N-substituted phenothiazine derivatives: How the stability of the neutral and radical cation forms affects overcharge performance in lithium-ion batteries

Article abstract of DOI:10.1002/cphc.201402674

Phenothiazine and five N-substituted derivatives were evaluated as electrolyte additives for overcharge protection in LiFePO₄/synthetic graphite lithium-ion batteries. We report on the stability and reactivity of both the neutral and radical-cation forms of these six compounds. While three of the compounds show extensive overcharge protection, the remaining three last for only one to a few cycles. UV/Vis studies of redox shuttle stability in the radical cation form are consistent with the overcharge performance: redox shuttles with spectra that show little change over time exhibit extensive overcharge performance, whereas those with changing spectra have limited overcharge protection. In one case, we determined that a C-N bond cleaves upon oxidation, forming the phenothiazine radical cation and leading to premature overcharge protection failure; in another case, poor solubility appears to limit protection.

Full text of DOI:10.1002/cphc.201402674

CHEMPHYSCHEM
ARTICLES
DOI: 10.1002/cphc.201402674
N-Substituted Phenothiazine Derivatives: How the
Stability of the Neutral and Radical Cation Forms Affects
Overcharge Performance in Lithium-Ion Batteries
Kishore Anand Narayana,[a] Matthew D. Casselman,[a] Corrine F. Elliott,[a] Selin Ergun,[a]
Sean R. Parkin,[a] Chad Risko,*[a, b] and Susan A. Odom*[a]
Phenothiazine and five N-substituted derivatives were evaluat-
ed as electrolyte additives for overcharge protection in
LiFePO4/synthetic graphite lithium-ion batteries. We report on
the stability and reactivity of both the neutral and radical-
cation forms of these six compounds. While three of the com-
pounds show extensive overcharge protection, the remaining
three last for only one to a few cycles. UV/Vis studies of redox
shuttle stability in the radical cation form are consistent with
the overcharge performance: redox shuttles with spectra that
show little change over time exhibit extensive overcharge per-
formance, whereas those with changing spectra have limited
overcharge protection. In one case, we determined that a CÀN
bond cleaves upon oxidation, forming the phenothiazine radi-
cal cation and leading to premature overcharge protection fail-
ure; in another case, poor solubility appears to limit protec-
tion.
1. Introduction
The lifetimes of lithium-ion batteries (LIBs) are limited by nu-
merous chemical and mechanical degradation mechanisms,
one of which occurs when the voltage of a battery rises past
its end-of-charge potential, a condition called overcharge.
Compounds called redox shuttles can be incorporated into the
battery electrolyte to limit cell voltage and thus prevent over-
charge.[1] These compounds shuttle current through the bat-
tery electrolyte by oxidizing at the cathode/electrolyte inter-
face, then diffusing to the anode/electrolyte interface, where
reduction occurs, regenerating the neutral form. Various organ-
ic compounds and a few metallocenes and inorganic salts
have been shown to protect batteries from overcharge with
varying degrees of success.[2]
lithium difluoro(oxalato)borate[2l] derivatives have also been re-
ported as multifunctional electrolyte additives for overcharge
protection. Except for derivatives reported by our group,[2m,n]
PT remains an untapped aromatic core.
Although alkylated phenothiazine additives were tested with
LiFePO4 cathodes, the low oxidation potentials of N-substituted
PT derivatives render them impractical for commercial applica-
tion for overcharge in LIBs, and the incorporation of electron-
withdrawing substituents is necessary to produce derivatives
with practical (larger) oxidation potentials, as has been report-
ed with dimethoxybenzene[2g,k] and ferrocene derivatives.[2p]
Modification of the PT core is facile, and transformations can
be high yielding, especially in the synthesis of derivatives with
substituents at the N position and positions para to the nitro-
gen atom.
N-Substituted phenothiazines (PTs) were originally reported
as redox shuttles for overcharge protection of LIBs in 2006 by
Dahn and coworkers,[2c,3] with some derivatives exhibiting
about 150 cycles of 100% overcharge when incorporated in
the electrolyte of LiFePO4/Li4/3Ti5/3O4 coin cells. Many recent
studies have focused on overcharge protection using dialkoxy-
benzene derivatives[2e,h,4] including some with electron-with-
drawing substituents to match end-of-charge potentials for
higher voltage cathodes.[2d,g] Fluorinated boronic ester[5] and
Using PT as a common redox-active core, our group has
studied the overcharge performance of a variety of disubstitut-
ed derivatives of N-ethylphenothiazine (EPT).[2m,n,6] Although
some EPT derivatives with substituents at the 3- and 7-posi-
tions exhibit extensive overcharge protection,[2m,n,6] we were
unsure if our choice of the ethyl substituent at the N position
was the best choice for long-lived redox shuttles. Dahn report-
ed PT derivatives with methyl (MPT), ethyl (EPT), and isopropyl
(iPrPT) substituents (Figure 1); little variation in performance
was observed when the compounds were evaluated in side-
by-side experiments in LiFePO4/Li4/3Ti5/3O4 coin cells.[3a] To iden-
tify the optimal substituent for the N position, we wanted to
compare the overcharge performance of the reported PT deriv-
atives and the unreported redox shuttle candidates PT, N-tert-
butylphenothiazine (tBuPT), and N-phenylphenothiazine (PhPT)
(Figure 1) in LiFePO4/synthetic graphite coin cells.
[a] K. A. Narayana, Dr. M. D. Casselman, C. F. Elliott, Dr. S. Ergun,
Dr. S. R. Parkin, Prof. C. Risko, Prof. S. A. Odom
Department of Chemistry, University of Kentucky
Lexington, KY, 40506-0055 (USA)
E-mail: chad.risko@uky.edu
susan.odom@uky.edu
[b] Prof. C. Risko
Center for Applied Energy Research, Lexington, KY, 40511 (USA)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cphc.201402674.
In addition to evaluating the overcharge performance, we
wanted to determine whether the stability of the neutral and
An invited contribution to a Special Issue on Organic Electronics
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
1
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
Table 1. Oxidation potentials and number of protection cycles survived
in LiFePO4/synthetic graphite coin cell batteries.
