Journal Pre-proof

Correlation between Microstructure and Fracture Mechanism of 960 MPa Grade

Ultra-High Toughness Steel

Hongshan Zhao, Chen Ge, Huan Teng, Jianwen Fan, Liya Guo, Han Dong

PII: S0921-5093(24)00623-3
DOI: https://doi.org/10.1016/j.msea.2024.146692
Reference: MSA 146692

To appear in: Materials Science & Engineering A

Received Date: 15 January 2024

Revised Date: 14 May 2024
Accepted Date: 20 May 2024

Please cite this article as: H. Zhao, C. Ge, H. Teng, J. Fan, L. Guo, H. Dong, Correlation between
Microstructure and Fracture Mechanism of 960 MPa Grade Ultra-High Toughness Steel, Materials
Science & Engineering A, https://doi.org/10.1016/j.msea.2024.146692.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2024 Published by Elsevier B.V.

 1 Correlation between Microstructure and Fracture
2 Mechanism of 960 MPa Grade Ultra-High Toughness Steel
3 Hongshan Zhaoa,b, Chen Gea, Huan Tenga, Jianwen Fana, Liya Guoa*, Han Donga,b
4 a State Key Laboratory of Advanced Special Steel, Shanghai Key Laboratory of
5 Advanced Ferrometallurgy, School of Materials Science and Engineering, Shanghai
6 University, Shanghai, 200444, PR China
7 b Zhejiang lnstitute of Advanced Materials, SHU, Zhejiang 314100, PR China
8 *Corresponding author e-mail addresses: [email protected]

9 Abstract
10 This study investigated the correlation between microstructure and fracture

f
oo
11 mechanisms in a 960 MPa grade ultra-high toughness high strength low alloy steel
12 produced through controlled rolling and off-line heat treatment. The steel features a
13 low-carbon composition (≤0.1 wt.%) and a fine-grained(≤10 µm) bainitic

r
14
15
-p
microstructure. Through microstructural characterization and fractographic analysis,
extensive quantitative analysis of grain, packet, dimple, and cleavage facet sizes was
re
16 conducted. Fracture surfaces were examined to identify crack initiation sites. The
17 findings indicate that grain refinement significantly enhances toughness. In
lP

18 fine-grained conditions, bainitic packets that approximate the size of the grains
19 modulate crack propagation pathways. The near-oriented bainite across adjacent
20 grains forms structures comparable to abnormally large grains, crucial for crack
na

21 nucleation. Moreover, martensite/austenite (M/A) islands, while crucial to the

22 initiation of cracks, require the synergistic interaction with grain boundaries and
ur

23 inclusions to significantly influence fracture initiation. This investigation elucidates

24 the complex interplay between microstructural refinement and fracture mechanisms.
Jo

25 Keywords: HSLA steel, microstructure, fracture mechanism, toughness

26 1. Introduction
27 Steels continue to progress towards higher levels of strength, posing a formidable
28 challenge to achieve matched or even enhanced toughness [1, 2]. In general, for high
29 strength low alloy (HSLA) steels used in construction, containers, and industrial
30 machinery with a yield strength less than 1000 MPa, the optimal matching of strength
31 and toughness is typically found in pipeline steels [3]. These pipeline steels, reaching
32 grades such as X80 (~552 MPa) and X100 (~690 MPa), exhibit low-temperature
33 toughness levels of 328 J (-40 °C) and 216 J (0 ℃) [4, 5], respectively. As strength
34 increases, there is a noticeable decrease in toughness. The low-temperature toughness
35 of 700-900 MPa HSLA steels typically ranges from 250 to 150 J [6-8], while the
36 toughness level for 900-1000 MPa grade steel is generally around 100 J. Through
37 accelerated cooling (ACC), direct quenching (DQ) or heat treatment on-line process
38 (HOP) [7, 9-12]. The current emphasis is on achieving further breakthroughs in
39 toughness, aiming for 250 J at this strength level.
40 The HSLA steels with ultra-high toughness at the 900 MPa grade are typically
41 designed with ultra-low carbon content (≤0.1 wt.%) and utilize thermal mechanical
42 control process (TMCP) techniques for enhanced toughness [13-17]. The
43 microstructure consists mainly of bainite, martensite/austenite (M/A) islands, and
44 carbides, featuring pancake-like grains. Extensive research has analyzed the impact of
45 rolling and cooling process parameters on the microstructure and mechanical
46 properties [18-21]. Studies indicate that bainite, M/A islands, and carbides are crucial
47 factors influencing toughness, particularly in the nucleation and propagation of cracks
48 during the quasi-cleavage stage [22-25]. Therefore, improving toughness involves
49 refining the effective grain, controlling the quantity, size and morphology of M/A
50 islands, and enhancing the uniform dispersion of fine cementite [7, 26-28]. It is worth

f
51 noting that the pancake-like grain characteristics of this microstructure lead to

oo
52 significant variations in material properties along the rolling and transverse
53 directions[29], rendering it unsuitable for applications requiring uniform isotropic

r
54 performance. However, the deformation energy and high density dislocation of
55
56
-p
pancake-like grains during the reheating process enhance the driving force and
nucleation sites, thus allowing further grain refinement in reheated samples [30].
re
57 Nonetheless, as grain size decreases, the number of bainite packets diminishes until a
58 single grain comprises a single packet [22, 25], resulting in a decrease in its ability to
lP

59 impede crack propagation. Additionally, although extensive research suggests that

60 M/A islands serve as crack sources [19, 22, 31, 32], upon grain refinement, it remains
na

61 to be studied whether cracks under different impact temperatures are controlled solely
62 by M/A islands or involve multifactor interactions including grain size, bainitic
microstructure characteristics, and carbides. Therefore, the relationship between
ur

63
64 microstructure and fracture mechanism in ultra-low carbon bainite-based fine-grained
Jo

65 high strength low alloy steels still requires further investigation.

