Tritium Diffusion Pathways in γ-LiAlO2 Pellets Used in TPBAR: A First

Apr 18, 2018 - (23) Tritium diffusion and recovery processes have been studied experimentally in ceramic breeders irradiated with neutrons by Okuno an...
0 downloads 4 Views 3MB Size
Subscriber access provided by UNIV OF DURHAM

C: Energy Conversion and Storage; Energy and Charge Transport 2

Tritium Diffusion Pathways in #-LiAlO Pellets Used in TPBAR: A First-Principles Density Functional Theory Investigation Hari P. Paudel, Yueh-Lin Lee, David J. Senor, and Yuhua Duan J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.8b01108 • Publication Date (Web): 18 Apr 2018 Downloaded from http://pubs.acs.org on April 19, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Tritium Diffusion Pathways in γ-LiAlO2 Pellets Used in TPBAR: a FirstPrinciples Density Functional Theory Investigation

a

Hari P. Paudel a, Yueh-Lin Lee a, David J. Senor b, Yuhua Duan a* National Energy Technology Laboratory, United States Department of Energy, Pittsburgh, Pennsylvania 15236, USA b Pacific Northwest National Laboratory, Richland, Washington 99354, USA

Abstract With superior thermo-physical and thermo-chemical properties, γLiAlO2 has high compatibility with other blanket materials and is used in the form of an annular pellet in tritium-producing burnable absorber rods (TPBARs) to produce tritium by thermal neutron irradiation of 6Li. In radiation damaged γ-LiAlO2, different types of vacancies, defects of its constituent elements and other trapping sites hinder the diffusion process of tritium. In this study, the first principles density functional theory approach is used to study the diffusion mechanisms of tritium defect and its species, such as interstitial and substitutional tritium defects, oxygen-tritium vacancy defects, and interaction of tritium with oxygen-vacancies in defective and non-defective γ-LiAlO2. The obtained results provide an understanding of how such defects hamper the diffusivity and solubility of tritium. By calculating several different diffusion pathways, our results show that the smallest activation energy barrier is 0.63 eV for substitutional tritium diffusion with a diffusion coefficient of 3.25x10-12 m2/s. The smallest oxygentritium diffusion barrier is found to be 2.17 eV which is around 3.5 times higher than tritium diffusion barrier alone.

*

Author to whom correspondence should be addressed. Tel. 412-386-5771, Fax 412-386-5990, E-mail: [email protected]

ACS Paragon Plus Environment

1

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 33

1. INTRODUCTION

Lithium aluminate (LiAlO2) is a wide band gap material currently being developed by researchers for different applications. It has been used as a suitable substrate for GaN epitaxial growth,1 coating in lithium (Li) electrodes

2-3

and as an additive in composite Li electrolytes.

4

Li

containing ceramics have been in the center of study in tritium science and technology as breeding materials owing to their superior thermo-physical and thermo-chemical properties. Most importantly, the high temperature phase (γ) of LiAlO2, has been a leading candidate as a tritium (3H, or T) breeding blanket for deuterium-tritium (D-T) fusion reactors.5-11 It has an excellent irradiation behavior at high temperature and is more swelling resistant than many other Li rich compounds. In addition to that, γLiAlO2 has high compatibility with other blanket materials, and displays required physical and chemical stability during operation at elevated temperature.12-13 While operating at higher temperature most of the other ceramic materials do not meet cumulative requirements of low vapor pressures, high melting point and acceptable tritium release behavior.9 The study of Li diffusivity in Li containing ceramics has also been a subject of interest in Li-ion battery and solid oxide fuel cells. Several experimental

14-16

and theoretical

17-21

efforts have been

made in attempts to understand Li diffusion mechanisms in γ-LiAlO2 and other ceramics. One of the powerful experimental techniques to understand atomic diffusion in crystals is nuclear magnetic resonance (NMR) measurement as a function of temperature. With the known value of spin-lattice relaxation rate from NMR measurements, it is possible to calculate the ionic jump rate (τ-1) that allows us to further calculate diffusion coefficients using the Einstein−Smoluchowski equation.22 Using NMR and dc conductivity measurements in γ- LiAlO2, it was revealed that the Li diffusion coefficient is in the range 10-20 to 10-13 m2/s over the temperature range of 400 K to 1000 K and the activation energy

ACS Paragon Plus Environment

2

Page 3 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

was found to vary between 0.74 to 1.14 eV.23 Tritium diffusion and recovery processes have been studied experimentally in ceramic breeders irradiated with neutrons by Okuno and Kudo.

24-25

They

found that tritium produced in the ceramic materials (Li2O, γ-LiAlO2, Li2SiO3, Li4SiO4, Li2ZrO3, Li8ZrO6) irradiated with neutrons was released mainly in the chemical form of tritiated water (HTO) when heated in vacuum.25 Employing density functional theory (DFT), Islam and Bredow calculated Li activation energy in γ-LiAlO2 to be 0.65 eV, 1.65 eV and 1.41 eV, respectively for the jump between the 1st, 2nd and 3rd nearest neighbor (NN).17 Understanding of Li diffusion in ceramic blanket material is crucial to address properly the tritium release behavior and performance of the material because tritium diffusivity can be correlated with the Li hoping mechanism. The 6Li is one of the stable isotopes of Li. It naturally occurs in an amount of around 7 % in Li containing ceramics such as γ- LiAlO2 with rest being 7Li. The 6Lienriched γ- LiAlO2 (6Li enrichment in the 25-30% range) ceramics are used in the form of annular pellets and located in between the zircaloy-4 liner and nickel-plated zircaloy-4 tritium getter in tritiumproducing burnable absorber rods (TPBARs).13,

26-27

When irradiated in a pressurized water reactor

(PWR), the 6Li absorbs neutrons, simulating the nuclear characteristics of a burnable absorber rod, and produces 3H along with alpha particles and large amount of energy. The reaction is given as Li + n →  H

+ α + 4.8 MeV. The initial kinetic energy of 3H is, therefore, a few MeVs and is expected to

thermalize as an impurity after moving several microns of distance in the bulk crystal. Significant radiation damage can also occur due to the neutron flux, and the alpha particles that sweep through the crystal.19,

28-29

After leaving the ceramic pellets, any tritiated water is reduced by liner and the 3H

chemically reacts with the metal getter, which upon capture of 3H becomes a metal hydride. In the ceramic pellets, there can be Li vacancies (VLi), O vacancies and other trapping sites such as F+ centers after the damage by radiation that can hinder the diffusion process of 3H. A recent molecular dynamics simulation on defect accumulation and ion transport in LiAlO2 revealed that at 573 K a rapid lattice disorder takes place at a dose between 0.1 and 0.2 displacements per atom leading to