+/0
+/0
Redox
shuttle
candidate
E1/2
E1/2
Number of
(vs. Cp2Fe+/0) [V]
in 0.1 M nBu4NPF6
in DCM
(vs. Li+/0) [V]
in 1.2 M LiPF6
in EC/EMC (3:7)
cycles of 100%
overcharge
protection[a]
PT
MPT
EPT
iPrPT
tBuPT
PhPT
0.17
0.31
0.26
0.33
0.53
0.26
3.45
3.55
3.51
3.59
3.45
3.52
1, 3
3, 5
127, 139
117, 118
2, 2
128, 161
Figure 1. Phenothiazine (R=H) and its N-substituted derivatives (R=CH3,
[a] The two numbers represent the values for two batteries cycled in
overcharge.
CH2CH3, CH(CH3)2, C(CH3)3, and C6H5) examined in this study.
oxidized forms of the redox shuttles as dilute solutions in rela-
tively inert organic solvents correlated with the extent of over-
charge protection observed. This goal is related to our interest
in predicting overcharge performance of redox shuttle candi-
dates prior to battery fabrication.[2n,7] Understanding the stabili-
ty and reactivity of redox shuttles in their neutral and oxidized
states in different solvents—not necessarily those specific to
the battery environment—may help to predict those candi-
dates that will exhibit extensive overcharge protection. In addi-
tion to using UV/Vis absorption spectroscopy to monitor redox
shuttle concentration in solution over time, we made use of
electron paramagnetic resonance (EPR) spectroscopy and mass
spectrometry to identify some of the byproducts of radical
cation formation—an area that we think deserves more atten-
tion in the redox shuttle litera-
stituent. Trends in oxidation potentials generally correlate in
value from one electrolyte to the other with one exception
(tBuPT). In nBu4NPF6/DCM, the oxidation potential of tBuPT is
higher than all of the derivatives, but in LiPF6/EC/EMC, it is tied
with PT for the lowest value.
2.2. Overcharge Performance
The overcharge performance of N-substituted phenothiazine
derivatives was analyzed in LiFePO4/synthetic graphite coin cell
batteries containing 0.08m redox shuttle in the electrolyte
1.2m LiPF6 in EC/EMC (3:7 wt. ratio). The redox shuttles (all
solids) fully dissolved within a few minutes of sonication
except for tBuPT, which took ca. 10 min to dissolve. Over-
ture—and density functional
theory (DFT) calculations to cal-
culate the energies of reaction
for decomposition pathways
that are consistent with experi-
mental observations.
2. Results and
Discussion
2.1. Cyclic Voltammetry
We recorded cyclic voltammo-
grams of the N-substituted PT
derivatives in 0.1m nBu4NPF6 in
dichloromethane (DCM) and in
1.2m LiPF6 in ethylene carbon-
ate/ethyl methyl carbonate (EC/
EMC, 3:7 wt. ratio), a common
electrolyte for LIBs. The voltam-
mograms obtained in each elec-
trolyte are shown in Figure 2.
Voltammograms recorded at
100 mVsÀ1 show reversible first
oxidations for all derivatives.
Figure 2. Cyclic voltammograms recorded at 100 mVsÀ1 showing the first oxidation observed for PT, MPT, EPT,
Half-wave oxidation potentials
(Table 1) varied slightly with sub- nPrPT, iPrPT, tBuPT, and PhPT at 3ꢁ10À4 m in 0.1m nBu4NPF6 in DCM (left) and in 1.2m LiPF6 in EC/EMC (right).
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
2
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
Figure 3. Voltage versus time plots of selected cycles for LiFePO4/synthetic graphite coin cell batteries cycled to 100% overcharge containing PT, MPT, EPT,
iPrPT, tBuPT, or PhPT at 0.08m in 1.2m LiPF6 in EC/EMC.
charge cycling was performed by charging the coin cells to
200% of the nominal cell capacity, or 100% overcharge, at
a C/10 charging rate. Specifically, this means that the coin cells
were programmed to charge for 20 h (10 h in charge, 10 h in
overcharge) or until 5.0 V was reached and then to discharge
for 10 h or until 3.0 V was reached. Two batteries were made
for each redox shuttle. Selected cycles of overcharge are
shown in Figure 3, plotted as voltage vs. time. The number of
cycles that each redox shuttle survived in overcharge is sum-
marized in Table 1.
derstand the difference in relative oxidation potential for
tBuPT between the two electrolytes.
2.3. Analysis of Cyclic Voltammetry and Overcharge
We examined the trends in oxidation potential for the PT de-
rivatives. A regression line fit by least-squares to a plot of oxi-
dation potentials in battery electrolyte versus those observed
in nBu4NPF6/DCM shows good correlation between the oxida-
tion potentials in each electrolyte (R2 =0.98) except for tBuPT,
which was excluded from the fit (Figure 4a). As mentioned in
Section 2.1, the oxidation potential of tBuPT in battery electro-
lyte was unexpectedly low and, notably, matched that of PT in
battery electrolyte. We were prompted by this result to further
explore the electrochemical properties of PT and tBuPT.
It was no surprise that PT survived for only one to three
cycles of overcharge protection, as we expected this com-
pound to be unstable as a radical cation. Aromatic cores with
NÀH bonds can dimerize upon oxidation to form new NÀN
bonds[8] or react with aromatic rings.[9] Of the N-substituted
phenothiazines, EPT, iPrPT, and PhPT showed the most exten-
sive overcharge cycling with>100 cycles at 100% overcharge.
In contrast, MPT and tBuPT—the derivatives with the smallest
and largest steric substitution adjacent to the N-substituent—
survived for five or fewer cycles of overcharge protection.
