pubs.acs.

org/JPCB Article

Effect of Concentration and Temperature on the Structure and Ion

Transport in Diglyme-Based Sodium-Ion Electrolyte
Shylendran Ardhra, Prabhat Prakash, Rabin Siva Dev, and Arun Venkatnathan*
Cite This: J. Phys. Chem. B 2022, 126, 2119−2129 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
Downloaded via INDIAN INST OF TECH GANDHINAGAR on December 28, 2023 at 14:29:29 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Glyme-based sodium electrolytes show excellent electrochemical properties and

good chemical and thermal stability compared with existing carbonate-based battery
electrolytes. In this investigation, we perform classical molecular dynamics (MD) simulations
to examine the effect of concentration and temperature on ion−ion interactions and ion−
solvent interactions via radial distribution functions (RDFs), mean residence time, ion cluster
analysis, diffusion coefficients, and ionic conductivity in sodium hexafluorophosphate (NaPF6)
salt in diglyme mixtures. The results from MD simulations show the following trends with
concentration and temperature: The Na+---O(diglyme) interactions increase with concentration
and decrease with temperature, while the Na+---F(PF6−) interactions increase with
concentration and temperature. The mean residence time suggests that Na+---O(diglyme)
are significantly longer lived compared with that of Na+---F(PF6−) and H (diglyme)---F(PF6−),
which shows the affinity of diglyme to the Na+ ions. The ion cluster analysis suggests that the
Na+ ions largely exist as solvated ions (coordinated to diglyme molecules), whereas some
fractions exist as contact-ion pairs, and negligible fractions as aggregated ion pairs, with the
latter two increasing slightly with temperature and more with ion concentration. The magnitude of the diffusion coefficients of Na+
and PF6− ions decreases with concentration and increases with temperature, where the Na+ ion has slightly lower mobility compared
with the PF6− anion. The simulated total ionic conductivities show qualitative trends comparable to experimental data and highlight
the need for the inclusion of ion−ion correlations in the Nernst−Einstein equation, especially at higher concentrations and lower
temperatures.

1. INTRODUCTION oxygen atoms of the ether groups in glyme and carbonyl-

Sodium-ion batteries have attracted attention due to the functional group-based solvents coordinate with the alkali
abundance of sodium and similarities with lithium. A wide metal ions leading to ionic dissociation and dissolution of the
range of electrolytes used in sodium-ion batteries (SIBs) are salt even at low dielectric constant20 (diglyme, ε = 7.4) of the
ionic liquids,1 aqueous electrolytes,2 organic electrolytes,3,4 solvent.21,22 The ability of glymes to coordinate with the Na+
solid-polymer electrolytes,5,6 and inorganic solid electrolytes.7,8 ions in an octahedral framework plays an important role in
Sodium salts dissolved in the mixtures of nonaqueous organic stability compared with carbonate solvents.23−25 The Na+-
solvents such as ethylene carbonate (EC), propylene O(glyme) coordination is similar to solvated ionic liquids
carbonates (PC), and dimethylene carbonate (DMC)9,10 are having [metal(glyme)]+ cation with a counter anion.26 A wide
traditional used electrolytes with high ion conductivity (∼10−3 electrochemical window of 4 V and high ionic conductivity (1
S/cm).11 Unfortunately, these solvents show high flammability mS/cm) further increases the scope of glyme electrolyte
and pose a significant drawback especially to technologies like applicability as solvents.27
electric vehicles.12 Moreover, the chemical reaction of organic The choice of solvent also depends on its affinity or
solvents like alkyl carbonates with electrodes leads to a interaction with the electrodes. For example, the commonly
formation of an unstable solid-electrolyte interface (SEI) used anode in LIBs is graphite28 although this is found to be
composed of carbonates, fluorides, chlorides, etc.13 In contrast mostly inactive in SIBs. This is due to the distance between the
to lithium-ion batteries (LIB), SEI dissolves rapidly in two graphite layers being smaller than the atomic size of Na
SIBs,13,14 and this leads to a continuous degradation of the
electrolyte with a loss of capacity and reduced cycle life.15,16
As an alternative to conventionally studied organic solvents, Received: January 23, 2022
glycol dimethoxy ethers, or glymes [R-(OCH2CH2)n-OR] Published: March 4, 2022
(Gn), a family of stable and amphiphilic ether solvents are
potential choices due to high electrochemical stability, wide
temperature window (>200°C), high donor numbers, and
relatively strong solvation for alkali metal ions.17−19 The