ACS Paragon Plus Environment

3

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 33

amorphization of the lattice and an increase in Li diffusivity by 30 times.19 Shah et al. performed ab initio calculations to understand the temperature dependence of the 3H diffusion mechanism in Li2O, and predicted that an upper limit of 3H diffusion barrier is 0.45 eV corresponding to interstitial 3H diffusion.30 However, the 3H transport through the ceramic pellets and the barrier/cladding system is still hampered by a lack of fundamental data such as hydrogen isotope solubility and diffusivity. In addition to that, it is still unclear whether 3H exists as an interstitial or a substitutional defect after it recoils losing reaction induced kinetic energy as heat. Despite the availability of Li diffusivity data, so far, the 3H diffusion mechanism in the ceramic pellets is not well understood. In order to identify the mechanisms associated with atomic  T formation, diffusion, transport, deposition, and the kinetics in TPBAR pellets, here in this study, we present the results of firstprinciples DFT calculations of interstitial and substitutional 3H defects, hydroxide (OT) vacancy defects, interaction of 3H with O-vacancies in ceramic γ-LiAlO2, and provide an understanding of how such defects hamper the diffusivity and solubility of 3H. This study focuses on the diffusion of 3H and its species such as T2O in bulk γ-LiAlO2 in presence of Li and O defects. In particular, we focus on the diffusion of substitutional and interstitial 3H, its species (such as T2O) and Li diffusion pathways in pristine and defective γ-LiAlO2 crystal. By considering several different diffusion pathways, we examine possible activation energy barrier of 3H, and O and 3H as a single entity. Our results show that the smallest activation energy barrier corresponds to the substitutional 3H diffusion with barrier height of 0.63 eV. The smallest OT diffusion barrier is found to be 2.17 eV which is three times higher than the energy barrier of 3H diffusing alone.31 We analyze diffusion pathways for each case in detail, and provide the values of corresponding diffusion coefficients. The other non-bulk pathways such as grain boundary diffusion, diffusion through line or planar defects such as dislocation and stacking faults will be subjects of further works.

ACS Paragon Plus Environment

4

Page 5 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

2. THEORETICAL METHODS

To perform the calculations, we use first principles DFT approach with plane-wave basis sets and the pseudopotential to describe the electron-ion interactions. The Vienna ab initio Simulation Package (VASP).32-33 was employed to calculate the electronic structures and diffusion pathways of lithium aluminate. In this study, all calculations have been done using the projector augmented wave (PAW) pseudo-potentials and the Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional. 3435

Plane wave basis sets were used with a cutoff energy of 520 eV and a kinetic energy cutoff for

augmentation charges of 605.4 eV. For higher cut-off energy, lattice parameters are changed by less than 0.02 Å suggesting that an acceptable total energy convergence within 0.01 eV has been met. The reciprocal space integration is performed with the Monkhorst-Pack grid of 7x7x7 for a unit cell structure. Energy convergence criterion of 10-6 eV per cell has been achieved with the MonkhorstPack grid of 7x7x7. Diffusion calculations were performed using a 2x2x2 supercell (corresponding to Li32Al32O64 system) for which a 2x2x2 Monkhorst-Pack grid was used to sample the Brillouin zone. The climbing-image nudged elastic band approach (cNEB) was used to map the diffusion pathways among different local minima. 36-37 To catch the effects of 6Li and 3H isotopes on the lattice dynamics and thermodynamic properties, the pseudo-potentials of 6Li and 3H are obtained by modifying the masses in the standard 7Li and 1H pseudo-potentials. Obviously, under the standard DFT scheme and the pseudo-potentials used in this study, the isotope does not affect the electronic structural properties as presented in following sections. In this study, we focus on the neutral defects. The charge defects are not included. In that case, the energy correction term contributed by charged defects in a periodic supercell vanishes 38.

ACS Paragon Plus Environment

5

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 33

3. RESULTS AND DISCUSSION

Using the powder neutron and single crystal X-ray techniques, the crystal structure of γ-LiAlO2 has been shown to have a tetragonal symmetry with space group P41212.39 It has four formula unit in a unit cell structure as shown in Fig. 1(a). As summarized in Table 1, the calculated values of symmetry positions agree well with the experimentally reported data.39 The optimized lattice parameters a and c are found to be higher by 0.05 Å and 0.04 Å, respectively, than the experimental values of 5.168 Å and 6.267 Å. The volume of the unit cell structure is calculated by fitting with the Birch-Murnaghan equation of state,40 and is obtained as 171.89 Å3, corresponding to minimum energy of -106.314 eV/unit cell. The calculated value of the volume is within 4 % of the experimental value of 167.45 Å3 as shown in the Table 1. A three-dimensional (3D) structure consists of periodic arrangement of slightly distorted tetrahedra XO4 (X = Li, Al) as viewed along c- direction. Each of the two Li-O bonds in LiO4 tetrahedra and Al-O bonds in AlO4 tetrahedra lying in the same plane is equal. The calculated values of the Li-O bond lengths are 1.961 Å (out of plane) and 2.07 Å (in plane), and that of Al-O bond lengths are 1.77 Å (out of plane) and 1.78 Å (in plane). In both cases, our calculated values are within 1 % of the experimentally reported values.39 The six vertices of XO4 tetrahedra form a distorted 3D hexagonal geometry around the center of the unit cell as shown in Fig. 1(a). The distorted 3D hexagonal geometry consists of two opposite facets of nearly rectangular shape with different metal tetrahedra sharing one edge diagonally as shown in the Fig. 1(b). The calculated value of the shared edge is 2.75 Å, which is well in agreement with the experimental value of 2.73 Å.

ACS Paragon Plus Environment

6

Page 7 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 1. Unit cell of γ –LiAlO2 (a), the shared edges of XO4 (X = Li, Al) tetrahedra (green color-Li tetrahedra, and light blue color-Al tetrahedra) are highlighted with yellow color and lie opposite facet of the 3D hexagonal structure (b).

Table 1. Structural lattice parameters, bond lengths, bond angles, and symmetry positions of γ –LiAlO2 (see Figure 1). Experimental values are from ref.39. tetragonal Crystal structure P41212 Space group calculated experiments Lattice parameters (Å) a 5.224 5.169 c 6.297 6.267 Volume (Å3) 171.89 167.45 LiO4 tetrahedron Bond lengths (Å) O1-Li1, O2-Li1 2.070 2.059 O4-Li1, O3-Li1 1.961 1.947 AlO4 tetrahedron O7-Al1, O5-Al1 1.784 1.766 O6-Al1, O4-Al1 1.771 1.755 Bond angles (Degrees) O-Li-O 111.90 110.6 O-Al-O 109.68 110.7 Li-O-Al 114.96 115.6 Li-O-Li 101.10 100.9 Al-O-Li 124.40 124.6 Symmetry position Li 4a (0.8134, 0.8134, 0) (0.812, 0.812, 0.0) Al 4a (0.1761, 0.1761, 0) (0.175, 0.175, 0.0) O 8b (0.3379, 0.2916, 0.7717) (0.3369, 0.290, 0.771)

ACS Paragon Plus Environment

7

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 33

Neutron irradiation on γ-LiAlO2 can induce several types of defects and vacancies. An important question is to understand how 3H diffuses in the defected crystal. When a Li atom is substituted by 3H as a substitutional defect (Ts), it makes a hydroxide type of bond with a nearby O atom. This is the most stable position of the Ts. The calculated distance of the OT bond in this case is 0.98 Å and is directed toward the VLi. As Ts is associated with the VLi, even a small rotation of OT bond around O increases the repulsive energy. We find that rotation of OT bond in (101) plane by 25° increases the repulsive interaction by 0.2 eV. Similarly, an interstitial 3H (Ti) and O bond has no rotational symmetry around O, and therefore, Ti cannot freely move around O unlike one found in case of Li2O system.30 Here, we have three species of atoms that create a less symmetric environment of force field in the crystal. Based on Figure 1, we created interstitial and substitutional tritium defects, hydroxide (oxygen-tritium) vacancy defects, and interaction of tritium with oxygen-vacancies in γ- LiAlO2, then explored the 3H diffusion pathways through these defective structures. In Figure 2(a), we show the creation of vacancies by removing Li atom at the 1st, 2nd and 3rd nearest neighbor (NN) positions (yellow spheres) which are located at distances of 3.12, 4.18 and 4.94 Å, respectively from the reference Li atom marked with yellow cross lines. Fig. 2(a) also shows the geometry of a hexagon marked by blue lines. The 1st NN vacancy is located at the adjacent corner of the hexagon, whereas the 2nd NN vacancy is located in the diagonally opposite corner of the hexagon. The 3rd NN vacancy is located in the neighboring hexagon on the upper right. The VLi at 1st, 2nd and 3rd NN position is created one by one. When a 3H atom is substituted in the VLi, position, 3H atom moves by a distance of 0.9 Å, and form a bond with the nearby O atom as shown in Fig. 2(b). In this case, OT bond is aligned toward VLi. In the following sections, we replace a Li atom by a 3H atom in the different NN position in step, and analyze the diffusion barriers and the pathways.