We were surprised that MPT exhibited so few overcharge
protection cycles because Dahn’s report indicated that MPT
survived approximately the same number of cycles of protec-
tion (155) as EPT (150) and iPrPT (>162) in LiFePO4/Li4/3Ti5/3O4
batteries. These results stimulated us to perform additional ex-
periments to understand why MPT and tBuPT failed so rapidly
in overcharge compared to the other compounds, and to un-
In addition to scanning through the first oxidation of each
compound in the battery electrolyte, we continued scanning
to the edge of the solvent window for PT and tBuPT and plot-
ted the superimposed voltammograms in Figure 4b. When the
cyclic voltammograms of PT and tBuPT are compared directly
(first oxidation only, as well as both first and second oxida-
tions), one can see that they are nearly identical. Additionally,
the voltage vs. time plots from overcharge protection tests of
PT and tBuPT show similarities in the behavior of these batter-
ies (Figure 4c). Both results are consistent with tBuPT decom-
posing in the battery electrolyte to form PT, but not decom-
posing as a neutral compound in nBu4PF6/DCM.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
3
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
shuttle and can be (or could
have been) eliminated from bat-
tery testing. Since we have only
applied this method of analysis
to a few reported redox shuttles,
we performed the test with the
intent of fabricating batteries
from all of the compounds—re-
gardless of the result—to see
whether our predictions were
useful.
In this experiment, we gener-
ated radical cations in dilute sol-
utions by treating solutions of
neutral compound in DCM with
0.1 equivalents of the oxidant
tris(4-bromophenyl)aminium
hexachloroantimonate (TBPA* +),
which was added as a more con-
centrated solution in DCM. An
excess of neutral compound was
used to ensure that the reaction
with the oxidant was driven to
completion. Based on our previ-
ous analysis of UV/Vis spectra of
Figure 4. a) A plot of first oxidation potentials in 1.2m LiPF6 in EC/EMC (vs. Li+/0) at 0 V versus first oxidation po-
tentials in 0.1m nBu4NPF6 in DCM (vs. Cp2Fe+/0) at 0 V. Cyclic voltammograms of PT and tBuPT in 1.2m LiPF6 in EC/
EMC scanning to the first oxidations and to the second oxidations (b). Overcharge cycling of PT and tBuPT overlaid
(c).
2.4. Redox Shuttle UV/Vis
the EPT radical cation,[7] we chose to work with a radical cation
concentration of 1.6ꢁ10À4 m; assuming complete electron
transfer, this would also produce one equivalent of neutral
tris(4-bromophenyl)amine, and nine equivalents of neutral PT
or PT derivative would remain. These neutral compounds do
not absorb beyond 400 nm, allowing the region from 400 to
1100 nm to be used for analysis of radical cation absorption
spectra. The UV/Vis spectra obtained between 0 and 5 h after
We further explored our hypothesis by analyzing the stability
of the PT derivatives by UV/Vis spectroscopy. The compounds
were dissolved at 3.0ꢁ10À4 m in DCM, EC/EMC (3:7) or 1.2m
LiPF6 in EC/EMC (3:7). The UV/Vis spectra are shown in
Figure 5.
Analysis of the UV/Vis spectra showed that although the ab-
sorption spectra of most compounds were similar in shape
and position in all solvents, the absorption spectrum of tBuPT
differed significantly in LiPF6/EC/EMC as compared to DCM or
EC/EMC. In battery electrolyte, the absorption of the sample
that contained tBuPT was red-shifted compared to its position
in DCM or EC/EMC, instead absorbing most intensely at nearly
the same wavelength as PT. As with cyclic voltammetry, the
UV/Vis data suggest that tBuPT decomposes in battery electro-
lyte and forms PT; from these results, we can further infer that
PT is responsible for overcharge protection in batteries fabri-
cated with tBuPT, not the alkylated phenothiazine (tBuPT)
itself.
*
+
generation of the radical cations (symbolized by
following
the compound abbreviation) in DCM are shown in Figure 6.
*
*
+
EPT* +, iPrPT +, and PhPT showed relatively little change
in their spectra over the course of 5 h, indicating that their rad-
ical cations are relatively stable. These compounds survive ex-
tended overcharge cycling in LiFePO4/synthetic graphite bat-
*
*
+
teries. In contrast, PT* +, MPT +, and tBuPT all show changes
within 5 h. The spectra for MPT* lost intensity over time, al-
+
though their shapes did not change. We observed the forma-
tion of a fine precipitate in the cuvette from the MPT experi-
ment.
The shapes of the absorption spectra for both PT* and
+
tBuPT* change over time, suggesting the formation of one or
+
2.5. Radical-Cation UV/Vis
more new materials that absorb in the visible region. The initial
+
Although cyclic voltammetry and UV/Vis spectroscopy provid-
ed an explanation for what happened to tBuPT in the battery
electrolyte, the fate of MPT remained unclear. We thought the
answer might lie in the stability not of its neutral form but in-
stead of its radical-cation state. Additionally, we wanted to de-
termine whether the stability of the radical cations in this
series could be used to predict or explain redox shuttle per-
formance across the entire series, independent of the battery
environment. We suspect that if a radical cation is unstable in
a solvent like DCM, then it is unlikely to be an effective redox
absorption spectrum of tBuPT* is similar to the spectra of the
radical cations of N-alkylated PT derivatives although it is
slightly red-shifted, consistent with the formation of tBuPT*
.
+
However, within the course of 1 h, no appreciable amount of
this radical cation remains. Instead, new peaks appear that
show remarkable similarity to the peaks present in spectra of
*
PT* +. This led us to the conclusion that tBuPT decomposes
+
into PT* when it is oxidized. However, the absorption spec-
+
trum of PT* also changes; this indicates that, although the
+
species may be present upon decomposition of tBuPT* and
+
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
4
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
collected as soon as possible after radical cation generation
and another set collected at 10 min after generation are
shown for both samples in Figure 7.
In these plots, one can see that although the EPR spectra—
plotted as a derivative of signal intensity—are considerably dif-
ferent upon initial generation, the spectra at 10 min resemble
each other. Although the intensities of the peaks do not match
exactly, the number and position of peaks are generally similar
between the samples, again consistent with the hypothesis
that tBuPT* decomposes to form PT*
.
+
+
2.7. Bulk Electrolysis and Mass Spectrometry
Although EPR spectroscopy is useful for comparing samples
and analyzing solutions containing primarily one material, we
needed more specific experiments for structural identification
of the product(s) that led to the peak at about 650 nm in the
UV/Vis spectra. To analyze the products that form upon de-
composition of tBuPT* +, we used mass spectrometry; we
thought this technique would be most appropriate for a poten-
tial combination of products that would not necessarily be
formed in high yield or be easy to separate. Although we
could use the remnants from UV/Vis or EPR experiments for
mass spectrometry, these samples were complicated by having
an excess of neutral PT derivative present and by containing
the neutral form of the chemical oxidant. We therefore elected
to use bulk electrolysis to produce samples for mass spectrom-
etry. In this way, we could also analyze larger amounts of
sample without requiring large amounts of oxidant.
