Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Journal of Materials Science & Technology 92 (2021) 148–158

Contents lists available at ScienceDirect

Journal of Materials Science & Technology


journal homepage: www.elsevier.com/locate/jmst

Research Article

Density functional theory study on the role of ternary alloying


elements in TiFe-based hydrogen storage alloys
Won-Seok Ko a,∗, Ki Beom Park b, Hyung-Ki Park b,∗
a
School of Materials Science and Engineering, University of Ulsan, Ulsan 44610, Republic of Korea
b
Functional Materials and Components R&D Group, Korea Institute of Industrial Technology, Gangneung 25440, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: The role of additional ternary alloying elements on the performance of stationary TiFe-based hydrogen
Received 16 January 2021 storage alloys was investigated based on first-principles density functional theory calculations. As a ba-
Revised 24 March 2021
sic step for examinations, the site preference of each alloying element in the stoichiometric and non-
Accepted 24 March 2021
stoichiometric B2 TiFe compounds was clarified considering possible anti-site defects. Based on the re-
Available online 8 May 2021
vealed site preference, the effect of various possible ternary elements on the hydrogen storage was exam-
Keywords: ined by focusing on the formation enthalpies of TiFeH and TiFeH2 hydrides, which were closely related to
Hydrogen storage alloy the change in the location of plateaus in the pressure−composition−temperature curve. Several physical
Titanium-iron properties such as the volume expansion due to hydride formation were also examined to provide ad-
Density functional theory calculation ditional criteria for selecting optimum alloying conditions in future alloying design processes. Candidate
Metallic hydride alloying elements that maximize the grain boundary embrittlement due to the solute segregation were
proposed for the enhanced initial activation of TiFe-based hydrogen storage alloys.
© 2021 Published by Elsevier Ltd on behalf of Chinese Society for Metals.

1. Introduction phase), are formed [1] according to the following reactions:

2.13FeTiH0.1 (α ) + 1.001H2 → 2.13FeTiH1.04 (β ) (1)


Hydrogen is considered one of the most promising clean, re-
newable, and sustainable energy sources as it is not accompanying
2.20FeTiH1.04 (β ) + 0.932H2 → 2.13FeTiH1.95 (γ ) (2)
any polluting and greenhouse gasses, unlike fossil fuels. An impor-
tant issue for revitalizing the hydrogen economy is the successful This alloy system is especially well suited for stationary applica-
control of hydrogen energy associated with its storage and trans- tions due to its excellent reversible hydrogen storage capacity per
portation. Among available methods, hydrogen storage and trans- unit volume; therefore, its properties have been actively investi-
portation by metallic hydrides based on intermetallic compounds gated in both experimental and theoretical works [4-10].
have received great interest due to advantages such as safety. Sev- Despite the many advantages, a major drawback of this alloy
eral requirements should be met when considering material for system is its poor activation performance, which must be over-
hydrogen storage, but the most important requirement is whether come for practical applications [11, 12]. During the manufactur-
it enables fast and reversible hydrogenation and dehydrogenation ing process in air, TiFe-based alloys usually suffer from poisoning
processes near ambient conditions with sufficient economic feasi- problems by gaseous elements such as oxygen. Because this be-
bility. Among several materials suggested for hydrogen storage, in- havior significantly degrades the hydrogen sorption property, an
termetallic TiFe compounds and their alloys are one of the most additional activation treatment is required for a long time at a
promising candidates due to their low costs and desirable hydro- high temperature under a hydrogen atmosphere. Several attempts
gen absorption and desorption properties near ambient tempera- have been made to overcome this difficulty. For example, mechan-
ture and under mild pressure conditions [1-3]. If hydrogen inter- ical treatments such as severe deformation [13] and ball milling
acts with the intermetallic TiFe compound, hydrides at two dif- [14] were performed to improve the hydrogenation properties of
ferent compositions, e.g., TiFeH∼1.0 (β phase) and TiFeH∼2.0 (γ TiFe alloys [15–17]. However, the quantity of materials that can be
treated by those special processes is greatly limited, and the price
of final products is increased by additional processing. Alternately,

Corresponding author. there have been efforts to enhance the initial hydrogen sorption
E-mail addresses: wonsko@ulsan.ac.kr (W.-S. Ko), mse03@kitech.re.kr (H.-K. property by including additional alloying elements other than Ti
Park). and Fe. For example, the substitution of the constituent elements

https://doi.org/10.1016/j.jmst.2021.03.042
1005-0302/© 2021 Published by Elsevier Ltd on behalf of Chinese Society for Metals.
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 1
Dimensions of supercells and corresponding k-point meshes used in the present DFT calculations.

Structure Purpose Number of atoms Dimension of a perfect stoichiometric compound (Å) k-points

TiFe (B2) Solution energy, Site preference 128 11.796 × 11.796 × 11.796 4 × 4 × 4
TiFe (B2) GB segregation, GB embrittlement 50 7.224 × 20.46∗ × 4.171 4 × 1 × 6
TiFe (B2) Enthalpy of hydride formation 8 4.171 × 5.898 × 4.171 9 × 7 × 9
TiFeH Enthalpy of hydride formation 12 4.263 × 5.797 × 4.548 9 × 7 × 9
TiFeH2 Enthalpy of hydride formation 16 6.109 × 7.024 × 2.768 6 × 5 × 13

The dimension of the supercell including the vacuum region is 33.46 Å.

(Ti and Fe) by ternary alloying elements (Al, Co, Cu, Cr, Fe, Mn, Mo,
Ni, Pd, V, Zr [18–27]) and mish-metals [28, 29] has been investi-
gated in hopes of improving the hydrogen sorption and diffusion
properties of surface oxides.
However, the selection of alloying elements in previous works
was quite arbitrary and was not the result of a systematic analysis.
In fact, the addition of alloying elements can significantly change
other important properties of TiFe-based hydrogen storage alloys,
such as hydrogen sorption and desorption pressures of TiFeH and
TiFeH2 hydrides. Besides, there may be other alloying conditions
that are much more effective than previously explored conditions
for the overall performance of hydrogen storage. Therefore, more
in-depth analyses on the effects of various alloying elements in
TiFe-based alloys are required to optimize alloying conditions for
the best performance of hydrogen storage. On the theoretical side,
the first-principles density functional theory (DFT) calculation is
effective in exploring the roles of various additional elements that
are not readily accessible by usual experimental studies. For in-
stance, the effect of ternary elements on the enthalpy of hydride
formation, which is a key quantity for the control of pressure and
temperature conditions of the hydrogen storage alloys, is readily
accessible by DFT calculations. In previous studies [30], DFT cal-
culations were successfully used to examine the enthalpy of for-
mation for TiFeH and TiFeH2 hydrides based on the total energy
Fig. 1. Atomic structure of a B2 TiFe-based bi-crystal cell associated with the
calculation of atomic configurations.  3(111)[11̄0] STGB used in the present DFT calculations for GB-related properties
The present study aims to provide crucial information to guide (the solute-GB binding energy and the GB decohesion). Atomic sites for the substi-
the design of stationary TiFe-based hydrogen storage alloys by tution of Ti and Fe atoms are denoted by numbers (Ti0, Ti1, Ti2, Fe0, Fe1).
considering various ternary alloying elements. Extensive DFT cal-
culations were performed to provide a large database that can
be utilized for future investigations. We examined the formation
riety of possible ternary elements in the TiFe alloy (Al, As, Ba, Be,
enthalpies of TiFeH and TiFeH2 hydrides to evaluate the effect
Ca, Cd, Ce, Co, Cr, Cs, Cu, Ga, Ge, Hf, K, La, Li, Mg, Mn, Mo, Na,
of ternary alloying elements on the location of plateaus in the
Nb, Ni, P, Pd, Rb, Rh, Ru, S, Sc, Se, Si, Sr, Ta, Tc, V, W, Y, Zn and
pressure−composition−temperature (PCT) curve. In addition, sev-
Zr) were considered in the present calculations using the recom-
eral physical features such as the site preference, the solution en-
mended pseudo potentials in the VASP library. Only possible sub-
ergy, the segregation tendency to grain boundary (GB), and the GB
stitutional alloying elements in the TiFe alloy were considered be-
decohesion tendency of alloying elements were examined to pro-
cause interstitial alloying elements can cause an unintended degra-
vide further selection criteria for the future design process of TiFe-
dation of the hydrogen diffusion and the hydrogen storage proper-
based hydrogen storage alloys. Especially, alloying elements that
ties by affecting the occupation of interstitial hydrogen. Magnetism
maximize the grain boundary embrittlement were suggested for
was included by considering spin-polarized calculations except for
the enhanced activation of hydrogen storage alloys.
atomic configurations with the GB.
To investigate the GB segregation and decohesion tendencies,
2. Methodology a representative GB configuration was selected for the TiFe stoi-
chiometric compound. The selected GB is  3(111)[11̄0] symmet-
The present DFT calculations were performed using the VASP rical tilt GB (STGB), where the plane normal to GB, the axis for
code [31–33] and the projector-augmented wave (PAW) method rotation of grains, and the misorientation angle are (111), [11̄0],
[34] within the Perdew-Burke-Ernzerhof generalized gradient ap- and 109.5°, respectively. This STGB was selected because it can be
proximation (GGA) [35] for the exchange-correlation functional. made small enough to be used in the DFT calculation, and it ex-
Cutoff energy of 400 eV for the plane-wave basis set and the hibits similar GB energy and segregation trends to those of usual
Methfessel-Paxton smearing method with a width of 0.1 eV were high-angle GBs [36, 37]. In the present calculation, a bi-crystal con-
used. Brillouin-zone integrations were performed for suitably large figuration consisting of 50 atoms with the  3(111)[11̄0] STGB was
sets of  -centered k-point meshes of each supercell as summarized used as shown in Fig. 1.
in Table 1. If not designated, all calculations were performed con- Initially, a bi-crystal cell consisting of only Ti and Fe atoms was
sidering full relaxation of atomic positions and cell shapes. Dur- obtained by performing a possible rigid-body translation of each
ing the ionic relaxation, criteria for the convergence of energy and grain along with lateral directions of the GB. The most stable cell
forces were set to 10−6 eV and 10−2 eV/Å, respectively. A wide va- with the minimum energy was then selected and used for further