ACS Paragon Plus Environment

8

Page 9 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 2. Lithium vacancies (VLi) are located at the 1st, 2nd and 3rd nearest neighbour (NN) positions, respectively, at distance of 3.12, 4.18 and 4.94 Å from a reference Li atom marked with the yellow cross lines (a). The 1st and 2nd NN VLi lie in the same hexagon whereas the 3rd NN VLi lies in the adjacent hexagon. A substitutional 3Hatom is shifted by 0.9 Å from the VLi position, and is shown to make a bond with nearby O atom. In this case, OT bond is aligned toward the VLi. The colour code of each atom is as shown in bottom of the figure.

To minimize the effect due to size of supercell, we calculate the vacancy formation energy by removing a Li atom form supercell of varying sizes 1x1x1, 2x2x2 and 3x3x3. By doing that, we vary defect concentration from ¼ to 1/108 of Li content in the supercell. Defective supercells are generated from the fully relaxed and optimized non-defective supercells coordinates in each case. To calculate the vacancy formation energy, we use the expression ∆E  E V + ELi  E , where ∆E, E V, ELi, and E are vacancy formation energy, total energy for supercell with one Li atom removed, total energy of Li atom in bulk lithium in a body-centered cubic phase, and total energy of non-defective supercell, respectively. The value of ∆E for neutral Li defect converges within 0.03 eV for supercell of size 2x2x2 Rest of the calculations for diffusion mechanisms are performed by taking 2x2x2 supercell. ACS Paragon Plus Environment

9

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 33

A. Substitutional 3H (Ts) migration The analysis of migration properties of 3H at low levels of tritium concentration (3.12 %) can provide a valuable insight, and can be directly correlated with future experimental observations on 3H diffusion properties. This has been done by substituting a Ts in a VLi position and creating another VLi in a neighboring location in the 2x2x2 supercell (3.12 % concentration) containing a total of 128 atoms as shown in Fig. 3. In fact, the defective supercell contains two VLi, one of which is substituted by a 3H atom to form substitutional 3H migration (Ts). In Fig. 3 (a), (b) and (c), we show a substitution of 3H atom and a creation of VLi at the 1st, 2nd and 3rd NN positions in 2x2x2 supercells. The Ts associated with a particular VLi in one of the LiO4 tetrahedra in the distorted hexagonal structure can migrate to another empty LiO4 tetrahedron either in the same hexagonal structure or in the one laying adjacent to it. In the case of Ts migrating to 1st NN, the OT bond is pointing away from VLi in the initial state whereas its aligned toward VLi in the final state, as can be seen from Figure 3 (a) and (d). In the final state, Ts forms a OT bond with an O atom associated with the Al tetrahedron within the same hexagon. Similarly, Figure 3 (b) and (c) show the initial and final state of the Ts migrating to 2nd NN VLi position. In this case, Ts migrates from one O atom to the other O atom both are bonded with the same metal ion of the hexagon. The migration pathway is found to be curved while Ts migrates to the 3rd NN VLi position as shown in Figure 3 (c) and (f) as the migration happens from one hexagon to the other adjacent to the first one.

ACS Paragon Plus Environment

10

Page 11 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 3. Initial (upper panel) and final (lower panel) states of Ts diffusion to the 1st (a, d) 2nd (b, e) and 3rd (c, f) NN VLi position in 2x2x2 supercells. In (a), Ts is diffusing to the1st NN VLi located at the adjacent corner of the hexagon. In (b), Ts is diffusing to the 2nd NN VLi located at the diagonally opposite corner. In (c), Ts is diffusing to the 3rd NN VLi located in the neighboring hexagon. Hexagons are marked with the blue lines in final states (d), (e) and (f). The vertical arrow on the upper left side shows the direction of c- axis. When Ts moves to VLi in the 1st NN position, the most probable pathway is to migrate along the facet of the distorted hexagon. In this case, we calculate the activation energy barrier (E ) to be 0.63 eV as shown in Figure 4 (a). An energy barrier of 0.65 eV for Li migration in γ-LiAlO2 has been reported.17 In Figure 4 (b) and (c), we also show E when Ts migrates to VLi position located at the 2nd and 3rd NN position whose optimized distances are 4.18 Å and 4.94 Å, respectively from the reference VLi (this is the location where we substitute 3H). The E for the 3H migration to the 2nd NN VLi position is found to be 1.51 eV. In the first step, Ts migrates to the adjacent Al tetrahedron within the same hexagon and forms a OT bond with O atom lying on the upper facet (see Figure 3 (b) and (e)). In the second step, it is attached to the O atom lying on the lower facet in the same Al ACS Paragon Plus Environment

11

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 33

tetrahedron. Finally, it migrates to VLi lying on the opposite corner and stabilizes with the OT bond that aligns toward the vacancy. The peak in the barrier, as shown in Figure 4 (b), while migrating to 2nd NN VLi position arises as the 3D hexagon further distorted due to an inward pulling of OT bond. During the Ts diffusion, the OT bond rotates by an angle of 48° in the (101) plane around O, a process that requires an energy of 0.85 eV. The migration pathways of Ts diffusing to the 3rd NN VLi position is more curved as compared to the diffusion to the 1st and 2nd NN VLi positions. It passes between two adjacent hexagons, as shown in Figure 3 (c) and (f). It is to be noted that there are crystal sites of higher repulsion in between two adjacent hexagons through which Ts moves while migrating to 3rd NN VLi position. The OT bond rotates by 41° before it aligns toward the vacancy with a stable configuration in the final state as shown in Figure 4 (c).

Figure 4. Activation energy barriers and transition pathways for Ts migration to the 1st (a), 2nd (b), and 3rd (c) NN VLi position for which the distances are 3.12, 4.18 and 4.94 Å, respectively. In the inset of each figure we present the transition states. The purple arrow indicates a transition direction in each initial state. In the Figure, we also show the tetrahedra around each Li vacancy.

ACS Paragon Plus Environment

12

Page 13 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

As one can see from Fig. 4, the Ts migration pathway to the 1st NN VLi position is found to be the most probable pathway among the pathways discussed above. The activation energy barrier for 3H migration to the 1st NN VLi position is 0.63 eV which is 0.88 and 0.23 eV smaller than the migration to the 2nd NN and 3rd NN VLi positions, respectively. We have summarized the activation energy barriers for Ts diffusion to different VLi NN sites in Table 2. Okuno and Kudo found that about 94 % of tritium produced in ! LiAlO2 and Li2O is released in the chemical form of tritiated water (HTO).10,

25-26

They found that HTO released is controlled by T+ diffusion in the material and reported that the activation energy for T+ is 0.94 eV in ! LiAlO2 at temperature range of 630 to 930 K. This value of activation energy is higher by 0.31 eV than our calculated value at 0 K.