Bulk electrolysis experiments of tBuPT were performed in
nBu4NPF6/DCM solutions with a reticulated vitreous carbon
(RVC) working electrode, Pt coiled-wire counter electrode and
Ag/AgCl reference electrode. A potential was applied until 1.00
eÀ/mol was consumed, after which N,N-dimethylformamide
(DMF) was added to the solution to serve as a base. After isola-
tion of the organic products and purification by column chro-
Figure 5. UV/Vis spectra of PT, MPT, EPT, iPrPT, tBuPT, and PhPT at
3.0ꢁ10À4 m in DCM, EC/EMC (3/7), or 1.2m LiPF6 in EC/EMC (3/7). The black
arrow indicates the lmax of solutions in which PT was dissolved, and the blue
arrow indicates the lmax of solutions in which tBuPT was dissolved.
1
matography, we identified one fraction as PT by H NMR; an-
other fraction that contained a mixture of products, which we
analyzed by electrospray ionization mass spectrometry (ESI-
MS). We chose this technique because it tends to not fragment
ions, rendering it useful for the analysis of samples in which
multiple components may be present.
after oxidation of PT, it does not remain unchanged in solution.
Specifically the new peak at ca. 650 nm in both samples is not
consistent with tBuPT* or PT* +. Our next experiments sought
+
to gain more evidence that PT* forms after tBuPT* genera-
tion, and to determine what decomposition products formed
afterward.
+
+
The mass spectrum collected from this sample is shown in
Figure 8 along with a proposed decomposition mechanism of
tBuPT* +. Significant observed peaks have mass-to-charge (m/z)
ratios of 395, 451, and 592, which are consistent with the for-
mation of a dimer of PT, a PT dimer containing a tert-butyl
group, and a PT trimer, respectively. Structures of possible de-
composition products are shown below; without further char-
acterization, we cannot know if these structures are correct or
if the products are isomers of those drawn in Figure 8, but the
2.6. Radical-Cation EPR
One experiment is undeniably conclusive for the presence of
radical species: electron paramagnetic resonance (EPR). In radi-
cal cations, if the lone electron has hyperfine coupling to spin-
active nuclei, unique spectra can be observed. We used EPR
spectroscopy to determine whether we had indeed produced
radical species when oxidizing PT and tBuPT, and to analyze
the spectra over time.
structures are consistent with those isolated or predicted by
others studying reactions of PT*
.
+ [9]
We hypothesize that the reason for the limited performance
+
+
+
To a solution of neutral PT or neutral tBuPT, we added a solu-
tion of TBPA* +, thereby generating a solution of about 1.6ꢁ
10À4 m in DCM (assuming complete electron transfer). Spectra
of MPT* is different from that of tBuPT* as the MPT* UV/
Vis absorption spectrum does not change shape over time, in
contrast to tBuPT* where a new species appears. For MPT*
,
+
+
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
5
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
it was unclear if the loss in inten-
sity is due to limited stability or
solubility. From UV/Vis data, we
know that it is not decomposing
to the PT* (at least, not at an
+
appreciable rate) because we do
not see the peaks characteristic
of this species, which were ob-
served in the decomposition of
tBuPT* +. We thought that bulk
electrolysis at higher concentra-
tions in battery electrolyte
would lead us to a clearer pic-
ture of what was happening
with MPT. We dissolved MPT at
0.08m in 1.2m LiPF6 in EC/EMC
(the same concentration as we
employ in our batteries) and car-
ried out bulk electrolysis at 3.8 V
(vs. Li+/0) until 1 eÀ/mol was
passed through the solution. A
dark solid formed and was isolat-
ed as a finely divided green,
leaving behind a deep red solu-
tion. The organic solution was
analyzed by TLC and GC-MS; by
either technique, only MPT was
observed. A similar experiment
was conducted on EPT. No solid
formed during bulk electrolysis,
and TLC analysis showed no
new products at the conclusion
of the reaction. Thus neither
Figure 6. UV/Vis of the products of the reaction of the chemical oxidant TBPA* with solutions of PT (a), MPT (b),
EPT (c), iPrPT (d), tBuPT (e), and PhPT (f) at 0, 1, 2, and 5 h, radical cations generated in DCM at 1.6ꢁ10À4 m.
+
MPT* nor EPT* form new
soluble products during bulk
electrolysis in battery electrolyte,
+
+
but MPT* precipitates where
+
EPT* does not at the same con-
+
centration; these results suggest
that MPT* is less soluble than
+
EPT* +, which could explain why
MPT does not survive as long in
overcharge.
2.8. Molecular Geometry and
Reactivity
To better understand the stabili-
ty and reactivity of the PT deriv-
atives, specifically tBuPT, we un-
dertook a study of the neutral
and radical-cation forms using
density functional theory (DFT)
at the B3LYP/6-311G(d,p) level.
All PT derivatives are bent sym-
metrically through the N and S
atoms, a trait referred to as the
Figure 7. EPR spectra of dichloromethane solutions of tBuPT and PT generated at 1.6ꢁ10À4 m after treatment with
the chemical oxidant TBPA* at about 1 min after mixing (tBuPT, a; PT, c) and about 10 min after mixing (tBuPT, b;
+
PT, d).
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
6
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
Figure 8. Proposed decomposition of tBuPT* and ESI mass spectrum of products isolated following bulk electrolysis.
+
“butterfly angle” (defined as the angle formed by the intersec-
tion of the aryl-ring planes). Notably, the DFT-determined mo-
lecular (neutral and radical cation) geometries of PT and its al-
kylated derivatives closely match those of reported crystal
structures[10] and of crystals grown in our laboratory. Represen-
tative crystal-structure and DFT-derived molecular geometries
for the neutral and radical cation forms of EPT and tBuPT are
shown in Figure 9.
large tert-butyl group hinders the PT ring system from moving
to the planar configuration favored by other PT radical cations,
which constrained-geometry calculations reveal to be an ener-
getically unfavorable state that results in an uncharacteristically
long CÀN bond between PT and the tert-butyl group (see Fig-
ure S1 and the SI for more details).