© 2022 American Chemical Society https://doi.org/10.1021/acs.jpcb.2c00557

2119 J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

atom.29 First principle investigations have confirmed the Previous experimental and theoretical studies have focused
inactivity of graphite electrodes with Na ions due to the largely on the examination of properties on 1 M concentration.
formation of energetically unstable Na-graphite intercalation Furthermore, NaPF6/diglyme as an electrolyte has been
compounds.30 This unfavorable mismatch between Na+ ions investigated only by experiments. Hence, in this current
and metal lattices can be overcome by metal co-intercalation investigation, we have chosen a wide range of electrolyte
phenomena, which is difficult to achieve in carbonate-based concentrations (0.75, 1.0, 1.25, 1.5, 1.75, and 2.0 M) and
electrolytes.31 Jache and Adelhelm32 and Kim et al.33 have temperatures (298, 308, 318, 328, 338, 348, 378, 400 K) to
observed that diethylene glycol dimethyl ether (diglyme) assist examine various structural and dynamical properties in NaPF6/
in intercalation of Na+ ions into the graphite layers to form diglyme system. The computational details are presented in the
ternary-graphite intercalation compounds. The intercalation next section. Section 3 contains results and discussions on the
also shows a reversible charge storage for Na+ ions on the structure and ion dynamics, and a summary of key findings
electrode with exceptional stability. Seh et al.34 showed that concludes this paper.
NaPF6/diglyme: Na0 anode results in the formation of a
uniform SEI with high reversibility and nondendritic growth at 2. COMPUTATIONAL DETAILS
room temperatures. In glyme-based solvents, it has also been 2.1. Classical MD Simulations. An initial set of
shown that the formation of solvate ionic liquid with the coordinates for optimized single molecular structures of
insertion of more ions increases electrochemical stability (by NaPF6 and diglyme was calculated (Figure 1) from quantum
∼0.5 V).25,26,35 Westman et al.27 reported that an electro-
chemically (0−4.4 V) and thermally stable SEI forms on the
use of NaPF6/diglyme as an electrolyte. Various sodium salts
are known to show different behaviors toward the formation of
SEI,36−38 suggesting the choice of an ideal solvent and anion of
salt can lead to the formation of stable SEI. Certain sodium
salts based on fluorinated organic anions like TFSI− (bis- Figure 1. Chemical structures of sodium (Na+), hexafluorophosphate
(trifluoromethanesulfonylimide)), FSI− (bis-(fluorosulfonyli- (PF6−), and diglyme.
mide)) form SEI and though it leads to continuous
degradation of the electrode.36 However, other salts based
on OTf− (trifluoromethanesulfonate) and PF6− (hexafluor- calculations on Gaussian 09, Revision A.0245 software package.
ophosphate) anions in diglyme exhibit much better stability These optimized molecular geometries were used to prepare
with graphite.38 Goktas et al.37 suggested that in NaOTf salt, a initial inputs for MD simulations. A cubic simulation box of
reversible co-intercalation of sodium in graphite results in an side length ∼70 Å was constructed and packed with the
SEI-free operation. optimized geometries of NaPF6 and diglyme using Packmol.46
Computational methods such as ab-initio and classical Each simulation box consists of fixed 1400 diglyme molecules
molecular dynamics (MD) simulations are powerful techni- with a varying number of Na+ and PF6− ions corresponding to
ques to examine the structure and dynamics of ion transport, the required concentration as seen in Table S1 of Supporting
which can enhance the understanding the chemistry of Information.
Classical MD simulations were performed using the
electrolytes. Several computational studies have focused on
GROMACS-5.0.7 code47 with the force-field parameters for
the solvation of ions, trends of ion association in lithium/
diglyme taken from OPLS-AA and Barbosa et al.48 and for Na+
sodium salts immersed in glyme solutions, which can be
and PF6− ions, from Jensen et al.49 The cutoff for the
correlated with observations from spectroscopic methods such
nonbonded interactions was fixed to 12 Å. The equations of
as FTIR.39,40 Jónsson and Johansson41 studied the role of motion were integrated by a velocity-verlet algorithm using a
anion structure and charge delocalization toward ion time step of 1 fs. The input configurations of NaPF6 in diglyme
association using density functional theory (DFT) calculations. at various concentrations were energy minimized using the
The authors investigated ion-pair dissociation, interionic steepest descent algorithm,50 followed by simulated annealing
structures, and the effect of DFT functionals using finite- (NPT ensemble) for 20 ns. During annealing, the system was
scale geometries. Dhumal et al.42,43 performed DFT calcu- heated from 300 to 420 K and then cooled back to 300 K (at a
lations on varying chain length on glyme and examined heating and cooling rate of 20 K/ns) followed by a 5 ns
interactions between cations (Li+, Na+, K+, etc.) and anions equilibration run at 300 K. The final configurations after
(PF6−, TFSA). These studies have focused on the prediction of annealing at every concentration were simulated for another 10
correct IR shifts observed from vibrational calculations. ns equilibration under NPT ensemble (at 298, 308, 318, 328,
Tsuzuki et al.44 modeled the salt−glyme interaction using 338, 348 378, and 400 K), followed by a 100 ns production
MD simulations and predicted that the length of a glyme run. The velocity-rescale thermostat51 and Berendsen baro-
molecule affects both ion-glyme coordination and ion stat52 with a coupling time constant of 0.1 and 1.0 ps,
dynamics. Particularly, a longer glyme molecule surrounds respectively, were used to maintain applied ensemble
and dissociates the cation more than a shorter glyme and conditions. The average density at every concentration for
decreases interionic (cation−anion) interactions resulting in a the temperatures such as 298, 308, 318, 328, and 338 K,
higher ionic mobility. Li et al.23 examined the behavior of ion obtained from the equilibration run (see Figure S1 in the
pairs (contact vs noncontact) of NaTFSA in various glyme Supporting Information) shows a deviation of only <2% from
solvents using FTIR, NMR, and atomistic simulations. The experimental values.
authors showed that mono-glymes and diglymes show the best 2.2. DFT-MD Simulations. Quantum mechanics-based
stabilization toward sodium/superoxide ion pair at the DFT-MD simulations were performed using CP2K 6.1.0
cathodic interface. code53 using Quickstep module. The forces and energies on
2120 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

each atom of the system were calculated with hybrid Gaussian compared with 6 from classical MD. The peak intensity shows
plane wave (GPW) approach using triple zeta basis sets a slightly more diffused solvation sphere from DFT-MD
(TZV2P-GTH)54 along with Perdew−Burke−Ernzerhof compared with classical MD. Such differences in the peak
(PBE)55 exchange correlation functional and the correspond- intensities have also been reported earlier in the work of
ing Goedecker−Teter−Hutter (GTH) pseudopotentials.56,57 Wahlers et al.39 These differences may arise due to smaller
The input geometries of 1 M NaPF6/diglyme (with four system sizes and shorter time scales of the computationally
molecules of NaPF6 and 28 molecules of diglyme) for DFT- demanding DFT-MD calculations and/or lack of polarization
MD simulations have been obtained from equilibrated classical effects in classical MD.
MD outputs from a 10 ns production trajectory at 400 K. The We further analyzed the DFT-MD trajectory to compare the
box length was set to 19.92 Å. The time step for the simulation accuracy of ionic mobility from classical MD simulations. The
was set to be 0.5 fs. We have performed a 20 ps equilibration calculated VACF (Figure 2b) and MSD (Figure 2c) at T = 400
(NPT) followed by a 15 ps production (NVT) at 400 K using K show an excellent agreement between the results from both
the Nosé−Hoover chain thermostat.58,59 Radial distribution methods.
functions (RDFs), mean square displacement (MSD), and Unfortunately, the computationally prohibitive cost of DFT-
velocity autocorrelation functions (VACF) have been calcu- MD simulations limits sampling to small system sizes and
lated using the TRAVIS60 program. The equations for these shorter simulation time scales. However, classical MD
properties are given in Supporting Information. simulations can sample large system sizes and longer time
scales (hundreds of nanoseconds). Hence, the remaining part
3. RESULTS AND DISCUSSION of this work presents results from the classical MD simulations.
3.2. Classical MD Simulations. 3.2.1. Structure. The
3.1. DFT-MD Simulations. The DFT-MD simulations solvated structures of Na+ ion in diglyme primarily form with
were performed for a comparison with the classical MD the coordination of Na+ ions with -O- (ether) groups, as
simulations as a validation on the ability of the force-field used observed in other poly glycols.39 A visual inspection (Figure 3)
to mimic the physical properties such as solvation structure
and local dynamics. These simulations were performed for 20
ps and at an elevated temperature T = 400 K (for sufficient
statistical sampling) for a system consisting of four NaPF6 ion
pairs in 28 diglyme molecules (equivalent to 1 M NaPF6/
diglyme). The calculated RDFs from DFT-MD at 400 K
(Figure 2a) show the maxima for Na+---O at 2.4 Å and are in
excellent agreement with the peak maxima of 2.3 Å obtained
from classical MD simulations. At first minima, 3 Å, the
number of O atoms around Na+ is ∼5 from DFT-MD

Figure 3. Final snapshots from 100 ns production run for 0.75 M

(left) and 2.0 M (right) concentrations of NaPF6/diglyme at T = 298
K. Color scheme: green (lines)diglyme; cyan (polyhedral)PF6−;
yellow (spheres)Na+; red (tubes)Na+---O bonds (<2.5 Å); green
(tubes)Na+---F (< 2.5 Å).

of the trajectory (extracted from simulations at room

temperature) shows complete solvation of Na+ ions with O
atoms of diglyme for all the concentrations. A range of
snapshots showed that the Na+ ions form four to six bonds
with O atoms (red tubes in the snapshots) within a distance of
2.5 Å. In contrast, fewer Na+---F bonds (green tubes) were
observed visually within 2.5 Å. We further analyzed the
qualitative structure with respect to temperature and
concentrations in the next subsection.
3.2.1.1. Effect of Concentration on the Solvation
Structure. From the production run of simulations, the
RDFs at 298 K and the corresponding coordination number
(CN) are computed upto the first solvation shell minima as
shown in Figure 4. For Na+---O interactions, RDF shows a
maximum at 2.3 Å for the first solvation shell for all the
concentrations (Figure 4a). The position of first maxima and
the first minima slightly increases with the concentration from
0.75 to 2.0 M. The first minima for Na+---O solvation shell in
NaPF6/diglyme system at all concentration is found smaller
Figure 2. Comparison of results of 1 M NaPF6/diglyme system from than the other crystallized systems like [Na(G4)1][TFSA] and
DFT-MD and classical MD simulations at T = 400 K: (a) RDF of Na [Na(G4)1]ClO4 (2.4−2.5 Å).61 In the present system of
and O (diglyme), (b) VACF of Na+ ions, and (c) MSD of Na+ ions. NaPF6/diglyme, the calculated CNNa−O (at first minima)
2121 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

Figure 4. RDFs (solid lines) for (a) Na+---O(diglyme) with coordination numbers (dashed lines), (b) Na+---F(PF6−), at different concentrations of
NaPF6/diglyme at 298 K.