Table 2. Activation energy (Eact) for a substitutional 3H Nearest Neighbour (NN) 1st NN 2nd NN 3rd NN Distance (Å) Eact (eV)

3.12 0.63

4.18 1.51

4.94 0.86

Exp. 25 (630–930 K) 0.93

B. Interstitial 3H (Ti) migration In Figure 5 (a), we present a defective supercell with a 3H substituted Frenkel pair (Ti + VLi). It also shows Ti migration pathways. In the case of Ti migration with a Frenkel defect, Ti passes through the center of the hexagon which is a site of higher symmetry. In this case, one of the facets of the hexagon is distorted due to a Li atom repelled in the [-100] direction by the migrating Ti with distance of 0.96 Å, as shown in Figure 5 (a). As Ti passes through the center, first O moves inward as it forms a bond with Ti stretching O-Li bonds by around 1 Å, and then the OT bond weakens as Ti moves further. This increases a repulsive interaction of Ti with neighboring Li atoms giving rise to a peak barrier of 1 eV (peak A in Figure 6. In the next step, Ti migrates to the next O atom within the same Li tetrahedron. It migrates out of the distorted hexagon as shown by the shaded portion in the transition image in the Figure 6 (a) where it experiences a higher repulsion. As the Ti moves through shaded ACS Paragon Plus Environment

13

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 33

portion in Figure 6, OT bond is weakened, and bond length is increased by 0.15 Å. This results in an increase of the activation energy barrier as indicated by the peak B with a barrier height of 0.62 eV. In Figure 5 (a) shows hexagonal geometries overdrawn on the crystal structure, which are helpful to understand the migration pathways of Ti. It is seen that Ti is migrating from one hexagonal site to the other hexagonal site adjacent to the first.

Figure 5. A 3H atom is located in the interstitial position in the defective supercell with one Li defect (a). In the initial states, 3H forms a bond with nearby O atom repealing the Li atom (original Li position is shown by a yellow sphere with cross lines) by a distance of 0.96 Å. The purple arrows represent the diffusion pathways. Related hexagons in the pathways are patterned. Final state (b) is stable with OT bond align towards the vacancy. The vertical arrow on the upper left side shows the direction of c-axis.

ACS Paragon Plus Environment

14

Page 15 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 6. Interstitial migration of Ti in a supercell with one Frenkel pair. Ti is migrating from one hexagonal site to the other hexagonal site in adjacent to the first. The shaded portion with orange color in the transition state image is a crystal site with higher repulsion.

C. Li-3H correlated migration Next, we proceed to calculate the migration pathways for the correlated motion of Li and Ts. This can be studied by swapping Li-Ts positions simultaneously. In Figure 7 we show supercells with Ts and migration pathways when it swaps position with a Li atom located at the 2nd (Figure 7 (a)) and 3rd (Figure 7 (b)) NN. To study their migration pathways, a VLi is created when the Li atom moves out from its original site. Then, Ts starts its migration to the VLi. The repulsion between two migrating ions can lead to a structural deformation if both Li and Ts start migrating at the same time from their original sites. Therefore, we assume that Ts migration is delayed following the Li migration so that the repulsive interaction between migrating ions is reduced. The pathways for such concerted migration are found to be highly curved as to reduce the repulsion (Figure 7 (a) and (b)).

ACS Paragon Plus Environment

15

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 33

Figure 7. Initial (upper panel) and final (lower panel) states of Ts migration by swapping position with a Li atom located initially at the position of 2nd (a, c) and the 3rd (b, d) NN in a 2x2x2 supercell. The migration pathways are indicated by dotted blue arrows for Ts and dotted green arrows for Li atom. The Li atom marked by yellow cross lines in each Figure is a swapping pair of Ts, and it migrates simultaneously with Ts. In Figure 8 we present the energy profile for a Li and a Ts swapping their positions when 3H is substituted in the 2nd and 3rd NN VLi positions. Figure 8 also shows the transition image of each migrating ion and the local atom environment. First, the Li moves, and a vacancy is created in each case, and then Ts migrates to that VLi position. In the case of 2nd NN migration as shown in Figure 8 (a), the Ts migrates by forming a OT bond that lies out of the hexagon while the migrating Li takes pathways through the middle of the hexagon. The calculated activation energy barrier is 1.87 eV. This energy barrier is higher by 0.77 eV than the Ts swapping position with the Li atom at the 1st NN (not shown in the Figure). The pathway is curved with a Ts diffusing outside of the hexagonal crystal site while Li is migrating diagonally though the middle of the hexagon. Other possibilities would be Ts taking a diagonal pathway while Li move by a curved path. This can further increase the repulsive ACS Paragon Plus Environment

16

Page 17 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

interaction. Similarly, in the case of Ts migration to the 3rd NN Li vacancy, a VLi is created due to the Li atom diffusing first. Again, due to a repulsion between migrating ions, Ts takes curved pathways while the Li atom diffuses through center of the hexagon as shown in Figure 8 (b). We calculate the activation energy barrier in this case to be 1.51 eV, which is smaller by 0.36 eV than the Ts swapping position with the 2nd NN Li atom (Figure 8 (a)) and higher by 0.33 eV than the Ts swapping position with the 1st NN Li atom.

Figure 8. The schematic of activation energy barrier for Li and T swapping their positions at (a) 2nd NN (a), and 3rd NN (b) Li vacancy positions. The activation energy barriers are 2.78 eV (a), and 1.51 eV (b). The migration pathways are shown with the dotted curves.

D. OT migration So far, we have discussed the defects and diffusion mechanisms for substitutional, interstitial, and Li-3H correlated migrations. The other important type of vacancy created by radiation damage in the crystal is a bound O and Li vacancy (VOLi). These vacancies can be occupied by migrating hydroxide (OT) from neighbor sites depending on the activation barriers. To study the OT diffusion ACS Paragon Plus Environment

17

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 33

mechanism and explore the pathways, we perform calculations for O and substituted T (Ts) migrating as a bound entity to the 1st, 2nd and 3rd NN VOLi positions (It is to be noted that NNs are defined with respect to the original position of Li atom). As demonstrated in Figure 9, we first create a pair of VLi in a pristine 2x2x2 supercell. One of the VLi is substituted by 3H. Then, we create a VO by removing the O atom bonded with the remaining VLi. Therefore, our initial state in the defective supercell consists of a Ts bonded with the O atom and a Li and O vacancy (VOLi). We calculate the diffusion pathways of the bound entity OTs for three cases: diffusing to VOLi located at the 1st 2nd and 3rd NN position, as shown in Figures 9 and 10. Figures 9 (a), (b) and (c) show the schematic of the initial state and the migration pathways of OTs, and Figures 9 (d), (e), and (f) show the corresponding final states. Obviously, these migration mechanisms involve rotation of the hydroxyl bond.

Figure 9. Schematic for Li and O vacancy creation and OTs diffusion pathways to the 1st (a, d), 2nd (b, e) and 3rd (c, f) NN vacancy positions in 2x2x2 supercells. Pathways are shown by the dotted lines with arrow.