149
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 2
Lattice constants (a, b, c), the formation energy (Ef ) of the B2 compound, the formation energies (Ef ) of hydrides per one mol of H2 (Eqs. (7)–(9)),
and the unit cell volume obtained by the present DFT calculations as compared with reported experimental and DFT results. Results for the B2 TiFe
compound and hydrides (TiFeH and TiFeH2 ) are presented.

Compound Space group Site Atomic position Property Experiment DFT DFT(Present)
c
TiFe (B2) Pm3̄m Fe (1a) (0, 0, 0) Lattice constants, a (Å) 2.972 2.945 f , 2.9437 g 2.949
Ti (1b) (0.5, 0.5, 0.5) Unit cell volume (Å3 ) 26.25 c
25.54 f , 25.51 g 25.65
Ef ∗ (kJ/gram-atom) −40.60
TiFeH P2221 Fe (2c) (0, 0.206, 0.25) a Lattice constants, a (Å) 2.956 d 2.889 f , 2.909 g
2.898
Ti (2d) (0.5, 0.25, 0.75 a Lattice constants, b (Å) 4.543 d 4.529 f , 4.507 g
4.548
H (2a) (0, 0, 0) a Lattice constants, c (Å) 4.388 d 4.264 f , 4.284 g
4.263
Unit cell volume (Å3 ) 58.93 d 55.79 f , 56.17 g
56.19
Ef (kJ/mol H2 ) −28.1 e −25.28 f −23.08
TiFeH2 Cmmm Fe (4i) (0, 0.288, 0) b Lattice constants, a (Å) 7.029 b 6.957 f , 6.962 g
7.024
Ti (4 h) (0.2232, 0, 0.5) b Lattice constants, b (Å) 6.233 b 6.071 f , 6.121 g
6.109
H1 (4e) (0.25, 0.25, 0) b Lattice constants, c (Å) 2.835 b 2.769 f , 2.795 g
2.768
H2 (2c) (0.5, 0, 0.5) b Unit cell volume (Å3 ) 124.2 b 117.0 f , 119.1 118.8
H3 (2a) (0, 0, 0) b Ef (kJ/mol H2 ) −33.7 – −31.0 e
−26.86 f −26.09

Reference states are hcp Ti and bcc Fe.
a
Previous experimental value from Ref. [54].
b
Previous experimental value from Ref. [48].
c
Previous experimental value from Ref. [42].
d
Previous experimental value from Ref. [47].
e
Previous experimental value from Ref. [1].
f
Previous DFT calculation result from Ref. [30].
g
Previous DFT calculation result from Ref. [55].

investigations. A vacuum region of at least 13 Å was presented in cube corner and the center. The lattice constant of the stoichio-
the direction normal to the GB normal direction (y-direction in metric B2 TiFe compound obtained by the present DFT calculation
Fig. 1) with periodic boundary conditions for all directions, thereby at 0 K is 2.949 Å, which is in sufficient agreement with the re-
introducing uppermost and lowermost free surface regions into the ported experimental result of 2.972 Å at room temperature [42]. At
bi-crystal cell. The dimensions of the cell along directions parallel the initial stage of hydrogen storage before the formation of metal-
to the GB were fixed based on a calculated lattice constant of the lic hydrides, hydrogen atoms occupy the interstitial site of the B2
bulk TiFe stoichiometric compound. The atomic positions of two TiFe compound. Based on the present DFT calculation, it was ex-
outermost layers for both free surface regions were fixed during pected that the preferred interstitial site of hydrogen is the octahe-
the simulation. These constraints were indispensable to examine dral site rather than the tetrahedral site, which is consistent with
the GB-related properties in a bulk-like environment using a lim- previous DFT results [43]. That is, a configuration with a hydrogen
ited size of DFT supercells. The bi-crystal cell with an optimized atom located at the octahedral site is more stable than that with a
GB structure was finally obtained by minimizing the stress applied hydrogen atom located at the tetrahedral site. The octahedral site
in the direction of the GB normal (y-direction in Fig. 1). This was in the B2 TiFe compound is distinguished into two different sites
done by relaxing positions of atoms except for those in the fixed depending on the configuration of surrounding atoms. One site is
boundary regions and adjusting the dimension of the supercell in surrounded by 4 Ti and 2 Fe atoms (Ti4Fe2), and the other site is
the direction of the GB normal. The total energy values of selected surrounded by 2 Ti and 4 Fe atoms (Ti2Fe4). Therefore, we exam-
configurations, which adjust the cell dimension along the direction ined the solution energy of a hydrogen atom located in these sites
normal to the GB, were obtained in a functional form of that di- as follows:
mension. A configuration with minimum total energy was finally
Esol = ETi64 Fe64 H1 − ETi64 Fe64 − 0.5EH2 (3)
selected as the equilibrium GB configuration.
To gain insight into how ternary alloying elements affect the GB As listed in Table 3, hydrogen prefers the site surrounded by
decohesion tendency, the charge density difference (CDD) was an- 4 Ti and 2 Fe atoms (Ti4Fe2), which is associated with a smaller
alyzed. We considered the bonding charge density (ρ ) [38–40], volume expansion due to the presence of hydrogen compared to
defined as the difference between the electron density of a self- the other configuration. This result is consistent with previous DFT
consistent calculation and the electron density associated with un- results on the B2 TiFe compound [43-46].
bound atoms of a non-self-consistent calculation [16,17]. To ana- The TiFe alloys with equiatomic and nearly equiatomic compo-
lyze the characteristic of the atomic bonds, isosurface structures sitions involve the formation of two hydrides at different hydrogen
with different values of ρ are visualized using VESTA [41]. concentrations, e.g., TiFeH (β phase) and TiFeH2 (γ phase) [1]. The
TiFeH hydride exhibits an orthorhombic symmetry corresponding
3. Results and discussion to P 2221 (space group 17) [47]. The TiFeH2 hydride exhibits an or-
thorhombic symmetry corresponding to Cmmm (space group 65)
3.1. Physical properties of the stoichiometric tife compound and [48]. The TiFeH hydride corresponds to a structure where hydro-
hydrides gen atoms occupy the more stable octahedral sites (Ti4Fe2) of the
B2 TiFe compound [44]. The TiFeH2 hydride corresponds to a struc-
To obtain a reference for investigating the influence of alloying ture where hydrogen atoms occupy both kinds of octahedral sites
elements on the properties related to hydrogen storage, it is help- (Ti4Fe2 and Ti2Fe4) [44]. Details of the crystal structures and cal-
ful to analyze the properties of the stoichiometric TiFe compound culated lattice constants of B2 TiFe compound and TiFe-based hy-
and hydrides. We first examined the lattice constants and the en- drides (TiFeH and TiFeH2 ) are summarized in Table 2. The present
thalpy of hydride formation as listed in Table 2. The equiatomic DFT calculation exhibits a sufficient accuracy in reproducing the
TiFe compound exhibits a CsCl-type (B2) structure corresponding structural parameters of both hydrides as they are in fairly good
to the P m3̄m (221) space group, where Ti and Fe atoms occupy the agreement with the reported experimental values. For the TiFeH

150
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 3
The solution energy (Eq. (3)) and the volume expansion due to the presence of an H atom on each interstitial site obtained by the present DFT
calculations.