66 This study has achieved a 960 MPa ultra-high toughness steel by employing a
67 combination of low-carbon design, controlled rolling (pancake-like grain), and
68 off-line quenching and tempering processes (fine equiaxial grain). The low
69 temperature impact toughness at -40°C can reach a maximum of 250 J. However, a
70 comprehensive understanding of the underlying micromechanism of fracture is
71 imperative. A key emphasis is placed on the correlation study between microstructure
72 and fracture mechanisms. This involves the fractographic analysis to explore
73 relationships such as grain size, the multilevel microstructure of bainite, dimple sizes,
74 and cleavage facets sizes. Furthermore, there is a critical need for analyzing and
75 statistically confirming crack initiation sites in the tough-brittle transition zone and
76 the lower platform, identifying the weakest link component in the microstructure will
77 trigger cleavage fracture [22, 23]. This work will provide insights into the complexity
78 of microstructure-fracture interactions in HSLA steels and will be helpful to unveil the
79 strength and toughness mechanism of this material, which will guide future
80 optimization of strength-toughness balance via purposeful microstructural design and
81 process control.
 82 2. Material and methods
83 The main chemical composition of the designed steel is shown in Table 1. The
84 steel plate was produced through smelting in an 80-ton converter and then rolled and
85 cut into a 15×2500×6000 mm steel plate using a 3500 mm mill. Grain size refinement
86 was mainly achieved through non-recrystallization zone rolling, resulting in the
87 formation of a pancake-like microstructure. The hot-rolled steel plate subsequent
88 was subjected to austenitization at a temperature ranging from approximately 880 to
89 920 ℃, followed by final tempering at around 200 ℃. The reheating process aims to
90 further refine the grains, while tempering at 200 ℃ serves the purpose of residual
91 stress relief.
92 Table 1. The chemical composition of the designed steel（wt.%）
C Si Mn P S Cr Ni Mo Nb+V+Ti B, ppm CEV

f
oo
0.08 0.25 1.20 0.011 0.002 0.50 0.50 0.30 0.20 19 0.50

93 The Instron5982 tensile testing machine was utilized to conduct the tensile tests.

r
94 Three tensile specimens were taken perpendicular to the rolling direction for testing.
95
96
-p
The gauge length of the tensile samples was 25 mm, and the testing speed was set to 2
mm/min. Fifteen Charpy V-notch impact specimens(parallel to the rolling direction)
re
97 with dimension of 10×10×55 mm were tested at a temperature of -40 ℃, with impact
98 energy ranging from 245 to 271 J. The temperature of -40 °C is of particular interest
lP

99 for engineering steel or armor steel plates used in extreme environments. To further
100 investigate the fracture mechanism of this steel, impact experiments were conducted
na

101 on samples oriented both along and perpendicular to the rolling direction, with
102 V-notches made alone the thickness direction. This resulted in crack propagation
occurring along and perpendicular to the rolling direction for samples aligned
ur

103
104 perpendicular and along, respectively. These experiments were carried out using
Jo

105 an instrumented impact testing equipment, specifically an Instron 750 MPX type
106 pendulum impact testing machine, covering temperatures from room temperature(RT)
107 to -196 ℃.
108 The prior-austenite grain boundaries of this steel were revealed by etching with an
109 aqueous solution of saturated picric acid, 1% HCl, and a wetting agent.
110 Metallographic microstructures of this steel, along with a detailed multilevel
111 microstructure of bainite matrix and the characteristics of the M/A islands, were
112 observed using the ZEISS Axio Imager.M2m metallographic microscope and the
113 Zeiss Sigma300 scanning electron microscope (SEM) equipped with Bruker eFlash
114 Energy Dispersive Spectroscopy (EDS) and the Oxford electron backscatter
115 diffraction (EBSD) camera. EBSD analysis results and the reconstruction of
116 martensite variants were conducted using the open-source crystallographic toolbox
117 MTEX 5.8 [33, 34]. The fracture morphology was observed using the Zeiss Axiocam
118 208 color stereo microscope and a detailed fractographic examination was performed
119 by SEM on the tested Charpy samples. This analysis aimed to classify and evaluate
120 possible cleavage crack-initiation sites and microstructural features in their vicinity.
121 Histogram charts and box plots are utilized for the statistical analysis of grain,
122 packet, dimple, and cleavage facet sizes. The histogram charts highlight differences in
123 data size across categories, aiding in comparisons. Box plots, on the other hand, show
124 data distribution including medians, quartiles (25%-75%, reflecting data
125 concentration), and extremes(10%-90%, indicating data range) and outliers (＞90%).
126 They effectively reveal the dispersion of data and help identify outliers, which may
127 potentially lead to scatter in impact values [35] and are closely associated with the
128 micromechanisms of quasi-cleavage fracture [22].

129 3. Results
130 3.1 Mechanical Properties and Ductile to Brittle Transition
131 The yield strength of the steel plate reaches 1048 MPa, with a tensile strength of
132 1217 MPa and an elongation of 15.3%, as shown in figure 1(a). Temperature-impact

f
oo
133 energy curves for longitudinal and transverse V-notch Charpy impact specimens are
134 depicted in Figure 1(b). This steel grade demonstrated CVN up to 230 J at -40°C.
135 Additionally, there is minimal variance in impact energy exhibited between

r
136
137
longitudinally and transversely oriented impact specimens.
Therefore, to further determine the temperature point of the ductile-to-brittle
-p
re
138 transition, according to the description in reference [36], the Boltzmann function was
139 employed to regress the curve of the ductile-to-brittle transition temperature. This
lP

140 approach exhibits a good correlation and small residuals, effectively describing the
141 upper plateau, lower plateau, and transition temperature region of the curve. The
142 Boltzmann function is formulated as follows:
na

𝐸1 − 𝐸2
143 𝐴k = + 𝐸2 (1 − 1)
1 + e(𝑡−𝑡f)/𝑡r
ur

144 Where Ak represents the impact energy, t is the test temperature, E1 is the upper
145 plateau energy, E2 is the lower plateau energy, 𝑡f is the ductile-to-brittle transition
Jo

146 temperature, and 𝑡r is the temperature range of the transition region. The fitting
147 parameters are shown in Table 2.
1400
300
(a) (b)
1200
Charpy Impact energy,CVN (J)

250
Engineering stress (MPa)

1000
200
800

150
600

100
400
Longitudinal
Transversal
200 50 Boltzmann fitting
95% Confidence Interval
Boltzmann fitting
95% Confidence Interval
0 0
0 2 4 6 8 10 12 14 -200 -180 -160 -140 -120 -100 -80 -60 -40 -20 0 20 40