Experimental work shows that tBuPT* is the least stable
+
radical cation in the series, perhaps because it has the most
constrained radical-cation geometry. We calculated the ener-
gies of reaction for several potential decomposition pathways
and, based on the reaction energy being the most negative,
The neutral PT derivatives are substantially bent, with but-
terfly angles ranging from 1358 to 1508. The radical cations, on
the other hand, are more planar, with butterfly angles that
we propose that tBuPT* readily decomposes via a low-energy
+
range from 1488 to 1808 (fully planar). (We note that crystal
structures are not available for iPrPT* +, tBuPT* +, or PhPT*
;
fragmentation pathway into PT* and isobutylene (Scheme 1,
+
+
see Table S2 for a summary of calculated and experimental
butterfly angles.) Although the geometries of most PT deriva-
tives show a fairly large difference between the neutral and
radical cation butterfly angles (i.e. as much as 308 for PT and
PhPT), the butterfly angle in tBuPT, the derivative with the
bulkiest alkyl substituent considered, changes by only 148
(from 1348 neutral to 1488 radical cation) upon oxidation. The
first reaction). A comparison of the (gas-phase) alkene decom-
position reactions across the alkylated PT series reveals that
the only energetically favorable reaction is the formation of
the isobutylene byproduct with a zero-point energy [ZPE] cor-
rected reaction energy of DE=À6.35 kcalmolÀ1 (see Table S3
in the SI for a table of energies of reaction). Use of a solvent-
continuum model (formic acid, e=51.1) to represent the polar-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
7
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
radical of the tert-butyl substituent (Scheme 1, second and
third reactions) were also considered, but these cleavage reac-
tions (in both the gas-phase and solvent continuum) are signif-
icantly higher-energy decomposition pathways.
The UV/Vis and EPR studies of the radial-cations reveal that
tBuPT, in stark contrast to the other alkylated PTs, decomposes
to a product that has many features reminiscent of PT* +. Con-
sistent with these results, the theoretical analysis of potential
decomposition pathways suggests that the tert-butyl group
may be readily eliminated as isobutylene after oxidation. The
combined experimental and theoretical results point to impor-
tant considerations concerning the chemistry of redox shuttles,
which need to be contemplated in future molecular design
protocols.
3. Conclusions
Subtle changes in structure can have dramatic effects on the
performance of a redox shuttle during overcharge protection
of LIBs. Although it is widely agreed that the stability of the
radical cation form of a redox shuttle is the key factor govern-
ing the stability of the compound in battery electrolyte, we
found that in one case, a redox
Figure 9. Crystal structures (left) and optimized molecular geometries at the
B3LYP/6-311G(d,p) level of theory (right) for the neutral and radical-cation
forms of EPT and tBuPT. A crystal structure for the tBuPT radical cation has
not been reported.
shuttle candidate decomposed
in battery electrolyte before ever
experiencing overcharge, yet
protected for as many cycles at
the same potential as the prod-
uct we identified that it forms;
hence, the stability of the neu-
tral redox shuttle is also impor-
tant to consider. In three cases,
radical-cation instability was con-
sistent with limited overcharge
cycling, and in the remaining
three cases, compounds with
highly stable radical cations
showed extended overcharge
cycling.
Although not explored in this
publication, additional important
factors for redox shuttles include
diffusion coefficients and solubil-
Scheme 1. Proposed decomposition pathways of tBuPT* +. Zero-point corrected energies of reaction (gas-phase)
computed at the B3LPY/6-311G(d,p) level of theory are also provided.
ity, both of which are often sen-
sitive to the electrolytes em-
ployed and the concentration of
electrolyte salts. None of these
izable environment of the battery electrolyte resulted in
a downward shift of the energies for all reactions considered,
factors can be ignored; reaching an ideal solution may not be
a matter of focusing on one problem, but rather balancing sta-
bility and physical properties. Although most of the decompo-
sition products of aromatic redox shuttles are redox active, it is
difficult to know if the products are responsible for overcharge
protection unless they are synthesized and tested separately.
From the substituents investigated at the N position of PT,
we have learned that while it is important to have a non-hy-
drogen substituent, it is important to consider the chemical
nature of the substituent, that is, the substituent shoud not be
as expected, with the PT* and isobutylene decomposition
+
products being the most favorable. Notably, the decomposi-
tion reaction of iPrPT to PT* and propene becomes modestly
+
favorable (DE=À0.68 kcalmolÀ1) when the solvent-continuum
is taken into account; this result correlates with the iPrPT radi-
cal cation having a comparatively bent structure, with a butter-
fly angle of 1618. The energies for alternate decomposition
pathways that would produce either the carbocation or the
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
8
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
trations of 3.0ꢁ10À4 m in analyte and at 100 mVsÀ1. Ferrocene was
added as an internal standard in all cases. In both electrolytes,
glassy carbon was the working electrode, and platinum was the
counter electrode. In 0.1m nBu4NPF6 in anhydrous DCM, freshly
anodized Ag/AgCl was used as the reference electrode, and vol-
a tert-butyl group as steric bulk prevents the formation of
a stable, alkylated PT radical cation (although this result does
not mean tert-butyl groups should not be used in other sys-
tems). Notably, the as yet unexplored phenyl substituent leads
to the most robust redox shuttle under consideration. As
a result of these experiments, we plan to incorporate the
phenyl substituent onto other PT derivatives that have shown
extended overcharge performance with ethyl substituents at
the N position in order to determine whether derivatives of
these redox shuttles will likewise display extended overcharge
protection.
tammograms were referenced to ferrocenium/ferrocene (Cp2Fe+/0
)
at 0 V. For voltammograms obtained in 1.2m LiPF6 in EC/EMC (3:7
wt. ratio), experiments were performed in an argon-filled glove
box, the reference was Li metal, and voltammograms were refer-
enced to Li+/0 at 0 V.
