7] Y. Miyajima, M. Mitsuhara, S. Hata, H. Nakashima, N. Tsuji, Quantification of
internal dislocation density using scanning transmission electron microscopy in
and they contained most dislocations visible at a low index zone axis. ultrafine grained pure aluminium fabricated by severe plastic deformation, Mat.
The background of the STEM images was more homogenous than Sci. Eng. A Struct. 528 (2) (2010) 776–779.
that of the CTEM images. [8] P.J. Phillips, M.C. Brandes, M.J. Mills, M.D. Graef, Diffraction contrast STEM of
dislocations: imaging and simulations, Ultramicroscopy 111 (9–10) (2011)
2. The STEM BF, ADF and ABF images at a low index zone axis were 1483–1487.
suitable for automatic intersection counting of dislocations. [9] P.J. Phillips, M.J. Mills, M.D. Graef, Systematic row and zone axis STEM defect
3. The self-designed program could perform the automatic intersection image simulations, Philos. Mag. 91 (16) (2011) 2081–2101.
[10] M. Tanaka, K. Higashida, K. Kaneko, S. Hata, M. Mitsuhara, Crack tip dislocations
counting and thus calculation of the dislocation density. The auto revealed by electron tomography in silicon single crystal, Scr. Mater. 59 (8) (2008)
matic intersection counting results of dislocations were in good 901–904.
consistence with the manual counting.