Académique Documents
Professionnel Documents
Culture Documents
Simulations
ABSTRACT
Molecular dynamics simulations are performed, modelling the initial charging of a Li-ion
nanobattery with a graphite anode and lithium hexaflourphosphate (LiPF6) salt dissolved in
ethylene carbonate (CO3C2H4) solvent as the electrolyte solution. The charging was achieved
through the application of continuous external electric fields simulating voltage sources.
A variety of force fields were combined to simulate the materials of the nanobattery, including
the solid electrolyte interphase, metal collectors, and insulator cover. Some of the force field
parameters were estimated using ab initio methods and others were taken from the literature.
We studied the behavior of Li-ions traveling from cathode to anode through the electrolyte
solution of different concentration; 1.15 M and 3.36 M.. A variety of force fields were
combined to simulate the materials of the nanobattery, including the solid electrolyte interphase,
metal collectors, and insulator cover. Some of the force field parameters were estimated using
ab initio methods and others were taken from the literature. Time-dependent variables such as
energy, temperature, volume, polarization, and mean square displacement are reported; a few of
these variables, as well as others such as current, resistance, current density, conductivity, and
resistivity, are reported as a function of the external field or charging voltage. A SEI layer is
added to the model to study the mechanism behind the diffusion of the Li-ion through the SEI.
As the battery is being charged, the depletion of Li atoms in the cathode and their accumulation
in the anode follow a linear increase of the polarizability in the solvent, until reaching a
saturation point after which the charging of the battery stops, i.e., the energy provided by the
external source goes to zero. The nanobattery model containing the most common materials of
a commercial lithium-ion battery is very useful to determine atomistic information that is
difficult or too expensive to obtain experimentally. Finally, the electrical properties calculated in
this model are in good agreement with the experimental calculations.
Introduction
Li-ion batteries are one of the most popular electrochemical systems. Li has one of the lowest
ionization energies in the Periodic Table, 5.39 eV1-2 and thus is the easiest to ionize among the
first 10 atoms; the atomic radii of Li-ion can be as low as 0.73 , almost half the radii of the
1
neutral atom. In 1976, Whittingham developed the LiTiS2 cell, the first attempt of a lithium-
metal battery3, but after 100 or 200 charge/discharge cycles at 100% capacity4, the lithium
electrode formed long dendrites across the electrolyte, causing internal short circuits. In 1977
Whittingham studied several alkali metal intercalates and found that LiTiS2 was the best cathode
to be used as host of Li5. In 1980 Goodenough6 developed a fully working, rechargeable Li-ion
battery, in which he used Li-metal for the anode, 1.15 M LiBF4 solution in propylene carbonate
for the electrolyte, and a layered lithium cobalt oxide (LiCoO2) cathode he developed; nowadays,
LiCoO2 is the most successful material for positive cathode electrodes7 with an initial discharge
capacity of 115 mAh/g, a 66% of capacity retention after 100 cycles at 0.2 C,8 they can yield up
to 30% of their maximum capacity at discharge currents of 10 mA/cm2 and at more moderate
rates of dischargecharge, the capacity practically remains constant over thousands of cycles.9
In 1991, Sony commercialized the lithium-ion battery, using anodes of coke.10 Since 1997, most
Li-ion battery manufactures have shifted to graphite and nowadays it is the most used anode
material in Li-ion cells. Its layered shape allows for the storage of lithium atoms during
charging, with the anode lithium reduction reaction 6C + Li+ + e- LiC6.11 Ethylene carbonate
(EC) is also the most popular electrolyte solvent used in Li-ion batteries because of its high
dielectric constant of 90.512-13 and high ionic conductivity when mixed with other additives to
increase performance, i.e., ethylene carbonate in a mole fraction of 0.98 at 293K mixed with 1-
ethyl-3-methylimidazolium bis[(trifluoromethyl)sulfonyl]imide 0.62 S/m,14 or a mixture of
propylene carbonate (10 vol %), ethylene carbonate (27 vol %) and dimethyl carbonate (63 vol
%) with LiPF6 1M at 293K 1.07 S/m.15 The presence of EC supports the formation of a stable
solid electrolyte interface (SEI)16 on the surface of graphite. During charging, around 75% of the
graphite converts to LiC6.17 The theoretical specific capacity of graphite is nF/3.6M mAh/g; thus
for LiC6, n = 1/6, F = 96,500 C/mol and M = 12.0107 g/mol, yielding 372 mAh/g.18 Graphite is
used as the anode material because of its reliability19 for commercial Li-ion rechargeable
batteries. The successful use of just a few materials for the anode, cathode, and electrolyte
solution requires a better understanding of atomistic details of the battery constituents in order to
improve them further and develop new materials. In this work, we present the atomistic
modeling of the charge of a full Li-ion nanobattery using classical molecular simulations and
supported in few cases by density functional theory (DFT) calculations. This study includes the
modeling of the anode, electrolyte, and solid electrolyte interface (SEI) as well as the effects of
an external electric field to simulate the charging of the cell. In the next few sections we explain
the methodology and models used, the results and discussion, and our conclusions.
Methodology
The Model
Classical molecular dynamics (MD) simulations, using the Large-Scale Atomic/Molecular
Massively Parallel Simulator (LAMMPS) program developed by Plimpton et al.,20 are performed
to calculate global properties and predict behavior of the anode and electrolyte solution of a
nanobattery (Figure 1a); as well as nanosensors, proteins and other specific nanotechnology
applications.21-26 The Verlet leapfrog algorithm is used to integrate Newtons equations of
motion.27-35 The initial simulation box size is 166.6 27.0 58.2 3 (Figure 1b), defined with
periodic boundary conditions (PBCs) in the three directions. As part of the electrolyte solution
there are 21 LiPF6 dissolved in 310 EC molecules, yielding a 1.15 molar (1.15M) electrolyte
2
solution. In addition, we have included the study of a solid electrolyte interface (SEI) formed
during the first charging cycles between the electrolyte and the anode and containing 1050 units
of Li2CoO3. The initial atomic electrolyte coordinates are generated by the Packing
Optimization for Molecular Dynamics Simulations (Packmol).36
`
(i) (ii) (iii) (iv) (v) (vi) (vii)
(a)
(b)
(c)
Figure 1. (a) Components of the nanobattery: (i) ethylene carbonate, (ii) lithium ion, (iii)
hexafluorophosphate, (iv) graphite (v) lithium cobalt oxide (vi) silicon oxide-like insulator, and (vii)
copper-like. (b) Li-ion battery uncharged and equilibrated at 293 K after an equilibration of 1 ns at 5 K,
followed by a rise of temperature to 293 K in 4 ns and an equilibration for 10 ns under the NVT
ensemble. The cathode (K) on the left and the anode (A) on the right are LiCoO2 and graphite,
respectively. They are connected to the metallic current collectors (gold). The electrolyte solution is
1.15M LiPF6 in ethylene carbonate. The anode, electrolyte, and cathode are isolated with a layer of a
3
hypothetical insulator with a stoichiometry of AX2. (c) Li-ion nanobattery (insulator hidden) during
equilibration at 293 K; during charge, oxidation Li Li+ + e- takes place in the blue box and the
reduction, Li+ + e- Li, in the red one.
