Computational Materials Science 192 (2021) 110319
Contents lists available at ScienceDirect
Computational Materials Science
journal homepage: www.elsevier.com/locate/commatsci
First-principles study of tensile and shear strength of an Fe2Al5//
Fe interface
Muhammad Zeeshan Khalid a, b, *, Jesper Friis c, d, Per Harald Ninive a, Knut Marthinsen b,
Inga Gudem Ringdalen c, Are Strandlie a
a
Department of Manufacturing and Civil Engineering, Norwegian University of Science and Technology, Gjøvik 2815, Norway
Department of Materials Science and Engineering, Norwegian University of Science and Technology (NTNU), Norway
SINTEF Materials and Chemistry, Trondheim, Norway
d
Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway
b
c
A R T I C L E I N F O
A B S T R A C T
Keywords:
Fe-Al intermetallics
Tensile strength
Mechanical strength
Atomistic simulations
Face-to-face matching
Welding
The interfacial strengths of a low misfit Fe2Al5//Fe interface structure found at aluminum-steel joints has been
studied using density functional theory. An interface between Fe and Fe2Al5 was selected based on a criteria of
low lattice misfit and number of atoms. Through virtual tensile testing of bulk Fe2Al5 and the interface structures
we show that the energy-displacement curve can be well described by including extra polynomial terms in the
Universal Binding Energy Relation (UBER). It is shown that the Fe2Al5//Fe interface has a higher tensile strength
than the bulk Fe2Al5 phase. We also find that the shear deformation process potentially can be initiated from an
Fe-terminated interface.
1. Introduction
Owing to the increased interest in light-weight and environmentalfriendly technology, Fe-Al compounds have been gaining increased industrial interest due to their light-weight, corrosion resistance and hightemperature resistance behavior [1–3]. However, the joining of
aluminum and steel by traditional fusion welding techniques has been
considered a main challenge due to the significant differences between
their physical and chemical properties [4,5]. The intermetallic compounds (IMCs) which develop at the interface are normally not wanted,
but unavoidable when welding aluminum and steel.
Various methods have been proposed and studied to join aluminum
and steel [3,6]. For any method which requires high temperatures, a
brittle layer of different types of Fe-Al IMCs is developed at the joint,
making it difficult to obtain the desired joint strength. Although solidstate welding techniques can suppress the formation of Fe-Al IMCs at
joints due to the low temperature, these methods can still not completely
limit the formation of IMCs and can thus only produce Fe-Al joints with
limited strength.
The thickness of the IMC layers also plays an important role in the
strengthening of Fe-Al joints. It has been reported that the thickness of
Fe-Al IMC layers formed in a brazed interface can be limited to less than
10 μm, which is considered as the critical thickness of a Fe-Al IMC layer
for Fe-Al joints with good mechanical strength [7]. Analyses of Fe-Al
joints suggest that the micro-structures and distribution of Fe-Al IMCs
at the interface are dependent on heat input, and play an important role
in determining the mechanical and/or corrosion behavior of the joints
[8,9]. In general, most of the experimental and theoretical studies on FeAl IMC layers focus on, (i) heat input and thickness of the IMC layers
[10] (ii) welding methodology [11,12] (iii) tensile and shear strength of
IMC layers at the joint [3] and (iv) extended isothermal treatment
[13–15].
Despite all these studies, the interfacial strength of intermetallics
such as Fe2Al5//Fe has not been studied much in literature. Since it is
thermodynamically possible to produce a range of Fe-Al compounds at
the interface [16,17], it is necessary to understand the basic mechanical
and interfacial strength of all these compounds to clarify their roles for
the joint strength. The lack of convincing results for the interfacial
strength is not due to a lack of academic and industrial interests on this
important subject. However, due to the small thickness (2.3 ± 0.6 μm) of
the IMC layers [18], it is very difficult to experimentally predict the
interface strength of these compounds.
The above brief review indicates that the understanding of the
behavior and strength of the individual interfaces of these compounds is
* Corresponding author at: Department of Manufacturing and Civil Engineering, Norwegian University of Science and Technology, Gjøvik 2815, Norway.
E-mail address: zeeshan.khalid039@gmail.com (M.Z. Khalid).
https://doi.org/10.1016/j.commatsci.2021.110319
Received 9 October 2020; Received in revised form 14 January 2021; Accepted 15 January 2021
Available online 18 February 2021
0927-0256/© 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
far from complete, and it is evident that an atomistic study of these interfaces could provide useful new insight. The lack of atomistic studies is
due to the complex atomic structure of the intermetallic compounds. It
is, therefore, challenging to develop an interface model which is periodic, simple and has a low lattice misfit. There have been many studies
in literature for the determination of the interface structures between
two bulk phases, e.g. [19–24]. Many different approaches such as Olattice theory [25,26], the edge to edge model [27], the Conincidence of
Reciprocal lattice point (CRLP) model [28] based on the Zur algorithm
[22] have been developed to find and characterize the OR and conincidence of lattice between phases and grains. A major disadvantage of
O-lattice theory is the lack of predictive capabilities, however the other
approaches can successfully predict the OR, but they do not match the
full structures. The edge to edge model considers the high density or
nearly closed pack planes and direction and the CRLP and Zur algorithm
match the underlying lattices of the structure ignoring the atoms inside
them.
In this work, we have used a face-to-face matching technique to
predict a possible Orientation Relationship (OR) between Fe2Al5 and Fe
suitable for atomistic calculations.
This work is a follow-up of a project working on the role of IMC
layers on the joining strength of aluminum and steel. Many distinct IMC
layers have been observed at aluminium and steel joints, and computational calculations on several other Fe-Al IMC layers have already been
published [29,30]. In order to make a consistent comparison between
different Fe-Al IMC layers, the same assumption, methodology, and
computational techniques were applied in this study and as in the other
studies related to the Fe-Al IMCs [29,30].
The scope of this paper is limited to establish and test the modeling
methodology for finding a good atomistic interface structure and to
study the mechanical and interfacial properties of the relevant Fe2Al5//
Fe interface. The structure of the paper is as follows. First, we present the
procedure for finding a low misfit interface structure between Fe and
Fe2Al5. In Section 3 we present the calculation methodology and procedure for performing virtual tensile calculations. In Section 4, we
present results of the strength of the bulk Fe2Al5 as well as the Fe//
Fe2Al5 interface structure. In the last section, we discuss the results
before presenting a summary and conclusions.
Table 1
Calculated equilibrium lattice constants, cohesive energy (Ec ) and formation
energy (ΔHR ) for bulk Fe2Al5.
Fe2Al5
a
b
c
Reference
a0(Å)
b0(Å)
c0(Å)
This work
DFTa
EAMb
Exp.c
7.418
7.466
7.622
7.675
6.428
6.181
6.323
6.403
4.103
4.808
4.178
4.203
Ec(eV/
atom)
ΔHR (eV/
atom)
−7.364
−13.728
−7.352
−8.352
[38]
[39]
[40]
occupancy factor of 1/6 for each of them [36]. We performed ground
state energy calculations to find a stable crystal structure by calculating
the formation enthalphy (ΔHR) and cohesive energy (Ec) and used this
structure further for bulk and interfacial calculations. The results of the
bulk strength calculations of Fe2Al5 have been reported in a previous
work [37] and can also be seen in Table 1.