Figure 5. RDFs (solid lines) for (a) Na+---O (diglyme) with coordination numbers (dashed lines), (b) Na+---F(PF6−), at different temperatures of
1 M concentration of NaPF6/diglyme.

decreases from 6.7 (0.75 M) to 6.2 (2.0 M). However, the The only effect of temperature can be seen as the calculated
change in CNNa−O is much smaller compared with other CNNa−O at 3.0 Å, where the CNNa−O decreases from 6.5 (298
electrolytes such as Na+-triflate/diglyme where the CN K) to 5.8 (400 K). In contrast, the peak intensity for Na+---F
decreases from 6 to 4 with the increase in concentration due RDF increases with temperature (Figure 5b), suggesting the
to a partial participation of O atoms from anion (of triflate, presence of more PF6− ions with the removal of diglyme in the
TFSI, etc.) to the Na+ solvation sphere.39 solvation shell of Na+ ions. It can be concluded that the Na+
The interaction between Na+ ion and PF6− anion has also solvation sphere replaces diglyme with PF6− ions with
been analyzed as a comparison to the Na+---O interactions. increasing temperature.
The F (PF6−) forms very few bonds and also less frequently 3.2.2. Dynamics. 3.2.2.1. Ion−Diglyme and Ion−Ion
with Na+ having peaks at 2.4 and 4.4 Å as seen in the Na+---F Lifetimes. The strength of ion−diglyme and ion−ion
RDFs in Figure 4b. While the Na+---F RDF shows a sharp first interactions can be quantified as the “uninterrupted residence
maxima at 2.4 Å, the calculated coordination numbers are very time,” which is calculated in the form of a residence correlation
small (≪1), which suggests a poor frequency of their function, C(t),62 where
occurrence. This is later analyzed as ion-pair residence times
in the dynamics section. Another solvation shell is observed at ⟨H(t )H(0)⟩
C(t ) =
6.4 Å in Na+---F RDF suggesting that the PF6− ion also ⟨H(0)H(0)⟩ (1)
coordinates with the [Na-diglyme]+ complex, and the peak
intensity reduces with increasing concentration. Our results The occurrence operator in the above expression is defined as
show excellent agreement with the partial-pair distribution follows: if an anion (PF6−) or the diglyme molecules are
functions calculated experimentally from the empirical present at a cutoff distance of r from the Na+ cation, then the
potential structure refinement (EPSR) model by Jensen et operator H(t) = 1 and 0 otherwise. We chose the position of
al.49 Further, we observed the possible ion−dipole interactions the first minima from the respective RDFs (Na+---O and Na+---
as seen from O---H (diglyme) and F(PF6−)---H(diglyme) F, as shown in Figure 3) as the cutoff r for this calculation. The
RDFs (Figure S2 of Supporting Information). These low- mean residence lifetime (τ)62 is then obtained as:
intensity RDFs support the absence of anion−diglyme and ∞
inter- or intra-atomic O---H interactions, as reported earlier.49 τ= ∫0 C(t )dt
(2)
3.2.1.2. Effect of Temperature on the Solvation Structure.
The calculated RDFs for Na+---O (diglyme) interactions To illustrate the effect of concentration and temperature on
(Figure 5a) in 1.0 M NaPF6/diglyme system suggest that the the residence time, we have calculated C(t) all the
peak height and position of the first maxima do not change concentrations and temperatures. The C(t) vs t plots for
significantly with temperature. temperatures T = 298 and 400 K and concentrations 0.75 and
2122 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

Figure 6. The residence correlation function for different interactions for (a) 0.75 M at 298 K, (b) 2.0 M at 298 K, (c) 0.75 M at 400 K, and (d)
2.0 M at 400 K in NaPF6/diglyme.

Table 1. Mean Residence Life Times (τ) for the Ion−Ion (Na+ and PF6−) and Ion−Solvent Interactions (Na+ and Diglyme
Molecules, and PF6− and Diglyme Molecules) Calculated from an Exponential Fit to C(t) Shown in Figure 6
298 K 400 K
τNa‑‑‑O (ns) τNa‑‑‑F (ns) τH‑‑‑F (ns) τNa‑‑‑O (ns) τNa‑‑‑F (ns) τH‑‑‑F (ns)
0.75 M 28.07 2.49 2.17 4.06 0.23 0.22
2.0 M 92.39 25.49 15.53 10.19 1.02 0.62

Figure 7. (a) Na+---P(PF6−) RDFs, at 298 K and at various concentrations of NaPF6/diglyme. (b) Percentage distributions of various ion clusters
with respect to the concentration of NaPF6/diglyme at T = 298 K. SolI = solvated ions, CIP = contact-ion pairs, and AGXIP = aggregated ion pairs
of size X. (c) Representative snapshot of a solvated Na+ ion, referred to as SolI. (d) Representative snapshot of a CIP. (e) Representative snapshot
of a AG4IP from the trajectory of 2.0 M concentration system. Exact population of various ion speciation is provided in Table S3 of the Supporting
Information. Color scheme in (b−d): Blue Na, tan P, pink F, red O, cyan C, white H.

2.0 M are provided in Figure 6. For [2.0 M, 298 K], the shows that Na---O (diglyme) pairs are extremely long-lived
simulation runs were performed for 200 ns to ensure a compared with Na---F (PF6−) and H (diglyme) ---F (PF6−),
sufficient decay of the tail part of C(t). An examination of C(t) which is further supported by the calculated mean residence
2123 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

Figure 8. (a) Distribution histogram of SolI for various concentrations of NaPF6/diglyme at T = 298 K; (b) a representative SSIP with four PF6−
neighbors observed at 2.0 M, 298 K.