ACS Paragon Plus Environment

18

Page 19 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

We present the migration energy profiles of OTs along with the transition state images locally as shown in Figure 10. The activation energy barrier is found to be 2.17 eV for the OTs migration to the VOLi located at the 1st NN position (Figure10 (a)), while it is 4.23 eV and 3.44 eV, respectively for the migration to the VOLi located at the 2nd (Figure 10 (b)) and 3rd NN (Figure10 (c)) positions. It should be noted that while OTs migrating to the VOLi at the 2nd NN position, the barrier is higher than that for the migration to the VOLi at 1st and 3rd NN positions. This is because OTs passes through the higher symmetry sites (around the center of hexagon) along with a rotation of the OTs bond. As OTs migrates to VOLi located at the 3rd NN position, it passes through the crystal sites between the two hexagons where it experiences a repulsive interaction from the nearby ions. This ionic repulsion slightly distorts the lattice and increases the activation barrier.

Figure 10. The OT diffusion to the bound Li and O vacancy (VOLi) at a) 1st NN, b) 2nd NN, and c) 3rd NN. The energy activation barriers are 2.17 eV, 4.23 eV, and 3.44 eV, respectively, in each above case. The Figures also show initial, transition and final states of OT migration.

ACS Paragon Plus Environment

19

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 33

E. Higher concentrations of Li defect and 3H The activation energy barrier for 3H diffusion can change with the increasing number of VLi and 3

H in the system. In experiments, irradiation can create randomly distributed Li vacancies, and after

recoil 3H can occupy either one of those vacancies or the interstitial sites. At thermodynamic equilibrium, the occupation probability of the vacancies and interstitial sites by atoms is directly −E / K T proportional to the Boltzmann factor, e , where KB is the Boltzmann constant and E is the B

formation energy, and the number of available vacancies and other empty sites around the diffusing atoms. The occupation probability can be affected by the interaction energy of diffusing atoms with the surrounding environment as occupancy can decrease with increasing repulsive interaction and vice versa. The other key factor is that a higher defect concentration can reduce the symmetry of the system. In a system with lower symmetry, interaction can be significantly increased due to anisotropy in force, leading to a change in the energy barrier for diffusion. It is computationally intensive in DFT to calculate with a higher concentration and large defective supercell size. It is possible to obtain an insight on the effect of increasing number of defects on the diffusion barriers and the pathways. The interaction energy between the defects quantifies the effect of increasing defect concentration. We calculate the energetics of transition states in the case of two Frenkel pairs (2VLi + 2Ti) in which both Ti are migrating simultaneously to VLi located at a distance of 3.12 Å apart (1st NN position). The energy barrier profile for two Ti migrations can be useful to understand the influence of migration of second Ti on the energy barrier for the first Ti. The initial state in this case is higher by 0.85 eV than single Ti migration as can be seen from the energy difference of the initial images in Figures 3 (a) and 12 (a). The activation barrier in case of a Ti pair migration with two Frenkel pairs is calculated to be 1.29 eV, which is 0.39 eV higher than one Ti migration with one Frenkel pair. When a

ACS Paragon Plus Environment

20

Page 21 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

pair of Ts which are 1st NN to each other diffuse to the VLi positions each of which is a 3rd NN from each Ts, the diffusion barrier is calculated to be 1.54 eV. This barrier is higher by 0.25 eV than the barrier for a Ti pair migration in two Frenkel pairs.

Figure 11. Schematic of diffusion pathways for Ti in a supercell with four Frenkel pairs. The diffusion pathways are shown with the dotted lines with arrow (a). The Ti diffuses to the nearby VLi and form OT bonds (b). To see the effect of increasing Li defects on the activation energy profile, we calculate the diffusion barrier for the recombination of four Frenkel pairs (4VLi + 4Ti) in a 2x2x2 supercell, as shown in Figure 11 (a). Each of the Ti diffuses to nearby VLi, the pathway for which has a lowest energy barrier or less repulsive (Figure 11 (b)). The calculated activation energy barrier is found to be 3.41 eV, as shown in Figure 12 (b). The images in Figure 12 (b) also show the recombination of four Frenkel pairs which are labeled as (aT, aV), (bT, bV), (cT, cV) and (dT, dV). The calculated energy barrier with four Frenkel pairs is higher by 2.41 eV for Ti diffusion in the case of one Frenkel pair and by 2.12 eV for Ti diffusion in the case of two Frenkel pairs (2Ti + 2VLi).

ACS Paragon Plus Environment

21

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 33

Figure 12. Interstitial migration of Ti in two (a), and four (b) Frenkel pairs. In case of four Frenkel pairs, for clarity recombining partners are labeled as (aT, aV), (bT, bV), (cT, cV) and (dT, dV), where aT recombines with aV, bT recombines with bV, cT recombines with cV and dT recombines with dV after migration. Purple arrows in the initial states are guide to eyes for Ti migration.

F. Diffusion Coefficient Using the Einstein-Smoluchowski relation, it is possible theoretically to calculate the diffusivity and probability of jump per second for 3H while diffusing to the nearest empty site.22 The diffusion coefficient can be written as "

#$ %&

(1)

where ' and τ are the average distance and time between two jumps, respectively, and c is a parameter with the value 2 in one-dimensional (1D), 4 in two-dimensional (2D) and 6 in threedimensional (3D) diffusion. The quantity ( is proportional to Boltzmann’s factor that gives a probability of jump from one site to other. It is to be noted that interaction between defects are

ACS Paragon Plus Environment

22

Page 23 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

neglected in the equation (1). This is valid in the dilute defect limit. Therefore, the diffusion coefficient can be written as − a2 D≈ R o ze 6

( E act + QV ) k BT

(2)

where R0 is attempt frequency, and z is coordination number. As an example, in 3D there are four 1st NN of a Li atom, and therefore, we take z=4.The quantities Eact and Qv are the energies for migration barrier and vacancy formation, respectively. The probability of jump can be obtained from

1 τ ≈ e Ro z

( E act

+ QV k BT

)

(3)

Suppose we create several vacancies in the crystal and mix throughout it. The change in entropy (∂S) per vacancy added, ∂Smix/∂Nν, due to creation of vacancies after mixing is proportional to a factor -ln(Nν/(1-Nν)), where Nv is the concerntration of vacancy. As Nν0, we get ∂Smix/∂Nν∞. This clearly tells that when the equilibrium is reached, there is always certain number of vacancies in the crystal. We assume that at equilibrium we have vacancies in LiAlO2 crystal and therefore, Qv = 0. This study is done at 0 K. At elevated temperature there could be a slight change in diffusion coefficient than we provide here due to contribution from the vibrational motion. The activation energy barrier is nearly an insensitive quantity with respect to temperature change up to certain temperature as shown in the work done by Mantina et al. 41 In Figure 13 we show the diffusion coefficient by assuming that we have a crystal with a Li vacancy (Qv = 0),42 and using the smallest energy barrier of 0.63 eV obtained for a substitutional 3H migrating to the 1st NN as shown in Table 3. We obtain about 1400 jumps/s at 300 K , whereas about 2 million jumps/s at 600 K. The diffusion coefficients at 300 and 600 K are 1.6 × 10- . and 3.25 ×

10- 2 m2/s, respectively. Our calculated diffusion coefficient for smallest activation energy barrier is two orders of magnitude smaller than the diffusion coefficient measured (D = 3 x 10-10 m2/s) in experiment by Okuno and Kudo in the temperature range of 630 to 930 K

ACS Paragon Plus Environment

25

. In this work, we only

23

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 33

focused on studying the bulk diffusion pathways. The porosity and non-bulk pathways, such as grain boundary diffusion and diffusion through line, could increase the apparent diffusion coefficient experimentally. As the calculated value of the activation energy barrier is smaller than the experimentally observed value, the discrepancy in the diffusion coefficient most probably is attributed to the difference in calculated and the observed jump rate. At higher temperature, the jump rate ( can differ significantly.