2.3. Prediction of orientation relationships
In order to create a good representative periodic interface structure
for the DFT calculations, a common supercell of two crystal surfaces
forming an interface is required. However, in the general case the two
crystals have different lattice constants, it is necessary to rotate the two
crystals relative to each other in order to obtain an interface with as little
strain as possible in order to get an as realistic as possible interface.it is
challenging to find low strain interface structure without an excessive
number of atoms in the supercell due to the huge number of possible
orientation relationships and orientations of the habit plane. The benefit
with our algorithm is that it actually finds, in a very efficient way, the
optimal OR and habit plane orientation with minimum strain and a
manageable number size of the supercell. Stradi et al. [21] developed an
algorithm for the efficient and systematic search for common supercells
between two crystalline surfaces. The method presented in this work is
based on the same principles as presented by Stradi et al. First the
equivalent directions of the two crystal surfaces are determined and
rotated to match at the interface. Then, both crystals are equally strained
to match at the interface. This results in different interface structures
and ORs based on the number of atoms and low lattice misfit as presented in Table 6. This method provides an advantage of predicting a
number of ORs between two crystals without an excessive number of
atoms and low lattice misfit interface.
The first step in creating the interface structure is to establish an OR
between the two phases in question. We modeled the interface as an
atomically sharp defect-free interface between two crystals 1 (Fe) and 2
(Fe2Al5). To find possible ORs, a large number of possible sets of crystallographic directions were explored. The possible interface planes in
Fe are defined by all pairs of lattice vectors, u1 and v1 , in Fe. Similarly, u2
and v2 define all possible interface planes of Fe2Al5. To obtain a periodic
interface, the following relations must be fulfilled:
2. Calculation methods and model
2.1. First-principles calculations
The first-principle calculations based on DFT were performed using
the Vienna ab initio Simulation Package (VASP) [31]. The exchange–correlation energy was evaluated using the Generalized Gradient
Approximation (GGA) by Perdew, Burke and Ernzerhof (PBE) [32] and
with the Projector Augmented Wave (PAW) [33] method. By using the
method proposed by Monkhorst–Pack to characterize energy integration
as the first irreducible Brillouin zone [34] mesh size of 9 × 5 × 2 for bulk
Fe2Al5 and 9 × 5 × 1 for Fe2Al5//Fe interface structures.Maximum energy cutoff value of 450 eV was used for the plane wave expansion in
reciprocal space. During the optimization process, the change in total
energy were converged to 10−5 eV. Furthermore, the average force per
atom was reduced to 0.009 eV/Å using a smearing factor of 0.2 and firstorder Meth-Paxton for the smearing of the partial occupation. Due to the
magnetic behavior of Fe atoms, spin-polarized calculations were performed for the interface structures and bulk Fe by specifying the initial
local magnetic moment of Fe.
|u1 | = |u2 |
|v1 | = |v2 )|
γ1 = γ2
(1)
where ∠ γ n = ∠(un , vn ), with n = 1, 2 for crystal 1 and 2, respectively,
and it is defined as the angle between vector directions u and v. We have
added a vacuum layer along the normal direction to avoid periodic
interaction. For this reason, angles ∠ αn = ∠(vn , wn ) and ∠ βn = ∠(un , wn )
are not relevant, as the interface structures do not need to be periodic
along the normal direction to the interface.
In the general case, it is not possible to find an OR satisfying these
conditions exactly. The resulting interface structure depends on how
well these conditions are fulfilled using the strains along direction u and
2.2. Determination of bulk Fe2Al5
Fe2Al5 has an orthorhombic unit cell which contains single crystallographic Fe sites (four per cell) and three Al sites [35]. The Al1 site,
which contains eight atoms per cell, is fully occupied, while Al2 and Al3
are too close to be occupied simultaneously, resulting in a partial
2
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Table 2
Work of separation values for different ORs of Fe2Al5//Fe interface structures.
Interface
Work of separation (J/m2)
Fe2Al5(0–20)//Fe(-121)
Fe2Al5(-1–20)//Fe(-343)
Fe2Al5(-200)//Fe(-343)
Fe2Al5(-200)//Fe(-323)
Fe2Al5(-110)//Fe(-121)
Fe2Al5(-120)//Fe(1–11)
3.58
1.25
0.32
1.72
3.28
2.40
an assortment of low misfit ORs. The calculated values are shown in
Table 2. We finally selected the interface structure that has a low misfit,
the least number of atoms and and the largest value of the work of
separation (3.58 J/m2) for further investigations.
2.4. Fe2Al5 (020)//Fe(121) interface
The atomic structure of Fe2Al5//Fe was constructed using the procedure described above (Sec. 2.3). To ensure the bulk-like interior of
atomic interfaces, six layers of Fe and Fe2Al5 were tested. It is worth
mentioning that Fe2Al5 can be terminated either by Al or Fe at the
interface. Both terminations were used for the interfaces shown in Fig. 2.
To avoid periodic interactions, a vacuum layer of >10 Å was added
along the normal direction to remove the effect of the two artificial interfaces. For the strength calculations, relaxed interface structures were
used as an input for virtual tensile and shear test calculations.
Fig. 1. A possible 3D interface model between crystal 1 and crystal 2. The
crystals are slightly strained in order for them to match.
direction v:
∊u =
||u2 | − |u1 ||
|u1 | + |u2 |
(2)
∊v =
||v2 | − |v1 ||
|v1 | + |v2 |
(3)
2.5. Virtual tensile test calculations
Ab-initio virtual tensile calculations of the Fe2Al5//Fe interface were
carried out in the framework of the Rigid Grain Shift (RGS) and RGS +
relaxation methodology [41,42]. In this approach, the equilibrium
structure was separated along the [020] direction. For each displacement, two kinds of calculations were performed: (1) RGS, without any
atomic relaxations, and (2) RGS followed by atomic relaxations with a
fixed supercell. We did not consider Poisson’s effect in this study [43].
The top two layers are fixed, while the remaining middle layers are
allowed to relax during the RGS + relaxation procedure, and a vacuum is
added at the interface to imitate the tensile tests, as illustrated in Fig. 3.
The same procedure was applied for both bulk and interface structures.
In the RGS approach, the interface structure was modeled by rigidly
separating the Fe slab along the normal c direction at the interface and
performing static calculations without any electronic and atomic relaxations, while in the RGS + relaxation method, atoms were allowed to
relax. The slabs were initially separated by gradually adding vacuum at
the interface in steps of 0.2 Å. The tensile displacement step size was
selected based on the following criteria: (i) try to sample fairly dense
near zero displacement to get a good estimate for the second derivative,
(ii) try to sample fairly dense near where we expect the inflection point
to be (iii) have at least one point at high displacements for good determination of the binding energy. Due to the computational cost of the
RGS + relaxation methodology, a non-uniform step size was selected at
the higher separation distances to find the fracture zone of the interface
structures. The fitting of calculated values with the analytical expression
provides a reasonable approach for reducing computational cost by
reducing the considered step sizes and it can also further be useful for
providing qualitative comparative analyses with other Fe-Al IMC interfaces [29,30].