Battery Fabrication
We hope that our future work will allow us to hone in on
the most important factors contributing to long-term over-
charge performance. We realize that some of these results will
impact not only additives for overcharge protection but may
also be helpful for other energy-storage applications such as
non-aqueous redox flow batteries, for which similar com-
pounds are of interest as electro-active materials for the posi-
tive side of the battery.[11]
Lithium iron phosphate (LiFePO4) cathodes were purchased from
Piotrek (Japan). The synthetic graphite anode, also termed as ’Gen
2 anode’, was supplied by Argonne National Laboratory, along
with 2032 coin cell battery components including battery case
(upper and lower caps), spacers and gaskets. Gen-2 anode was
composed of 92 wt.% MAG-10 graphite (Hitachi) as the active ma-
terial and 8 wt.% polyvinylidene fluoride as the binder. The trilayer
polymer separator Celgard 2325 was donated by Celgard (Char-
lotte, USA). 2032 LiFePO4/synthetic graphite coin cell batteries
were assembled in an argon-filled glove box and contained an
electrolyte consisting of 0.08m redox shuttle in 1.2m LiPF6 in EC/
EMC (3:7 wt. ratio).
Experimental Section
General
PT and MPT were purchased from Sigma Aldrich and were crystal-
lized from pentane. EPT was synthesized as previously reported.[7]
Bis(2-bromophenyl)sulfane[12] and PhPT[13] were synthesized follow-
ing procedures similar to those used for the previously-reported
compounds. iPrPT and tBuPT were synthesized from bis(2-bromo-
phenyl)sulfane using a modified Buchwald-Hartwig coupling reac-
tion.[14] See the SI for synthetic procedures.
Overcharge Cycling
Battery cycling experiments were performed on a Maccor 4200 Bat-
tery Cycler. The battery cycling procedure involved charging the
coin cells with constant current C/10 for 20 h (100% overcharge)
or until a specific upper voltage (5.0 V) was reached. If the voltage
of the coin cell did not reach 5.0 V after 20 h, the charging step
was followed by a 30 s rest and discharging to 3.0 V with the same
current rate. If a coin cell reached an upper voltage of 5.0 V, cycling
was stopped.
Sodium hydride was purchased from Alfa Aesar. Potassium carbon-
ate, copper powder, tetra(n-butyl)ammonium hexafluorophosphate
(nBu4NPF6), and tris(4-bromophenyl)aminium hexachloroantimo-
nate were purchased from Sigma Aldrich. Iodobenzene, isopropyl
amine, tert-butyl amine, sodium sulfide nonahydrate, 1-bromo-2-io-
dobenzene, bis(dibenzylideneacetone)-palladium (Pd(dba)2), Æ-2,2’-
bis(diphenylphosphino)-1,1’-binaphthyl (Æ-BINAP), and copper(I)
iodide were purchased from Acros Organics. Silica gel (65ꢁ250
mesh) was purchased from Sorbent Technologies. Anhydrous sol-
vents and solvents for product purification were purchased from
Fisher Scientific. Ethylene carbonate, ethyl methyl carbonate, and
lithium hexafluorophosphate were purchased from BASF Corpora-
tion (NJ, USA). 1H and 13C NMR spectra were obtained on Varian
spectrometers in [D6]DMSO or CDCl3 from Cambridge Isotope Lab-
oratories. Mass spectra were obtained on an Agilent 5973 Network
mass selective detector attached to Agilent 6890N Network GC
system or through ESI for which samples were dissolved in acetoni-
trile/water (2:1) before analysis. Electrospray ionization (ESI) mass
spectra were obtained on a Thermo Finnigan LTQ (ion trap mass
spectrometer), with sample introduction by direct infusion (syringe
pump) at 3 mLminÀ1. Full scan mass spectra were recorded in posi-
tive ion mode. Instrument parameters included spray voltage:
3.5 kV, capillary temperature: 1858C, capillary voltage: 50 V, and
tube lens voltage: 80 V.
UV/Vis Spectra
UV/Vis spectra were obtained using optical glass cuvettes (Starna)
with 1 cm path length on an Agilent 8453 diode array spectrome-
ter. UV/Vis spectra for the neutral compounds were obtained in
DCM, in 0.1m EC/EMC (3:7 wt. ratio), and in 1.2m LiPF6 in EC/EMC
(3:7 wt. ratio) at 3.0ꢁ10À4 m.
UV/Vis spectra of the radical cations were obtained in anhydrous
DCM. In each case, 1.0 mL of a solution of TBPA* in DCM (5.0ꢁ
+
10À4 m) was added to 2.0 mL of a solution of phenothiazine in
DCM (2.5ꢁ10À3 m) to make a final concentration in each cuvette of
1.6ꢁ10À4 m oxidant and 1.6ꢁ10À3 m analyte. The cuvette was im-
mediately capped with a Teflon stopper and rotated to distribute
the oxidant throughout the sample. Assuming complete electron
transfer, the combination of oxidant and analyze would produce
1.6ꢁ10À4 m in radical cation, 1.6ꢁ10À4 m in neutral oxidant, and
1.44ꢁ10À3 m remaining neutral phenothiazine. Spectra were re-
corded at various times from 0–5 h.
Electron Paramagnetic Resonance (EPR) spectroscopy
EPR spectra were obtained using 4 mm quartz EPR tubes (Wilmad)
on an X-band Bruker EPR spectrometer. For generating the radical
cations of PT and tBuPT, the same amounts and concentrations of
neutral redox shuttle and oxidant were combined and transferred
Cyclic Voltammetry
Cyclic voltammetry experiments were performed using a three-
electrode setup on a CH Instruments 600D potentiostat at concen-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
9
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
to the EPR tube as were used in the UV/Vis studies of radical cat-
ions. Initial spectra were obtained in about 1 min of oxidant addi-
tion, which was followed by an acquisition at about 10 min follow-
ing oxidant addition.
Optimizations of restricted geometries were performed with the
B3LYP functional alternately using 6–31G(d,p), 6–311G(d,p) and cc-
pVTZ basis sets. For each of these calculations, the dihedral angles
on the phenothiazine core of tBuPT (radical cation) were frozen at
Æ1808 (using opt=modredundant). Bond lengths and angles
within the phenothiazine core were allowed to optimize, as was
the entirety of the alkyl group.