Figure 9. The displacement and lifetime distribution of various SSIPs for (a) 0.75 M, (b) 1.0 M, (c) 1.25 M, (d) 1.5 M, (e) 1.75 M, and (f) 2.0 M
systems at T = 298 K.

times τNa‑‑‑O, τNa‑‑‑F and τH‑‑‑F shown in Table 1. Such longer 3.2.2.2. Ion Cluster Analysis. To understand the nature of
mean residence times support a high CN observed from Na+--- interionic interactions and ion-pair formation in the system, we
O interactions compared with Na+---F. At [0.75 M, 298 K], the calculated the fraction of various types of ion clusters in the
values of τNa‑‑‑O and τH‑‑‑F are an order of magnitude smaller NaPF6/diglyme system. The formation of these various
compared with [2.0 M, 298 K], whereas τNa‑‑‑F is lower by clusters depends on two pair interactions, Na+---O(diglyme)
∼two orders of magnitude. Similarly, mean residence time and Na+---PF6−. Apart from the Na+---O(diglyme) interaction,
reduces with temperature from 298 to 400 K for all the Na---P(PF6−) RDF (Figure 7a) suggests that there are two
peaks for this pair interaction. The first peak represents a
concentrations. The trends of mean residence times in C(t)
primary solvation shell due to direct contact of Na---P(PF6−)
and τ for other concentrations and temperatures are seen in (r < 4 Å). The second peak arises due to contact of P(PF6−)
Figure S3A−C and Table S2 in the Supporting Information. with solvated Na+ ions (6 Å < r < 9 Å). The distribution
This suggests that high ionic concentration and low temper- histogram of various ionic speciation is calculated from the
ature lead to longer residence times, which also restricts the production trajectory of each NaPF6/diglyme system at T =
mobility of ions, which can be further examined by the 298 K and shown in Figure 7b. The distribution of these ion
calculation of diffusion coefficients as described in the clusters at other temperatures is provided as Table S3 in the
subsequent subsection. Supporting Information. The definition of various ion clusters
2124 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

in the system is as follows: all the Na+ ions either coordinate traveling this distance together. At higher concentrations, some
with diglyme molecules or with PF6− ions. The “solvated ions” of the SSIP travel even up to 9 Å before the breakdown. The
(SolI) (Figure 7c) are essentially Na+ ions solvated with lifetime of SSIP increases with concentration due to relatively
diglyme molecules. The SolI can exist either as free ions (FI) fewer diglyme molecules at higher concentrations. At high
or as solvent-separated ion pairs (SSIPs). The other ion-pair concentrations, the exchange of PF6− ions with solvated Na+
clusters are contact-ion pairs (CIP) (Figure 7d) and ions reduces and hence leads to a long-lived SSIP. The
aggregated ion pairs (AGIP), respectively. Any Na+ ion intensity of occurrence (in the boundaries of red color) in
coordinating with two PF6− ions or a PF6− ion coordinating Figure 9 shows that the majority of SSIP have a lifetime of
with two Na+ ions is considered as an AGIP of size three ∼100 ps for 0.75 M, ∼150 ps at 1.0 M, and ∼350 ps for other
(AG3IP). Similarly, AG4IP (Figure 7e) are ion aggregates of higher concentrations. Some of the SSIPs (shaded in green
size four. yellow and bluerelative occurrence ∼25%) exhibit a lifetime
The population analysis of various ion clusters in the system of >1 ns for concentrated solutions like 1.75 and 2.0 M.
suggests that SolI are the most dominant clusters in the Previous studies, based on the integration of RDFs63 to
NaPF6/diglyme system, from 79% (at high concentration, 2.0 calculate coordination numbers, and elucidation and integra-
M) to 94% (at low concentration, 0.75 M). A subsequent tion of the FTIR peaks23,39 to calculate the fractions of
paragraph follows the discussion about these clusters. The next speciation, provided much approximated and limited informa-
predominant clusters are CIP, which form from 5% (at low tion about the complex nature of the ion clusters. The results
concentration, 0.75 M) to 14% (at high concentration, 2.0 M). from our study provide a comprehensive molecular-level
The other higher-order ion clusters are AG3IPwhich form insight about the spatial and temporal behavior of various
from 1 to 7% from low to high concentration, and AG4IP forms of ionic clusters using dimer-displacement and dimer-
which are <1% in occurrence. Further, an increase in lifetime functions. The most important feature of the dimer-
temperature leads to the breakdown of some SolI, as the distribution approach used here is not to treat all the SSIPs in
fluctuations result in a lower ion−solvent lifetime (seen from an averaged manner and obtain a more detailed view of these
the calculated τNa‑‑‑O, as seen in Table 1). The breakdown of disparate ion clusters compared with earlier studies.23,39,63
SolI with temperature results in more accessibility for PF6− 3.2.2.3. Self-Diffusion Coefficient. The mobility of Na+
ions to form CIP and AGIP, and hence the occurrence of CIP ions, PF6− ions, and diglyme molecules is seen from the
and AGIP increases with temperature for all concentrations. mean squared displacement (MSD) vs time plots calculated
The SolI, as discussed above, is further divided into two from the production run trajectories (see Figures S5 and S6 in
types of clusters: FI and SSIP. The more abundance of free the Supporting Information). The self-diffusion coefficients, Di,
diglyme molecules allows the Na+ ions to be completely are calculated from the linear regime of the MSD using
surrounded by more shells of diglyme molecules, and hence Einstein’s relationship.64 An examination of MSD of Na+ and
these Na+ ions do not form an SSIP, especially at low PF6− ions shows a subdiffusive behavior for the initial few
concentrations. Unlike ion−solvent and CIP, the SSIPs are hundred picoseconds especially at low temperatures (T = 298
more complex clusters. For example, they may form with more and 308 K), and hence Di could not be reported. At other
than one neighbor PF6− ions after solvent separation, which temperatures (T ≥ 318 K), a linear regime from the MSD was
shows a large distribution of the varying number of neighbors obtained, and the calculated self-diffusion coefficients for Na+
PF6− and diglyme molecules. In order to calculate the ions, D+, PF6− ions, D−, and diglyme molecules Dsol are shown
distribution of SSIPs, we performed a dimer-distribution, in Figure 10.
dimer-displacement, and dimer-lifetime analysis60 for Na+--- The effect of temperature on the magnitude of the diffusion
P(PF6−) pairs for all concentrations and at T = 298 K. The coefficient is significant between T = 348 K and T = 400 K. For
dimer-distribution histogram (Figure 8a) counts the occur- example, at 0.75 M concentration, D+ and D− are ∼ two times
rences of various PF6− ions around solvated Na+ ions averaged larger at 400 K compared with 348 K. The order of mobility is
over the production trajectory runtime. If the number of PF6− Dsol > D− > D+ and is independent of the temperature and
ions is zero (on the X-axis), the SolI is defined as an FI. The concentration, where the transference number of anions ≥0.5.
FIs are more abundant at low concentrations (∼5% for 0.75 The decrease in diffusion coefficients for ions and diglyme at
M) compared with high concentrations (∼0.07% for 2.0 M). higher concentration of NaPF6/diglyme can be attributed to
Further, if one or more PF6− ions approach the Na+ ions, the the strong interaction of diglyme and sodium ions and higher
SolI is defined as SSIP. Figure 8b shows a representative SSIP viscosity in the concentrated electrolyte, which restricts the
with four neighbor PF6− ions. As the concentration of NaPF6 diffusive motion of ions. These results are consistent with the
in diglyme increases, more PF6− ions surround a solvated Na+ previous findings of Kumar and coworkers23,39,65 where the
ion forming an SSIP. It can also be seen that the size of the authors showed that an increase in the concentration of the salt
most abundant SSIP increases with increasing concentration and the length of the glyme chain leads to reduced diffusion
(from 2 for 0.75 M to 4.5 for 2.0 M). coefficients.
To further understand the SSIP formation and lifetime, Kumar and coworkers23,39 observed that D+ and D− were
dimer-displacement and dimer-lifetime analysis for Na+--- similar in magnitude and decreasing considerably with the
P(PF6−) pairs was performed, and the net displacement and concentration in the sodium triflate/diglyme system. The
lifetime with respect to relative occurrence intensities are decrease was more dominant at concentrations higher than 1.5
shown in Figure 9. The relative occurrence of all the possible M, which implies the possibility of the formation of CIP. An
Na+---PF6− SSIP is plotted with respect to displacement and examination of D+, D−, and Dsol from our work shows a
lifetime. The lifetimes of the SSIP are calculated from decrease in concentration (see Figure 10a−c). The large
uninterrupted ion pairs and considering the solvent exchanges associative behavior between the cations and anions is also
in the system. For all the concentrations, the most probable evident from their similar magnitudes of self-diffusion
displacement of an SSIP is 2.3 Å,that is, an SSIP breaks before coefficients. Furthermore, the residence autocorrelation times
2125 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