Force Fields
The electrolyte solution is modeled with LiPF6 salt, which dissolves into Li+ ions and PF6
counter ions37 in ethylene carbonate (EC). The intramolecular interactions are taken from the
CFF93 force field;38 however, specific parameters, bonds, angles, dihedrals, and improper
dihedrals for EC and LiPF6 are obtained through quantum chemical calculations using Gaussian-
09 program39 with the B3PW91 functional40-42 and the cc-pVTZ basis set43-45 (Table 1). The
intermolecular force field parameters for Lennard-Jones (LJ) interactions between the atoms of
EC, Li in LiCoO2, PF6-, and graphene are taken from the UFF46 using their averaging rules for
pairs with different type of atoms (Table 2); however, some of the specific pair interactions were
adjusted instead of using the usual average quantities in order to prevent EC or PF6- from going
inside the anode or Li in LiCoO2 from going inside the electrolyte while no electric fields are
being applied. Mulliken populations were also obtained to represent charges for coulombic
interactions between atoms; they are justified because the potential calculated using Mulliken
populations matches very well with the one produced by natural bond orbital (NBO) charges and
also with the potential calculated directly solving the Schrodinger equation.47 On the other hand,
fluorine atoms, could form -holes,48-50 a noncovalent interaction between a covalently-bonded
halogen and a negative site; however, the fraction of covalent bonding in PF6 is only ~0.4 due
to the large differences in electronegatives of P and F atoms (2.1 and 4, respectively), thus the
possibility of -holes is negligible.
Table 1. Geometry Optimized Parameters for Ethylene Carbonate Obtained with the B3PW91/cc-pVTZ
Level of Theory.
Atom 1 Atom 2 Atom 3 Atom 4 Bond 1-2 () Angle 1-2-3 () Dihedral 1-2-3-4 ()
Ox Cs Os Os/Ch 1.186 124.8 180.0/-172.8
Os Cs Os Ch 1.354 110.3 7.2
Os Ch Ch Os 1.426 103.0 20.3
Os Ch Ch H 1.426 103.0 117.2/-96.8/138.0
Ch Ch H H 1.525 112.7 126.9
Cs Os Ch Ch 1.354 109.6 -17.2
Cs Os Ch H 1.354 109.6 102.6/-138.1
Os Ch H H 1.426 108.9 119.0
H Ch Ch H 1.090 112.7 20.8/-104.4/146.0
H Ch H 1.090 109.5
The anode is built from a pristine crystal of graphite. The 3-body Tersoff potential of Lindsay et
al.51 is used to model the interaction among carbons within a layer, while a Lennard-Jones 6-12
potential models the interlayer interactions. Interaction between graphite and the solvent
solution as well as all other nonbonded interactions are modeled with LJ 6-12 potentials. The
depth and length parameters are listed in Table 2. Other pairs of atoms not shown in Table 2
were taken from the UFF.46 Combination rules, 12 = (1 + 2 )/2 and 12 = 1 2 are used
to estimate the parameters for atom-pairs not listed in Table 2.
4
In addition to the electrolyte solution and anode, we used much simpler models for the current
collectors, cathode, and insulator cover as the focus of this work centers on the electrolyte
solution and anode. The cathode, which in this work is simply a source of Li-ions when an
electric field is applied, is modeled with the structure of lithium cobalt oxide7 (blue box, Figure
1b), with harmonic bond and angle energies Eb = kb(r-r0)2 and Ea = ka( - 0)2, respectively,
where kb = 40 kcal/mol2, ka = 139.354 kcal/mol(radian)2, r0 = 1.92 , and 0 = 90. The
current collectors, whose main function in these simulations is to avoid the shifting of the anode
and cathode due to PBC along the longitudinal direction, are modeled as an inert hypothetical
conductor material, (Figure 1b) with large depth energy of 230 kcal/mol in order to have a robust
structure. Similarly, to cover the nanobattery on the lateral sides, we use a hypothetical insulator
that prevents solvent molecules and ions from leaving the regions of interest. This hypothetical
insulator helps to maintain the periodicity along the two non-axial directions and responds to a
stoichiometry of SiO2 using the Tersoff potential.52 However, the interactions of Si and O with
other components of the nanobattery have been exaggerated such that the insulator can function
as a cover with the parameters shown in Table 2. The force fields for the SEI were taken from
the CFF93 force field38 for the carbonate (CO3) and the Li-O bond from the harmonic class 2
with r0 = 1.953 and the force constant was taken from the C-O bond of EC. The Mulliken
populations for the SEI were calculated using the NBO using the B3PW91 functional40-42 and the
cc-pVTZ basis set43-45 (Table 3).
5
(1) Hypothetical materials: eventually AX2 becomes SiO2 and M becomes Cu (anode side) or Al (cathode
site).
Table 3. Mulliken Populations and NBO Populations Obtained with the B3PW91/cc-pVTZ Level of
Theory.