Rose et al. [44] observed that the separation energy of metals has a
universal form;
and the difference in angles γ between the lattice directions;
Δγ = |γ2 − γ 1 |
(4)
These angles between two crystals are illustrated in Fig. 1. The two
structures (red and brown) are strained to match the angles to form a
coherent interface structure (γ1,2 ∕
= 90◦ ).
In general an interface structure has 9 degrees of freedom (3 degrees
related to the possible OR, 2 degrees for the possible interface plane, 2
for lateral translation along the interface plane and 2 degrees for position of where the interface cuts each phase). Ideally, all Δγ = 0, but
when these conditions are not fulfilled, the minimum difference between angles (min Δγ) can be considered.
To construct good interface models, ORs are obtained by looping
through all possible combinations of orientations up to a given crystal
lattice vector length and testing them against the criteria listed above.
We can thereby choose an interface structure with a low misfit and a
corresponding supercell structure with low enough number of atoms so
that DFT calculations are feasible.
By using the methodology presented above, we have predicted the
possible interface structures between Fe2Al5 and Fe. The DFT-relaxed
bulk structure of Fe2Al5 was used as input for finding the interface
structures. To reduce computational cost, we only considered interface
structures where the number of atoms and misfit are relatively small (see
Appendix Table 6). We considered different orientation relationships
based on their compact planes and directions and calculated the work of
separation. Since some of the interfaces are quite large, optimizing all
structures with atomic relaxations is a computationally expensive and
difficult task. However, to make reasonable comparisons and to find the
best low energy interface OR, we built all interfaces using the same
criteria, these criteria where that the interplanar spacing between two
bulk phases at the interface were set to be 2.86 Å and the transverse
layers were assured to have a thickness of 10 Å on each side of the phase.
In addition a vacuum layer of 10 Å was added to avoid periodic interaction along the normal direction. In order to motivate the selection of a
good representative interface, we calculated the work of separation for
Eb (d) = |Eeb |⋅g(a)
(5)
where Eeb is the separation energy of the equilibrium structure, d is the
displacement defined with respect to the equilibrium structure and a is
the re-scaled displacement, given by a = d/l, where l is a characteristic
3
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Fig. 2. Virtual tensile tests for the Fe2Al5(020)// Fe (-121) interface structure: (a, b) shows the Fe and Al-terminated relaxed equilibrium structures, (c) Feterminated virtual tensile test, and (d) Al-terminated virtual tensile test.
parameters Eeb , and Eb′ . This virtual tensile testing provides separation
energy versus tensile displacement. The results obtained from these
calculations can then be fitted to the UBER curve using Eqs. (5) and (7)
(below). As Rose et al. observed, the metallic bonding-energy curve can
be approximately scaled into the universal binding energy relation for
the following cases: (i) metallic or bimetallic adhesion (ii) chemisorption on a metal surface, and (iii) cohesion of bulk metals [44]. Although
UBER describes well separation energy versus displacement for unrelaxed metal surfaces, it is unable to describe the behavior of tightly
bound intermetallics [45]. To find a good fit which captures the
behavior of the separation energy versus displacement curve, we used a
generalized form which includes two polynomials [46,30]:
For the hydro-static compression/expansion, g(a) was determined to
be [46]:
′
g(a) = − (1 + a + P(a))e−a−Q(a)
(7)
where a is the rescaled displacement and P and Q are polynomials of
order two or larger with positive (leading) coefficients. This expression
for g(a) ensures that g(0) = −1, g(a → ∞) = 0 and g′ (0) = 0. The firstorder terms are excluded from P and Q since they are related to each
other as well as to the characteristic length.
To ensure that the fitting behaves well, one should only include oddorder terms in the polynomials P and Q and make sure that all coefficients are zero or positive.
By differentiating the fitted energy-displacement curve, the theoretical tensile strength of the atomic structures can be evaluated [47];
Fig. 3. Schematic illustration of the virtual tensile tests procedure.
σ th =
length which can be approximated by the curvature of the energydisplacement curve at its minimum. Eq. (6) is used as a starting point
for the fitting procedure,
√̅̅̅̅̅̅̅̅̅̅̅̅
|Eeb |
l=
(6)
E′′b (0)
∂Eb
∂d
(8)
The theoretical strength σ th at its maximum value is defined as the
Ultimate Tensile Strength (σUTS ). The value of d at σUTS is defined as the
critical length dc .
If the functional form g(a) is known, we can determine the theoretical strength and critical displacement of any material from the
4
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Fig. 4. Virtual tensile tests for the bulk Fe2Al5(020) structure: (a) relaxed equilibrium structure (b) Al-terminated virtual tensile test, and (c) Fe-terminated virtual
tensile test.
3. Results and discussion
3.2. Interface strength
3.1. Bulk strength
3.2.1. Energy-displacement curves
Energy-displacement curves are shown in Fig. 7 for Al-terminated
and Fe-terminated interface structures using the above-mentioned
fitting technique. Fig. 7 (a) and (c) show the energy-displacement
curve for RGS and Fig. 7 (b) and (d) show the same curves for the
RGS + relaxation methodology. A steep and continuous curve is obtained for the RGS methodology without any atomic relaxations, which
can be fitted well by using Eq. (5). As can be seen from Fig. 7(b) and (d),
with increasing in tensile displacement, the energy required to fracture
the interface structure decreases until the structure separates into two
surfaces at larger displacements (>3 Å). The separation length at this
point is defined as the final fracture length (df ). Even though there is no
unique way of determining df , we here define it to be at the point where
the binding energy curve reaches −0.003 eV/Å2 [30].
The minimum value of the binding energy gives -Eb(0) = Wsep for
RGS and -Eb(0) = Wad for the RGS + relaxation methodology. Table 4
lists the Wsep and Wad values for the Fe2Al5//Fe interface. As given in
Table 4, the Al-terminated interface shows higher Wsep (4.45 J/m2) as
compared to the Fe-terminated interface (3.82 J/m2). Lazar [41]
postulated the rough approximation that Wsep = 1.06 Wad by linear
fitting of DFT results of RGS and RGS + relaxation methodologies for
different compounds and materials. This fits perfectly for the Alterminated interface but less so for the Fe-terminated interface.
An optimal fit for the relaxed surfaces is shown in Fig. 7 (b) and (d).
For the relaxed-type virtual tensile tests, crack opening is initiated by
separating two blocks by introducing vacuum and subsequently allowing atoms to relax while keeping the plane area fixed. The initial crack
introduced during RGS can potentially be healed by atomic relaxations if
the separation between the two blocks is smaller than the critical length
(dc ) [47]. In Fig. 7 (b) and (d), dc is located at the border of Region I
(d < dc ). Table 4 lists the critical (dc ) and fracture lengths (df ) for the
two relevant interface structures.