Property Octahedral site (surrounded by 4 Ti and 2 Fe atoms) Octahedral site (surrounded by 2 Ti and 4 Fe atoms)

Solution energy (eV) 0.203 1.102


Volume expansion by H (%) 0.149 0.211

hydride, the calculated unit cell parameters a, b and c deviated atomic energies of pure Ti, Fe, and element X are hcp, bcc, and a
from the experimental values by 2.0%, 0.1% and 2.8%, and for the single atom, respectively. Then, the site preference of alloying el-
TiFeH2 hydride, they deviated by 0.1%, 2.0% and 2.4%, respectively. ements can be determined by the difference between the defect
We further examined the formation energy of the TiFe-based formation energies of each substitution site as follows:
hydrides (TiFeH and TiFeH2 ) considering the following partial and
ETi site→Fe site = EX→Fe site − EX→Ti site (12)
total hydrogenation reactions:
Table 4 lists the calculated standard defect formation energies
2TiFe + H2 → 2TiFeH(partial hydrogenation ) (4)
and the resulting site preferences for selected alloying elements.
Although results obtained by the standard defect formation en-
2TiFeH + H2 → 2TiFeH2 (partial hydrogenation ) (5)
ergy are generally applicable, this method has a weakness that the
prediction is critically dependent on the arbitrary choice of refer-
TiFe + H2 → TiFeH2 (total hydrogenation ) (6)
ence states for constitutive Ti and Fe. This issue is even more crit-
The formation energies of hydrides corresponding to these re- ical if one wants to predict the site preference of ternary elements
actions are expressed as follows: in non-stoichiometric compounds. If the composition of the target
compound significantly deviates from the stoichiometric composi-
Efpartial (TiFeH ) = 2ETiFeH − 2ETiFe − 1EH2 (7) tion, existing defects in the ordered structure play a decisive role in
the site occupation of alloying elements. For example, if the Ti-rich
Efpartial (TiFeH2 ) = 2ETiFeH2 − 2ETiFeH − 1EH2 (8)
B2 compound (or Fe-rich B2 compound) is stable, a certain propor-
tion of anti-site defects or vacancies in the Fe sublattice (or the Ti
Eftotal (TiFeH2 ) = 1ETiFeH2 − 2ETiFe − 1EH2 (9)
sublattice) must be present. Then, the site preference of ternary al-
In these equations, all the calculated values were defined to loying elements can be driven by competition with those defects.
represent the formation energy per one mol of H2 gas for a direct The contribution by the vacancies is expected to be considerable
comparison between different reactions. For the reference energy only at very high temperatures, where the equilibrium concentra-
of H2 , the entropy term was neglected as it was approximated to tion of vacancies remains high. Instead, the presence of anti-site
the energy of a H2 molecule in vacuum at 0 K. Therefore, the de- defects is expected to account for the non-stoichiometric composi-
termined formation energies do not exactly match the actual for- tion of the ordered B2 alloy under usual conditions. We thus ex-
mation energies at finite temperatures, but the relative differences amined the site preference of ternary alloying elements by com-
indicate that they provide a good approximation. The calculated paring the formation energies of competing configurations of the
formation energies of the stoichiometric TiFe-based hydrides are same composition using the following equations:
listed in Table 2. The present DFT calculation reproduces an exper-
(Ti−rich )
imental trend in the relative stability between both hydrides, i.e., ETi site→Fe site
= Ef [(Ti64 ), (Fe63 X1 )] − Ef [(Ti63 X1 ), (Fe63 Ti1 )]
a more negative formation energy of the TiFeH2 hydride than that (13)
of the TiFeH hydride. Sufficient agreement between the calculated
(Fe−rich )
and the experimental values indeed demonstrates that the present ETi site→Fe site
= Ef [(Ti63 Fe1 ), (Fe63 X1 )] − Ef [(Ti63 X1 ), (Fe64 )]
DFT calculation can be further used to examine the effect of vari-
(14)
ous ternary elements on the properties of the TiFe-based hydrogen
storage alloys. In Eq. (13), Ef [(Ti64 ), (Fe63 X1 )] is the formation energy of a
Ti-rich supercell formed by the substitution of a Fe site by a X
3.2. Site preference of ternary elements in the tife-based hydrogen atom (Fe63 X1 ). Ef [(Ti63 X1 ), (Fe63 Ti1 )] is the formation energy of
storage alloys another Ti-rich supercell with the same composition formed by the
substitution of a Ti site by an X atom (Ti63 X1 ) and the substi-
The preference of an alloying element to go to the Ti or Fe tution of a Fe site by an anti-site Ti atom (Fe63 Ti1 ). In Eq. (14),
sublattice site must be identified before examining the change in Ef [(Ti63 X1 ), (Fe64 )] is the formation energy of a Fe-rich super-
physical properties of the TiFe-base hydrogen storage alloys due to cell formed by the substitution of a Ti site by an X atom (Ti63 X1 ).
the addition of ternary elements. To examine the site preference Ef [(Ti63 Fe1 ), (Fe63 X1 )] is the formation energy of another Fe-rich
of ternary elements in stoichiometric compounds, the standard de- supercell with the same composition formed by the substitution of
fect formation energies [49] associated with the substitution of a a Fe site by an X atom (Fe63 X1 ) and the substitution of a Ti site
Ti site (EX→Ti site ) and a Fe site (EX→Fe site ) were calculated based by an anti-site Fe atom (Ti63 Fe1 ). The formation energy of each
on DFT calculations of 4 × 4 × 4 supercells with 128 atoms using configuration was obtained using the following equation:
the following equations:
Ef = E (Tia Feb Xc ) − aE (Ti ) − bE (Fe ) − cE (X ) (15)
EX→Ti site = [(E (Ti63 X1 Fe64 ) + ETi ) − (E (Ti64 Fe64 ) + EX )] (10)
By combining Eqs. (13-15), the composition-dependent site
preference can be calculated using only the total energies of the
EX→Fe site = [(E (Ti64 Fe63 X1 ) + EFe ) − (E (Ti64 Fe64 ) + EX )] (11)
defect configurations as follows:
Here, E (Ti63 X1 Fe64 ) is the energy of a supercell with the substi- (Ti−rich )
ETi site→Fe site
= E [(Ti64 ), (Fe63 X1 )] − E [(Ti63 X1 ), (Fe63 Ti1 )] (16)
tution of a Ti site by the element X, Ti64 Fe63 X1 is the energy of a
supercell with the substitution of a Fe site by the element X, and
(Fe−rich )
EX is the atomic energy of element X. The reference states for the ETi site→Fe site
= E [(Ti63 Fe1 ), (Fe63 X1 )] − E [(Ti63 X1 ), (Fe64 )] (17)

151
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 4
The site preference of additional alloying elements in the B2 TiFe compound obtained from the standard defect formation energy Eqs. (10)-((12)) and the defect formation
energy based on non-stoichiometric compounds Eqs. (16)-((18)). The reference states for the atomic energies of pure Ti, Fe, and element X are hcp, bcc, and a single atom,
respectively. Previous predictions of the site preference via the Bozzolo–Ferrante–Smith (BFS) method [51] (in parentheses) are compared to the present results.