Figure 13. Diffusion coefficient calculated for the lowest energy barrier of 0.63 eV for LiAlO2. The diffusion coefficients along the y- axis are marked corresponding two temperatures, 300 K and 600 K.

In Table 3, we have given the calculated values of diffusion coefficients at 600 K for possible pathways that were investigated in the previous sections. For the Ts migration to VLi located at the position of the 2nd NN, the diffusion coefficient is found to be 6.45 × 10-23 m2/s, whereas the Ts migration to VLi located at the position of the 3rd NN it is found to be 3.74 × 10- 5 m2/s. This shows that diffusivity for Ts migration to the nearest neighbor can be ordered as 1st > 3rd> 2nd. The diffusivity

ACS Paragon Plus Environment

24

Page 25 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

for Ti interstitial migration is found to be 1.72x10-14 m2/s which is two orders of magnitude smaller than the diffusion coefficient for Ts migration to VLi at the 1st NN position and six orders of magnitude larger than Ts migration to the VLi at 2nd NN position, whereas it is in the same order of magnitude for Ts migration to the VLi at the 3rd NN position.

Table 3. Summary of 3H and OT activation energy barriers (Eact), diffusion coefficients (D), and migration distance (L) Type of diffusion Location D at 600 K L Eact of VLi (eV) (m2/s) (Å) st Substitutional 1 NN 3.12 0.63 3.25 × 10- 2 nd 2 NN 4.18 1.51 6.45 × 10-23 rd 3 NN 4.94 0.86 3.74 × 10- 5 Interstitial 6.30 0.9 1.72 × 10- 5 Li-T correlated migration 1st NN 3.12 1.18 3.82 × 10- . nd (swapping position) 2 NN 4.18 1.87 5.96 × 10-2 rd 3 NN 4.94 1.52 1.60 × 10-2 OT migration 1st NN 3.12 2.17 3.56 × 10-27 nd 2 NN 4.18 4.23 1.17 × 10-52 rd 3 NN 4.94 3.44 7.19 × 10- Two Ti migration L(VLi) = 3.12 Å 6.30 1.29 8.99 × 10- 8 (Two Frenkel pairs) Two Ts migration L(VLi) = 3.12 Å 4.94 1.54 3.53 × 10-23 Four Ti migration VLi in random 3.41 6.03 × 10- (4 Frenkel pairs) Exp. Data (630-930K) 25 3.0 x 10-10

For the Li-3H correlated migration, the diffusion coefficient as can be seen from Table 3 at 600 K, is found to be 3.82x10-17 m2/s when Ts swaps its position with a 1st NN Li atom. When Ts swaps position with a Li atom at the 2nd and the 3rd NN position, the obtained values are 5.96x10-23 and 1.60x10-21 m2/s, respectively. By comparing the diffusion coefficient of the correlated migration with those of substitutional and interstitial migration, it is seen that the correlated migration is less probable mechanism of diffusion at 600 K. Similarly, OT migration is found to be a highly less probable diffusion mechanism than those discussed above as the values of the diffusion coefficients for OT ACS Paragon Plus Environment

25

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 33

migrating to the VOLi position located at 1st, 2nd and 3rd NN are respectively, 3.56x10-25, 1.17x10-42, and 7.19x10-36 m2/s, which are much lower than other diffusion mechanisms studied here. Similarly, the diffusion coefficient for Ti migration simultaneously in the case of two Frenkel pairs with a distance of 3.12 Å between two VLi is found to be 8.99 × 10- 8 m2/s, which is four orders of magnitude smaller than the single Ti migration. When there are four Frenkel pairs and migrating Ti’s, the diffusion coefficient is found to be drastically reduced to 6.03 × 10- m2/s. The Table 3 also shows diffusion coefficient for a pair of Ts which are initially located at a distance of 3.12 Å to each other and diffusing to the VLi at the 3rd NN position. The diffusion coefficient is found to be 3.53 × 10-23 m2/s, which is two orders of magnitude smaller than the Ti migration in the case of two Frenkel pairs. There could be two possible limitations of using DFT to estimate diffusion barriers and diffusion coefficients: a) DFT results are more accurate at 0 K with experimental results obtained at low temperature. At elevated temperature, diffusion coefficient is expected to deviate due to contribution from the entropy term. However, temperature dependence of enthalpy and entropy is shown to be nearly cancelled out to each other.41 b) Due to inherent limitation in correlation treatment with DFT, there could still be error exists while obtaining the diffusion barriers by subtracting transition image energy from the final state energy. Nevertheless, our obtained result for diffusion barriers and diffusion coefficients provide good estimates, and understanding of main diffusion mechanisms that exist in radiation damaged and pristine LiAlO2 pellets.

4.

CONCLUSIONS

In this work, we studied the diffusion mechanisms of 3H and OT species in order to provide an understanding of the effects on diffusion due to the presence of interstitial and substitutional Li defects, hydroxide (O-T) vacancy defect, and of the interactions of 3H with O-vacancies in γ-LiAlO2. ACS Paragon Plus Environment

26

Page 27 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

By considering different possible diffusion pathways, we calculated the activation energy barrier of 3H as well as of O-T as a single entity while migrating as interstitial defects or as substitutional defects in a 2x2x2 supercell. Our results showed that the smallest activation energy barrier and the diffusion coefficient at 600 K are 0.63 eV and 3.25x10-12 m2/s, respectively, corresponding to the substitutional 3

H diffusion to the 1st NN Li vacancies in LiAlO2. The smallest energy barrier for OT migration was

found to be 2.17 eV corresponding to diffusion to the 1st NN Li vacancies. This diffusion coefficient in this case was calculated to 3.56x10-25 m2/s at 600 K. We also studied the correlated motion of 3H and Li by swapping their positions. We found that the smallest activation energy barrier is 1.18 eV and the diffusion coefficient is 3.82 × 10- . m2/s at 600 K, corresponding to Li-3H diffusion to the 1st NN position of VLi. We also studied the higher concentrations of the Li defects and 3H but found that the diffusivities to be several orders of magnitude smaller than single defect in supercell of the same size. Two major avenues can be chosen to explore further the diffusion mechanisms in blanket materials (such as γ-LiAlO2, Li2ZrO3, Li4SiO4, Li8ZrO6, etc.) First, it is still necessary to explore the effects of higher concentrations of Li, O and Al defects on the 3H diffusivity and solubility in γLiAlO2. After irradiation, due to large amount of kinetic energy of neutron beams, the recoiled helium nuclei can create defect rich regions within grain boundaries. These defects can trap or slow down the released molecular 3H which may have different diffusion mechanism. Second, it is essential to understand the surface and interface diffusion mechanisms of various chemical species such as 3H, OT-, and H2O. Once the 3H diffuses from the bulk to the surface/interface, surface chemistry plays a key role. The mechanisms such as adsorption/desorption and recombination/dissociation at given temperatures are key factors controlling formation of 3H. In addition to these, the lattice oxygen could become bonded to 3H, leading to formation of T2O or OT- species on γ-LiAlO2 surface, from where desorption can take place. These studies could provide a further understanding of the various physical and chemical processes occurring in the blanket materials, and hence assist in modelling and to improve the performance of TPBAR systems.