Atom Q/e
Ox(EC) -0.46 a
Cx(EC) 0.81 a
Os(EC) -0.45 a
Ch(EC) -0.12 a
H(EC) 0.20 a
Li+( LiPF6) 1.00 a
P(LiPF6) 1.34 a
F(LiPF6) -0.39 a
C(Li2CO3) 0.84b
O(Li2CO3) -0.28b
Li(Li2CO3) 0.00
a
Mulliken Populations
b
NBO Populations
Simulations
After a geometry minimization, all MD simulations included a temperature equilibration at 5 K
for 1 ns under the NVT ensemble53 to eliminate any hot spots in the initial geometry.
Temperature is maintained using the Nose-Hoover thermostat54. After this initial equilibration at
5 K, the temperature is increased from 5 K to 293 K in 4 ns and then is equilibrated at 293 K for
10 ns under the NVT ensemble. After the second equilibration, several values of electric field
are applied along the longitudinal direction of the nanobattery under the NPT ensemble. During
charging, Li-ions travel from the cathode to the anode due to the action of the external electric
field. Once a Li-ion arrives at the anode, it is reduced and a Li atom is oxidized in the cathode at
the same time, keeping the molarity of Li-ions in the electrolyte constant. The reduction takes
place in the red box and the oxidation in the blue box (Figure 1c). In addition to the 1.15M
battery we also constructed a 3.36M with 50 LiPF6 dissolved in 240 ECs for additional testing.
The visualization of trajectories is performed using the program Visualization Molecular
Dynamics (VMD) by Humphrey et al.55 During the simulations, the molarity remains constant
for both cases, 3.36 M and 1.15 M.
6
Figure 2. Total energy evolution during the MD equilibration and heating runs.
(a)
7
(b)
Figure 3. Snapshots taken during the first charge of the nanobattery. (a) The Li-ion battery at 7.5 ps
begins to polarize due to the effect of the electric field, showing the Li-atom interstitial and centered in a
carbon ring. (b) Snapshot of an overcharged battery at 10 ps.
The total energy and temperature of the battery reaches an equilibrium value that is proportional Formatted: Font: Italic
to the applied E (Figure 4a). We suggest that these constant values are reached when the rate of
energy supplied by the external power source equals the energy lost by the cooling and
dissipation processes in the battery. Therefore, at this point, no more energy is used to further
charge the battery, just an increase in atom vibrations and thus increases in temperature (Figure
4b).
During charging, the number of Li-ions arriving to the anode increases gradually; thus, the ionic
current at steady conditions can be obtained (Figure 4c). In order to have enough time for the Li-
ions to travel through the electrolyte and reach the anode in computationally practical times,
simulations were performed for up to 400 ps for e = 0.2, 0.4, and 0.6 V/; however, for E = 0.8,
1.0, 1.2, and 1.5 V/, only 20 ps were needed to have an statistically acceptable number of ions
arriving to the anode. For e = 0.2 V/, we displacement of Li-ions were difficult to obserb,
except for small fluctuations around their equilibrium positions.
8
= 0.3 kcal/mol), is stronger than the PF-6Co ( = 0.026 kcal/mol); Li-ions are more attracted to
the cathode than the PF-6. All these values result in the electrolyte polarization before the
application of an external field. In addition, the stronger inertial forces for the Li-ions compared
with those for the anions results in an additional contribution to the dipole in the negative
direction. This polarization provides useful information to understand the reactivities when the
battery is left unused for a long time. During the anode lithiation, the volume of the Li-ion
battery increases 0.6% (Figure 4f) while running the simulations under the NPT with an applied
E. The drift velocity (vd) when the electric field is Ei (i = 0.4, 0.6, 0.8, 1.0, 1.2 and 1.5) is
calculated from vd = JiVi/Niq56, where J = current density, V = electrolyte volume, N = total
number of ions in the electrolyte volume, and q = ion charge. The differential mobility comes
from the slope in Figure 4g (vd/E=3.3610-7 m2/Vs) as we have a nonlinear system with a dead
zone. The static ion mobility ( = vd/E) after a threshold field of 0.5 V/ increases linearly with
the field at a rate of 2.3910-7 m2/Vs (Figure 4h). J was obtained directly from the current in the
battery and the current was obtained from the plot of the number of Li atoms in the anode vs.
time.
(a)
(b)
9
(c)
0.6
(d)
(e)
4.0
vd = 3,36E - 1,875103m/s
vd(103m/s)
1.5
V(nm3)
266 3.0
1.2
265 1.0 2.0
0.8
264 0.6 1.0
0.4
263
0.2
0.0 E (V/)
0.0
0 t(ps) 10 20 0 0.5 1 1.5 2
(f) (g)
10
= 2.39E - 1.2310-7m2/sV
(10-7m2/sV)
2.5
2.0
1.5
1.0
0.5
E (V/)
0.0
0 0.5 1 1.5 2
(h)
Figure 4. Effect of the electric field on the time evolution of (a) Energy, (b) Temperature, (c) Number of
Li atoms arriving to the anode, (d) longitudinal dipole moment, px, (e) lateral electric dipole moments, py,
pz, (f) volume, (g) drift velocity as a function of E and (h) Li-ion mobility as a function of E.
The potential energy between one of the lithium-ions and its neighboring atoms inside a sphere
of 6 radius for E = 1 V/ is shown in Figure 5. Region 1 (cathode) shows the Li-ion in the
LiCoO2 crystal with a potential energy average of -255 kcal/mol. After 4.62 ps the Li-ion has
moved to the interface between regions 1 and 2 (electrolyte). An average energy of -36.7
kcal/mol is calculated in region 2, which represents the binding of the Li-ion with the electrolyte
solution. The energy for the Li-ion to leave the cathode-electrolyte interface is 218.3 kcal/mol;
this energy is supplied by the electric field. After 7.9 ps the Li-ion has gone through the
electrolyte, entered the anode, and been reduced; its potential energy average in region 3 is 2.1
kcal/mol and the binding energy of the Li-ion at the anode-electrolyte interface is 38.8 kcal/mol.
Notice that this is the energetics for one particular Li-ion for which we follow its trajectory
through the battery; these energies are not representative averages as they simply allow us to
obtain typical geometric structures of the Li-ion local environment.
200
kcal/mol
100
-100 1 3
-200
2
-300
-400
t(ps)
-500
3 4 5 6 7 8 9 10
Figure 5. Interaction energy (E) of one Li-ion with its neighbors within a sphere of 6 radius during
charge with E = 1.0 V/. Notice that this is one particular trajectory of a Li-ion.