where N is the number of ion pairs, q+ and q− represent the

charge on the cations and anions, respectively, V is the volume
of the system, kB is the Boltzmann constant, T is temperature,
and D+ and D− represent the self-diffusion coefficients of
cations and anions. The calculated σNE vs concentration for
each temperature is shown in Figure 11a. At the highest
temperature (i.e., 400 K), the value of σNE attains a maximum
value of 3.49 S/m for 1.0 M and subsequently decreases with
further increase in concentration with similar trends observed
at other temperatures. However, the experimental work of
Westman et al.27 showed that the ionic conductivity at T = 298
K increases with increasing concentration and finally attains a
saturation at 1.5 M. The trends of the calculated σNE with the
experimental values could also be due to the exclusion of
correlated motions of ions. The contributions from cation−
cation (+ +), anion−anion (− −), and cation−anion (+ −)
correlated interactions play a significant role in the
determination of ionic conductivities. The classical MD
simulation investigations on ionic liquids by McDaniel and
Son,66 Afandak and Eslami,67 and Gudla et al.68 have shown
the need to include the contributions from various interionic
interactions to improve the accuracy of ionic conductivity.
Hence, we also calculate the total ionic conductivity, which
accounts for the various interactions that occur among and
between Na+ and PF6− ions. The total ionic conductivity
(σtot)66−68 is given by:
σtot = σNE + σcorr (4)

σcorr = σ++ + σ −− + σ+− (5)

Figure 10. Self-diffusion coefficients of (a) Na+ ions, (b) PF6− ions,
and (c) diglyme at various concentrations of NaPF6/diglyme.
The ion−ion correlation contributions to σtot is calculated
from the production run trajectory. However, the calculated
ion correlation as seen from eqs 678 (also called as collective
(discussed in the previous subsection) also infers that ionic MSDs) at longer time scales shows a nonlinear behavior (see
association could increase from 0.75 to 2.0 M, and hence the Figure S7A−C in the Supporting Information). The linear
average lifetime of an ion pair increases from less than 1 ns to regime to calculate σtot is observed only for a relatively short
greater than 10 ns. Furthermore, the magnitude of self- timescale.66 Hence, we have extracted the linear regime from
diffusion coefficients of ions and diglyme and the concen- ∼1 to 2 ns elapsed time from the entire trajectory. Figure 11b
tration and temperature dependence obtained from Green− shows a plot of σtot vs concentration. At 400 K, the value of σtot
Kubo relations are also similar to that of Einstein’s diffusion increases from 1.44 S/m (0.75 M) and attains a maximum
(Figure S6 in the Supporting Information). value of 2.26 S/m at 1.5 M and subsequently decreases with
3.2.2.4. Ionic Conductivity. The ionic conductivity of the further increase in concentration. Similar trends are observed
system is calculated using the Nernst−Einstein’s (NE) at other temperatures such as 348 and 378 K. Figure S7A−C
relationship and can be written as: shows that the contribution of ion−ion correlations to the total
N ionic conductivity is smaller compared with σNE and also a net
σNE = σ+ + σ − = (q 2D+ + q−2D−), negative contribution arises from anticorrelated cation−anion
VkBT + (3) interactionsespecially at higher temperatures. This leads to a

Figure 11. (a) σNE and (b) σtot vs concentration of NaPF6/diglyme at various temperatures.

2126 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

lower value of σtot compared with σNE and is consistent with classical MD simulations. Finally, the insights from this work
several earlier reports,66−68 which have shown that σNE with the effect of concentration and temperature on structure
overestimates ionic conductivity. In contrast to σNE, σtot and dynamics can lead to the development of alternatives to
reproduces the trends of optimal conductivity (1.5 M) with conventional battery electrolytes.
ion concentration similar to that observed from experiments27
especially at higher temperatures (T = 340, 378, 400 K).
The σ++, σ−−, and σ+− are cation−cation, anion−anion, and
■ ASSOCIATED CONTENT
* Supporting Information
sı
cation−anion correlated interactions can be written as:66
The Supporting Information is available free of charge at
n
1 +
https://pubs.acs.org/doi/10.1021/acs.jpcb.2c00557.
σ++ = lim ∑ ⟨(q+[R i(t ) − R i(0)])
6VkBT t →∞ i , j ≠ 1 Mass density, additional RDFs, MSD plots, residence
correlation functions, ion cluster analysis, mean
·(q+[R j(t ) − R j(0)])⟩ (6) residence times, Green−Kubo diffusion, and ion
n
correlation plots are provided in the Supporting
−
1 Information (PDF)
σ −− = lim ∑ ⟨(q−[R i(t ) − R i(0)])

■
6VkBT t →∞ i , j ≠ 1
AUTHOR INFORMATION
·(q−[R j(t ) − R j(0)])⟩ (7)
Corresponding Author
2
n+ n− Arun Venkatnathan − Department of Chemistry and Centre
σ+− = lim ∑ ∑ ⟨(q+[R i(t ) − R i(0)]) for Energy Science, Indian Institute of Science Education and
6VkBT t →∞ i j Research Pune, Pune 411008, India; orcid.org/0000-
0001-8450-5417; Phone: +91-20-2590-8085;
·(q−[R j(t ) − R j(0)])⟩ (8) Email: [email protected]; Fax: +91-20-2586-5315
where n+ and n− are the number of cations and anions with Authors
charges q+ and q−, and Ri and Rj are positions for any atom or
Shylendran Ardhra − Department of Chemistry and Centre
ion i and j, respectively.
for Energy Science, Indian Institute of Science Education and
4. CONCLUSIONS Research Pune, Pune 411008, India
Prabhat Prakash − Department of Chemistry and Centre for
The effect of concentration and temperature on structure and Energy Science, Indian Institute of Science Education and
dynamics in NaPF6/diglyme system is explored in this Research Pune, Pune 411008, India; Materials Science and
investigation using classical MD simulations. The properties Engineering, Indian Institute of Technology Gandhinagar,
examined are RDFs and mean residence time corresponding to Gandhinagar 382355 Gujarat, India; orcid.org/0000-
ion−ion and ion−diglyme interactions, population analysis of 0003-1430-2379
ion clusters, diffusion of ions, and diglyme and ionic Rabin Siva Dev − Department of Chemistry and Centre for
conductivity. The RDFs validate some of the experimental Energy Science, Indian Institute of Science Education and
hypotheses, for example, weak interactions between H- Research Pune, Pune 411008, India
(diglyme) and F(PF6−) atoms of anion show a good agreement
with experimental RDF obtained from the structure factor. The Complete contact information is available at:
results from mean residence time calculations suggest that high https://pubs.acs.org/10.1021/acs.jpcb.2c00557
ionic concentration and low temperature lead to a longer
residence lifetime. The RDFs and the mean residence time Notes
suggest that Na+ ions interact more prominently with the The authors declare no competing financial interest.
solvent molecules than the PF6− forming solvent-separated
species even for 2.0 M concentration at 298 K. These findings
are supported by the population analysis of various ion clusters
■ ACKNOWLEDGMENTS
The authors acknowledge the National Supercomputing
in the system, suggesting that the speciation of ions at different Mission (NSM) ‘PARAM Brahma’ at IISER Pune, which is
concentrations follows the trend as SolI ≫ CIPs > AGIPs. The implemented by C-DAC and supported by the Ministry of
CIP and AGIP clusters increase with the increase in Electronics and Information Technology (MeitY) and Depart-
temperature and concentration. The distribution plots for ment of Science and Technology (DST), Government of
displacement and lifetime of SSIP provided in this work India. AV acknowledges SERB DST CRG/2018/001536 grant
provide a complete view of the various ionic clusters compared for funding. SA acknowledges DST-INSPIRE for a graduate
with the previously reported work based on RDFs63 and fellowship. The authors also thank Dr. Jesse G. McDaniel from
spectroscopy.23,39The calculated self-diffusion coefficients the Georgia Institute of Technology for providing the code for
show that for any given temperature and concentration; the the analysis of ionic conductivity.