Bulk Electrolysis
tBuPT (25.5 mg, 0.100 mmol) was dissolved in a 50 mm solution of
nBu4NPF6 in DCM (50 mL). Bulk electrolysis of the sample was con-
ducted using an RVC working electrode, a Pt coiled-wire counter
electrode and a freshly anodized Ag/AgCl reference electrode. A
potential of 0.95 V (relative to Ag/AgCl) was applied until
1.00 FmolÀ1 was consumed. During this time, the solution turned
dark red/brown in color. Following electrolysis, DMF (5 mL) was
added to the reaction mixture. The solution turned deep green in
color. The reaction was stirred for 30 min at room temperature.
The organic mixture was then washed with saturated aqueous
sodium thiosulfate and brine, and then concentrated to a reduced
volume (ca. 5 mL). Diethyl ether (50 mL) was added to precipitate
salts, and the sample was dried over MgSO4, filtered, and concen-
trated to dryness. The residue was purified by column chromatog-
raphy using 0–10% MeOH/DCM eluent to afford unsubstituted
phenothiazine (14 mg). A fraction containing a mixture of intense-
ly-colored compounds and TBA salts was also obtained and ana-
lyzed by ESI-MS.
Acknowledgements
We thank the University of Kentucky’s Office for the Vice Presi-
dent for Research and the College of Arts and Sciences, the Amer-
ican Chemical Petroleum Research Fund, and the National Sci-
ence Foundation, Division of Chemistry for Award Number CHE-
1300653 for funding for this research. We thank the University of
Kentucky’s High Performance Computing Facility for supercom-
puter access. We also thank Andrew Jansen and Bryant Polzin at
the Cell Manufacturing and Modeling Center at Argonne National
Laboratory for battery cycler supplies, battery electrodes, and
helpful advice.
Keywords: overcharge protection · phenothiazine · radical
cations · redox shuttles · substituent effects
MPT (203 mg) was dissolved at 0.08m in 1.2m LiPF6 in EC:EMC (3:7,
12 mL). A fritted glass tube was placed in a vial and 1 mL of solu-
tion was added to the fritted tube, and the vial was filled with so-
lution to the same level as that of the tube. A platinum coil was
used as a working electrode in the fritted tube; the counter elec-
trode was a platinum wire and the reference electrode was lithium
metal. Bulk electrolysis at 3.8 V (vs. Li+/0) was carried out until
7.72 C (1 equiv eÀ/mol) of electricity was passed through solution.
A dark solid formed in the fritted tube. The radical cation solution
and solid were pipetted out. The solid was filtered to afford
a finely divided green solid. The solution was deep red in color. An
aliquot of the solution was diluted with dichloromethane and
washed with aqueous NaHSO3. The organic layer was analyzed by
TLC and GC-MS. No new products were formed. The same experi-
ment was repeated with EPT. No solid formed during bulk electrol-
ysis and TLC analysis showed no new products at the conclusion
of the reaction.
[1] a) Q. Wang, S. M. Zakeeruddin, I. Exnar, M. Graetzel, Electrochem.
Commun. 2008, 10, 651–654; b) Z. Chen, Y. Qin, K. Amine, Electrochim.
Acta 2009, 54, 5605–5613.
[2] a) C. Buhrmester, L. M. Moshurchak, R. L. Wang, J. R. Dahn, J. Electro-
chem. Soc. 2006, 153, A1800–A1804; b) J. K. Feng, X. P. Ai, Y. L. Cao,
H. X. Yang, Electrochem. Commun. 2007, 9, 25–30; c) L. M. Moshurchak,
C. Buhrmester, R. L. Wang, J. R. Dahn, Electrochim. Acta 2007, 52, 3779–
3784; d) L. Zhang, Z. Zhang, H. Wu, K. Amine, Energy Environ. Sci. 2011,
4, 2858–2862; e) W. Weng, Y. Tao, Z. Zhang, P. C. Redfern, L. A. Curtiss,
K. Amine, J. Electrochem. Soc. 2013, 160, A1711–A1715; f) Y. Zhang, L.
Wang, A. Zhang, C. Zhang, P. Zhang, Ionics 2011, 17, 421–427; g) L. M.
Moshurchak, W. M. Lamanna, M. Bulinski, R. L. Wang, R. R. Garsuch, J.
Jiang, D. Magnuson, M. Triemert, J. R. Dahn, J. Electrochem. Soc. 2009,
156, A309–A312; h) L. Zhang, Z. Zhang, P. C. Redfern, L. A. Curtiss, K.
Amine, Energy Environ. Sci. 2012, 5, 8204–8207; i) M. N. Golovin, D. P.
Wilkinson, J. T. Dudley, D. Holonko, S. Woo, J. Electrochem. Soc. 1992,
139, 5–10; j) Z. Chen, J. Liu, A. N. Jansen, G. GirishKumar, B. Casteel, K.
Amine, Electrochem. Solid-State Lett. 2010, 13, A39–A42; k) Z. Chen,
A. N. Jansen, K. Amine, Energy Environ. Sci. 2011, 4, 4567–4571; l) Z.
Chen, Y. Ren, A. N. Jansen, C.-k. Lin, W. Weng, K. Amine, Nat. Commun.
2013, 4, 1513; m) S. Ergun, C. F. Elliott, A. P. Kaur, S. R. Parkin, S. A.
Odom, Chem. Commun. 2014, 50, 5339–5341; n) S. Ergun, C. F. Elliott,
A. P. Kaur, S. R. Parkin, S. A. Odom, J. Phys. Chem. C 2014, 118, 14824–
14832; o) K. M. Abraham, D. M. Pasquariello, E. B. Willstaedt, J. Electro-
chem. Soc. 1990, 137, 1856–1857; p) K. M. Abraham, J. Electrochem. Soc.
2012, 159, A1057–A1064.
X-ray Crystallography
Crystal structures were obtained by single-crystal X-ray diffraction
for EPT, iPrPT, tBuPT, and PhPT. Crystallographic data for some neu-
tral compounds were obtained from published data including
those of PT,[15] MPT,[16] EPT,[17] iPrPT,[18] and PhPT[19] as well as those
of the radical cations of PT,[10a,b] MPT,[10c] and EPT.[10d]
[3] a) C. Buhrmester, L. Moshurchak, R. L. Wang, J. R. Dahn, J. Electrochem.