148 Engineering strain (%) Temperature(℃)

149 Figure 1. (a) Stress-Strain Curve and (b) Impact Energy at Various Temperatures for
150 the Steel.
151
152 Table 2. The fitting parameters for the Bolzmann function of the designed steel
Direction E1(J) E2(J) 𝑡f (℃) 𝑡r (℃)
Longitudinal 240±10.5 12±4.3 -102±5.5 19.9±3.3
Transverse 234±3.2 14±1.5 -75±3.5 11.8±1.7

153 According to the fitting results, the upper plateau energy of the longitudinally
154 sampled impact test specimen is 240 J, the lower plateau energy is 12 J, and the
155 ductile-to-brittle transition temperature is -102 °C, with a transition region
156 temperature range of approximately 20 °C. The upper and lower plateau energies for
157 the transverse specimen are similar to those of the longitudinal specimen, and the
158 ductile-to-brittle transition temperature is approximately -75 °C.
159 The typical load-deflection curves of longitudinal specimens are shown in figure
160 2(a). Instrumented impact data facilitates the identification of the ductile-to-brittle
transition temperature [37]. Consistently, load-deflection curves from room

f
161

oo
162 temperature to -80 °C exhibit no steep drop in force, indicative of stable (ductile)
163 crack propagation, which is a typical characteristic of high toughness curves. In figure
2(b), the load-deflection curves of three impact specimens tested at -100 °C are

r
164
165
166
-p
displayed. These curves represent stable crack propagation, varying degrees of stable
and unstable crack propagation, and solely unstable crack propagation, respectively,
re
167 demonstrating the characteristics of the ductile-to-brittle transition. This temperature
168 aligns with the Boltzmann fitting value in Table 2. Based on the fractographic analysis
lP

169 of crack initiation sites in the discussion below, the specimen tested at -100 ℃ with
170 relatively higher toughness (98 J) initiated cracks at carbides located in near-oriented
bainite (NOB) grains or abnormally large grains (ALG). Conversely, the specimen
na

171
172 with lower toughness (54 J) showed crack initiation at M/A islands within NOB or
173 ALG. For impact specimens tested between -120 °C and -196 °C, only unstable crack
ur

174 propagation occurs, accompanied by a gradual decrease in maximum force.

Jo

(a)35 RT (b) 35 -100℃-1

-40℃ -100℃-2
30 30
-60℃ -100℃-3
-80℃
25 25
-100℃ Specimen-1: 98 J
-120℃ Specimen-2: 216 J
Load (kN)

Load (kN)

20
20 -140℃ Specimen-3: 54 J
-160℃ 15
15 -180℃
-196℃ 10
10
5
5
0

0 -5
0 5 10 15 20 25 30 0 5 10 15 20 25 30

175 Deflection (mm) Deflection (mm)

176 Figure 2. (a) Load-deflection curves from instrumented impact tests on longitudinal
177 specimens at temperatures ranging from room temperature to -196 °C, (b)
178 load-deflecton curves of specimens test at -100 °C
179 3.2 Microstructure Characterization
180 Figure 3 presents SEM images of the microstructure and EBSD maps depicting
181 the distribution of grain boundary misorientation for the designed steel. Three
182 methods were employed to distinguish between bainite and martensite, 1) Analysis
183 based on the corrosion levels and the orientation relationship between carbides and
184 the matrix [25], as observed in SEM images (see Figure 3(a)). 2) The use of the band
185 contrast peak method in EBSD [38, 39]. 3) The reconstruction method based on the
186 EBSD data by MTEX [33, 34]. Method 2 and 3 are discussed further in the discussion
187 section. The microstructure of the steel mainly consists of bainite, martensite (M/A
188 islands), and carbides. Overall, the grains are fine, but a small number of abnormally
189 large grains are present. Notably, the size of bainitic packets within these fine grains
190 closely matches the grain size, consistent with prior research, suggesting that as the
191 grain size decreases, the orientation of bainitic laths becomes more uniform, and the
192 number of bainite variants within an austenite grain decreases until the packet size is
193 comparable to the grain size [22, 25]. In Figure 3(b), it can be observed that the

f
194 designed steel has fewer low-angle grain boundaries, indicating a lower presence of

oo
195 substructures such as packets and blocks within the grains. Consequently, the material
196 exhibits reduced capability for obstacles inside the grains to divert crack propagation,

r
197 leading to easier crack propagation within the grains and less propensity for cracks to
198 traverse grain boundaries.
-p
re
lP
na
ur
Jo

199
200 Figure 3. (a) The SEM images of the microstructure and (b) EBSD maps of grain
201 boundary misorientation distribution for the designed steel, where black and red lines
202 indicate the high misorientation angle boundaries (θ≥15°) and low angle boundaries
203 (2°≤θ＜15°)
204 The grain size and packet size were statistically analyzed using the
205 cross-sectional method, as shown in figure 4. The histograms and box plots
206 illustrating the length and width distributions are presented in figure 4 (c)-(f). It is
207 worth noting that due to higher corrosion resistance of martensite, the analyzed packet
208 sizes predominantly pertain to bainite's packet sizes. The average value of the length
209 and width of grains are 4.8±2.7 μm and 3.6±1.9 μm, respectively. The peak length of
210 the grain size falls within the range of 3-4 μm, while the peak width is observed
211 between 2-3 μm. The length distribution of grain spans from 3.1 μm to 5.7 μm for the
212 25% to 75%, with a few individual grains exceeding 15 μm. In terms of grain width,
213 the distribution ranges from 2.4 μm to 4.3 μm for the 25% to 75%, with a few
214 individual grains larger than 10 μm.
215 Similarly, for packets, the average length and width are 3.7±1.4 μm and 2.8±1.1
216 μm, respectively, with a length peak at 3-4 μm and a width peak at 2-3 μm. The 25%
217 to 75% range for packet length is 2.7-4.3 μm, with a few individual packets larger
218 than 8 μm. Regarding packet width, the 25% to 75% range is 2.0-3.3 μm, with a few
219 individual packets larger than 6 μm.