11
With the electric currents shown in Table 4, calculated using = (e = electron charge,
nLi = number of Li-atoms arriving at the anode,), and the external voltages, calculated using V =
E ( = longitudinal length of the nanocell), which yields Vip = V/nip (nip = constant number of
ion-pairs in the electrolyte), we determine electrical parameters and quantities for the Li-ion
battery such as resistance, current density conductivity and resistivity (Table 4). Notice that the
voltages per ion-pair should correspond to those used to charge Li-ion batteries. The time-step
have to be reduced for fields larger than 1 V/ in order to follow the faster dynamics of the Li-
ions. Except for the first point, which corresponds to a very low voltage and the final one which
corresponds to a very high one, the conductivity of the solvent in the nanobattery follows a
practically linear behavior with the applied field. From 0.8 to 1.0 V/ , the conductivity features
a large increase, from 3.49 to 11.47 S/m; this is because for values greater than 0.8 V/, the
volume tends to increase as the applied electric field increases, creating more space between the
solvent molecules, allowing a faster diffusion of the Li-ions through the electrolyte. The
conductance of this battery reaches a value of 0.87 at charging voltages of 1.83 V per ion pair.
Experimental values are around 1 S/m even for mixed electrolytes, e.g., 1.07 S/m with small
amount of EC.15
Table 4. Electrical Properties Calculated in the Electrolyte Solution of 1.15M LiPF6 in Ethylene
Carbonate
E a(V/) b(fs) toc(ps) Tod(K) Tfe(K) Vipf Ig(A) Jh(A/m2) i(S/m)
0.4 1 14.87 297 298 1.16 0.00006 8.62106 0.002
0.6 1 2.86 321 303 1.74 0.00128 1.83108 0.030
0.8 1 0.90 324 366 2.32 0.16 2.301010 2.874
1.0 1 0.61 342 509 2.90 0.66 9.481010 9.483
1.2 0.2 0.37 348 1128 3.49 1.33 1.911011 15.92
1.5 0.2 0.28 380 1525 4.36 1.90 2.731011 18.20
a
E = Electric field.
b
= Time step.
c
t0 = Time needed for the nearest Li-ions to reach the anode after application of E.
d
T0 = Temperature at to.
e
Tf = maximum temperature during the simulation.
f
Vip = V/n = voltage drop per ion pair in the electrolyte.
g
I = Li-ion current.
h
J = current density.
i
=Electrolyte conductivity = J/E.
j
= Electrolyte resistivity.
We also analyzed an identical nanobattery with the only difference of having a 3.36 M solution
composed of 56 LiPF6 and 240 EC instead of a 1.15 M. The results are consistent with the
1.15M results. The results using the 3.36 M are consistent with the 1.15 M results: The total
energies of the simulation practically follow the same trend; the similarity of the results is very
close at very low electrical fields (Figure 6a) and very different at high external electrics. The
rate of energy supplied by the external power source equals the energy lost by the cooling and
dissipation processes in the battery; for E less than 1 V/, the equilibrium in the 1.15M is 40%
smaller than in the 3.36M; on the other hand, for E greater or equal than 1 V/ in 3.36M is 300%
12
higher than 1.15M; for E greater than 1 V/, Li-ions have enough energy to pass through the
electrolyte yielding higher velocities and increasing the temperature. (Figure 6a-4a)
The temperatures right after start applying E at 10 ns {you have to indicate exactly at what Formatted: Not Highlight
moment after applying field this is observed as the percentage depends on the applied voltage
(5% and 12%)} are very similar in its equilibrium value which arriving time depends on the Formatted: Font: Italic, Not Highlight
applied E (Figure 4b and 6b) for both samples no more than 5% higher except for the 1.5 V/ Formatted: Not Highlight
that is 12% higher since the kinetic energy increases due to stronger atomic fluctuations and Formatted: Font: Algerian, Not Highlight
higher Li-ions velocities. As expected the temperature increases to extremely high values even Formatted: Not Highlight
at fields of ~1 V/ (Figure 6b). The temperatures right after equilibration, once the electric fields
are applied, in the case of 1M, the temperature reaches the steady state at ~ 5ps for E 1 V/,
when the electric field is greater than 1 V/A the steady state takes more time to get reached, ~ 10
ps. (Figure 4b) The same behavior occurs for the 3.36M concentration battery, the temperature
reaches a steady state at ~ 5ps for E 0.8 V/, when the electric field is greater than 0.8 V/A
the steady state takes more time to get reached, ~15 ps. (Figure 6b) The final temperatures are
higher for the 3.36M (Figure 6b) when the step size is of 1 fs but the temperature decreases when
the step size is smaller than 1 fs. This is because the temperature control is more often for
shorter times of step size; all runs were set to regulate the temperature every 50 steps.
Temperature control reduces from 50 to 5 fs when step time reduces to 0.1 fs from 1 fs), thus
cooling is more effective at shorter step sizes.
The temperatures right after equilibration at 10 ns {you have to indicate exactly at what moment
after applying field this is observed as the percentage depends on the applied voltage (5% and
12%)} are very similar (Figure 4b and 6b) for both samples no more than 5% higher except for
the 1.5 V/ that is 12% higher since the kinetic energy increases due to stronger atomic
fluctuations and higher Li-ions velocities. The number of Li-ions (Figure 6c) arriving to the
anode during charge increases by a factor of 3.2 and 3 for the E = 1.0 and 1.2 V/, respectively,
at 5 ps. Due to the higher concentration of Li-ions, the 3.36 M sample, is charged at higher
current (charge / time) than the 1 M sample.
The dipoles in the x-direction (field direction, Figure 6d) are larger than their counterparts in the
1.15M case (Figure 4d) by ~90% due to the higher molarity. The fluctuating dipoles (Figure 6e)
have the same behavior with those of the 1.15M (Figure 4e) solution with values are around 0
Debye for the y-direction; however, in the z-direction, the electrolyte is between two
hypothetical insulators. Fluctuations in the y-direction are very quite similar with a slide
difference larger by 0.1 units for the 3.36 M than the 1.15 M sample. in the y. I would say
slightly larger by 0.1 units larger in the 3.2M case. Also, those ofthe dipole on the z directions is
the 3.2 M are twice larger and not symmetric, more towards the positive side, in the the 3.36 M
than the 1 M sample.pz directions and not symmetric, more towards the positive side.