Region II is defined for separations dc < d < df . In this region, the
structure is neither separated nor being able to heal by elastic relaxations, which is why it is defined as the instability region. The range
of this instability region is determined by taking the difference between
df and dc . The width of Region II is related to the brittleness/ductility of
the interface structure [41]. For the Al-terminated interface structure,
To compare the bulk and interface structures, we also calculated the
tensile properties of Fe2Al5(020) using the rigid shift (RGS) and RGS +
relaxation methodology as explained in Sec. 3.1. We studied the virtual
tensile strength of both the Al- and the Fe-terminated Fe2Al5 structures
as shown in Fig. 4.
Fig. 5 (a) and (b) show the separation energy versus tensile
displacement curve for Al and Fe-terminated fractures using the RGS and
RGS + relaxation methodologies, respectively. In the stable configuration of the Fe2Al5 phase, the Fe-Al bond distance is 2.50 Å and the Fe-Fe
bond distance is 2.96 Å. During the virtual tensile testing, this bond
distance at the cutting plane is stretched further until the bulk structure
fractures and separates into two free surfaces. Fig. 4 (b) and (c) show the
procedure for introducing a crack with Al and Fe-terminations. Table 3
lists the work of separation (Wsep ) and the work of adhesion (Wad ). The
former is defined as the work needed to separate a bulk phase without
atomic relaxations, and the latter is the energy needed to separate a bulk
interface into two relaxed surfaces [48].
The binding energy increases with tensile displacement. RGS without
atomic relaxation produces a steeper curve which was fitted using Eqs.
(5) and (7). During tensile displacement, the separation energy increases
sharply until it stabilizes at larger displacements (> ̃5 Å).
Table 3 lists the calculated values of σUTS . Fig. 6 shows the stress–strain curves for Fe2Al5 along with bulk strengths for the RGS and RGS +
relaxation methodologies. With increasing tensile strain, the tensile
stress increases until its maximum value (σ UTS ). One can note that σ UTS
calculated with the RGS + relaxation methodology is lower than that for
the RGS methodology. For comparisons, we also present the strength of
the bulk Fe (111) plane. The Al-terminated Fe2Al5 bulk phase shows
higher strength (20.09 GPa for RGS and 15.48 GPa for RGS + relaxation)
as compared to the Fe-terminated structure (17.72 GPa for RGS and
13.28 GPa for RGS + relaxation). Moreover, the bulk Fe structure shows
higher values of Wsep and σUTS , which signify the higher strength of bulk
Fe than that of the Fe2Al5 phase. A lower strength of the Fe-terminated
bulk Fe2Al5 structure indicates a weaker bonding between Fe-Fe atoms
which will be discussed in sub-Section 4.3. Besides, the long bonding
distance between Fe-Fe also contributes to the weakening of the bond.
5
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Fig. 6. Virtual tensile tests stress–strain curve for the bulk Fe2Al5(020) structure with both Al and Fe-terminations calculated with the RGS and RGS +
relaxation methods.
At longer separation distances (d > df ), the interface structures are
completely separated into two relaxed bulk surfaces. This region is
defined as Region III in light grey color (Fig. 7 (b) and (d)). In this region,
there is no interaction at the interface, and relaxation of the atomic
positions relaxes the bulk surfaces into stable configurations. For this
reason, the binding energy versus separation curve stabilizes, and no
further increase in binding energy can be seen.
3.2.2. Tensile strength
Table 4 lists σ UTS of Fe2Al5// Fe interface structures for both terminations. Since RGS + relaxation calculations were performed with
atomic relaxations, σ UTS calculated from this approach provides more
realistic values than those for the RGS calculations. Based on the RGS +
relaxations virtual tensile tests, the Al-terminated interface shows lower
strength (23.88 GPa) as compared to the Fe-terminated interface (31.48
GPa). Overall, the interface structures show higher σ UTS values than bulk
Fe2Al5. The Fe-terminated interface shows the highest strength (31.48
GPa) and Fe-terminated bulk Fe2Al5 the lowest strength (17.72 GPa).
In order to elucidate the bonding characteristics of the interfacial and
bulk atoms, total charge density isosurfaces and charge density difference plots for all surfaces were constructed as shown in Fig. 8. A high
charge density cloud (labeled as B in Fig. 8(c)) can be seen for the Feterminated interface as compared to the Al-terminated interface
(labeled as A in Fig. 8(a)). Moreover, there is a higher charge transfer for
the Fe-terminated interface, while there is a weak charge transfer zone
for the Al-terminated interface as shown in yellow color in Fig. 8(b). This
high charge density and transfer rate at B indicates stronger bonding
between interfacial Fe-Fe atoms at the Fe-terminated interface, which
explains the higher σUTS for this interface as compared to the Alterminated interface.
For the bulk Fe2Al5 structure as shown in Fig. 8(e-f), Fe-Fe bonding
(labeled as C) was found to be weaker than the Al-Fe bonding (labeled as
D). This observation is consistent with the lower σ UTS for the Feterminated bulk structure. Generally, Fe-Al atoms are found to have
higher charge density and charge transfer regions at the interfaces and in
the bulk structures. However, in the Al-terminated interface, the Al
atoms move towards the Fe atoms and develops a bond at the interface
by compromising the bonding strength at the first layer of the Fe2Al5
side, labeled as I in Fig. 8(a). This fracture plane can be a weak link of the
overall Al-terminated interface structure.
Fig. 5. Separation energy versus displacement for virtual tensile tests of the
bulk Fe2Al5(020) structure: (a) Al-terminated virtual tensile test, and (b) Feterminated virtual tensile test.
Table 3
Calculated ultimate tensile strength, Wsep and Wad of bulk Fe2Al5 and bcc Fe.
Structure
σUTS (RGS)
σUTS (RGS +
relaxation) (GPa)
Wsep (J/
m2)
Wad (J/
m2 )
Al-terminated
Fe2Al5
Fe-terminated
Fe2Al5
Fe
20.09
15.48
6.16
5.81
17.72
13.28
5.54
5.16
27.7a <
111 >
–
(GPa)
28.5b,c <
111 >
a
b
c
6.09 <
121 >
–
[49]
[50]
[51]
the length of the instability region is approximated to be 0.84 Å, while
for the Fe-terminated interface, it is 0.79 Å. The shorter range of the
instability region for the Fe-terminated interface indicates a more brittle
fracture than that of the Al-terminated interface.
3.2.3. Ideal shear strength
To calculate the ideal shear strength a series of incremental shear
6
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Fig. 7. Energy-displacement curves resulting from virtual tensile tests for the Fe2Al5(020)//Fe interface structure with both Al and Fe terminations, (a) and (c) show
the virtual tensile test results for the RGS methodology and (b) and (d) for the RGS + relaxation methodology. Red points show DFT calculation results and the blue
solid line is the fitted curve.
The shear stress is given by
Table 4
Calculated ultimate tensile strengths, Wsep and Wad values of the Fe//Fe2Al5
interface structure.
Structure
σUTS
dc
(Å)
df
(Å)
σUTS (RGS +
relaxation)
(GPa)
Wsep
(J/
m2)
Wad
(J/
m2)
29.56
1.80
2.64
23.88
4.45
3.04
24.50
1.51
2.30
31.48
3.82
3.36
(RGS)
(GPa)
Al-terminated
interface
Fe-terminated
interface
γs =
∞
∑
[An cos(kn d) + Bn sin(kn d)]
(10)
where A is the interface area. The maximum value in the resulting sheardisplacement curve corresponds to the ideal shear strength, which is
defined as the interface resistance to the shear displacement after which
it starts to deform.