Standard defect formation Defect formation energy based on non-stoichiometric compounds


energy (eV) (eV) Site preference
Average (Ti−rich ) (Fe−rich )
ETi site→Fe site ETi site→Fe site
ETi site→Fe site
ETi site→Fe site
Equiatomic alloy Ti-rich alloy Fe-rich alloy

Al 1.039 0.771 −0.682 2.224 Ti site Fe site (Fe site) Ti site (Ti site)
As 0.228 −0.018 −1.455 1.418 Fe site Fe site Ti site
Ba 1.484 1.652 −0.202 3.507 Ti site Fe site Ti site
Be −0.032 −0.334 −1.842 1.173 Fe site Fe site Ti site
Ca 1.647 1.366 −0.128 2.859 Ti site Fe site Ti site
Cd 0.711 0.439 −1.047 1.924 Ti site Fe site Ti site
Ce 2.036 2.185 0.241 4.130 Ti site Ti site Ti site
Co −1.335 −1.621 −3.089 −0.154 Fe site Fe site (Fe site) Fe site (Fe site)
Cr −0.959 −1.189 −2.765 0.387 Fe site Fe site (Fe site) Ti site (Ti site)
Cs 0.788 0.981 −0.997 2.958 Ti site Fe site Ti site
Cu −0.321 −0.613 −2.145 0.920 Fe site Fe site Ti site
Ga 0.539 0.209 −1.311 1.729 Ti site Fe site (Fe site) Ti site (Ti site)
Ge 0.526 0.205 −1.294 1.705 Ti site Fe site Ti site
Hf 2.550 2.343 0.883 3.803 Ti site Ti site (Fe site) Ti site (Ti site)
K 0.657 0.380 −1.121 1.880 Ti site Fe site Ti site
La 2.043 1.719 0.366 3.072 Ti site Ti site Ti site
Li −0.177 −0.464 −2.003 1.075 Fe site Fe site Ti site
Mg 0.965 0.671 −0.822 2.163 Ti site Fe site Ti site
Mn −1.089 −1.362 −2.882 0.158 Fe site Fe site Ti site
Mo −0.611 −0.830 −2.387 0.727 Fe site Fe site (Ti site) Ti site (Ti site)
Na 0.208 −0.029 −1.575 1.518 Fe site Fe site Ti site
Nb 0.947 1.211 −0.709 3.132 Ti site Fe site (Ti site) Ti site (Ti site)
Ni −1.115 −1.414 −2.911 0.082 Fe site Fe site (Fe site) Ti site (Ti site)
P 0.083 −0.161 −1.608 1.285 Fe site Fe site Ti site
Pd −0.901 −1.209 −2.693 0.274 Fe site Fe site Ti site
Rb 0.693 0.419 −1.051 1.888 Ti site Fe site Ti site
Rh −2.060 −2.371 −4.172 −0.886 Fe site Fe site Fe site
Ru −2.416 −2.694 −3.857 −1.2167 Fe site Fe site (Fe site) Fe site (Fe site)
S −1.005 −1.250 −2.695 0.196 Fe site Fe site Ti site
Sc 2.101 1.847 0.396 3.299 Ti site Ti site Ti site
Se −0.364 −0.604 −2.044 0.837 Fe site Fe site Ti site
Si 0.777 0.518 −0.918 1.953 Ti site Fe site (Ti site) Ti site (Ti site)
Sr 1.761 1.428 −0.111 2.966 Ti site Fe site Ti site
Ta 1.591 1.445 −0.068 2.957 Ti site Fe site (Ti site) Ti site (Ti site)
Tc −1.868 −2.136 −3.654 −0.617 Fe site Fe site Fe site
V 0.569 0.820 −1.104 2.745 Ti site Fe site (Fe site) Ti site (Ti site)
W 0.007 −0.151 −1.653 1.350 Fe site Fe site (Ti site) Ti site (Ti site)
Y 2.386 2.146 0.693 3.599 Ti site Ti site Ti site
Zn 0.262 0.481 −1.527 2.488 Ti site Fe site Ti site
Zr 2.242 2.513 0.577 4.449 Ti site Ti site (Fe site) Ti site (Ti site)

Results obtained from Eqs. (16) and (17) indicate the site pref- sub-lattice site in practical B2 TiFe compounds. If we distinguish
erences predicted for the non-stoichiometric Ti-rich and Fe-rich the site preferences of these elements, even considering small en-
B2 compounds, respectively. For example, a negative value from ergy differences, the standard defect formation energy cannot pro-
Eqs. (16) and (17) means that the alloying element prefers the Fe vide consistent information because the prediction critically de-
site rather than the Ti site for the substitution, and vice versa. Even pends on the choice of reference states for Ti and Fe. There-
though Eqs. (16) and (17) were derived for the non-stoichiometric fore, we focused on the defect formation energy based on non-
compounds, the averaging of quantities obtained by both equations stoichiometric compounds (Eq. (18)) to determine the site prefer-
provides an additional criterion to determine the site preference of ences of the equiatomic alloy in further investigations.
the stoichiometric B2 compound as follows: Table 4 also lists the site preferences of ternary alloying ele-
  ments in the B2 TiFe with non-stoichiometric Ti-rich and Fe-rich
Average ( Ti−rich ) (Fe−rich )
ETi site→Fe site
= ETi site→Fe site
+ ETi site→Fe site
/2 (18) compositions. At first glance, the site preference for the Fe-rich
compound may seem impractical because the binary phase dia-
This equation provides estimates similar to Eq. (12), but it is gram of the Ti-Fe system shows significant off-symmetric solubil-
more advantageous because the results are independent of the ity limits of the B2 compound for only Ti-rich compositions [50].
choice of reference states for Ti and Fe. Finally, determined defect However, a recent experiment on Ti-Fe-V hydrogen storage alloys
formation energies and resulting site preferences for ternary alloy- [27] reported that a B2 compound with a significant Fe-rich com-
ing elements are listed in Table 4. position (Ti: 46.0, Fe: 49.5, V: 4.5 at.%) can be obtained by adding
By comparing different predictions of the equiatomic alloy 5 at.% V to a Fe-rich alloy (Ti45 Fe50 V5 ). This implies that the solu-
based on the standard defect formation energy (Eq. (12)) and bility limits of the B2 compound can be extended to the Fe-rich re-
the defect formation energy of non-stoichiometric compounds gion by adding relevant ternary alloying elements. We thus present
(Eq. (18)), there is general agreement between both predictions, results for the Fe-rich composition as well as those for the Ti-rich
but with inconsistencies present in As, Na, P and W. These ele- and equiatomic compositions in Table 4.
ments commonly indicate defect formation energies close to zero, We also compared our results with previous theoretical re-
implying that these elements do not strongly favor a single kind of sults via the Bozzolo–Ferrante–Smith (BFS) method [51]. Simi-

152
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 5
The role of additional alloying elements on the formation energies (Ef ) of hydrides (TiFeH and TiFeH2 ) per one mol of H2 and the volume expansion (V ) considering
partial Eqs. (7) and ((8)) and total (Eq. (9)) hydrogenation reactions obtained by the present DFT calculations. The results for stoichiometric hydrides without alloying
elements are highlighted in bold. Underlined values indicate that the results correspond to the investigated site preference of each alloying element in the equiatomic TiFe
compound (Table 4).

Ti site Fe site
Ef (kJ / mol H2 ) V (%) Ef (kJ / mol H2 ) V (%)
Element TiFeH TiFeH2 TiFeH2 TiFeH TiFeH2 TiFeH2 TiFeH TiFeH2 TiFeH2 TiFeH TiFeH2 TiFeH2
(partial) (partial) (total) (partial) (partial) (total) (partial) (partial) (total) (partial) (partial) (total)