On the other hand, for the 3.36M sample (Figure 6f), the total volume increased 3.2% more than
the volume of the 1.15M sample due to the lower molarity, total volume in 1.15M before
applying E is 3% less than the volume in the 3.36M. In both cases, 1.15 M and 3.36 M, the
volume of the cell increases if the electric field applied increases (Figure 6f). Up to E = 0.8 V/
13
the change in volume is not that critical, less than 0.18%. However, for values greater than 0.8
V/A, the change in volume is greater, this increase on volume can be explained because the
temperature exceeds 516 K which is the experimental boiling point51 of the EC, changing the
solvent from a liquid to a pre-gaseous phase.
(a)
(b)
(c) (d)
(e)
14
(f)
Figure 6. Effects of the electrical field (tagged in each curve in V/) on the 3.36M Li-ion battery (a)
Energy, (b) Temperature, (c) Number of Li atoms in the anode, and (d) electric dipole moment in the x
direction, px, (e) the lateral components py, and pz and the (f) Volume.
Fluctuations in the y-direction are very similar in the y. I would say slightly larger by 0.1 units
larger in the 3.2M case. Also, those of the 3.2 M are twice larger in the pz directions and not
symmetric, more towards the positive side.
The comparison between the 1.15M and 3.36M yields interesting differences (Table 4 and 5).
First, the observed delay, time required for the first Li-ion to arrive to the anode, increases from
1.4 for the 1.15 M sample to 2.4 for the larger 3.36M than for the 1Msample. This is due to the
presence of more PF6- ions, which are the biggest in size and mass in the solvent solution and
they do not let the Li-ions to diffuse easily. This behavior is in agreement with the experiment
by Valen et. al.,15 in which the diffusion coefficient of the Li-ions decreased while the
concentration increased.
The temperatures right after equilibration are very similar for both samples, no more than 5%
higher; however, after applying E, the temperature raises to its equilibrium {explain this better as
the T was in equilibrium during equilibration} The rate of energy supplied by the external power
source equals the energy lost by the cooling and dissipation processes in the battery; for E less
than 1 V/, the equilibrium in the 1.15M is 40% smaller than in the 3.36M; on the other hand,
for E greater or equal than 1 V/ in 3.36M is 300% higher than 1.15M; for E greater than 1 V/,
Li-ions have enough energy to pass through the electrolyte yielding higher velocities and
increasing the temperature. The final temperatures are higher for the 3.36M (Figure 6b) when
the step size is of 1 fs but the temperature decreases when the step size is smaller than 1 fs. This
is because the temperature control is more often for shorter times of step size; all runs were set to
regulate the temperature every 50 steps. Temperature control reduces from 50 to 5 fs when step
time reduces to 0.1 fs from 1 fs), thus cooling is more effective at shorter step sizes.
Based on the temperature plots for both samples,(Figure 6b-4b) we can identify that when, due to
the action of the applied electric field, the temperature present an abrupt change exceeding
~600K at the same time an increase on the volume it is observed. This information points out
that something is occurring at ~600K, we can infer that the solvent is starting to change its phase,
the change on volume is not high enough to conclude that it a change into a gas phase is
occurring, however is a hint that at least a change into a pre-boiling phase is happening. The 3.36
M sample needs the application of a lower electric field in comparison of the 1.15 M sample to
15
exceed ~600K, because Also in the 1.15M sample, an electric field greater or equal than 1.2 V/
is needed to exceed the 516 K changing the phase of the solvent into a gas, {how this was
observed? Did the volume increase tremendously? perhaps what you want to say is that the
material goes to a pre-boiling phase or the like but still no gas phase. Also we do not know the
boiling point of our model} for the 3.36M, the electric field only needs to be greater or equal
than 1.0 V/ to exceed the 516K {complete the sentence}. The explanation {of what?} relays
on the concentration of the salt, even we are using a weaker electric field for the 3.36M ( 1.0
V/) than for the 1.15M ( 1.2 V/) {this is confusing: we said weaker but the inequalities do
not guaranteed that, except between 1 and 1.2}, the number of ion and counter-ions moving
through the electrolyte is greater for the 3.362 M sample, hitting each other and also with the
ECs and, increasing the kinetic energy and therefore, the temperature. increases.
The current relation follows an odd behavior,behavior; it is very largealmost 300 at low Efields Formatted: Not Highlight
but decreases to 2.53 from almost 300 at very lowhigh Efields ; it could be due to the numbers of Formatted: Not Highlight
Li-ions in the battery at 3.36M is greater than 1.15M, consequently the Li-ions in the battery at Formatted: Not Highlight
3.36M need less energy than 1.15M to arrive to the anode.{I do not from where this come from;
please explain in detail}.
The ratio of currents is equal to the ratio of current densities since the conductivity is calculated
from I/A/E where A is the cross-sectional area.
The current follows an odd behavior, it is very large at low fields but decreases to 2.53 from
almost 300 at very low fields {I do not from where this come from; please explain in detail}.
The ratio of currents is equal to the ratio of current densities since the conductivity is calculated
from I/A/E where A is the cross-sectional area.
16
The conductivities calculated for both samples are compared to those from Dudley et al.57
(Figure 7); in this experiment, the electrolyte conductivities were measured in a 2-electrode glass
cell with platinum electrodes in a EC/PC solvent with LiPF6 salt under 1M concentration.
Conductivities obtained in the present work, especially for values lower than 350 K, are
compatible in the range order of magnitude obtained by Dudley experiment. The differences can
be explained with the fact that their solvent was a 50-50 EC/PC mixture and fixed temperature
{how they fixed the T?}for low temperature (-233K) using nitrogen liquid and ethanol.fixed
temperature {how they fixed the T?}
Figure 7. Conductivities calculated in the present work for the 1.0M and 3.2M samples and those from
Dudley et al.57 experiment
17
the cluster is oxidize, the binding changes to ~50 kcal/mol, understanding that during the short
life time of the negative ions the other ECs in the cluster are very tied to it.