■
order of mobility of the species is Dsol > D− > D+. The self-
diffusion coefficients of ions decrease with increasing ion REFERENCES
concentration and increase with increasing temperature. The (1) Yang, Q.; Zhang, Z.; Sun, X.-G.; Hu, Y.-S.; Xing, H.; Dai, S. Ionic
inclusion of correlated motions of ions in the calculation of Liquids and Derived Materials for Lithium and Sodium Batteries.
ionic conductivity (σtot) shows similar trends to experiments Chem. Soc. Rev. 2018, 47, 2020−2064.
especially at elevated temperatures. However, the quantitative (2) Bin, D.; Wang, F.; Tamirat, A. G.; Suo, L.; Wang, Y.; Wang, C.;
differences in σtot from the simulations and experiments can be Xia, Y. Progress in Aqueous Rechargeable Sodium-Ion Batteries. Adv.
reduced by the development of polarizable force fields for Energy Mater. 2018, 8, No. 1703008.

2127 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

(3) Wang, J.; Yamada, Y.; Sodeyama, K.; Watanabe, E.; Takada, K.; Triglyme, or Tetraglyme+n-Heptane) at Several Temperatures. J.
Tateyama, Y.; Yamada, A. Fire-Extinguishing Organic Electrolytes for Chem. Thermodyn. 2011, 43, 275−283.
Safe Batteries. Nat. Energy 2018, 3, 22−29. (23) Li, K.; Galle Kankanamge, S. R.; Weldeghiorghis, T. K.; Jorn,
(4) Xu, Z.-L.; Yoon, G.; Park, K.-Y.; Park, H.; Tamwattana, O.; Joo R.; Kuroda, D. G.; Kumar, R. Predicting Ion Association in Sodium
Kim, S.; Seong, W. M.; Kang, K. Tailoring Sodium Intercalation in Electrolytes: A Transferrable Model for Investigating Glymes. J. Phys.
Graphite for High Energy and Power Sodium Ion Batteries. Nat. Chem. C 2018, 122, 4747−4756.
Commun. 2019, 10, 2598. (24) Tsuzuki, S.; Mandai, T.; Suzuki, S.; Shinoda, W.; Nakamura, T.;
(5) Hayashi, A.; Noi, K.; Sakuda, A.; Tatsumisago, M. Superionic Morishita, T.; Ueno, K.; Seki, S.; Umebayashi, Y.; Dokko, K.; et al.
Glass-Ceramic Electrolytes for Room-Temperature Rechargeable Effect of the Cation on the Stability of Cation−Glyme Complexes and
Sodium Batteries. Nat. Commun. 2012, 3, 856. Their Interactions with the [TFSA] − Anion. Phys. Chem. Chem. Phys.
(6) Ponrouch, A.; Monti, D.; Boschin, A.; Steen, B.; Johansson, P.; 2017, 19, 18262−18272.
Palacín, M. R. Non-Aqueous Electrolytes for Sodium-Ion Batteries. J. (25) Yoshida, K.; Nakamura, M.; Kazue, Y.; Tachikawa, N.; Tsuzuki,
Mater. Chem. A 2015, 3, 22−42. S.; Seki, S.; Dokko, K.; Watanabe, M. Oxidative-Stability Enhance-
(7) Famprikis, T.; Canepa, P.; Dawson, J. A.; Islam, M. S.; ment and Charge Transport Mechanism in Glyme−Lithium Salt
Masquelier, C. Fundamentals of Inorganic Solid-State Electrolytes for Equimolar Complexes. J. Am. Chem. Soc. 2011, 133, 13121−13129.
Batteries. Nat. Mater. 2019, 18, 1278−1291. (26) Mandai, T.; Yoshida, K.; Tsuzuki, S.; Nozawa, R.; Masu, H.;
(8) Wang, Y.; Song, S.; Xu, C.; Hu, N.; Molenda, J.; Lu, L. Ueno, K.; Dokko, K.; Watanabe, M. Effect of Ionic Size on Solvate
Development of Solid-State Electrolytes for Sodium-Ion Battery−A Stability of Glyme-Based Solvate Ionic Liquids. J. Phys. Chem. B 2015,
Short Review. Nano Mater. Sci. 2019, 1, 91−100. 119, 1523−1534.
(9) Cresce, A. V.; Russell, S. M.; Borodin, O.; Allen, J. A.; Schroeder, (27) Westman, K.; Dugas, R.; Jankowski, P.; Wieczorek, W.; Gachot,
M. A.; Dai, M.; Peng, J.; Gobet, M. P.; Greenbaum, S. G.; Rogers, R. G.; Morcrette, M.; Irisarri, E.; Ponrouch, A.; Palacín, M. R.; Tarascon,
E.; et al. Solvation Behavior of Carbonate-Based Electrolytes in J. M.; et al. Diglyme Based Electrolytes for Sodium-Ion Batteries. ACS
Sodium Ion Batteries. Phys. Chem. Chem. Phys. 2017, 19, 574−586. Appl. Energy Mater. 2018, 1, 2671−2680.
(10) Shakourian-Fard, M.; Kamath, G.; Smith, K.; Xiong, H.; (28) Li, Y.; Lu, Y.; Adelhelm, P.; Titirici, M.-M.; Hu, Y.-S.
Sankaranarayanan, S. K. R. S. Trends in Na-Ion Solvation with Alkyl- Intercalation Chemistry of Graphite: Alkali Metal Ions and Beyond.
Carbonate Electrolytes for Sodium-Ion Batteries: Insights from First- Chem. Soc. Rev. 2019, 48, 4655−4687.
Principles Calculations. J. Phys. Chem. C 2015, 119, 22747−22759. (29) Stevens, D. A.; Dahn, J. R. High Capacity Anode Materials for
(11) Abd Azes, N. I.; Yusoff, N. H.; Najmi, N.; Sulaiman, K. S.; Rechargeable Sodium-Ion Batteries. J. Electrochem. Soc. 2000, 147,
Sulaiman, M. A. Conductivity and Electrochemical Study of Non 1271.
(30) Nobuhara, K.; Nakayama, H.; Nose, M.; Nakanishi, S.; Iba, H.
Aqueous NaPF6 Electrolyte in Organic Solvent Mixture for Sodium-
First-Principles Study of Alkali Metal-Graphite Intercalation Com-
Ion Batteries. J. Ind. Technol. 2017, 25, 19−28.