220

f
100

oo
(c) Length (d) 20
90 Width 25%~75%
18 10%~90%
80
16 Median

r
70 Mean
14
60 -p Outlier
Count

12
50
Range

10
re
40
8
30
6
lP

20
4
10

0 2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
na

0
221 Grain size(um) Size (length) Size (width) length/width

100
(e) length (f) 20
90 25%~75%
ur

width
18 10%~90%
80
16
Median
70 Mean
Jo

14 Outlier
60
Count

12
50
Range

10
40
8
30
6
20
4
10

0 2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
222 Packet sizes(um) Size (length) Size (width) length/width

223 Figure 4. The grain size (a), packet size (b) and statistical data of histograms and box
224 plots (c)-(f) of the designed steel by cross-sectional method. Outliers in the box plots
225 refer to data points lying outside the whiskers(10%-90%)
226 The aspect ratio of grain length to width and the aspect ratio of packet length to
227 width have similar averages of 1.35±0.33 and 1.36±0.41, respectively, indicating
228 minimal differences between the two.
229 Through comparison, both the length and width of packets are smaller by
230 approximately 1 μm compared to grain dimensions, with less data dispersion observed.
231 This suggests that larger grains struggle to exist as individual packet bundles, while
232 smaller grains tend to have a higher occurrence of single packet sizes.
233 3.3 Fracture Surface Morphology
234 The macroscopic fracture morphology of some typical impact specimens are
235 illustrated in Figure 5. At temperatures ranging from RT down to -80 °C, the
236 macroscopic fracture surface is primarily composed of fibrous regions and shear lips.
237 Additionally, distinctive striped features are present within the fibrous regions.
238 Similarly, one of the fracture surfaces at -100 °C, corresponding to higher toughness,
239 exhibits characteristics consistent with those observed at -80 °C. Conversely, the other
240 fracture surface at -100 °C, indicative of lower toughness, consists mainly of flat
241 radial regions separated by macroscopic ridges, although shear lips still occupy a
242 certain area. At temperatures ranging from -120 °C to -196 °C, the impact fracture
243 indicates the nearly complete absence of shear lips, with the fracture surface

f
244 predominantly consisting of radial regions.

oo
245

r
-p
re
246
lP
na
ur

247
Jo

248 Figure 5. The macroscopic fracture morphology of impact specimens. (a) RT, (b)
249 -40 ℃, (c) -80 ℃, (d) -100℃ (216 J), (e) -100 ℃ (54 J), (f) -120 ℃, (g) -160 ℃, (h)
250 -196 ℃. The white lines are used to delineate different regions. Additionally, the
251 white lines in the radial region indicate the direction of the cracks.
252 Macroscopic and microscopic fracture morphology at RT is depicted in figure 6,
253 revealing large elongated primary dimples, followed by smaller primary and
254 secondary dimples. Consequently, the fracture mechanism involves a combined
255 mechanism of primary dimple growth and coalescence, along with secondary dimple
256 expansion. The larger primary dimples are primarily located within the fibrous bands
257 of the macroscopic fracture surface, characterized by their large and deep nature,
258 which are typical fracture features of high-toughness samples.
 f
259

oo
260 Figure 6. Macroscopic and microscopic fracture morphology of the impact specimen
261 at RT.

r
262
263
-p
Using the cross-sectional method, the lengths and widths of primary and
secondary dimples at different locations on the RT and -40 °C impact fracture surfaces
re
264 were statistically analyzed and compared, as shown in figure 7. At the fibrous region
265 of the RT specimen, the average lengths and widths of the larger primary dimples are
lP

266 37.3±13.5 and 28.8±9.8 μm, respectively. The 25%-75% range for lengths is
267 concentrated between 27.3-41.6 μm, with a few dimples exceeding 70 μm. The
268 25%-75% range for widths is centered between 21.6-33.7 μm, with a few dimples
na

269 exceeding 50 μm. These dimple sizes are much larger than the grain size.
270 For slightly smaller primary dimples in other locations, the average lengths and
ur

271 widths are 15.5±4.9 and 12.8±4.5 μm, respectively. The 25%-75% range for lengths is
272 concentrated between 12.3-18.0 μm, with a few dimples exceeding 20 μm. The
Jo

273 25%-75% range for widths is centered between 9.8-15.0 μm, with a few dimples
274 exceeding 20 μm. These dimple sizes are comparable to the abnormal grain sizes.
275 The average lengths and widths of secondary dimples are 2.8±1.0 and 2.2±0.8
276 μm, respectively. The 25%-75% range for lengths is concentrated between 2.2-3.5 μm,
277 and for widths, it is centered between 1.6-2.6 μm. Secondary dimple lengths and
278 widths are noticeably smaller than the original grain size and are largely aligned with
279 the 25%-75% range of packet sizes.
280 It should be noted that, given the lack of one-to-one correspondence between
281 dimple size, grain size, and packet size, the analysis correlating dimple size with grain
282 size and packet size is based solely on mathematical statistical analysis results. The
283 analysis of cleavage facet dimensions encounters a similar scenario, as described in
284 the subsequence.
285

f
r oo
-p
re
286
lP

(f) 100 25%~75%

10%~90%
Median
80
Mean
na

Outlier

60
Size(um)
ur

40
Jo

20

RT-priamry voids RT-priamry voids RT-secondary -40℃-priamry voids -40℃-secondary

287 (large) (small) voids (large) voids

288 Figure 7. Analysis of dimple dimensions on impact fracture surfaces at RT and -40 °C.
289 (a)-(c) RT impact specimen, (d)-(e) -40 ℃ impact specimen, (f) the statistical data of
290 the box plot of dimple size distribution for RT and -40 ℃ samples
291 By contrasting the characteristics of matched fracture surfaces with cleavage
292 facets, a statistical analysis of the cleavage facets in quasi-cleavage fracture was
293 performed using the cross-sectional method. The size distribution histograms and box
294 plots of their lengths and widths are presented in figure 8. The average lengths and
295 widths of the cleavage surfaces are 4.0±1.8 μm and 3.0±1.2 μm, respectively, with
296 length peaks centered around 3-4 μm and width peaks around 2-3 μm. The 25%-75%
297 range for length sizes is concentrated between 2.8-5.0 μm, with a few surfaces
298 exceeding 10 μm. The 25%-75% range for width sizes is centered between 2.1-3.7 μm,
299 with a few surfaces exceeding 5 μm. Upon comparison, it is observed that the
300 cleavage surface sizes closely coincide with the grain sizes (except for a few
301 extremely large size outliers). It should be noted that, as mentioned earlier, it is
302 observed that in the case of fine-grained microstructures, the packet size of bainite in
303 this designed steel is roughly equivalent to the grain size. Given this observation, it is
304 uncertain whether the dimensions of cleavage facets correspond to grain size or
305 packet size.

f
oo
306
307

r
(c) 20
30

25
-p
Length
Width
(d)
18

16
25%~75%
10%~90%
Median
re
Mean
14 Outlier
20
12
lP
Count

Range

15 10

8
10
na

5 4

2
ur

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
308 Facets sizes(um) Size (length) Size (width) length/width
Jo

309 Figure 8. Size distribution histograms and box plots for the lengths and widths of the
310 cleavage facets in the impact specimen test at -100 ℃.