Table 6. HOMO LUMO, Number of Imaginary Frequencies, Frequency and Energy of EC forming
monomer, dimer, trimer, tetramer and pentamer, ionized and neutral.
-20
-30
-40
-50
Figure 8. Trend of the relative energy of negative clusters with respect to their corresponding neutral
18
Solid Electrolyte Interface
The SEI is formed during the first few charging cycles of the battery.58 Most experiments of
charge-discharge that are able to observe the growth of the SEI are performed at very slow rates
of input voltage (voltammetry experiments), which give enough time for the formation of the
SEI; however, in practical situations, batteries are charged by suddenly applying the full
charging voltage. In our model, a particular SEI, lithium carbonate (LiCO3) is already formed at
the electrolyte-anode interface (Figure 9) and we determine its effect on the nanobattery features,
assuming that the SEI layer does not grow during the charging period and also assuming that in
this stage, the electron leakage is very low in the electrolyte-SEI interface, yielding a very small
growth of the SEI,59-60 that we are not considering. Also, incorporate the physical can chemical
properties Previous studies have been done, using MD models, to study the effects of the SEI
thickness and composition on the electrolyte.61-62 However, in this particular work, we are
focused on the diffusion of the Li-ion through the SEI.
Figure 9. Nanobattery with SEI (Li2CO3) with an electrolyte solution of 1.15M concentration, simulated
under a three-dimensional PBC.
Li+ diffuses through SEI in the [010] direction between the Li2CoO2 planes.63 The RDF analyses
(Figure 10a) yield nearest neighbor (NN) distances of 1.25, 1.95 and 2.95 for C-O, Li-O and
Li-Li, respectively. However, the second NN distances are 3, 2.5, and 5 for C-O, Li-O, and
Li-Li, respectively; thus Li+ has more space to travel along the [010] direction than other
directions. Diffusion of Li-ions takes place as shown in Figure 10b, mostly closer to the O
(Figure 10c,d) due to their opposite charges. The diffusion of Li+ occurs continuously propelled
by the energy from the voltage source; this energy is higher than the migration barrier of Li+ in
Li2CO3 of 0.6 eV.63
19
200
g(r)
C-O(SEI)
150 Li-O(SEI)
Li-Li(SEI)
100
50
0
0 1 2 3 4 r() 5
(a) (b)
(c) (d)
Figure 10. (a) Radial distribution functions of the SEI. (b) Li2CO3 SEI (all grey) and moving Li+
(purple). Diffusion of Li+ through Li2CO3: (c) top view, (d) a lateral view.
Conclusions
We have developed a testbed for the design of new materials for batteries with better
performances in cycling, capacity, and safety. In this first attempt we use the standard materials
of a commercial Li-ion battery to determine several aspects about its performance that would be
very difficult to obtain with macroscopic models or even with sophisticated experiments. The
computational model also helps us to establish procedures to get a satisfactory simulation of a
complete functional nanobattery.
The LiPF6 salt dissolved in the EC solvent at a concentration of 1.15M yields conductivities from
0.002 to 18.2 S/m, depending on the temperature and applied charging voltage. This range of
values includes the experimental 1.07 S/m obtained with LiPF6 1M dissolved in a mixture of
20
propylene carbonate (10 vol %), ethylene carbonate (27 vol %) and dimethyl carbonate (63 vol
%) at 293K using a YSI model 35 conductance meter.15 The battery current and the polarization-
time depend directly on the applied electric field.
When the electric field is less than 0.2 V/ at room temperature, the Li-ions do not drift in the
solvent at velocities that are practical to be observed by the simulations; they just fluctuate
around their positions. Once the temperature is raised to 1000 K, the Li-ions acquire the required
kinetic energy to drift through the solvent. Increasing the temperature also affects the solvent
phase, which at room temperature is in the solid phase. As the temperature increases, the solvent
changes from a solid phase into a liquid phase, facilitating the diffusion of the Li-ions.
Since a 3.36M has a higher concentration of LiPF6 it is expected to charge faster than the 1.15M
of LiPF6 concentration in the battery. One of the consequences of increasing the concentration
of LiPF6 is that the range of operation temperature is higher in the 3.36M concentration. In both
cases, 3.36M and 1.15M, the battery reaches a steady state; for the 3.36M the steady state and the
polarization take more time due to the number of LiPF6 is higher than the 1.15M; however at
3.36M does not present very different behavior from the 1.15M talking about energy,
temperature and polarization and the charging time is way less than the 1.15M battery, so
increasing the molarity and using a better cooling-system could be a possible solution to improve
and develop new batteries.
Both samples, 1.15M and 3.36M, present a threshold value of electric field; exceeding these
values, 1 and 0.8 V/ for the 1.15M 3.36M, respectively, yields a big change of temperature and
energy. Applying electric fields below the threshold values, the temperature never exceeds
600K. We can infer that the solvent is in liquid phase and changes into a pre-gaseous phase when
the applied electric field exceeds the threshold value, yielding an abrupt temperature change.
The conductivities are in good agreement with experimental resutls, especially when the
temperature does not exceed 360 K. Therefore, even the value of the electric field applied can be
considered really high since is in the order of magnitude of V/, the overall effect of this electric
field on the Li-ions emulates the charging process accurately enough to get acceptable electrical
properties. The voltage drop per ion pair in the electrolyte are also of the same order as the
voltages used to charge commercial Li-ion batteries (~3.6 V).
In addition to the charging ending times due to depletion of Li in the cathode, the saturation of Li
in the anode, or the rise of temperature to excessive values that might change the phase or
properties of the electrolyte, a very important factor to consider is the time to reach maximum
polarizability in the electrolyte solution, after which the charging of the battery is not practical
and eventually takes place long before reaching any of the other three situations. This dipole
moment may have a strong effect on Li-plating on the SEI or even directly on the anode of a new
battery. Thus, the excessive polarization during charging may enhance the local formation of Li-
dendrites, which may cause the battery to short circuit.
Acknowledgements
We thank Mr. J.T. Austin for a thorough reading of the manuscript. We would like to
acknowledge the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of
21
Vehicle Technologies of the U.S. Department of Energy under Contract No. DE-AC02-
05CH11231, Subcontract No. 7060634 under the Advanced Batteries Materials Research (BMR)
Program and the Qatar National Research Fund (QNRF) through the National Priorities Research
Program (NPRP 7-162-2-077). We also appreciate the support of computational resources from
Texas A&M High Performance Research Computing and the Texas Advanced Computing Center
(TACC).