Fig. 9 shows the stress-displacement curve for the shear stress as a
function of shear displacement for both Al- and Fe-terminations.
Initially, stress increases with the increase in the shear displacement
until it reaches a maximum value for both cases, which is taken as the
ideal shear strength of the interface structure. Table 5 summarizes the
ideal shear strength of the Fe2Al5//Fe interface structure for the
different cases discussed in this work. Results are quite different for both
interface terminations. The Fe-terminated interface shows low shear
strength (0.97 GPa) along < 001 > and larger shear strength along <
100 > (4.74 GPa), while the Al-terminated interface shows high shear
strength (2.51 GPa) along the < 100 > direction and a slightly lower
shear strength along < 001 > (3.97 GPa). In general the Al-terminated
structure shows higher shear strength than the Fe-terminated interface
structure. These calculations, therefore, indicate that the Fe-terminated
< 001 > interface is more prone to shear failure than the Al-terminated
interface.
Comparing shear strength with tensile strength indicates that the Alterminated interface shows higher tensile and shear strength than the
Fe-terminated interface. From Table 4 and 5, it can be seen that shear
strains were applied to the Fe2Al5//Fe supercell. We moved the Fe
surface along the < 001 > and < 001 > shear directions. For these
calculations, six layers of Fe were sheared along the defined shear directions with respect to the Fe2Al5 atoms at the interface. Atoms were
allowed to relax along the normal direction to the interface to remove
any strain along that direction. The shear energies are defined in terms
of a Fourier series;
Es (d) = E0 +
1 ∂Es
A ∂d
(9)
n=1
where Es (d) and E0 are the energy of the displaced and unsheared
structure, respectively, d is the shear displacement, and kn = 2πλ n, where
λ is the periodicity along the shear direction. Appendix Table 5 and 6
gives the Fourier series coefficient values and the value of λ for both
interface structures.
7
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Fig. 8. Calculated total charge density isosurfaces (a,c,e) drawn at 0.03 e/Å3 and charge density difference plots (b,d,f) for (a-b) Al-terminated interface, (c-d) Feterminated interface and (e-f) bulk Fe2Al5. A, B, C, and D define the cutting planes for virtual tensile testing and I indicates the weak fracture plane for the Alterminated interface.
trend has been observed experimentally and theoretically in the literature [16,52,53].
Table 5
Calculated Ideal shear strength values of the Fe2Al5 (020)//Fe(121) interface,
directions are defined with respect to Fe2Al5.
Interface
<001> (GPa)
<100> (GPa)
Fe-termination
Al-termination
0.97
3.97
4.74
2.51
4. Discussion
Before discussing the implications of these results, some limitations
are worth to be mentioned. These simulations have been performed
without considering dislocations, micro-voids, and other effects occurring at larger length scales, that will obviously influence the strength of
real joints [54]. Hence, the calculated strengths are thus generally
overestimated. Still, these calculations provide important insights about
the crack formation mechanism of the interface structure at the atomic
instability can occur earlier than normal decohesion. This is consistent
with the experimental observations of an Al-Fe welded system [52]. The
shear strength calculated in this study for loading parallel to the interface is lower than the perpendicular loading direction (σ UTS ). The same
Fig. 9. Fitted shear stress-displacement curve of the Fe2Al5//Fe interface for (a) Al- and (b) Fe-terminations during the shear strength calculations as a function of
shear displacement along the < 001 > and < 100 > shear directions.
8
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
scale. The role of crystal defects on the mechanical properties is proposed to be a subject of future studies.
In this work we have studied the effect the Fe2Al5 intermetallic phases
has on the strength of an aluminium-steel joints. It is a very difficult task to
identify the fractured layer of aluminum and steel joints by experiments.
For this reason, to predict the weak zone of the Fe2Al5// Fe interface
structure, two zones were studied: (i) bulk Fe2Al5 and (ii) Fe2Al5//Fe
interface. Based on bulk and interface calculations, the interface between
Fe2Al5// Fe showed higher strength as compared to bulk Fe2Al5 and
smaller than bulk Fe [49]. Virtual tensile testing, therefore, indicates that
bulk Fe2Al5 is more prone to induce fracture than the interface and bulk Fe
side. Mechanical strength inferred from the virtual tensile calculations
indicates that fracture is most likely to be initiated from the Fe-terminated
side of the bulk Fe2Al5 due to weak bonding between Fe-Fe atoms. Shear
strength is seen to be lower than the tensile strength, which is also
consistent with the experimental observation of Fe2Al5 [55].
We have performed more calculations with the strained interface
structures to study the effects of elastic strain on work of separation. All
calculations were performed considering the optimized equilibrium
interface structures. As discussed in Section 2, both bulk slabs were
strained to match at the interface. In order to determine the elastic
contribution to the work of separation, calculations were performed to
determine the work of separation for the strained interface structures
and the results are compared with those of the equilibrium interface
structure. In the case of the Fe-terminated interface, the work of separation was reduced from 5.54 J/m2 to 3.58 J/m2. However, for the Alterminated interface the work of separation was drastically reduced:
from 6.16 J/m2 to 2.91 J/m2. This indicates that the failure mechanism
for the interface structure might be more complex than what can be
described using this method and should be investigated in more detail in
further work.
Moreover, the theoretical tensile strengths of an interface structure
depends on the number of crystallographic layers of the model. A recent
study [56] has indicated the decrease in fracture stress with increasing
supercell size with localized strain models. Effect of supercell size is
beyond the scope of this study. Moreover, as the main objective is to
make a comparative analysis of Fe-Al IMCs interfaces, we adopted a
consistent methodology and approach for all interface structures to
make appropriate qualitative comparisons. However, further investigations are needed to find the influence of the number of layers on
the strength values for the RGS + relaxation methodology. Since this
study is limited to the DFT methodology, it presents an extra challenge
of computational cost. This is why the values obtained from the RGS +
relaxation method only provides qualitative comparative strength of the
Fe//IMC interface as compared to the Al/IMC [29], IMC//Fe and pure
IMC//IMC interfaces [30]. Still we believe this provides useful insights
into the role of Fe-Al IMCs on the joining of aluminium and steel.
In general, these results have a particular significance for the welding
of aluminum and steel joints for different welding methodologies, where
the presence of an Fe2Al5 intermetallic layer has been reported along the
steel side. However, the defects at the IMC layers also play a significant
role in deteriorating the joint strength and have to be included in the
calculations to give more reliable predictions for real systems in the future.
shear strength of the Fe2Al5//Fe interface. The interface structure with
the lowest lattice misfit and number of atoms was selected for the DFT
calculations of this work. Virtual tensile tests were performed with the
rigid grain shift (RGS) methodology without atomic relaxations and
RGS + relaxation methodology with atomic relaxations. Polynomial
terms were introduced into the UBER to find a reasonable fit for the
tensile stresses. Based on RGS calculations, the Al-terminated interface
showed higher strength than bulk Fe2Al5 and the Fe-terminated interface structure. During the relaxation of atomic positions in the RGS +
relaxation methodology, the tensile strength decreased for all structures
except for the Fe-terminated interface. Moreover, the charge density
maps indicated a weaker bonding between Fe-Fe atoms in the bulk
Fe2Al5 structure, which contributed to a lower tensile strength. We also
analyzed the shear strength for the interface along < 001 > and <
100 > directions. We found that < 001 > has lower shear strength for
the Fe-terminated interface while it showed higher strength for the Alterminated interface.