ACS Paragon Plus Environment

27

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 33

Acknowledgements This research was supported by the National Nuclear Security Administration (NNSA) of the U.S. Department of Energy (DOE) through the Tritium Science Research Supporting the Tritium Sustainment Program (TSP) managed by Pacific Northwest National Laboratory (PNNL). H. P. and Y.-L. L. thank the National Energy Technology Laboratory (NETL) Research and Innovation Center (R&IC) providing Computational resources administered by the Oak Ridge Institute for Science and Education (ORISE). We also thank Dr. Dan Sorescu for fruitful discussions and suggestions. This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency hereof.

ACS Paragon Plus Environment

28

Page 29 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

References 1.

2. 3.

4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

15. 16.

17. 18. 19. 20. 21. 22. 23.

Waltereit, P.; Brandt, O.; Trampert, A.; Grahn, H. T.; Menniger, J.; Ramsteiner, M.; Reiche, M.; Ploog, K. H., Nitride Semiconductors Free of Electrostatic Fields for Efficient White Light-Emitting Diodes. Nature 2000, 406, 865-868. Cao, H.; Xia, B. J.; Zhang, Y.; Xu, N. X., LiAlO2-Coated LiCoO2 as Cathode Material for Lithium Ion Batteries. Solid State Ionics 2005, 176, 911-914. Li, L. J.; Chen, Z. Y.; Zhang, Q. B.; Xu, M.; Zhou, X.; Zhu, H. L.; Zhang, K. L., A HydrolysisHydrothermal Route for the Synthesis of Ultrathin LiAlO2-Inlaid LiNi0.5Co0.2Mn0.3O2 as a HighPerformance Cathode Material for Lithium Ion Batteries. J. Mater. Chem. A 2015, 3, 894-904. Dissanayake, M. A. K. L., Nano-Composite Solid Polymer Electrolytes for Solid State Ionic Devices. Ionics 2004, 10, 221-225. Botter, F.; Lefevre, F.; Rasneur, B.; Trotabas, M.; Roth, E., Effects of Radiation on Lithium Aluminate Samples Properties. J. Nucl. Mater. 1986, 141, 364-368. Arons, R. M.; Poeppel, R. B.; Tetenbaum, M.; Johnson, C. E., Preparation, Characterization, and Chemistry of Solid Ceramic Breeder Materials. J. Nucl. Mater. 1982, 103, 573-577. Johnson, C. E., Tritium Behavior in Lithium Ceramics. J. Nucl. Mater. 1999, 270, 212-220. Johnson, C. E.; Noda, K.; Roux, N., Ceramic Breeder Materials: Status and Needs. J. Nucl. Mater. 1998, 258, 140-148. Nishikawa, M.; Kinjyo, T.; Ishizaka, T.; Beloglazov, S.; Takeishi, T.; Enoeda, M.; Tanifuji, T., Release Behavior of Bred Tritium from LiAlO2. J. Nucl. Mater. 2004, 335, 70-76. Okuno, K.; Kudo, H., Diffusion-Controlled Tritium Release from Neutron-Irradiated Gamma-LiAlO2. J. Nucl. Mater. 1986, 138, 210-214. Roux, N.; Tanaka, S.; Johnson, C.; Verrall, R., Ceramic Breeder Material Development. Fusion Eng. Des. 1998, 41, 31-38. Roux, N.; Johnson, C.; Noda, K., Properties and Performance of Tritium Breeding Ceramics. J. Nucl. Mater. 1992, 191, 15-22. Senor, D. J. Recommendations for Science and Technology in Support of the Tritium Sustainment Program, Pnnl-27216; PNNL: 2017. Wohlmuth, D.; Epp, V.; Bottke, P.; Hanzu, I.; Bitschnau, B.; Letofsky-Papst, I.; Kriechbaum, M.; Amenitsch, H.; Hofer, F.; Wilkening, M., Order Vs. Disorder-a Huge Increase in Ionic Conductivity of Nanocrystalline LiAlO2 Embedded in an Amorphous-Like Matrix of Lithium Aluminate. J. Mater. Chem. A 2014, 2, 20295-20306. Hu, Q. W.; Lei, L.; Jiang, X. D.; Feng, Z. C.; Tang, M. J.; He, D. W., Li Ion Diffusion in LiAlO2 Investigated by Raman Spectroscopy. Solid State Sci. 2014, 37, 103-107. Wiedemann, D., Nakhal, S., Rahn, J., Witt, E., Islam, M. M., Zander, S., Heitjans, P., Schmidt, H., Bredow, T., Wilkening, M., Lerch, M., Unravelling Ultraslow Lithium-Ion Diffusion in Gamma-LiAlO2: Experiments with Tracers, Neutrons, and Charge Carriers. Chem. Mater. 2016, 28, 915-924. Islam, M. M.; Bredow, T., Interstitial Lithium Diffusion Pathways in Gamma-LiAlO2: A Computational Study. J. Phys. Chem. Lett. 2015, 6, 4622-4626. Jacobs, J. P.; SanMiguel, M. A.; Alvarez, L. J.; Giral, P. B., Lithium Diffusion in Gamma-LiAlO2, a Molecular Dynamics Simulation. J. Nucl. Mater. 1996, 232, 131-137. Setyawan, W.; Senor, D. J.; Devanathan, R., Insights on Amorphization of Lithium Aluminate from Atomistic Simulation. J. Phys. Chem. C 2017, 121, 7635-7642. Murphy, S. T., Tritium Solubility in Li2TiO3 from First-Principles Simulations. J. Phys. Chem. C 2014, 118, 29525-29532. Murphy, S. T.; Hine, N. D. M., Point Defects and Non-Stoichiometry in Li2TiO3. Chem. Mater. 2014, 26, 1629-1638. Shewmon, P., Diffusion in Solids; Springer International: Switzerland, 2016. Indris, S.; Heitjans, P.; Uecker, R.; Roling, B., Li Ion Dynamics in a LiAlO2 Single Crystal Studied by Li7 NMR Spectroscopy and Conductivity Measurements. J. Phy.s Chem. C 2012, 116, 14243-14247.