– −23.08 −29.11 −26.09 9.55 5.68 15.77 −23.08 −29.11 −26.09 9.55 5.68 15.77
Al −1.15 12.85 5.85 10.74 6.19 17.60 −35.21 −22.74 −28.98 7.56 6.46 14.51
As 7.71 65.09 36.40 13.78 7.55 22.36 −6.51 −16.68 −11.59 7.28 10.76 18.82
Ba −100.74 −77.01 −88.88 18.95 4.45 24.24 −15.23 −112.25 −63.74 8.90 3.06 12.23
Be −27.15 0.47 −13.34 11.84 5.79 18.32 −36.07 −23.35 −29.71 8.76 6.11 15.41
Ca −89.99 −53.15 −71.57 10.65 4.11 15.19 −64.82 −81.93 −73.37 12.11 4.63 17.29
Cd −44.52 3.36 −20.58 10.25 4.47 15.17 −42.48 −42.49 −42.48 7.72 5.35 13.48
Ce −49.59 −43.23 −46.41 9.44 5.04 14.96 −17.33 −109.08 −63.21 5.41 8.36 14.21
Co −53.48 −32.17 −42.82 10.99 6.22 17.90 −35.39 −32.92 −34.16 8.77 5.70 14.97
Cr −28.02 −38.18 −33.10 9.54 5.65 15.73 −41.24 −40.02 −40.63 9.95 4.63 15.03
Cs −90.03 −138.34 −114.18 26.91 11.14 41.05 −9.12 −125.15 −67.13 7.50 7.27 15.32
Cu −40.42 −24.10 −32.26 9.46 6.65 16.74 −49.15 −34.29 −41.72 8.79 5.56 14.83
Ga 0.76 23.90 12.33 12.81 5.07 18.53 −29.21 −17.64 −23.42 8.14 6.29 14.94
Ge 32.03 45.10 38.56 12.68 6.87 20.43 −17.04 −9.02 −13.03 6.46 7.72 14.68
Hf −30.13 −19.55 −24.84 8.97 5.11 14.54 −39.24 −85.56 −62.40 7.69 7.51 15.77
K −106.50 −115.16 −110.83 5.84 11.38 17.88 −65.48 −126.07 −95.78 3.25 6.32 9.77
La −72.21 −40.53 −56.37 14.40 1.49 16.10 −26.09 −97.55 −61.82 7.84 4.41 12.60
Li −77.83 −50.22 −64.03 9.57 6.03 16.18 −59.87 −47.63 −53.75 9.35 6.38 16.32
Mg −59.97 −21.76 −40.86 10.01 4.77 15.25 −47.21 −48.83 −48.02 8.89 6.07 15.49
Mn −33.91 −13.51 −23.71 11.96 2.26 14.49 −28.44 −29.44 −28.94 9.82 5.51 15.87
Mo −26.58 −40.91 −33.74 8.25 5.09 13.75 −40.25 −36.07 −38.16 10.30 4.52 15.29
Na −83.74 −76.31 −80.03 4.88 12.04 17.51 −63.75 −78.32 −71.04 6.49 9.04 16.12
Nb −28.30 −33.52 −30.91 8.47 5.52 14.46 −49.44 −59.03 −54.23 10.28 4.13 14.83
Ni −36.43 −26.29 −31.36 8.07 5.92 14.47 −45.90 −34.84 −40.37 8.93 5.40 14.82
P −0.97 83.38 41.20 14.86 12.05 28.70 −4.55 −8.55 −6.55 8.11 6.88 15.55
Pd −39.05 −22.77 −30.91 7.17 4.93 12.46 −38.63 −45.90 −42.27 8.56 5.01 14.00
Rb −103.45 −136.78 −120.12 8.72 12.45 22.26 −38.26 −143.42 −90.84 4.34 7.04 11.68
Rh −21.83 −32.77 −27.30 5.15 6.30 11.78 −22.95 −36.02 −29.49 8.88 5.33 14.68
Ru −20.09 −38.83 −29.46 7.77 3.70 11.76 −8.26 −25.52 −16.89 9.50 6.08 16.16
S −60.16 54.81 −2.67 12.15 14.81 28.77 −22.86 −58.08 −40.47 10.76 12.68 24.80
Sc −48.19 −35.02 −41.61 9.84 4.73 15.03 −49.29 −75.63 −62.46 9.69 4.27 14.38
Se −35.86 46.99 5.57 12.58 12.42 26.56 −8.94 −54.24 −31.59 9.00 11.10 21.10
Si 38.64 50.98 44.81 13.01 6.15 19.95 −18.95 −6.79 −12.87 6.59 7.18 14.24
Sr −103.44 −64.03 −83.74 12.72 0.67 13.47 −47.17 −91.25 −69.21 5.29 3.99 9.49
Ta −20.10 −20.77 −20.43 8.50 5.50 14.48 −46.62 −63.54 −55.08 10.12 4.03 14.55
Tc −25.86 −43.44 −34.65 6.96 5.74 13.09 −18.53 −20.84 −19.68 9.82 5.29 15.63
V −21.51 −34.47 −27.99 9.45 5.85 15.85 −49.11 −56.59 −52.85 9.17 4.68 14.28
W −15.37 −27.02 −21.20 8.20 5.18 13.81 −39.04 −38.78 −38.91 10.52 3.93 14.87
Y −67.24 −30.27 −48.76 11.77 2.11 14.13 −41.65 −87.20 −64.42 7.01 3.76 11.03
Zn −32.10 0.05 −16.03 10.66 5.67 16.94 −38.20 −28.80 −33.50 8.75 5.74 15.00
Zr −37.61 −28.51 −33.06 8.81 4.96 14.21 −41.43 −88.91 −65.17 7.28 7.91 15.77

lar to the present method, this study [51] provided predictions 3.3. Roles of ternary elements on the hydrogen storage
on the site preferences of several elements in various B2 com-
pounds using configurations with anti-site defects. The reported In the previous section, we examined the site preference of
predictions [51] considering non-stoichiometric compositions are each alloying element in the B2 TiFe compound. In this section,
listed and compared to our predictions in Table 4 (in parenthe- we extend the investigation to properties of the TiFe-based hy-
ses). There is general agreement between both results, but sev- drides with alloying elements based on the revealed tendency of
eral inconsistencies are present regarding the Ti-rich alloy. We the site preference. We analyzed the roles of ternary alloying ele-
attribute these deviations to the different methods of obtaining ments on the formation energies of the TiFe-based hydrides (TiFeH
the total energy of the atomic configurations. The BFS method and TiFeH2 ), which are closely connected with the alteration of re-
[51] is a more semi-empirical calculation, as it is a quantum ap- quired pressure conditions for the hydrogen sorption and desorp-
proximation technique based on the formation energy of the tar- tion processes. All calculation results were obtained using a fixed
get atomic configuration, given as the sum of individual atomic concentration where a solute atom substitutes the Ti or Fe sublat-
contributions. tice site of a B2 supercell composed of 4 Ti atoms and 4 Fe atoms
Although comparison of our results with experimental data is as listed in Table 1. Therefore, the present DFT calculations only
desirable, to the best of our knowledge there is an absence of di- provide overall trends for the change in formation energies as each
rect analyses of site preferences for ternary elements in TiFe alloys, ternary alloying element is added, and the estimates do not corre-
including investigations based on the atom probe tomography. Al- spond to the exact composition that can be practically realized.
ternately, we can justify our predictions in an indirect manner by Table 5 lists the calculated formation energies of the TiFe-
further considering the alloying effects on hydrogen storage prop- based hydrides with a ternary alloying element. The results for
erties, which will be discussed in the next section. the stoichiometric TiFeH and TiFeH2 compounds without any so-

153
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 6
The role of additional alloying elements in altering the plateau hydrogen pressure in the PCT curve. The predictions according to the present DFT calculations in Tables 4
and 5 are compared to the previous predictions via the Bozzolo–Ferrante–Smith (BFS) method [51] and previously reported experimental trends. The results for different
compound stoichiometries (Equiatomic, Ti-rich, and Fe-rich) are specified. “Stabilized” and “Destabilized” indicate a decrease and increase in the plateau hydrogenation
pressure, respectively.

Experiment Present prediction Prediction by BFS method [51]


Element Stoichiometry Stability of hydrides Site preference Stability of hydrides Site preference Stability of hydrides∗

Al Equiatomic Destabilized [25] Ti site Destabilized


(Ti47.5 Fe47.5 Al5 )
Co Ti-rich Stabilized [20] Fe site Stabilized Fe site Stabilized
(Ti50 Fe45 Co5 )
Co Equiatomic Stabilized [56] Fe site Stabilized Fe site Stabilized
(Ti49 Fe49 Co2 )
Cr Equiatomic Stabilized [25] Fe site Stabilized
(Ti48 Fe48 Cr4 )
Cr Ti-rich Stabilized [27] Fe site Stabilized Fe site Stabilized
(Ti47 Fe43.3 Cr9.7 ,
Ti47.1 Fe41 Cr11.9 )
Mn Ti-rich Stabilized [18] Fe site Stabilized
(Ti49 Fe46 Mn5 ,
Ti50 Fe40 Mn10 ,
Ti48 Fe40 Mn12 ,
Ti50 Fe35 Mn15 )
Nb Equiatomic Stabilized [56] Ti site Stabilized Ti site Stabilized
(Ti49 Fe49 Nb2 )
Ni Ti-rich Stabilized [20, 57] Fe site Stabilized Fe site Stabilized
(Ti50 Fe45 Ni5 ,
Ti50 Fe42.5 Ni7.5 ,
Ti50 Fe40 Ni10 )
Pd Ti-rich Stabilized [19] Fe site Stabilized
(Ti50 Fe45 Pd5 ,
Ti50 Fe40 Pd10 )
V Ti-rich Stabilized [26] Fe site Stabilized Fe site Stabilized
(Ti50 Fe47 V3 , Ti50 Fe45 V5 )
V Fe-rich Destabilized [26] Ti site Destabilized Ti site Destabilized
(Ti47 Fe50 V3 , Ti45 Fe50 V5 )
Zr Equiatomic Stabilized [58] Ti site Stabilized
(Ti45.1 Fe45.1 V9.8 )