pounds. J. Power Sources 2013, 243, 585−587.
(12) Hess, S.; Wohlfahrt-Mehrens, M.; Wachtler, M. Flammability of
(31) Jache, B.; Binder, J. O.; Abe, T.; Adelhelm, P. A Comparative
Li-Ion Battery Electrolytes: Flash Point and Self-Extinguishing Time
Study on the Impact of Different Glymes and Their Derivatives as
Measurements. J. Electrochem. Soc. 2015, 162, A3084−A3097.
Electrolyte Solvents for Graphite Co-Intercalation Electrodes in
(13) Mogensen, R.; Brandell, D.; Younesi, R. Solubility of the Solid
Lithium-Ion and Sodium-Ion Batteries. Phys. Chem. Chem. Phys. 2016,
Electrolyte Interphase (SEI) in Sodium Ion Batteries. ACS Energy
18, 14299−14316.
Lett. 2016, 1, 1173−1178.
(32) Jache, B.; Adelhelm, P. Use of Graphite as a Highly Reversible
(14) Ma, L. A.; Naylor, A. J.; Nyholm, L.; Younesi, R. Strategies for
Electrode with Superior Cycle Life for Sodium-Ion Batteries by
Mitigating Dissolution of Solid Electrolyte Interphases in Sodium-Ion Making Use of Co-Intercalation Phenomena. Angew. Chem. Int. Ed.
Batteries. Angew. Chem. Int. Ed. 2021, 133, 4905−4913. 2014, 126, 10333−10337.
(15) Kumar, H.; Detsi, E.; Abraham, D. P.; Shenoy, V. B. (33) Kim, H.; Hong, J.; Park, Y.-U.; Kim, J.; Hwang, I.; Kang, K.
Fundamental Mechanisms of Solvent Decomposition Involved in Sodium Storage Behavior in Natural Graphite Using Ether-Based
Solid-Electrolyte Interphase Formation in Sodium Ion Batteries. Electrolyte Systems. Adv. Funct. Mater. 2015, 25, 534−541.
Chem. Mater. 2016, 28, 8930−8941. (34) Seh, Z. W.; Sun, J.; Sun, Y.; Cui, Y. A Highly Reversible Room-
(16) Muñoz-Márquez, M. A.; Zarrabeitia, M.; Castillo-Martínez, E.; Temperature Sodium Metal Anode. ACS Cent. Sci. 2015, 1, 449−455.
Eguía-Barrio, A.; Rojo, T.; Casas-Cabanas, M. Composition and (35) Zhang, C.; Yamazaki, A.; Murai, J.; Park, J. W.; Mandai, T.;
Evolution of the Solid-Electrolyte Interphase in Na 2 Ti 3 O 7 Ueno, K.; Dokko, K.; Watanabe, M. Chelate Effects in Glyme/
Electrodes for Na-Ion Batteries: XPS and Auger Parameter Analysis. Lithium Bis(Trifluoromethanesulfonyl)Amide Solvate Ionic Liquids,
ACS Appl. Mater. Interfaces 2015, 7, 7801−7808. Part 2: Importance of Solvate-Structure Stability for Electrolytes of
(17) Yoshida, K.; Tsuchiya, M.; Tachikawa, N.; Dokko, K.; Lithium Batteries. J. Phys. Chem. C 2014, 118, 17362−17373.
Watanabe, M. Change from Glyme Solutions to Quasi-Ionic Liquids (36) Maibach, J.; Jeschull, F.; Brandell, D.; Edström, K.; Valvo, M.
for Binary Mixtures Consisting of Lithium Bis- Surface Layer Evolution on Graphite During Electrochemical Sodium-
(Trifluoromethanesulfonyl)Amide and Glymes. J. Phys. Chem. C Tetraglyme Co-Intercalation. ACS Appl. Mater. Interfaces 2017, 9,
2011, 115, 18384−18394. 12373−12381.
(18) Frech, R.; Huang, W. Conformational Changes in Diethylene (37) Goktas, M.; Bolli, C.; Berg, E. J.; Novák, P.; Pollok, K.;
Glycol Dimethyl Ether and Poly(Ethylene Oxide) Induced by Langenhorst, F.; Roeder, M. V.; Lenchuk, O.; Mollenhauer, D.;
Lithium Ion Complexation. Macromolecules 1995, 28, 1246−1251. Adelhelm, P. Graphite as Cointercalation Electrode for Sodium-Ion
(19) Tang, S.; Zhao, H. Glymes as Versatile Solvents for Chemical Batteries: Electrode Dynamics and the Missing Solid Electrolyte
Reactions and Processes: From the Laboratory to Industry. RSC Adv. Interphase (SEI). Adv. Energy Mater. 2018, 8, No. 1702724.
2014, 4, 11251. (38) Goktas, M.; Bolli, C.; Buchheim, J.; Berg, E. J.; Novák, P.;
(20) Hefter, G. T.; Salomon, M. Conductivities of 1?1 Salts in 2- Bonilla, F.; Rojo, T.; Komaba, S.; Kubota, K.; Adelhelm, P. Stable and
Cyanopyridine. J. Solution Chem. 1994, 23, 579−593. Unstable Diglyme-Based Electrolytes for Batteries with Sodium or
(21) Horwitz, G.; Factorovich, M.; Rodriguez, J.; Laria, D.; Corti, H. Graphite as Electrode. ACS Appl. Mater. Interfaces 2019, 11, 32844−
R. Ionic Transport and Speciation of Lithium Salts in Glymes: 32855.
Experimental and Theoretical Results for Electrolytes of Interest for (39) Wahlers, J.; Fulfer, K. D.; Harding, D. P.; Kuroda, D. G.;
Lithium-Air Batteries. ACS Omega 2018, 3, 11205−11215. Kumar, R.; Jorn, R. Solvation Structure and Concentration in Glyme-
(22) Riadigos, C. F.; Iglesias, R.; Rivas, M. A.; Iglesias, T. P. Based Sodium Electrolytes: A Combined Spectroscopic and
Permittivity and Density of the Systems (Monoglyme, Diglyme, Computational Study. J. Phys. Chem. C 2016, 120, 17949−17959.

2128 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129
The Journal of Physical Chemistry B pubs.acs.org/JPCB Article