311 4. Discussion
312 4.1 Crack Nucleation Sites Analysis
313 A comparative analysis and statistical assessment of crack nucleation sites were
314 conducted on impact specimens tested at temperatures ranging from -100 °C to
315 -196 °C, utilizing two matching sides of the fracture surface. A total of 12 impact
316 specimens were examined, with two specimens displaying multiple crack nucleation
317 sites (-160 °C, CVN=13 J, and -180 °C, CVN=15 J). When identifying crack initiation
318 sites, such as carbides, M/A islands, and inclusions, SEM imaging was primarily
319 utilized to highlight detailed morphological and size characteristics. EDS was
320 employed solely to corroborate these observations. Figure 9 displays the fracture
321 morphology and corresponding EDS line scanning images of crack nucleation sites
322 associated with M/A islands and carbides. SEM analysis distinctly shows carbides to
323 be considerably smaller than M/A islands, facilitating their differentiation based on
324 morphology and size. The EDS analysis confirms relatively higher carbon content in
325 both M/A islands and carbides compared to the matrix, aligning with SEM
326 observations.

327

f
80 120
(c) C

oo
1.15μm (d) C
70 110 0.14μm

60 100

r
90
50
-p
Pulses

Pulses

80
40
70
re
30
60
20
50
lP

10 40

0 30
0 1 2 3 4 5 0.0 0.5 1.0 1.5 2.0 2.5
na

328 Pos.[µm] Pos.[µm]

329 Figure 9. The (a) SEM image of the fracture surface with crack nucleated from
M/A islands at -120 °C. (b) SEM image of crack nucelated from carbides at -140 °C.
ur

330
331 (c) and (d) the carbon elemental line scan results via EDS, corresponding to M/A
Jo

332 islands and carbides, respectively.

333 Matching fracture surface morphologies within the temperature range of -100 to
334 -196 °C are illustrated in the figure 10-15. It can be seen that various types of
335 nucleation sites, such as abnormally large grain, carbides, inclusions, and M/A islands,
336 contribute to crack initiation. Abnormally large grain, if present around the cleavage
337 origin, allow the nucleated crack to grow larger than the critical size required for
338 unstable propagation before it can be obstructed by the grain boundary. As a result,
339 fracture occurs at lower stress than that required when finer grains are present around
340 the cleavage origin [35]. An analysis of the river-pattern morphology and overall
341 appearance of crack initiation region on the -100 ℃ specimen, coupled with grain size
342 statistics data, suggests that the crack initiation region consist of grains with closely
343 aligned bainite packet orientation, as shown in figure 10. Consequently, the proximity
344 of bainite packet orientations among adjacent grains is considered indicative of
345 abnormally large grains. Additionally, specimens exhibiting fractures at carbide
346 locations show relatively higher impact toughness (figure 10 and 12).
347

f
r oo
-p
re
348
349 Figure 10. (a) Appearance of crack initiation region on the -100 ℃ impact specimen,
lP

350 (b) and (c) the matching fracture surface morphologies at the crack initiation, flip
351 horizontal, (d) the depict of the crack initiation region.
na

352
ur
Jo

353
354
355
356 Figure 11. (a) Appearance of crack initiation region on the -120℃ impact specimens,
357 (b) nucleation sites of fracture crack initiation (ALG+inclusion) of the specimen of
358 -120℃-1.(c) and (d) nucleation at ALG or NOB+M/A of specimen of -120℃-2, (d) is

f
359 processed by flip horizontal.

oo
360 In the temperature range of -140 to -196°C in the lower plateau, crack initiation
361 is primarily caused by the separation of M/A islands from the matrix. However, at the

r
362
363
-p
crack initiation induced by the M/A islands, there may be multiple adjacent M/A
islands, the presence of nearby inclusions, or located in an abnormally large grain, or
re
364 multiple nucleation sources. Specifically, the -196 °C specimen (CVN=8 J) displays
365 fractures at a location where M/A islands, inclusion, and abnormally large grain (or
lP

366 multiple grains with closely aligned bainite packet orientations) coexist. .
na
ur
Jo

367
368 Figure 12.Morphology and nucleation sites of fracture crack initiation of matching
369 sides of the fracture surface of -140 ℃ impact specimens. (a) & (b) nucleation at
370 carbide. (b) is processed by flip horizontal.

371
372
373 Figure 13. Morphology and nucleation sites of fracture crack initiation of matching
374 sides of the fracture surface of -160 ℃ impact specimens. (a) & (b) nucleation at M/A,
375 (c) & (d) Grain boundary junctions. (b) and (d) are processed by flip horizontal.

f
r oo
-p
re
lP

376
na
ur
Jo

377
378 Figure 14. Morphology and nucleation sites of fracture crack initiation of matching
379 sides of the fracture surface of -180 ℃ impact specimens. (a) & (b) nucleation at M/A,
380 (c) & (d) M/A+inclusion. (b) and (d) are processed by flip horizontal.
381

f
r oo
-p
re
382
383 Figure 15. Morphology and nucleation sites of fracture crack initiation of matching
lP

384 sides of the fracture surface of -196 ℃ impact specimens. (a) & (b) nucleation at
385 inclusion, (c) & (d) M/A+inclusion (ALG or NOB). (b) and (d) are processed by flip
na

386 horizontal.
387 The results of crack nucleation site determination are presented in the Table 3.
ur

388 The primary location for crack initiation was the M/A islands, with 11 out of the 14
389 crack nucleation sites(from 12 specimens, including 2 specimens with two crack
Jo