Because the BFS method [51] provided only the site preference information, alteration of the plateau hydrogen pressure is obtained based on the present calculation
results for the hydride formation energies listed in Table 5.

lute elements (Table 2) are also listed for comparison. The ob- systems are usually compared by evaluating only the enthalpy
tained changes in the formation energy are fairly large due to a change associated with the (de)hydrogenation process [52]. There-
high concentration (12.5 at.% of solute atoms in the B2 structure) fore, a decrease or increase in the plateau hydrogenation pres-
of ternary alloying elements considered in the present calculation. sure (Peq ) indicates that the hydride phase is stabilized (more en-
Therefore, only the direction of changes, i.e., the stabilization or dothermic) or destabilized (more exothermic), respectively. When
destabilization of hydrides due to the addition of ternary elements neglecting the temperature dependence of the enthalpy differ-
should be of interest from the present calculation results. We cal- ence of each phase, the formation energies of hydrides (Ef ) at
culated and listed all formation energies considering possible sub- 0 K obtained by the present DFT calculations can be approxi-
stitutions of both Ti and Fe sites. This information should be used mated to be the formation enthalpy determined from the slope
in conjunction with the composition of practical alloys, such as (H 0 /R) of the experimental Van’t Hoff plot (logP vs. 1/T in
the equiatomic, Ti-rich, or Fe-rich compositions. As an example, Eq. (19)).
we specified predictions (underlined values in Table 5) that cor- The experimentally measured trends in the alteration of the
respond to the investigated site preference of alloying elements in plateau hydrogen pressure in the PCT curve by the addition of
the equiatomic TiFe compound (Table 4). ternary elements with detailed information on the bulk compo-
Then, predicted trends in the change of hydride formation en- sition are summarized and compared to the present results in
ergies due to ternary alloying elements are compared to reported Table 6. It should be noted that the B2 compound composition
experimental trends. The equilibrium conditions of pressure and can deviate somewhat from the bulk composition if the mate-
temperature where material can absorb and desorb hydrogen can rial is not composed of a single phase. In this table, “destabi-
be expressed by following the Van’t Hoff equation: lized” and “stabilized” indicate an increase and decrease in the
  plateau hydrogenation pressure, respectively, and these trends re-
Peq H 0 S 0 spectively correspond to the more positive and negative formation
ln = − (19)
P0 RT R enthalpies of hydrides. As listed, the present results considering
Here, Peq is the plateau (or equilibrium) hydrogen pressure in the stoichiometry of alloying elements are consistent with the re-
the PCT curve, P0 is the pressure at standard state (1 atm), R is ported experimental trends. In particular, it is worth noting the
the ideal gas constant, T is the temperature, H 0 is the change predictions for the V-containing alloy system. As mentioned in the
in enthalpy at standard state, and S0 is the change in entropy previous section, the recent experimental study on Ti-Fe-V hydro-
at standard state. Because the entropy change associated with the gen storage alloys [27] analyzed the effects of V addition on al-
(de)hydrogenation process is similar for usual metallic hydrides tering the plateau hydrogen pressure considering both the Ti-rich
and its changes due to the substitution of alloying elements are and Fe-rich compositions of the B2 compound. The reported re-
negligible, the equilibrium properties of various metal–hydrogen sults showed that the addition of V to the Ti-rich B2 compound

154
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Table 7
The solute-GB binding energy for possible segregation sites near the GB (Ti1, Ti2, and Fe1 shown in Fig. 1) and the maximum tensile stress
(σmax ) obtained by the present DFT calculation. In the solute-GB binding energy, the most probable site for the segregation of each ternary
element considering the preferred sublattice site of the equiatomic B2 structure (Table 4) is highlighted in bold. The maximum tensile
stress was obtained using the bi-crystal cell with an additional solute atom at the most probable site near the GB.

Ti1 (eV) Ti2 (eV) Fe1 (eV) σmax (GPa)


TiFe 31.48
Al 0.037 −0.191 30.89
As 0.487 29.13
Ba 1.602 1.220 12.69
Be 0.639 26.85
Ca 1.143 0.693 18.68
Cd 0.747 0.101 27.71
Ce 1.058 0.958 23.05
Co 0.120 28.59
Cr 0.104 29.35
Cs 1.450 1.061 12.65
Cu 0.339 24.28
Ga 0.101 −0.154 29.98
Ge −0.224 −0.131 30.03
Hf 0.503 0.208 29.28
K 1.375 1.055 13.54
La 1.339 1.081 19.00
Li 0.033 18.11
Mg 0.563 0.065 28.57
Mn 0.024 31.27
Mo −0.173 28.95
Na −0.225 13.03
Nb 0.356 0.139 31.55
Ni 0.300 27.20
P 0.743 27.56
Pd −0.010 24.47
Rb 1.439 1.184 12.85
Rh −0.235 28.49
Ru −0.327 30.71
S 0.790 20.11
Sc 0.580 0.247 28.37
Se 0.565 19.10
Si −0.516 −0.246 30.46
Sr 1.414 1.072 14.98
Ta 0.190 0.049 31.67
Tc −0.301 31.28
V −0.231 −0.050 30.13
W −0.054 31.91
Y 1.104 0.679 22.46
Zn 0.334 −0.093 30.54
Zr 0.651 0.324 28.92

causes a significant decrease in plateau hydrogenation pressure, 3.4. Roles of ternary elements on grain boundary embrittlement
while the addition of V to the Fe-rich B2 compound causes a slight
increase in plateau hydrogenation pressure. This is expected based GB embrittlement affected by the addition of ternary elements
on the present calculation results for the site preference and for- is another factor that can be considered in the future alloying de-
mation energies of hydrides listed in Tables 4 and 5. This com- sign process. GB embrittlement due to the segregation of impu-
parison justifies the present predictions of the hydride formation rities is not a desirable phenomenon in usual metallic materials.
energies as well as the predictions of the composition-dependent However, one can think of GB embrittlement as a possible way for
site preference of alloying elements, as explained in the previous enhancing activation of hydrogen storage alloys because it can pro-
section. vide native surface regions without oxidation through the fracture
In Table 5, we also provide the volume expansion due to the of initial hydrogen storage alloys with surface oxides. The virtues
hydride formation, i.e., the volume difference between the initial of possible impurities in the TiFe alloys that should be used for
B2 compound and the final hydride structures. In many metal– this purpose are the high segregation tendency to GB regions and
hydrogen systems including the TiFe system, hysteresis in the PCT the high decohesion tendency of atomic bonds.
curve is commonly presented as indicated by a higher transition Even considering the same misorientation angle and axis for ro-
pressure during the hydrogenation process than that during the tation, there are two types of  3(111)[11̄0] STGB, where Ti or Fe
dehydrogenation process. The origin of such hysteresis behavior atoms are located at the center of the GB region. We first con-
has not been fully understood, but a severe lattice expansion due firmed that the GB with Ti atoms in its center is more stable than
to the hydrogenation process is believed to be a possible factor the GB with Ti atoms in its center. Therefore, we focused on the
[53]. For example, the formation of hydrides is expected to cause more stable GB (Fig. 1) for further examining the segregation and
irreversible plastic deformation of the matrix phase. Then, the vol- decohesion tendencies. We then examined the GB segregation ten-
ume expansion due to the hydride formation and its alteration by dency of ternary elements in the B2 TiFe compound by examining
the addition of ternary elements can affect the amount of plastic the binding energy between the GB and a ternary solute atom. This
deformation and the resulting intensity of the hysteresis in the PCT binding energy is related to approximately the maximum amount
curve. of the solute atoms near the GB region that can be induced by