Overall the Fe bulk side was found to be the strongest zone of the
Fe2Al5//Fe interface structure followed by the interface and bulk Fe2Al5.
Based on these calculations, it can be anticipated that during a mechanical failure, fracture is most likely to be initiated at the bulk Fe2Al5
side. This study can potentially be the starting point for further investigations of the effects of crystal defects and temperature on the joint
strength of aluminum-steel joints.
CRediT authorship contribution statement
Muhammad Zeeshan Khalid: Conceptualization, Software, Data
curation, Visualization, Investigation, Validation, Formal analysis,
Writing - original draft, Writing - review & editing. Jesper Friis:
Methodology, Software, Data curation, Formal analysis, Supervision,
Writing - review & editing, Conceptualization, Validation. Per Harald
Ninive: Methodology, Software, Supervision, Resources, Writing - review & editing. Knut Marthinsen: Supervision, Project administration,
Writing - review & editing. Inga Gudem Ringdalen: Methodology,
Writing - review & editing. Are Strandlie: Conceptualization, Methodology, Supervision, Project administration, Resources, Writing - review
& editing, Funding acquisition.
Declaration of Competing Interest
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.
Acknowledgements
The work reported in this paper was based on activities within the
centre for research-based innovation SFI Manufacturing in Norway and
is partially funded by the Research Council of Norway under contract
number 237900. UNINETT Sigma2 AS (The Norwegian Metacenter for
High Performance Computing) provided computational resources
through Project NN9466K and NN9158K.
5. Summary and conclusions
To summarize, we have performed DFT calculations of tensile and
Appendix A
Appendix Table 6 lists the predicted ORs between Fe and Fe2Al5 by the face-to-face matching technique. In Table 6, m1 , m2 and m3 are the
components of a linear combination of vector u1 of crystal 1, similarly n1 , n2 and n3 are defined for crystal 2, and is given as;
u1 = m1 a1 + m2 b1 + m3 c1
u2 = n1 a2 + n2 b2 + n3 c2
(11)
9
M.Z. Khalid et al.
Computational Materials Science 192 (2021) 110319
Table 6
Some of the predicted ORs between Fe2Al5 and Fe atoms. m1 , m2 and m3 are the direction vectors for Fe2Al5 phase and n1 , n2 and n3 for Fe atoms. length and strain (%)
are the length of supercell and misfit percentage (as defined in Eq. (2) and (3)) of the interface structures respectively.
#
d
m1
m2
m3
1
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
u
v
w
h
0
1
0
0
0
2
−1
0
0
1
−2
−2
0
1
−2
−2
0
0.5
−2.5
−3
0
0.5
−2.5
−3
0
1
−2.5
−2
0
0.5
−1.5
−1
0
1
−1
0
0
1.5
0
1
0
1
0.5
3
1
−0.5
0
0
0
0.5
−2
−8
0
0
−2
−2
0
0
−2
−1
0
2
0
1
0
2
0
1
0
1.5
0.5
1
0
1.5
0.5
1
0
2
0.5
1
0
0.5
1.5
1
0
0
2
1
0
−0.5
2
3
1
−1
1.5
2
0
1.5
−2
−4
1
−1.5
0
−2
−1
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
−1
0
0
1
−1
0
0
2
1
−2
−1
0
2
3
3
−2
1
0
−1
2
3
4
5
6
7
8
9
10
11
12
13
length (Å)
∠γ
4.10
7.40
12.88
90◦
90◦
90◦
4.10
14.80
14.86
119.9◦
90◦
90◦
4.10
14.86
14.80
119.9
90◦
90◦
4.10
14.86
14.80
119.9◦
90◦
90◦
4.10
10.35
18.77
101.1
90◦
90◦
4.10
10.35
18.77
101.1◦
90◦
90◦
4.10
14.86
18.77
110◦
90◦
90◦
4.10
4.90
14.71
97.9◦
90◦
90◦
4.10
8.46
14.86
115.8◦
90◦
119.0◦
4.10
12.26
12.88
105.2
90◦
109.5◦
10.42
10.63
13.20
119.2◦
92◦
94.1◦
7.40
13.19
17.81
95.8◦
90◦
106.3◦
10.42
11.13
19.38
109.4◦
90◦
104.3◦
◦
◦
◦
n1
n2
n3
1
1.5
−2
−1
1
3.5
−3
−3
1
3.5
−3
−3
1
3
−3
−1
1
0.5
−3
−7
1
2.5
−1
−1
1
1
−4.5
−5
1
1
−2
−1
1
−1.5
3
5
1
−1
−3.5
−1
3
1
−2.5
−6
2.5
−0.5
−1
−5
3
−1.5
−2
−14
0
1.5
4
2
0
1.5
4
14
0
1.5
4
14
0
3
2
2
0
3.5
0
2
0
0.5
4
10
0
5.0
−0.5
2
0
1
3
2
0
2.5
−1
2
0
4
−1.5
0
0
−3
−0.5
−8
0.5
0.5
4
22
2
0
3
21
1
−1.5
2
1
1
−3.5
3
3
1
−3.5
3
3
1
−3
3
1
1
−0.5
3
7
1
−2.5
1
1
1
−1
4.5
5
1
−1
2
1
−1
0.5
3
5
−1
−1
3.5
1
−2
2
−3.5
−9
0.5
−4.5
1
3
0
−3.5
3
9
length (Å)
∠γ
4.06
7.46
14.06
90◦
90◦
90◦
4.06
14.84
16.73
119.8◦
90◦
90◦
4.06
14.84
16.73
119.8
90◦
90◦
4.06
14.91
13.46
119.5◦
90◦
90◦
4.06
10.25
12.18
101.4
90◦
90◦
4.06
10.25
12.18
101.4◦
90◦
90◦
4.06
14.91
18.32
110.3◦
90◦
90◦
4.06
4.97
11.83
98.0◦
90◦
90◦
4.06
8.49
12.51
115.2◦
90◦
118.6◦
4.06
12.18
14.84
105.9
90◦
109.5◦
10.35
10.74
12.43
119.6◦
91.8◦
94.3◦
7.46
137
12.18
95.9◦
90◦
105.9◦
10.35
112
13.46
109.5◦
90◦
104.6◦
◦
◦
◦
strain (%)
# atoms
0.89
0.79
64
0.56
0.89
0.39
130
0.52
0.89
0.90
130
0.56
0.89
0.38
116
0.63
0.89
0.96
98
0.71
0.89
0.96
98
0.71
0.89
0.38
165
0.60
0.89
1.35
41
0.80
0.89
0.40
56
0.87
0.89
0.67
98
0.84
0.68
11
192
0.79
0.92
215
0.69
0.71
0.94
230
0.65
Table 7
The fitting Fourier series coefficient values for the shear strength calculation of Al-terminated Fe2Al5//Fe interface.