ACS Paragon Plus Environment

29

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 33

24. Carrera, L. M.; Jimenez-Becerril, J.; Basurto, R.; Arenas, J.; Lopez, B. E.; Bulbulian, S.; Bosch, P., Tritium Recovery from Nanostructured LiAlO2. J. Nucl. Mater. 2001, 299, 242-249. 25. Okuno, K.; Kudo, H., Tritium Diffusivity in Lithium-Based Ceramic Breeders Irradiated with Neutrons. Fusion Eng. Des. 1989, 8, 355-358. 26. Jiang, W. L.; Zhang, J.; Edwards, D. J.; Overman, N. R.; Zhu, Z.; Price, L.; Gigax, J.; Castanon, E.; Shao, L.; Senor, D. J., Nanostructural Evolution and Behavior of H and Li in Ion-Implanted Gamma-LiAlO2. J. Nucl. Mater. 2017, 494, 411-421. 27. Jiang, W. L.; Zhang, J. D.; Kovarik, L.; Zhu, Z. H.; Price, L.; Gigax, J.; Castanon, E.; Wang, X. M.; Shao, L.; Senor, D. J., Irradiation Effects and Hydrogen Behavior in Hi and He Implanted Gamma-LiAlO2 Single Crystals. J. Nucl. Mater. 2017, 484, 374-381. 28. Okuno, K.; Kudo, H., Chemical-States of Tritium and Interaction with Radiation Damages in Li2O Crystals. J. Nucl. Mater. 1986, 138, 31-35. 29. Oyaidzu, M.; Takeda, T.; Kimura, H.; Yoshikawa, A.; Okada, M.; Munakata, K.; Nishikawa, M.; Okuno, K., Correlation between Annihilation of Radiation Defects and Tritium Release in Neutron-Irradiated LiAlO2. Fusion Sci. Technol. 2005, 48, 638-641. 30. Shah, R.; Devita, A.; Payne, M. C., Ab-Initio Study of Tritium Defects in Lithium-Oxide. J. Phys.Condens. Mat. 1995, 7, 6981-6992. 31. Paudel, H.; Lee, Y.-L.; Holber, J.; Sorescu, D. C.; Duan, Y. Fundamental Studies of Tritium Solubility and Diffusivity in LiAlO2 and Lithium Zirconates Pellets Used in TPBAR; DOE-National Energy Technology Laboratory, DOE/NETL-PUB-21464: 2017. 32. Kresse, G.; Hafner, J., Abinitio Molecular-Dynamics for Liquid-Metals. Phys. Rev. B 1993, 47, 558-561. 33. Kresse, G.; Hafner, J., Ab-Initio Molecular-Dynamics Simulation of the Liquid-Metal AmorphousSemiconductor Transition in Germanium. Phys. Rev. B 1994, 49, 14251-14269. 34. Blöchl, P. E., Projector Augmented-Wave Method. Phys. Rev. B 1994, 50, 17953-17979. 35. Perdew, J. P.; Burke, K.; Ernzerhof, M., Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865-3868. 36. Sheppard, D.; Terrell, R.; Henkelman, G., Optimization Methods for Finding Minimum Energy Paths. J. Chem. Phys. 2008, 128, 134106. 37. Sheppard, D.; Xiao, P. H.; Chemelewski, W.; Johnson, D. D.; Henkelman, G., A Generalized Solid-State Nudged Elastic Band Method. J. Chem. Phys. 2012, 136, 074103. 38. Freysoldt, C.; Neugebauer, J.; Van de Walle, C. G., Fully Ab Initio Finite-Size Corrections for ChargedDefect Supercell Calculations. Phys. Rev. Lett. 2009, 102, 016402. 39. Marezio, M., Crystal Structure and Anomalous Dispersion of Gamma-Lialo2. Acta Crystallogr. 1965, 19, 396-400. 40. Birch, F., Finite Elastic Strain of Cubic Crystals. Phys. Rev. 1947, 71, 809-824. 41. Mantina, M.; Wang, Y.; Arroyave, R.; Chen, L. Q.; Liu, Z. K.; Wolverton, C., First-Principles Calculation of Self-Diffusion Coefficients. Phys. Rev. Lett. 2008, 100, 215901. 42. Morgan, B. J., Lattice-Geometry Effects in Garnet Solid Electrolytes: A Lattice-Gas Monte Carlo Simulation Study. Royal Society Open Science 2017, 4, 170824.

ACS Paragon Plus Environment

30

Page 31 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure captions Figure 1. Unit cell of γ –LiAlO2 (a), the shared edges of XO4 (X = Li, Al) tetrahedra (green color-Li tetrahedra, and light blue color-Al tetrahedra) are highlighted with yellow color and lie opposite facet of the 3D hexagonal structure (b). Figure 2. Lithium vacancies (VLi) are located at the 1st, 2nd and 3rd nearest neighbour (NN) positions, respectively, at distance of 3.12, 4.18 and 4.94 Å from a reference Li atom marked with the yellow cross lines (a). The 1st and 2nd NN VLi lie in the same hexagon whereas the 3rd NN VLi lies in the adjacent hexagon. A substitutional 3Hatom is shifted by 0.9 Å from the VLi position, and is shown to make a bond with nearby O atom. In this case, OT bond is aligned toward the VLi. The colour code of each atom is as shown in bottom of the figure. Figure 3. Initial (upper panel) and final (lower panel) states of Ts diffusion to the 1st (a, d) 2nd (b, e) and 3rd (c, f) NN VLi position in 2x2x2 supercells. In (a), Ts is diffusing to the1st NN VLi located at the adjacent corner of the hexagon. In (b), Ts is diffusing to the 2nd NN VLi located at the diagonally opposite corner. In (c), Ts is diffusing to the 3rd NN VLi located in the neighboring hexagon. Hexagons are marked with the blue lines in final states (d), (e) and (f). The vertical arrow on the upper left side shows the direction of c- axis. Figure 4. Activation energy barriers and transition pathways for Ts migration to the 1st (a), 2nd (b), and 3rd (c) NN VLi position for which the distances are 3.12, 4.18 and 4.94 Å, respectively. In the inset of each figure we present the transition states. The purple arrow indicates a transition direction in each initial state. In the Figure, we also show the tetrahedra around each Li vacancy. Figure 5. A 3H atom is located in the interstitial position in the defective supercell with one Li defect (a). In the initial states, 3H forms a bond with nearby O atom repealing the Li atom (original Li position is shown by a yellow sphere with cross lines) by a distance of 0.96 Å. The purple arrows represent the diffusion pathways. Related hexagons in the pathways are patterned. Final state (b) is stable with OT bond align towards the vacancy. The vertical arrow on the upper left side shows the direction of c-axis. Figure 6. Interstitial migration of Ti in a supercell with one Frenkel pair. Ti is migrating from one hexagonal site to the other hexagonal site in adjacent to the first. The shaded portion with orange color in the transition state image is a crystal site with higher repulsion. Figure 7. Initial (upper panel) and final (lower panel) states of Ts migration by swapping position with a Li atom located initially at the position of 2nd (a, c) and the 3rd (b, d) NN in a 2x2x2 supercell. The migration pathways are indicated by dotted blue arrows for Ts and dotted green arrows for Li atom. The Li atom marked by yellow cross lines in each Figure is a swapping pair of Ts, and it migrates simultaneously with Ts. Figure 8. The schematic of activation energy barrier for Li and T swapping their positions at (a) 2nd NN (a), and 3rd NN (b) Li vacancy positions. The activation energy barriers are 2.78 eV (a), and 1.51 eV (b). The migration pathways are shown with the dotted curves. Figure 9. Schematic for Li and O vacancy creation and OTs diffusion pathways to the 1st (a, d), 2nd (b, e) and 3rd (c, f) NN vacancy positions in 2x2x2 supercells. Pathways are shown by the dotted lines with arrow. ACS Paragon Plus Environment

31

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 33

Figure 10. The OT diffusion to the bound Li and O vacancy (VOLi) at a) 1st NN, b) 2nd NN, and c) 3rd NN. The energy activation barriers are 2.17 eV, 4.23 eV, and 3.44 eV, respectively, in each above case. The Figures also show initial, transition and final states of OT migration. Figure 11. Schematic of diffusion pathways for Ti in a supercell with four Frenkel pairs. The diffusion pathways are shown with the dotted lines with arrow (a). The Ti diffuses to the nearby VLi and form OT bonds (b). Figure 12. Interstitial migration of Ti in two (a), and four (b) Frenkel pairs. In case of four Frenkel pairs, for clarity recombining partners are labeled as (aT, aV), (bT, bV), (cT, cV) and (dT, dV), where aT recombines with aV, bT recombines with bV, cT recombines with cV and dT recombines with dV after migration. Purple arrows in the initial states are guide to eyes for Ti migration. Figure 13. Diffusion coefficient calculated for the lowest energy barrier of 0.63 eV for LiAlO2. The diffusion coefficients along the y- axis are marked corresponding two temperatures, 300 K and 600 K.

ACS Paragon Plus Environment

32

Page 33 of 33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

TOC graphic

ACS Paragon Plus Environment

33