155
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

the intended pre-annealing process at relevant temperatures to en-


hance the segregation phenomenon. The binding energy between
two defects is defined as a difference in energies of two configura-
tions when defects are adjacent to each other and when they are
separated enough and non-interacting. In the present study, the
interaction energy between the  3(111)[11̄0] STGB and a ternary
solute atom placed in possible segregation sites was considered.
Those segregation sites are composed of two Ti sites (Ti1, Ti2) and
one Fe site (Fe1) near the GB region as shown in Fig. 1. Consider-
ing these sites, the binding energy between the GB and a ternary
solute atom was calculated about the configuration where a solute
atom substitutes for a Ti (Ti0) or a Fe (Fe0) site located far from
the GB region (Fig. 1). To calculate accurate binding energy values,
the minimization of the stress applied in the direction of the GB
normal (y-direction in Fig. 1) was performed after each substitu-
tion of solute atoms.
Table 7 lists the calculated binding energies between various
ternary elements and the GB. According to the definition of bind-
ing energy, a positive sign (negative sign) means that the GB and a
ternary solute atom are attracted to (repulsed by) each other. The
calculated binding energies are different for each GB site. These
values provide useful information in evaluating the most probable
site for the segregation of various ternary alloying elements near
the GB region. Among different values for each site, the most im-
portant is the largest binding energy given the preferred substitu-
tion site of each ternary alloying element, as highlighted in bold in
Table 7 (e.g., a value for the Ti1 site if Al is the allying element).
We also examined the role of ternary elements on the fracture
of the GB by performing a DFT tensile test. This method provides a
useful way to evaluate the rough embrittlement tendency focusing
on the decohesion of atomic bonds due to the addition of alloying
elements near the GB region. For the DFT tensile test, an initial bi-
crystal cell with a ternary solute atom located in the most stable
GB site (Table 7) was considered for each alloying element. After
the stable GB configuration was obtained by minimizing the stress
along the direction normal to GB (y-direction in Fig. 1), a uniaxial
tensile loading was applied along this direction by increasing the
distance between the uppermost and lowermost boundary regions
with an incremental step of 0.25 Å. Positions of all atoms except
those in the fixed boundary regions could fully relax. Dimensions
along the lateral directions of the GB were fixed during the tensile
loading to better reproduce a triaxial stress state near the crack tip
that presents in the fracture process of practical materials. After
a certain number of loading steps, the bond-breaking process and
resulting exposure of surfaces near the GB region were observed
while the simulation starts without any pre-defined fracture sur-
faces. The DFT tensile test provided a variation in the total energy
as a function of the distance between two fixed boundary regions.
The corresponding tensile stress (σ ) at a specific strain (ε ) was ob-
tained using the following equation:
1 ∂E Fig. 2. DFT calculation results for the tensile stress vs. loading distance responses
σ= (18)
(ε ) ∂ε of B2 TiFe-based bi-crystal cells with a ternary solute atom near the GB region. For
each alloying element, a bi-crystal cell with a ternary solute atom at the most prob-
Here, (ε ) is the volume and E is the total energy of the bi- able GB site (Table 7) is used for the calculation. The maximum tensile stress (GPa)
obtained from each response is represented in parenthesis. The reference response
crystal cell.
of the bi-crystal cell without ternary alloying elements is represented as “TiFe”.
The obtained relations between the tensile stress and the sepa-
ration distance of bi-crystal cells with various ternary alloying el-
ements are shown in Fig. 2. The response of a pristine bi-crystal
cell without any alloying elements is also presented for compar- elements located in their respective stable GB sites. As shown
ison. Each alloying element located near the GB region exhibited in the figure, there is a noticeable correlation between the CDD
significantly different decohesion tendencies of atomic bonds. near the ternary atom and the maximum tensile stress. Given that
Fig. 3 shows the calculated CDD of bi-crystal cells at the initial charge accumulation and depletion are direct indicators of the
stage of the tensile loading, as visualized by the isosurface struc- strengthening or weakening of atomic bonds, it is not surprising
tures with different values of ࢞ρ . We selected several ternary al- that the CDD is associated with the decohesion tendency of GBs.
loying elements that exhibit a wide range of the maximum tensile For alloying elements that exhibit relatively higher maximum ten-
stress values and compared the CDD of bi-crystal cells with those sile stress values (e.g., Nb and Mn), the CDD near a ternary atom is

156
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

Fig. 3. The calculated charge density difference (CDD) in the (11̄0 ) plane of the  3(111)[11̄0] STGB at the initial stage of tensile loading obtained using the present DFT
calculations. GBs with a solute atom substituting the Ti site (Nb, Sc, Cd, Y, Ca, and Sr) and those with a solute atom substituting the Fe site (Mn, P, Cu, S, Li, and Na) are
presented. The values in parentheses represent the maximum tensile stresses (GPa) listed in Table 7.

similar to that of the reference TiFe GB. For other alloying elements
with lower maximum tensile stress values, there is a significant
decrease in the charge density around the ternary atom, indicating
that the atomic bonds between the ternary atom and nearby Ti or
Fe atoms become weaker compared to those of the reference TiFe
GB.
As stated above, possible ternary alloying elements in the TiFe
alloys induce GB embrittlement if they have a high tendency to
segregate to the GB region and a high tendency to exhibit deco-
hesion of atomic bonds. To identify the combination of these con-
ditions more clearly, the correlation between the solute-GB bind-
ing energy (segregation tendency) and the maximum tensile stress
(decohesion tendency) was plotted for each ternary solute element,
as shown in Fig. 4. There are several candidate ternary alloying el-
ements that can maximize the GB embrittlement behavior as high-
lighted by the shaded area in Fig. 4.
These elements are expected to contribute to the exposure of
native surfaces without oxides through enhanced GB fracture if
proper processing conditions (e.g., pre-annealing process) to en-
hance the amount of the GB segregation is considered during the
manufacturing of the TiFe-based hydrogen storage alloys. Although
Fig. 4. Relations between the solute-GB binding energy (segregation tendency) and
the suggested alloying elements satisfy the selection criteria that the maximum tensile stress (decohesion tendency) for each alloying element ob-
focused on GB embrittlement, other factors could be important tained by the present DFT calculations. Among solute-vacancy binding energy val-
considering the practical alloying design process. For example, if ues for different segregation sites of each alloying element, the maximum value
the solubility of a ternary alloying element in the B2 TiFe com- considering the predicted site preference of each element (values in bold in Table 7)
is used.
pound is extremely low, the actual amount of the GB segregation
can be much less than expected. If so, alternative alloying elements
with sufficient solubility in the B2 TiFe compound with a moder- ing conditions is beyond the scope of the current work and is left
ate GB segregation tendency could be a better choice. Moreover, in to possible future experimental investigations.
the practical alloy design process, detailed conditions for alloying
elements and target concentrations should be decided considering 4. Conclusion
their roles on other important characteristics of hydrogen storage
alloys, e.g., the enthalpy of hydride formation and the amount of The role of various ternary elements on the performance of sta-
hysteresis in the PCT curve. The derivation of such detailed alloy- tionary TiFe-based hydrogen storage alloys has been investigated