Polynomial terms
A0
A1
A2
A3
B1
B2
B3
λ
< 100 >
−1107.76
110.59
1864.90
−646.56
271.21
11115
−803.98
77.85
< 001 >
110.59
−1107.76
1864.90
271.21
−646.56
11115
77.85
−803.98
Table 8
The fitting Fourier series coefficient values for the shear strength calculation of Fe-terminated Fe2Al5//Fe interface.
Polynomial terms
A0
A1
A2
A3
A4
A5
B1
B2
B3
B4
B5
λ
< 100 >
1660.13
−988.84
−3113.83
−1627.44
−357.08
−299.39
106.08
−0.469
−0.0014
0.052
3842.35
−1144.45
-
1794.44
−4305.2
0.058
1624.54
< 001 >
0.415
-
10
0.0699
0.0153
-
-
8.23
Computational Materials Science 192 (2021) 110319
M.Z. Khalid et al.
Normally m1 , m2 , m3 and n1 , n2 , n3 are integers, but due to sub-lattice translations in the conventional cell, fractions are also possible.
Tables 7 and 8 lists the Fourier series coefficient values for the shear strength calculations for both interface structures.
References
[29] M.Z. Khalid, J. Friis, P.H. Ninive, K. Marthinsen, A. Strandlie, Ab-initio study of
atomic structure and mechanical behaviour of Al/Fe intermetallic interfaces,
Comput. Mater. Sci. 174 (2020), 109481.
[30] M.Z. Khalid, J. Friis, P.H. Ninive, K. Marthinsen, A. Strandlie, First-principles study
of tensile and shear strength of Fe-Al and α-AlFeSi intermetallic compound
interfaces, Comput. Mater. Sci. 187 (2020), 110058.
[31] G. Kresse, J. Furthmüller, Vienna ab-initio simulation package (VASP), Vienna:
Vienna University.
[32] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made
simple, Phys. Rev. Lett. 77 (18) (1996) 3865.
[33] P.E. Blöchl, Projector augmented-wave method, Phys. Rev. B 50 (24) (1994)
17953.
[34] H.J. Monkhorst, J.D. Pack, Special points for brillouin-zone integrations, Phys.
Rev. B 13 (12) (1976) 5188.
[35] U. Burkhardt, Y. Grin, M. Ellner, K. Peters, Structure refinement of the
iron–aluminium phase with the approximate composition Fe2Al5, Acta
Crystallographica Section B 50 (3) (1994) 313–316.
[36] Fe2Al5 (FeAl2.8) crystal structure: Datasheet from ”pauling file multinaries edition
– 2012” in springermaterials (https://materials.springer.com/isp/
crystallographic/docs/sd_1201135), copyright 2016 Springer-Verlag, Berlin
Heidelberg & Material Phases Data System (MPDS), Switzerland & National
Institute for Materials Science (NIMS), Japan. https://materials.springer.com/isp/
crystallographic/docs/sd_1201135.
[37] M.Z. Khalid, J. Friis, P.H. Ninive, K. Marthinsen, A. Strandlie, DFT calculations
based insight into bonding character and strength of Fe2Al5 and Fe4Al13
intermetallics at Al-Fe joints, Procedia Manuf. 15C (2018) 1407–1415.
[38] Y. Liu, X. Chong, Y. Jiang, R. Zhou, J. Feng, Mechanical properties and electronic
structures of Fe-Al intermetallic, Physica B 506 (2017) 1–11.
[39] C.-H. Zhang, S. Huang, J. Shen, N.-X. Chen, Structural and mechanical properties of
Fe-Al compounds: An atomistic study by EAM simulation, Intermetallics 52 (2014)
86–91.
[40] H. Okamoto, T. Massalski, et al., Binary alloy phase diagrams, ASM International,
Materials Park, OH, USA.
[41] P. Lazar, R. Podloucky, Cleavage fracture of a crystal: Density functional theory
calculations based on a model which includes structural relaxations, Phys. Rev. B
78 (10) (2008), 104114.
[42] R. Janisch, N. Ahmed, A. Hartmaier, Ab initio tensile tests of Al bulk crystals and
grain boundaries: Universality of mechanical behavior, Phys. Rev. B 81 (18)
(2010), 184108.
[43] M. Yamaguchi, First-principles study on the grain boundary embrittlement of
metals by solute segregation: Part i. iron (Fe)-solute (B, C, P, and S) systems,
Metallurgical and Materials Transactions A 42 (2) (2011) 319–329.
[44] J.H. Rose, J.R. Smith, J. Ferrante, Universal features of bonding in metals, Phys.
Rev. B 28 (4) (1983) 1835.
[45] E.A. Jarvis, R.L. Hayes, E.A. Carter, Effects of oxidation on the nanoscale
mechanisms of crack formation in aluminum, ChemPhysChem 2 (1) (2001) 55–59.
[46] I. J. Jensen, J. Friis, C. D. Marioara, I. G. Ringdalen, The role of grain boundary
precipitates during intergranular fracture in 6XXX series Aluminium alloys, In
preparation.
[47] D. Zhao, O.M. Løvvik, K. Marthinsen, Y. Li, Segregation of Mg, Cu and their effects
on the strength of Al σ5 (210)[001] symmetrical tilt grain boundary, Acta Mater.
145 (2018) 235–246.
[48] R. Yang, S. Tanaka, M. Kohyama*, First-principles study on the tensile strength and
fracture of the Al-terminated stoichiometric α-Al2O3 (0001)/Cu (111) interface,
Philosophical Magazine 85 (25) (2005) 2961–2976.
[49] M. Černỳ, J. Pokluda, Ideal tensile strength of cubic crystals under superimposed
transverse biaxial stresses from first principles, Phys. Rev. B 82 (17) (2010),
174106.
[50] S. Ogata, J. Li, N. Hirosaki, Y. Shibutani, S. Yip, Ideal shear strain of metals and
ceramics, Phys. Rev. B 70 (10) (2004), 104104.
[51] M. Černỳ, P. Řehák, Y. Umeno, J. Pokluda, Stability and strength of covalent
crystals under uniaxial and triaxial loading from first principles, J. Phys.: Condens.
Matter 25 (3) (2012), 035401.
[52] P. Peyre, G. Sierra, F. Deschaux-Beaume, D. Stuart, G. Fras, Generation of
aluminium-steel joints with laser-induced reactive wetting, Mater. Sci. Eng.: A 444
(1–2) (2007) 327–338.
[53] Y.-J. Wang, C.-Y. Wang, Influence of the alloying element re on the ideal tensile
and shear strength of γ-Ni3Al, Scripta Mater. 61 (2) (2009) 197–200.
[54] J. Hirth, R. Pond, R. Hoagland, X.-Y. Liu, J. Wang, Interface defects, reference
spaces and the Frank-Bilby equation, Prog. Mater Sci. 58 (5) (2013) 749–823.