(40) Borodin, O.; Olguin, M.; Ganesh, P.; Kent, P. R. C.; Allen, J. L.; (60) Brehm, M.; Thomas, M.; Gehrke, S.; Kirchner, B. TRAVISA
Henderson, W. A. Competitive Lithium Solvation of Linear and Free Analyzer for Trajectories from Molecular Simulation. J. Chem.
Cyclic Carbonates from Quantum Chemistry. Phys. Chem. Chem. Phys. 2020, 152, 164105.
Phys. 2016, 18, 164−175. (61) Mandai, T.; Nozawa, R.; Tsuzuki, S.; Yoshida, K.; Ueno, K.;
(41) Jónsson, E.; Johansson, P. Modern Battery Electrolytes: Ion-Ion Dokko, K.; Watanabe, M. Phase Diagrams and Solvate Structures of
Interactions in Li +/Na + Conductors from DFT Calculations. Phys. Binary Mixtures of Glymes and Na Salts. J. Phys. Chem. B 2013, 117,
Chem. Chem. Phys. 2012, 14, 10774−10779. 15072−15085.
(42) Abroshan, H.; Dhumal, N. R.; Shim, Y.; Kim, H. J. Theoretical (62) Luzar, A. Resolving the Hydrogen Bond Dynamics Conun-
Study of Interactions of a Li + (CF 3 SO 2) 2 N − Ion Pair with CR 3 drum. J. Chem. Phys. 2000, 113, 10663−10675.
(OCR 2 CR 2) n OCR 3 (R = H or F). Phys. Chem. Chem. Phys. (63) Rajput, N. N.; Qu, X.; Sa, N.; Burrell, A. K.; Persson, K. A. The
2016, 18, 6754−6762. Coupling between Stability and Ion Pair Formation in Magnesium
(43) Dhumal, N. R.; Gejji, S. P. Theoretical Studies on Blue versus Electrolytes from First-Principles Quantum Mechanics and Classical
Molecular Dynamics. J. Am. Chem. Soc. 2015, 137, 3411−3420.
Red Shifts in Diglyme-M +-X- (M = Li, Na, and K and X = CF3so 3,
(64) Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids;
PF6, and (CF3so2)2N). J. Phys. Chem. A 2006, 110, 219−227. Oxford University Press: Oxford, 1987.
(44) Tsuzuki, S.; Shinoda, W.; Matsugami, M.; Umebayashi, Y.; (65) Li, K.; Subasinghege Don, V.; Gupta, C. S.; David, R.; Kumar,
Ueno, K.; Mandai, T.; Seki, S.; Dokko, K.; Watanabe, M. Structures of R. Effect of Anion Identity on Ion Association and Dynamics of
[Li(Glyme)]+ Complexes and Their Interactions with Anions in Sodium Ions in Non-Aqueous Glyme Based Electrolytes - OTf vs
Equimolar Mixtures of Glymes and Li[TFSA]: Analysis by Molecular TFSI. J. Chem. Phys. 2021, 154, 184505.
Dynamics Simulations. Phys. Chem. Chem. Phys. 2015, 17, 126−129. (66) McDaniel, J. G.; Son, C. Y. Ion Correlation and Collective
(45) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Dynamics in BMIM/BF 4 -Based Organic Electrolytes: From Dilute
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, Solutions to the Ionic Liquid Limit. J. Phys. Chem. B 2018, 122,
B.; Petersson, G. A., et al. Gaussian 09, Revision E.01; Gaussian Inc.: 7154−7169.
Wallingford, CT, 2009. (67) Afandak, A.; Eslami, H. Ion-Pairing and Electrical Conductivity
(46) Martínez, L.; Andrade, R.; Birgin, E. G.; Martínez, J. M. in the Ionic Liquid 1- n -Butyl-3-Methylimidazolium Methylsulfate
PACKMOL: A Package for Building Initial Configurations for [Bmim][MeSO 4 ]: Molecular Dynamics Simulation Study. J. Phys.
Molecular Dynamics Simulations. J. Comput. Chem. 2009, 30, Chem. B 2017, 121, 7699−7708.
2157−2164. (68) Gudla, H.; Shao, Y.; Phunnarungsi, S.; Brandell, D.; Zhang, C.
(47) Abraham, M. J.; Murtola, T.; Schulz, R.; Páll, S.; Smith, J. C.; Importance of the Ion-Pair Lifetime in Polymer Electrolytes. J. Phys.
Hess, B.; Lindahl, E. GROMACS: High Performance Molecular Chem. Lett. 2021, 12, 8460−8464.
Simulations through Multi-Level Parallelism from Laptops to
Supercomputers. SoftwareX 2015, 1-2, 19−25.
(48) Barbosa, N. S. V.; Zhang, Y.; Lima, E. R. A.; Tavares, F. W.;
Maginn, E. J. Development of an AMBER-Compatible Transferable
Force Field for Poly(Ethylene Glycol) Ethers (Glymes). J. Mol. Model.
2017, 23, 194.
(49) Jensen, A. C. S.; Au, H.; Gärtner, S.; Titirici, M.; Drew, A. J. Recommended by ACS
Solvation of NaPF 6 in Diglyme Solution for Battery Electrolytes.
Batter. Supercaps 2020, 3, 1306−1310. Dynamic Heterogeneity of Solvent Motion and Ion Transport
(50) Payne, M. C.; Teter, M. P.; Allan, D. C.; Arias, T. A.; in Concentrated Electrolytes
Joannopoulos, J. D. Iterative Minimization Techniques for Ab Initio
Chao Fang, Rui Wang, et al.
Total-Energy Calculations: Molecular Dynamics and Conjugate
FEBRUARY 17, 2023
Gradients. Rev. Mod. Phys. 1992, 64, 1045−1097. THE JOURNAL OF PHYSICAL CHEMISTRY B
(51) Bussi, G.; Donadio, D.; Parrinello, M. Canonical Sampling READ
through Velocity Rescaling. J. Chem. Phys. 2007, 126, No. 014101.
(52) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Ionic Association in
DiNola, A.; Haak, J. R. Molecular Dynamics with Coupling to an CH3–(CH2–CF2)n –CH3(PVDF)–Li+–(CF3SO2)2N– for n = 1, 4:
External Bath. J. Chem. Phys. 1984, 81, 3684−3690. A Computational Approach
(53) Kühne, T. D.; Iannuzzi, M.; Del Ben, M.; Rybkin, V. V.; Mathew Daniel, Nilesh R. Dhumal, et al.
Seewald, P.; Stein, F.; Laino, T.; Khaliullin, R. Z.; Schütt, O.; FEBRUARY 15, 2022
Schiffmann, F.; et al. CP2K: An Electronic Structure and Molecular ACS OMEGA READ
Dynamics Software Package - Quickstep: Efficient and Accurate
Electronic Structure Calculations. J. Chem. Phys. 2020, 152, 194103. Salt Activity Coefficient and Chain Statistics in Poly(ethylene
(54) VandeVondele, J.; Hutter, J. Gaussian Basis Sets for Accurate oxide)-Based Electrolytes
Calculations on Molecular Systems in Gas and Condensed Phases. J. Chao Fang, Rui Wang, et al.
Chem. Phys. 2007, 127, 114105. MARCH 02, 2021
(55) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient MACROMOLECULES READ
Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868.
(56) Goedecker, S.; Teter, M.; Hutter, J. Separable Dual-Space
Onsager Transport Coefficients and Transference Numbers
Gaussian Pseudopotentials. Phys. Rev. B 1996, 54, 1703−1710.
in Polyelectrolyte Solutions and Polymerized Ionic Liquids
(57) Krack, M. Pseudopotentials for H to Kr Optimized for
Gradient-Corrected Exchange-Correlation Functionals. Theor. Chem. Kara D. Fong, Kristin A. Persson, et al.
Acc. 2005, 114, 145−152. OCTOBER 21, 2020
(58) Martyna, G. J.; Klein, M. L.; Tuckerman, M. Nosé-Hoover MACROMOLECULES READ
Chains: The Canonical Ensemble via Continuous Dynamics. J. Chem.
Phys. 1992, 97, 2635−2643.
Get More Suggestions >
(59) Nosé, S. A Unified Formulation of the Constant Temperature
Molecular Dynamics Methods. J. Chem. Phys. 1984, 81, 511−519.

2129 https://doi.org/10.1021/acs.jpcb.2c00557
J. Phys. Chem. B 2022, 126, 2119−2129