{Make sure subscript and superscripts are use in the references, see 63 for example}
References
22
16. Zhang, S. S.; Jow, T. R.; Amine, K.; Henriksen, G. L., LiPF6-EC-EMC Electrolyte for Li-Ion
Battery. J. Power Sources 2002, 107 (1), 18-23.
17. Wang, X.-L.; An, K.; Cai, L.; Feng, Z.; Nagler, S. E.; Daniel, C.; Rhodes, K. J.; Stoica, A.
D.; Skorpenske, H. D.; Liang, C.; Zhang, W.; Kim, J.; Qi, Y.; Harris, S. J., Visualizing the
Chemistry and Structure Dynamics in Lithium-Ion Batteries by In-Situ Neutron Diffraction.
Sci. Rep. 2012, 2, 747-754.
18. Wakihara, M., Recent Developments in Lithium Ion Batteries. Mater. Sci. Eng. R-Rep. 2001,
33 (4), 109-134.
19. Kim, H.; Seo, D.-H.; Kim, S.-W.; Kim, J.; Kang, K., Highly Reversible Co3O4/Graphene
Hybrid Anode for Lithium Rechargeable Batteries. Carbon 2011, 49 (1), 326-332.
20. Plimpton, S.; Crozier, P.; Thompson, A., LAMMPS-Large-Scale Atomic/Molecular
Massively Parallel Simulator. Sandia National Laboratories 1995, 18.
21. Bellido, E. P.; Seminario, J. M., Molecular Dynamics Simulations of Ion Bombarded
Graphene. J. Phys. Chem. C 2012, 116 (6), 4044-4049.
22. Bellido, E. P.; Seminario, J. M., Graphene Based Vibronic Devices. J. Phys. Chem. C 2012,
116 (15), 8409-8416.
23. Bobadilla, A. D.; Samuel, E. L. G.; Tour, J. M.; Seminario, J. M., Calculating the
Hydrodynamic Volume of Poly(Ethylene Oxylated) Single-Walled Carbon Nanotubes and
Hydrophilic Carbon Clusters. J. Phys. Chem. B 2013, 117 (1), 343-354.
24. Bobadilla, A. D.; Seminario, J. M., Assembly of a Noncovalent DNA Junction on Graphene
Sheets and Electron Transport Characteristics. J. Phys. Chem. C 2013, 117 (50), 26441-
26453.
25. Bobadilla, A. D.; Seminario, J. M., Argon-Beam Induced Defects on Silica-Supported Single
Walled Carbon Nanotube. J. Phys. Chem. C 2014, 118 (48), 28299-28307.
26. Rodrguez-Jeangros, N.; Seminario, J. M., Density Functional Theory and Molecular
Dynamics Study of The Uranyl Ion (UO2)2+. J. Mol. Mod. 2014, 20, 1-12.
27. Yamabe, T.; Tanaka, K.; Ago, H.; Yoshizawa, K.; Yata, S., Proceedings of the International
Conference on Science and Technology of Synthetic Metals Structure and Properties of
Deeply Li-Doped Polyacenic Semiconductor (PAS). Synth. Met. 1997, 86 (1), 2411-2414.
28. Ago, H.; Kato, M.; Yahara, K.; Yoshizawa, K.; Tanaka, K.; Yamabe, T., Ab Initio Study on
Interaction and Stability of LithiumDoped Amorphous Carbons. J. Electrochem. Soc.
1999, 146 (4), 1262-1269.
29. Mrquez, A.; Vargas, A.; Balbuena, P. B., Computational Studies of Lithium Intercalation in
Model Graphite in the Presence of Tetrahydrofuran. J. Electrochem. Soc. 1998, 145 (10),
3328-3334.
30. Kohanoff, J.; Galli, G.; Parrinello, M., Theoretical Study of LiC6. J. Phys. IV 1991, 1 (C5),
351-356.
31. Komoda, S.; Watanabe, M.; Komaba, S.; Osaka, T.; Kikuyama, S.; Yuasa, K., Assessment of
Lithium Ion Doping Into Low Crystallized Carbonaceous Materials Using Molecular Orbital
Calculations. Electrochim. Acta 1998, 43 (2122), 3127-3133.
32. Holzwarth, N. A. W.; Louie, S. G.; Rabii, S., Lithium-Intercalated Graphite: Self-Consistent
Electronic Structure for Stages One, Two, and Three. Phys. Rev. B 1983, 28 (2), 1013-1025.
33. Scanlon, L.; Sandi, G., Layered Carbon Lattices and their Influence on the Nature of Lithium
Bonding in Lithium Intercalated Carbon Anodes. J. Power Sources 1999, 81, 176-181.
34. Li, T.; Balbuena, P. B., Theoretical Studies of Lithium Perchlorate in Ethylene Carbonate,
Propylene Carbonate, and Their Mixtures. J. Electrochem. Soc. 1999, 146 (10), 3613-3622.
23
35. Soetens, J.-C.; Millot, C.; Maigret, B., Molecular Dynamics Simulation of Li +BF4- in
Ethylene Carbonate, Propylene Carbonate, and Dimethyl Carbonate Solvents. J. Phys. Chem.
A 1998, 102 (7), 1055-1061.
36. Martnez, L.; Andrade, R.; Birgin, E. G.; Martnez, J. M., PACKMOL: A Package for
Building Initial Configurations for Molecular Dynamics Simulations. J. Comput. Chem.
2009, 30 (13), 2157-2164.
37. Tarascon, J. M.; Armand, M., Issues and Challenges Facing Rechargeable Lithium Batteries.
Nature 2001, 414 (6861), 359-367.
38. Sun, H.; Mumby, S. J.; Maple, J. R.; Hagler, A. T., An Ab Initio CFF93 All-Atom Force
Field for Polycarbonates. J. Am. Chem. Soc. 1994, 116 (7), 2978-2987.
39. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J.