390 nucleation sites each) being associated with them. However, it is important to note
391 that crack initiation is not solely attributed to the presence of M/A islands.
392 Additionally, crack initiation sites involving M/A islands are also found concurrently
393 with inclusions, carbides, grain boundaries, and abnormally large grains. This
394 suggests that the interaction among M/A islands and other nucleation factors
395 contributes to fracture nucleation. Similarly, in three instances, inclusions are
396 observed at the crack initiation sites. However, the inclusions are either located at
397 abnormally large grains (-120 °C, CVN=39 J) or in the vicinity of the M/A islands
398 (-180 °C, CVN=15 J, and -196 °C, CVN=14 J). No cases are observed where
399 impurities alone triggered cracking.
400 Although the impact of M/A islands on quasi-cleavage fracture has been widely
401 reported, the results of this study diverge from current research. For instance, Kim [33]
402 focused on coarse-grained regions of welded joints and primarily investigated the
403 influence of M/A islands on toughness reduction and the formation of local brittle
404 zones. They observed a significant decrease in material toughness with increasing
405 M/A island content, attributing M/A islands as the primary factor leading to
406 embrittlement in coarse-grained regions. However, Kim suggested that grain size had
407 minimal influence on local embrittlement due to the relatively large grain size (30-49
408 µm) in their study material.
409 Table 3. Statistical list of crack nucleation site for impact specimens tested from -100 ℃
410 to -196 ℃.
Test temperature(℃） CVN (J) Crack nucleation site
-100-1 98 NOB+Carbide
-100-2 54 ALG or NOB+M/A
-120-1 39 ALG+inclusion
-120-2 30 ALG or NOB+M/A
-140-1 74 Carbide
-140-2 49 Carbide+M/A
-160-1* 13 Grain boundary junctions

f
-160-1* 13 M/A

oo
-160-2 27 Grain boundary junctions+M/A
-180-1 20 M/A

r
-180-2* 15 M/A+inclusion
-180-2*
-196-1
15
14
-p Multiple M/A
M/A+inclusion
re
-196-2 8 ALG or NOB+M/A+inclusion
*multiple crack nucleation sites
lP

411

412 In contrast, the findings of this study on fine-grained low-carbon bainitic steel
indicate that while M/A islands remain influential in quasi-cleavage fracture, they are
na

413
414 not the sole determinants of fracture mechanisms. Instead, they interact with various
415 factors such as abnormally large grains, inclusions, grain boundaries, and M/A island
ur

416 clustering, collectively contributing to fracture initiation. The complex interplay of

417 these microstructural defects in fine-grained steels suggests that the mechanisms
Jo

418 underlying local embrittlement fracture may be more intricate than those in
419 coarse-grained regions, necessitating the development of new explanatory models.
420 4.2 Bainitic Matrix in Crack Initiation and Propagation
421 The band contrast (BC) microstructure obtained by EBSD and the reconstructed
422 microstructure performed using the open-source crystallographic toolbox MTEX 5.8
423 are shown in figure 16. Quantitative analysis of martensite and bainite was conducted
424 using the BC peak method in EBSD [38, 39]. The results indicate that the steel
425 possesses a bainite content ranging from 70% to 80%, with a corresponding
426 martensite (or M/A) content of 20-30%. Furthermore, the reconstruction
427 microstructures reveal that the steel is primarily composed of bainite, with a
428 predominant occurrence of singular bainite packet. Martensite(M/A) exhibits three
429 distribution states: within bainite, at grain boundaries, and as isolated grains.
430

f
r oo
-p
re
lP
na

431
432 Figure 16. The band contrast and reconstructed microstructure. (a) BC map, (b) the
reconstructed microstructure derived from (a), (c) reconstructed microstructure of
ur

433
434 another specimen, (d) representative grains and the corresponding phase compositions
Jo

435 In figure 17, observation and analysis of secondary cracks on the fracture surface
436 indicate that when there are cleavage steps or at interfaces (phase or grain boundaries),
437 secondary cracks tend to experience a certain degree of deviation, although the
438 deviation angle is relatively small. Simultaneously, the presence of long straight
439 cracks within the bainite suggests that the bainite laths have minimal influence on the
440 deviation of cracks [25]. Almost all observed secondary cracks cease their
441 propagation within the second grain, indicating that the critical event in the
442 micro-mechanisms of quasi-cleavage fracture is the propagation of cracks crossing
443 grain/grain boundaries and entering the adjacent grain [22]. These findings are
444 consistent with the results shown in Figures 3 and 4, indicating that the primary
445 contributor to the high toughness of this material is grain refinement, coupled with a
446 higher density of high-angle grain boundaries, effectively impeding crack propagation.
447 Additionally, it was initially thought that the bainitic substructure would also
448 contribute to hindering crack propagation. However, due to the lower number of
449 bainitic packets within grains under fine grain conditions, there are fewer high-angle
450 grain boundaries within the grains. Nevertheless, this also suggests that the ability to
451 impede crack propagation within the grains is limited. Therefore, to further improve
452 toughness, increasing the number of bainitic packets within grains under fine grain
453 conditions could be considered. This would necessitate further optimization of
454 composition and microstructural control methods.

455

f
r oo
-p
re
lP

456
na

457 Figure 17. Secondary cracks observed on the fracture surface of impact specimens
458 tested at different temperatures, (a) -140 ℃, (b) -160 ℃, (c) -180 ℃, (d) -196 ℃.
In summary, the designed steel exhibits extremely high toughness, with the
ur

459
460 transition temperature for ductile to brittle behavior in longitudinal impact specimens
Jo

461 being -100 °C. The microstructural characteristics of the material include fine grains
462 with a bainitic matrix. However, within the fine grains, the size of bainitic packets is
463 comparable to that of the grains, and there is a low density of high-angle grain
464 boundaries inside the grains. Additionally, there is a certain amount of M/A islands
465 present, distributed either within grains, at grain boundaries, or as isolated grains.
466 Moreover, there are small amounts of fine carbides present.
467 Under ductile fracture conditions, the fracture mechanism involves the growth
468 and coalescence of primary dimples, along with the expansion of secondary dimples,
469 as shown in figure 18(a). The larger primary dimples constitute a significant
470 proportion and their large and deep nature contributes effectively to the material's
471 impact toughness.
472 Under quasi-cleavage fracture conditions, the predominant sites of crack
473 initiation are typically M/A islands. Nevertheless, crack initiation is often associated
474 with other factors such as abnormally large grains, carbides, and inclusions. When the
475 microstructure predominantly consists of single bainitic phases within grains, cracks
476 cannot effectively be hindered from propagating. However, due to the small grain size
477 and the presence of a certain amount of fine M/A islands[24], cracks tend to deflect or
478 arrest within grain boundaries or within second phases, increasing the crack
479 propagation path, enhancing toughness, and indicating that the propagation of
480 quasi-cleavage cracks across grains is identified as the critical micro-mechanism
481 event, as shown in figure 18(b).