Soc. 2006, 153, A288–A294; b) J. R. Dahn, C. Buhrmester, R. L. Wang,
W. M. Lamanna, U. S. Patent US7615312 B2 2009.
[4] J. C. Forgie, D. Rochefort, RSC Adv. 2013, 3, 12035–12038.
[5] Z. Chen, K. Amine, Electrochem. Commun. 2007, 9, 703–707.
[6] A. P. Kaur, S. Ergun, C. F. Elliott, S. A. Odom, J. Mater. Chem. A, manu-
script accepted.
[7] S. A. Odom, S. Ergun, P. P. Poudel, S. R. Parkin, Energy Environ. Sci. 2014,
7, 760–767.
[8] S. R. Waldvogel, B. Janza, Angew. Chem. Int. Ed. 2014, 53, 7122–7123;
Angew. Chem. 2014, 126, 7248–7249.
[9] a) P. Hanson, R. O. C. Norman, J. Chem. Soc. Perkin Trans. 2 1973, 264–
271; b) O. B. Tomilin, E. P. Konovalova, V. N. Yuzhalkin, L. V. Ryabkina, E. P.
Sanaeva, Chem. Heterocyc. Compd. 1996, 32, 365–370; c) H. J. Shine, J. J.
Silber, R. J. Bussey, T. Okuyama, J. Org. Chem. 1972, 37, 2691–2697;
DFT Calculations
All density functional theory (DFT) calculations were performed
using the Gaussian09 (Revision A.02b) software suite.[20] Geometry
optimizations of the neutral and radical-cation states were carried
out with the B3LYP functional[21] and 6–311G(d,p) basis set. Fre-
quency analyses of all (fully relaxed) optimized geometries were
undertaken to ensure that the geometries were energetic minima.
The energy of each decomposition pathway was obtained from en-
ergies for each product/reactant computed at the same level of
theory.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&10
&
ÞÞ
These are not the final page numbers!

CHEMPHYSCHEM
ARTICLES
www.chemphyschem.org
d) L. G. Shagun, I. A. Dorofeev, I. A. Tokareva, V. I. Smirnov, V. A. Shagun,
M. G. Voronkov, Russ. J. Org. Chem. 2012, 48, 1263–1264.
[10] a) T. Uchida, M. Ito, K. Kozawa, Bull. Chem. Soc. Jpn. 1983, 56, 577–582;
b) S. V. Rosokha, J. K. Kochi, J. Am. Chem. Soc. 2007, 129, 3683–3697;
c) Q. Wang, L. Youcheng, D. Yangbing, L. Zhongli, Chin. J. Struct. Chem.
1988, 7, 153; d) L. Youcheng, D. Yanbing, L. Zhongli, W. Qiguang, Chin.
J. Struct. Chem. 1989, 8, 140.
[11] a) Z. Li, S. Li, S. Liu, K. Huang, D. Fang, F. Wang, S. Peng, Electrochem.
Solid-State Lett. 2011, 14, A171–A173; b) F. R. Brushett, J. T. Vaughey,
A. N. Jansen, Adv. Energy Mater. 2012, 2, 1390–1396; c) L. Su, M. Ferran-
don, J. A. Kowalski, J. T. Vaughey, F. R. Brushett, J. Electrochem. Soc.
2014, 161, A1905–A1914.
[12] Y. Li, C. Nie, H. Wang, X. Li, F. Verpoort, C. Duan, Euro. J. Org. Chem.
2011, 7331–7338.
[13] Y. Ahn, D. E. Jang, Y.-B. Cha, M. Kim, K.-H. Ahn, Y. C. Kim, Bull. Korean
Chem. Soc. 2013, 34, 107–111.
[14] S. Hong, S. Hwimin, Vol. US 2012/0059182A1, University of Florida Re-
search Foundation, United States, 2012.
[15] J. McDowell, Acta Crystallogr. Sect. B 1976, 32, 5–10.
[16] S. S. C. Chu, D. Van der Helm, Acta Crystallogr. Sect. B 1974, 30, 2489–
2490.
[18] S. S. C. Chu, D. Van der Helm, Acta Crystallogr. Sect. B 1976, 32, 1012–
1016.
[19] C. L. Klein, J. M. Conrad, III, S. A. Morris, Acta Crystallogr. Sect. C 1985,
41, 1202–1204.
[20] Gaussian 09 (Revision A.02b), M. J. Frisch, G. W. Trucks, H. B. Schlegel,
G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B.
Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratch-
ian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M.
Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.
Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr., J. E. Peralta,
F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Starover-
ov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant,
S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E.
Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski,
R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J.
Dannenberg, S. Dapprich, A. D. Daniels, ꢂ. Farkas, J. B. Foresman, J. V.
Ortiz, J. Cioslowski, D. J. Fox, Gaussian, Inc., Wallingford, CT, USA, 2009.
[21] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652; b) C. Lee, W. Yang,
R. G. Parr, Phys. Rev. B 1988, 37, 785–789.
[17] S. S. C. Chu, D. Van der Helm, Acta Crystallogr. Sect. B 1975, 31, 1179–
1183.
Received: September 26, 2014
Published online on && &&, 0000
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&
11
&
ÞÞ
These are not the final page numbers!

ARTICLES
K. A. Narayana, M. D. Casselman,
C. F. Elliott, S. Ergun, S. R. Parkin,
C. Risko,* S. A. Odom*
Redox shuttles are electrolyte additives
that can prevent overcharge in batter-
ies. They travel between the anode and
cathode in their neutral and oxidized
forms, mitigating excess current and
stabilizing cell voltage. However, if the
neutral or radical cation form of the
shuttle decomposes, products can form
that no longer protect the battery from
overcharge, as is observed in the case
of N-tert-butylphenothiazine.
&& – &&
N-Substituted Phenothiazine
Derivatives: How the Stability of the
Neutral and Radical Cation Forms
Affects Overcharge Performance in
Lithium-Ion Batteries
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 0000, 00, 1 – 12
&12
&
ÞÞ
These are not the final page numbers!