R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.;
Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;
Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.;
Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark,
M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.;
Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.;
Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.;
Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.;
Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.;
Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas; Foresman, J. B.; Ortiz, J. V.;
Cioslowski, J.; Fox, D. J., Gaussian 09, Revision B.01. Wallingford CT, 2010.
40. Becke, A. D., Density Functional Thermochemistry. III. The Role of Exact Exchange. J.
Chem. Phys. 1993, 98 (7), 5648-5652.
41. Perdew, J. P.; Burke, K.; Wang, Y., Generalized Gradient Approximation for The Exchange-
Correlation Hole of a Many-Electron System. Phys. Rev. B 1996, 54 (23), 16533-16539.
42. Perdew, J. P.; Chevary, J.; Vosko, S.; Jackson, K. A.; Pederson, M. R.; Singh, D.; Fiolhais,
C., Atoms, Molecules, Solids, and Surfaces: Applications of The Generalized Gradient
Approximation for Exchange and Correlation. Phys. Rev. B 1992, 46 (11), 6671-6687.
43. Dunning, T. H., Gaussian Basis Sets for Use in Correlated Molecular Calculations. I. The
Atoms Boron Through Neon and Hydrogen. J. Chem. Phys. 1989, 90 (2), 1007-1023.
44. Kendall, R. A.; Dunning, T. H.; Harrison, R. J., Electron Affinities of The FirstRow Atoms
Revisited. Systematic Basis Sets and Wave Functions. J. Chem. Phys. 1992, 96 (9), 6796-
6806.
45. Peterson, K. A.; Woon, D. E.; Dunning, T. H., Benchmark Calculations with Correlated
Molecular Wave Functions. IV. The Classical Barrier Height of The H+H2H2+H reaction.
J. Chem. Phys. 1994, 100 (10), 7410-7415.
46. Rapp, A. K.; Casewit, C. J.; Colwell, K.; Goddard III, W.; Skiff, W., UFF, A Full Periodic
Table Force Field for Molecular Mechanics and Molecular Dynamics Simulations. J. Am.
Chem. Soc. 1992, 114 (25), 10024-10035.
47. Kumar, N.; Seminario, J. M., Lithium-Ion Model Behavior in an Ethylene Carbonate
Electrolyte Using Molecular Dynamics. J. Phys. Chem. C 2016, 120 (30), 16322-16332.
48. Politzer, P.; Murray, J. S.; Clark, T., Halogen bonding and other -hole interactions: a
perspective. Phys. Chem. Chem. Phys. 2013, 15 (27), 11178-11189.
49. Clark, T.; Hennemann, M.; Murray, J. S.; Politzer, P., Halogen bonding: the -hole. J Mol
Model 2007, 13 (2), 291-296.
24
50. Murray, J. S.; Lane, P.; Politzer, P., Expansion of the -hole concept. J Mol Model 2009, 15
(6), 723-729.
51. Lindsay, L.; Broido, D. A., Optimized Tersoff and Brenner Empirical Potential Parameters
for Lattice Dynamics and Phonon Thermal Transport in Carbon Nanotubes and Graphene.
Phys. Rev. B 2010, 81 (20), 205441-205446.
52. Tersoff, J., Empirical interatomic potential for silicon with improved elastic properties. Phys.
Rev. B 1988, 38 (14), 9902-9905.
53. Nauchitel, V. V., Energy-Distribution Function for The NVT Canonical Ensemble. Mol.
Phys. 1981, 42 (5), 1259-1265.
54. Kalibaeva, G.; Ferrario, M.; Ciccotti, G., Constant Pressure-Constant Temperature Molecular
Dynamics: A Correct Constrained NPT Ensemble Using The Molecular Virial. Mol. Phys.
2003, 101 (6), 765-778.
55. Humphrey, W.; Dalke, A.; Schulten, K., VMD: Visual Molecular Dynamics. J. Mol.
Graphics 1996, 14 (1), 33-38.
56. Griffiths, D., Introduction to Electrodynamics. 3 ed.; Prentice-Hall: Upper Saddle River, NJ,
1999; p 289.
57. Dudley, J. T.; Wilkinson, D. P.; Thomas, G.; LeVae, R.; Woo, S.; Blom, H.; Horvath, C.;
Juzkow, M. W.; Denis, B.; Juric, P.; Aghakian, P.; Dahn, J. R., Conductivity of electrolytes
for rechargeable lithium batteries. J. Power Sources 1991, 35 (1), 59-82.
58. An, S. J.; Li, J.; Daniel, C.; Mohanty, D.; Nagpure, S.; Wood Iii, D. L., The state of
understanding of the lithium-ion-battery graphite solid electrolyte interphase (SEI) and its
relationship to formation cycling. Carbon 2016, 105, 52-76.
59. Benitez, L.; Seminario, J. M., Electron Transport and Electrolyte Reduction in the Solid-
Electrolyte Interphase of Rechargeable Lithium Ion Batteries with Silicon Anodes. J. Phys.
Chem. C 2016, 120 (32), 17978-17988.
60. Soto, F. A.; Ma, Y.; Martinez de la Hoz, J. M.; Seminario, J. M.; Balbuena, P. B., Formation
and Growth Mechanisms of Solid-Electrolyte Interphase Layers in Rechargeable Batteries.
Chem. Mater. 2015, 27 (23), 7990-8000.
61. Jorn, R.; Kumar, R.; Abraham, D. P.; Voth, G. A., Atomistic Modeling of the Electrode
Electrolyte Interface in Li-Ion Energy Storage Systems: Electrolyte Structuring. J. Phys.
Chem. C 2013, 117 (8), 3747-3761.
62. Ramos-Sanchez, G.; Soto, F. A.; Martinez de la Hoz, J. M.; Liu, Z.; Mukherjee, P. P.; El-
Mellouhi, F.; Seminario, J. M.; Balbuena, P. B., Computational Studies of Interfacial
Reactions at Anode Materials: Initial Stages of the Solid-Electrolyte-Interphase Layer
Formation. J. Electrochem. En. Conv. Stor. 2016, 13 (3), 031002-031002-10.
63. Iddir, H.; Curtiss, L. A., Li Ion Diffusion Mechanisms in Bulk Monoclinic Li 2CO3 Crystals
from Density Functional Studies. J. Phys. Chem. C 2010, 114 (48), 20903-20906.
25