f
r oo
-p
re
lP

482
483 Figure 18. Fracture surface characteristics and schematic illustrations of crack
484 propagation for (a) ductile and (b)quasi-cleavage fracture in the designed steel.
na

485 4. Conclusion
ur

486 An ultra-high toughness high strength low alloy steel is achieved through a
487 combination of low-carbon design, controlled rolling, and off-line quenching and
Jo

488 tempering processes, resulting in a yield strength of 1048 MPa, tensile strength of
489 1217 MPa, elongation of 15.3%, and Charpy impact energy of up to 250 J at -40 °C.
490 The difference between longitudinal and transverse impact specimens is minimal,
491 with the ductile-to-brittle transition temperatures approximately -100 °C and -75 °C,
492 respectively. Through microstructural characterization and fractographic analysis, the
493 correlation between the material's microstructure and fracture mechanism is examined.
494 The conclusions are as follows:
495 1) Fracture mechanisms of specimens tested at temperatures ranging from room
496 temperature to -80 °C involve the growth and coalescence of primary dimples,
497 accompanied by the expansion of secondary dimples. The larger primary dimples
498 constitute a significant proportion, and their large and deep nature contributes to
499 exceptionally high toughness.
500 2) The primary factor contributing to the high toughness of this material is grain
501 refinement. However, in the fine-grained microstructure, the bainitic packet size is
502 comparable to the grain size, which fails to effectively increase the pathways for crack
503 propagation within grains. Additionally, the similar orientation of bainite laths in
504 multiple adjacent grains is equivalent to abnormally large grains, making them
505 susceptible locations for crack nucleation.
506 3) Despite M/A islands being the predominant nucleation sites for cracks in the
507 fine-grained microstructure, they require the combined action of other factors such as
508 abnormally large grains, grain boundaries, and inclusions to contribute to fracture
509 nucleation. Thus, in the quasi-cleavage fracture mode, crack initiation is not solely
510 attributed to the presence of M/A islands..
511 4) The critical event in the micro-mechanisms of quisi-cleavage fracture for this
512 material is the propagation of cracks crossing grain/grain boundaries and entering the
513 adjacent grain.

514 Declaration of competing interest

515 None.

516 Acknowledgments

f
oo
517 The support from the GuiKeAA21077010 is gratefully acknowledged.

518 Data availability

r
519 References
-p
re
520 [1] Y. Weng, H. Dong, Y. Gan, Advanced steels: the recent scenario in steel science and technology,
521 Springer Science & Business Media, 2011.
lP

522 [2] R.O. Ritchie, The conflicts between strength and toughness, Nature materials, 10 (2011) 817-822.
523 [3] S. Igi, M. Miyake, Development of Thermo-Mechanical Control Process (TMCP) and High
na

524 Performance Steels in JFE Steel, JFE Tech. Rep, 26 (2021) 86-94.
525 [4] K. Hayashi, A. Nagao, Y. Matsuda, 550 and 610 MPa class high-strength steel plates with excellent
ur

526 toughness for tanks and penstocks produced using carbide morphology controlling technology, JFE
527 Technical Report, 11 (2008) 20.
Jo

528 [5] N. Shikanai, S. Mitao, S. Endo, Recent development in microstructural control technologies through
529 the thermo-mechanical control process (TMCP) with JFE steel’s high-performance plates, JFE technical
530 report, 11 (2008) 1-6.
531 [6] P. Bai, C. Shang, H.-H. Wu, G. Ma, S. Wang, G. Wu, J. Gao, Y. Chen, J. Zhang, J. Zhu, A review on the
532 advance of low-temperature toughness in pipeline steels, Journal of Materials Research and
533 Technology, (2023) 6949 e6964.
534 [7] A. Guo, R. Misra, J. Xu, B. Guo, S. Jansto, Ultrahigh strength and low yield ratio of
535 niobium-microalloyed 900 MPa pipeline steel with nano/ultrafine bainitic lath, Materials Science and
536 Engineering: A, 527 (2010) 3886-3892.
537 [8] J.-Y. Yoo, S.-S. Ahn, D.-H. Seo, W.-H. Song, K.-B. Kang, New development of high grade X80 to X120
538 pipeline steels, Materials and Manufacturing Processes, 26 (2011) 154-160.
539 [9] H. Kong, C. Xu, C. Bu, C. Da, J. Luan, Z. Jiao, G. Chen, C. Liu, Hardening mechanisms and impact
540 toughening of a high-strength steel containing low Ni and Cu additions, Acta Materialia, 172 (2019)
541 150-160.
542 [10] L. Kan, Q. Ye, Z. Wang, T. Zhao, Improvement of strength and toughness of 1 GPa Cu-bearing HSLA
543 steel by direct quenching, Materials Science and Engineering: A, 855 (2022) 143875.
544 [11] Y. Zhao, X. Tong, X. Wei, S. Xu, S. Lan, X.-L. Wang, Z. Zhang, Effects of microstructure on crack
545 resistance and low-temperature toughness of ultra-low carbon high strength steel, International
546 Journal of Plasticity, 116 (2019) 203-215.
547 [12] A. Kaijalainen, Effect of microstructure on the mechanical properties and bendability of
548 direct-quenched ultrahigh-strength steels, University of Oulu, Finland, 2016.
549 [13] C.S.f. Metals, C.A.o. Engineering, Z. Li, X. Sun, Z. Yang, Q. Yong, Grain Refinement and Toughening
550 of Low Carbon Low Alloy Martensitic Steel with Yield Strength 900 MPa Grade by Ausforming, HSLA
551 Steels 2015, Microalloying 2015 & Offshore Engineering Steels 2015: Conference Proceedings,
552 Springer, 2016, pp. 195-201.
553 [14] Y. Weng, Ultra-fine grained steels, Springer Science & Business Media, 2009.
554 [15] A. Ghosh, S. Das, S. Chatterjee, P.R. Rao, Effect of cooling rate on structure and properties of an
555 ultra-low carbon HSLA-100 grade steel, Materials Characterization, 56 (2006) 59-65.
556 [16] K. Zhang, Z.-D. Li, X.-J. Sun, Q.-L. Yong, J.-W. Yang, Y.-M. Li, P.-L. Zhao, Development of Ti–V–Mo