157
W.-S. Ko, K.B. Park and H.-K. Park Journal of Materials Science & Technology 92 (2021) 148–158

based on extensive DFT calculations. As a basic step for predictions, [18] S.V. Mitrokhin, V.N. Verbetsky, R.R. Kajumov, H. Cunmao, Z. Yufen, J. Alloys
the site preference of each alloying element in the stoichiometric Compd. 199 (1993) 155–160.
[19] I. Yamashita, H. Tanaka, H. Takeshita, N. Kuriyama, T. Sakai, I. Uehara, J. Alloys
and non-stoichiometric B2 TiFe compounds has been clarified con- Compd. 253-254 (1997) 238–240.
sidering the formation of anti-site defects. Based on the revealed [20] S.M. Lee, T.P. Perng, J. Alloys Compd. 291 (1999) 254–261.
site preference, the effects of a wide variety of possible alloying [21] B.K. Singh, A.K. Singh, C.S. Pandey, O.N. Srivastava, Int. J. Hydrogen Energy 24
(1999) 1077–1082.
elements have been examined by focusing on the formation en- [22] A. Szajek, M. Jurczyk, E. Jankowska, J. Alloys Compd. 348 (2003) 285–292.
thalpies of TiFeH and TiFeH2 hydrides, which are closely related [23] A. Kocjan, A. Gradišek, N. Daneu, T. Apih, P.J. McGuiness, S. Kobe, J. Magn.
to the changes in the locations of plateaus in the PCT curve. Addi- Magn. Mater. 324 (2012) 2043–2050.
[24] M.Y. Zadorozhnyi, S.D. Kaloshkin, S.N. Klyamkin, O.V. Bermesheva,
tional physical properties such as the volume expansion due to hy-
V.Y. Zadorozhnyi, Met. Sci. Heat Treat. 54 (2013) 461–465.
dride formation have been also examined to provide further crite- [25] V.Y. Zadorozhnyy, S.N. Klyamkin, M.Y. Zadorozhnyy, O.V. Bermesheva,
ria for selecting optimum alloying conditions in future alloying de- S.D. Kaloshkin, J. Alloys Compd. 586 (2014) S56–S60.
[26] J.Y. Jung, Y.S. Lee, J.Y. Suh, J.Y. Huh, Y.W. Cho, J. Alloys Compd. 854 (2021)
sign processes. In particular, alloying elements that maximize the
157263.
GB embrittlement due to the solute segregation phenomenon are [27] H. Kim, M. Faisal, S.I. Lee, J.Y. Jung, H.J. Kim, J. Hong, Y.S. Lee, J.H. Shim,
proposed for enhancing the initial activation of TiFe-based hydro- Y.W. Cho, D.H. Kim, J.Y. Suh, J. Alloys Compd. 864 (2021) 158876.
gen storage alloys. [28] B.K. Singh, A.K. Singh, O.N. Srivastava, Int. J. Hydrogen Energy 21 (1996)
111–117.
[29] J. Ma, H. Pan, X. Wang, C. Chen, Q. Wang, Int. J. Hydrogen Energy 25 (20 0 0)
Declaration of Competing Interest 779–782.
[30] A. Izanlou, M.K. Aydinol, Int. J. Hydrogen Energy 35 (2010) 1681–1692.
[31] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251–14269.
The authors declare that they have no known competing finan- [32] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15–50.
cial interests or personal relationships that could have appeared to [33] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169–11186.
influence the work reported in this paper. [34] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953–17979.
[35] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.
[36] D. Yeşilleten, T.A. Arias, Phys. Rev. B 64 (2001) 174101.
Acknowledgements [37] M. Rajagopalan, M.A. Tschopp, K.N. Solanki, JOM 66 (2014) 129–138.
[38] P.N.H. Nakashima, A.E. Smith, J. Etheridge, B.C. Muddle, Science 331 (2011)
1583–1586.
This research was supported by the National Research Founda-
[39] W.Y. Wang, S.L. Shang, Y. Wang, K.A. Darling, L.J. Kecskes, S.N. Mathaudhu,
tion of Korea (NRF) funded by Ministry of Science and ICT (Nos. X.D. Hui, Z.K. Liu, J. Alloys Compd. 586 (2014) 656–662.
NRF-2019M3E6A1103984 and NRF-2019M3D1A1079214). [40] W.Y. Wang, F. Xue, Y. Zhang, S.L. Shang, Y. Wang, K.A. Darling, L.J. Kecskes, J. Li,
X. Hui, Q. Feng, Z.K. Liu, Acta Mater 145 (2018) 30–40.
[41] K. Momma, F. Izumi, J. Appl. Crystall. 41 (2008) 653–658.
References
[42] A.M. Van der Kraan, K.H.J. Buschow, Phys. B+C 138 (1986) 55–62.
[43] A.V. Bakulin, S.S. Kulkov, S.E. Kulkova, S. Hocker, S. Schmauder, Int. J. Hydrogen
[1] J.J. Reilly, R.H. Wiswall, Inorg. Chem. 13 (1974) 218–222. Energy 39 (2014) 12213–12220.
[2] V. Güther, A. Otto, J. Alloys Compd. 293-295 (1999) 889–892. [44] A. Kinaci, M.K. Aydinol, Int. J. Hydrogen Energy 32 (2007) 2466–2474.
[3] G. Sandrock, J. Alloys Compd. 293-295 (1999) 877–888. [45] K. Takahashi, S. Isobe, Phys. Chem. Chem. Phys. 16 (2014) 16765–16770.
[4] H. Yukawa, Y. Takahashi, M. Morinaga, Comput. Mater. Sci. 14 (1999) 291–294. [46] Z.S. Nong, J.C. Zhu, X.W. Yang, Y. Cao, Z.H. Lai, Y. Liu, Comput. Mater. Sci. 81
[5] T. Nambu, H. Ezaki, H. Yukawa, M. Morinaga, J. Alloys Compd. 293-295 (1999) (2014) 517–523.
213–216. [47] P. Thompson, M.A. Pick, F. Reidinger, L.M. Corliss, J.M. Hastings, J.J. Reilly, J.
[6] A.A. Novakova, O.V. Agladze, S.V. Sveshnikov, B.P. Tarasov, Nanostruct. Mater. Phys. F: Met. Phys. 8 (1978) L75–L80.
10 (1998) 365–374. [48] P. Fischer, J. Schefer, K. Yvon, L. Schlapbach, T. Riesterer, J. Less Common Met.
[7] J.M. Marchetti, E. González, P. Jasen, G. Brizuela, A. Juan, Int. J. Hydrogen En- 129 (1987) 39–45.
ergy 36 (2011) 9037–9044. [49] C. Booth-Morrison, Z. Mao, R.D. Noebe, D.N. Seidman, Appl. Phys. Lett. 93
[8] E. González, P. Jasen, J.M. Marchetti, G. Brizuela, A. Juan, Int. J. Hydrogen En- (2008) 033103.
ergy 37 (2012) 2661–2668. [50] T.B. Massalski, H. Okamoto, P.R. Subramanian, L. Kacprzak, Binary Alloy Phase
[9] G. Lee, J.S. Kim, Y.M. Koo, S.E. Kulkova, Int. J. Hydrogen Energy 27 (2002) Diagram, ASM International, Materials Park, Ohio, 1990.
403–412. [51] G.H. Bozzolo, R.D. Noebe, C. Amador, Intermetallics 10 (2002) 149–159.
[10] S.E. Kulkova, S.S. Kulkov, A.V. Bakulin, S. Hocker, S. Schmauder, Int. J. Hydrogen [52] P. Ngene, A. Longo, L. Mooij, W. Bras, B. Dam, Nat. Commun. 8 (2017) 1846.
Energy 37 (2012) 6666–6673. [53] D.G. Ivey, D.O. Northwood, J. Mater. Sci. 18 (1983) 321–347.
[11] G.D. Sandrock, P.D. Goodell, J. Less Common Met. 73 (1980) 161–168. [54] D.G. Westlake, J. Mater. Sci. 19 (1984) 316–326.
[12] L. Jai-Young, C.N. Park, S.M. Pyun, J. Less Common Met. 89 (1983) 163–168. [55] K. Benyelloul, Y. Bouhadda, M. Bououdina, H.I. Faraoun, H. Aourag, L. Seddik,
[13] K. Edalati, J. Matsuda, A. Yanagida, E. Akiba, Z. Horita, Int. J. Hydrogen Energy Int. J. Hydrogen Energy 39 (2014) 12667–12675.
39 (2014) 15589–15594. [56] E.A. Berdonosova, V.Y. Zadorozhnyy, M.Y. Zadorozhnyy, K.V. Geodakian,
[14] H. Emami, K. Edalati, J. Matsuda, E. Akiba, Z. Horita, Acta Mater 88 (2015) M.V. Zheleznyi, A.A. Tsarkov, S.D. Kaloshkin, S.N. Klyamkin, Int. J. Hydrogen
190–195. Energy 44 (2019) 29159–29165.
[15] C.H. Chiang, Z.H. Chin, T.P. Perng, J. Alloys Compd. 307 (20 0 0) 259–265. [57] Y. Li, H. Shang, Y. Zhang, P. Li, Y. Qi, D. Zhao, Int. J. Hydrogen Energy 44 (2019)
[16] H. Hotta, M. Abe, T. Kuji, H. Uchida, J. Alloys Compd. 439 (2007) 221–226. 4240–4252.
[17] M. Abe, T. Kuji, J. Alloys Compd. 446-447 (2007) 200–203. [58] P. Lv, Z. Liu, J. Mater. Res. Technol. 8 (2019) 5972–5983.

158

You might also like