[55] W. Qian, X.-S. Leng, T.-H. Yang, J.-C. Yan, Effects of Fe—Al intermetallic
compounds on interfacial bonding of clad materials, Trans. Nonferrous Metals Soc.
China 24 (1) (2014) 279–284.
[56] M. Černỳ, J. Pokluda, On the effect of supercell size and strain localization in
computational tensile tests, Modell. Simul. Mater. Sci. Eng. 28 (6) (2020), 065011.
[1] A. Hotař, M. Palm, P. Kratochvíl, V. Vodičková, S. Daniš, High-temperature
oxidation behaviour of Zr alloyed Fe3Al-type iron aluminide, Corros. Sci. 63 (2012)
71–81.
[2] B.L. Silva, A. Garcia, J.E. Spinelli, The effects of microstructure and intermetallic
phases of directionally solidified Al-Fe alloys on microhardness, Mater. Lett. 89
(2012) 291–295.
[3] T. Sakiyama, G. Murayama, Y. Naito, K. Saita, Y. M. H. Oikawa, T. Nose, Dissimilar
metal joining technologies for steel sheet and aluminum alloy sheet in auto body,
Nippon Steel & Sumitomo Metal Corp., Tokyo, Nippon Steel Technical Report 103
(2013) 91–98.
[4] T. Ogura, Y. Saito, T. Nishida, H. Nishida, T. Yoshida, N. Omichi, M. Fujimoto,
A. Hirose, Partitioning evaluation of mechanical properties and the interfacial
microstructure in a friction stir welded aluminum alloy/stainless steel lap joint,
Scripta Mater. 66 (8) (2012) 531–534.
[5] H. Das, S. Basak, G. Das, T.K. Pal, Influence of energy induced from processing
parameters on the mechanical properties of friction stir welded lap joint of
aluminum to coated steel sheet, Int. J. Adv. Manuf. Technol. (2013) 1–9.
[6] S. Imaizumi, Welding of aluminium to dissimilar metals, Welding Int. 10 (8) (1996)
593–604.
[7] S. Yang, J. Zhang, J. Lian, Y. Lei, Welding of aluminum alloy to zinc coated steel by
cold metal transfer, Mater. Des. 49 (2013) 602–612.
[8] S. Niu, S. Chen, H. Dong, D. Zhao, X. Zhang, X. Guo, G. Wang, Microstructure and
properties of lap joint between aluminum alloy and galvanized steel by CMT,
J. Mater. Eng. Perform. 25 (5) (2016) 1839–1847.
[9] Y. Shi, J. Li, G. Zhang, J. Huang, Y. Gu, Corrosion behavior of aluminum-steel
weld-brazing joint, J. Mater. Eng. Perform. 25 (5) (2016) 1916–1923.
[10] G. Zhang, M. Chen, Y. Shi, J. Huang, F. Yang, Analysis and modeling of the growth
of intermetallic compounds in aluminum-steel joints, RSC Adv. 7 (60) (2017)
37797–37805.
[11] S. Lan, X. Liu, J. Ni, Microstructural evolution during friction stir welding of
dissimilar aluminum alloy to advanced high-strength steel, Int. J. Adv. Manuf.
Technol. 82 (9–12) (2016) 2183–2193.
[12] G. Sorger, H. Wang, P. Vilaça, T.G. Santos, FSW of aluminum AA5754 to steel DX54
with innovative overlap joint, Welding in the World 61 (2) (2017) 257–268.
[13] K.-K. Wang, L. Chang, D. Gan, H.-P. Wang, Heteroepitaxial growth of Fe2Al5
inhibition layer in hot-dip galvanizing of an interstitial-free steel, Thin Solid Films
518 (8) (2010) 1935–1942.
[14] A. Bouayad, C. Gerometta, A. Belkebir, A. Ambari, Kinetic interactions between
solid iron and molten aluminium, Mater. Sci. Eng.: A 363 (1) (2003) 53–61.
[15] D. Naoi, M. Kajihara, Growth behavior of Fe2Al5 during reactive diffusion between
Fe and Al at solid-state temperatures, Mater. Sci. Eng.: A 459 (1) (2007) 375–382.
[16] Y. Chen, A. Gholinia, P. Prangnell, Interface structure and bonding in abrasion
circle friction stir spot welding: a novel approach for rapid welding aluminium
alloy to steel automotive sheet, Mater. Chem. Phys. 134 (1) (2012) 459–463.
[17] W.-J. Cheng, C.-J. Wang, Study of microstructure and phase evolution of hotdipped aluminide mild steel during high-temperature diffusion using electron
backscatter diffraction, Appl. Surf. Sci. 257 (10) (2011) 4663–4668.
[18] L. Agudo, D. Eyidi, C.H. Schmaranzer, E. Arenholz, N. Jank, J. Bruckner, A.
R. Pyzalla, Intermetallic FexAly-phases in a Steel/Al-alloy fusion weld, J. Mater. Sci.
42 (12) (2007) 4205–4214.
[19] L. Jelver, P.M. Larsen, D. Stradi, K. Stokbro, K.W. Jacobsen, Determination of lowstrain interfaces via geometric matching, Phys. Rev. B 96 (8) (2017), 085306.
[20] K. Mathew, A.K. Singh, J.J. Gabriel, K. Choudhary, S.B. Sinnott, A.V. Davydov,
F. Tavazza, R.G. Hennig, Mpinterfaces: A materials project based python tool for
high-throughput computational screening of interfacial systems, Comput. Mater.
Sci. 122 (2016) 183–190.
[21] D. Stradi, L. Jelver, S. Smidstrup, K. Stokbro, Method for determining optimal
supercell representation of interfaces, J. Phys.: Condens. Matter 29 (18) (2017),
185901.
[22] A. Zur, T. McGill, Lattice match: An application to heteroepitaxy, J. Appl. Phys. 55
(2) (1984) 378–386.
[23] N.T. Taylor, F.H. Davies, I.E.M. Rudkin, C.J. Price, T.H. Chan, S.P. Hepplestone,
Artemis: Ab initio restructuring tool enabling the modelling of interface structures,
Comput. Phys. Commun. 257 (2020), 107515.
[24] F. Therrien, P. Graf, V. Stevanović, Matching crystal structures atom-to-atom,
J. Chem. Phys. 152 (7) (2020), 074106.
[25] W. Bollmann, Crystal defects and crystalline interfaces, Springer Science &
Business Media, 2012.
[26] W. Bollmann, O-lattice calculation of an FCC–BCC interface, Physica Status Solidi
(a) 21 (2) (1974) 543–550.
[27] M.-X. Zhang, P.M. Kelly, M.A. Easton, J.A. Taylor, Crystallographic study of grain
refinement in aluminum alloys using the edge-to-edge matching model, Acta
Mater. 53 (5) (2005) 1427–1438.
[28] Y. Ikuhara, P. Pirouz, Orientation relationship in large mismatched bicrystals and
coincidence of reciprocal lattice points (CRLP), in: Materials Science Forum, Vol.
207, Trans Tech Publications, 1996, pp. 121–124.
11