f
557 complex microalloyed hot-rolled 900-MPa-grade high-strength steel, Acta Metallurgica Sinica (English

oo
558 Letters), 28 (2015) 641-648.
559 [17] S. Thompson, Interrelationships between yield strength, low-temperature impact toughness, and

r
560 microstructure in low-carbon, copper-precipitation-strengthened, high-strength low-alloy plate steels,
561
562
-p
Materials Science and Engineering: A, 711 (2018) 424-433.
[18] W.-T. Zhu, J.-J. Cui, Z.-Y. Chen, Y. Zhao, L.-Q. Chen, Correlation of microstructure feature with
re
563 impact fracture behavior in a TMCP processed high strength low alloy construction steel, Acta
564 Metallurgica Sinica (English Letters), 35 (2022) 527-536.
lP

565 [19] H. Xie, L.-X. Du, J. Hu, R.D.K. Misra, Microstructure and mechanical properties of a novel 1000
566 MPa grade TMCP low carbon microalloyed steel with combination of high strength and excellent
na

567 toughness, Materials Science and Engineering: A, 612 (2014) 123-130.

568 [20] B. Hwang, C.G. Lee, S.-J. Kim, Low-temperature toughening mechanism in thermomechanically
ur

569 processed high-strength low-alloy steels, Metallurgical and Materials Transactions A, 42 (2011)
570 717-728.
Jo

571 [21] H. Zhao, E.J. Palmiere, Influence of cooling rate on the grain-refining effect of austenite
572 deformation in a HSLA steel, Materials Characterization, 158 (2019) 109990.
573 [22] J. Chen, R. Cao, Micromechanism of cleavage fracture of metals: a comprehensive microphysical
574 model for cleavage cracking in metals, Butterworth-Heinemann, 2014.
575 [23] A. Pineau, A.A. Benzerga, T. Pardoen, Failure of metals I: Brittle and ductile fracture, Acta
576 Materialia, 107 (2016) 424-483.
577 [24] H. Lan, L. Du, R. Misra, Effect of microstructural constituents on strength–toughness combination
578 in a low carbon bainitic steel, Materials Science and Engineering: A, 611 (2014) 194-200.
579 [25] H. Bhadeshia, Bainite in steels: Theory and Practice, CRC Press, London, 2015.
580 [26] B. Wang, J. Lian, Effect of microstructure on low-temperature toughness of a low carbon Nb–V–Ti
581 microalloyed pipeline steel, Materials Science and Engineering: A, 592 (2014) 50-56.
582 [27] T. Baker, Microalloyed steels, Ironmaking & Steelmaking, 43 (2016) 264-307.
583 [28] J.-m. Zhang, W.-h. Sun, H. Sun, Mechanical properties and microstructure of X120 grade high
584 strength pipeline steel, Journal of Iron and Steel Research International, 17 (2010) 63-67.
585 [29] M.S. Joo, D.W. Suh, H.K.D.H. Bhadeshia, Mechanical anisotropy in steels for pipelines, ISIJ
586 international, 53 (2013) 1305-1314.
587 [30] Y. Han, J. Shi, L. Xu, W. Cao, H. Dong, Effect of hot rolling temperature on grain size and
588 precipitation hardening in a Ti-microalloyed low-carbon martensitic steel, Materials Science and
589 Engineering: A, 553 (2012) 192-199.
590 [31] B. Kim, S. Lee, N. Kim, D. Lee, Microstructure and local brittle zone phenomena in high-strength
591 low-alloy steel welds, Metallurgical Transactions A, 22 (1991) 139-149.
592 [32] X. Luo, X. Chen, T. Wang, S. Pan, Z. Wang, Effect of morphologies of martensite–austenite
593 constituents on impact toughness in intercritically reheated coarse-grained heat-affected zone of HSLA
594 steel, Materials Science and Engineering: A, 710 (2018) 192-199.
595 [33] F. Niessen, T. Nyyssönen, A.A. Gazder, R. Hielscher, Parent grain reconstruction from partially or
596 fully transformed microstructures in MTEX, Journal of applied crystallography, 55 (2022) 180-194.
597 [34] T. Yin, Y. Shen, N. Jia, Y. Li, W. Xue, Controllable selection of martensitic variant enables concurrent
598 enhancement of strength and ductility in a low-carbon steel, International Journal of Plasticity, 168
599 (2023) 103704.

f
600 [35] D. Chakrabarti, M. Strangwood, C. Davis, Effect of bimodal grain size distribution on scatter in

oo
601 toughness, Metallurgical and Materials Transactions A, 40 (2009) 780-795.
602 [36] L. Cao, S. Wu, P. Flewitt, Comparison of ductile-to-brittle transition curve fitting approaches,

r
603 International journal of pressure vessels and piping, 93 (2012) 12-16.
604
605
-p
[37] E. Lucon, M.L. Martin, B.L. Calvo, X. Chen, M.A. Sokolov, Assessment of Different Approaches for
Measuring Shear Fracture Appearance in Charpy Tests, Journal of Testing and Evaluation, 51 (2023)
re
606 2833-2846.
607 [38] M.-S. Baek, K.-S. Kim, T.-W. Park, J. Ham, K.-A. Lee, Quantitative phase analysis of
lP

608 martensite-bainite steel using EBSD and its microstructure, tensile and high-cycle fatigue behaviors,
609 Materials Science and Engineering: A, 785 (2020) 139375.
na

610 [39] Y. Wang, J. Hua, M. Kong, Y. Zeng, J. Liu, Z. Liu, Quantitative analysis of martensite and bainite
611 microstructures using electron backscatter diffraction, Microscopy Research and Technique, 79 (2016)
ur

612 814-819.
613
Jo
 Highlights
⚫ Achieved 960 MPa yield strength grade steel and 250 J toughness at -40°C.
⚫ Grain refinement enhances toughness, but comparable bainitic packet size limits
crack propagation pathways.
⚫ M/A islands are predominant nucleation sites, but require combined factors for
fracture initiation.
⚫ Critical micro-mechanism is cleavage crack crossing grain boundaries.

f
r oo
-p
re
lP
na
ur
Jo
Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests:

of
ro
-p
re
lP
na
ur
Jo