Journal of Electroanalytical Chemistry 717–718 (2014) 177–188

Contents lists available at ScienceDirect

Journal of Electroanalytical Chemistry

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

Determination of the ‘‘NiOOH’’ charge and discharge mechanisms

at ideal activity
Matthew Merrill ⇑, Marcus Worsley, Arne Wittstock, Juergen Biener, Michael Stadermann
Chemical Sciences Division, Lawrence Livermore National Laboratory, Livermore, CA 94550, United States

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

Article history: Optimization of electrodeposition conditions produced Ni(OH)2 deposits chargeable up to 1.84 ± 0.02 e
Received 26 March 2013 per Ni on and the resulting nickel oxide/hydroxide active material could subsequently deliver 1.58 ± 0.02
Received in revised form 11 January 2014 e per Ni ion (462 mA h/g) over a potential range <0.2 V. The ability of the ‘‘NiOOH’’ active material to
Accepted 16 January 2014
deliver an approximately ideal charge and discharge facilitated a coulometric and thermodynamic anal-
Available online 24 January 2014
ysis through which the charge/discharge mechanisms were determined from known enthalpies of forma-
tion. The (dis)charge states were confirmed with in situ Raman spectroscopy. The mechanisms were
Keywords:
additionally evaluated with respect to pH and potential dependence, charge quantities, hysteresis, and
NiOOH
Ni(OH)2
fluoride ion partial inhibition of the charge mechanism. The results indicate that the ‘‘NiOOH’’ (dis)-
Nickel electrode charges as a solid-state system with mechanisms consistent with known nickel and oxygen redox reac-
Nickel oxy-hydroxide tions. A defect chemistry mechanism known for the LiNiO2 system also occurs for ‘‘NiOOH’’ to cause both
Active oxygen high activity and hysteresis. Similar to other cation insertion nickel oxides, the activity of the ‘‘NiOOH’’
mechanism is predominantly due to oxygen redox activity and does not involve the Ni4+ oxidation state.
The ‘‘NiOOH’’ was produced from cathodic electrodeposition of Ni(OH)2 from nickel nitrate solutions onto
highly oriented pyrolytic graphite at ideal electrodeposition current efficiencies and the deposition mech-
anism was also characterized.
Published by Elsevier B.V.

1. Introduction Recent developments in spectroscopic analysis, theoretical

Nickel oxides have been the dominant cathode material for sec-
ondary batteries over many years and Li-ion batteries have only re-
cently emerged as a superior alternative due to their higher energy
densities. Still, batteries with nickel oxide cathodes have several
features that are superior to Li-ion batteries with regards to higher
power densities, non-flammable electrolytes, and less costs [1,2].
In practice, the nickel oxide material is the limiting electrode for
energy density in the battery with an activity of 0.4 e per Ni ion
or 115 mA h/g Ni(OH)2 [2,3]. Nickel oxide materials have been re-
ported in the literature to reach high activities of up to 1.65 e per
Ni ion or 475 mA h/g Ni(OH)2 yet this higher performance has
never been realized in practical batteries [2–4]. An improvement
in the energy density of nickel oxides by a factor of 3–4 could re-
store interest in this battery material. Both the highly active nickel
oxide mechanisms and the mechanisms for producing the highly
active nickel oxide materials are further investigated here in order
to facilitate the practical implementation of more active nickel
oxide materials.
⇑ Corresponding author. Tel.: +1 925 422 9262.
E-mail address: merrill10@llnl.gov (M. Merrill).
modeling, and the analysis of nickel oxide materials in Li-ion bat-
teries challenge the conventionally assumed nickel oxide mecha-
nism in aqueous conditions and indicate that the mechanism
needs revision. The conventionally assumed mechanism for the en-
ergy producing reaction in nickel oxide/hydroxide batteries is
based upon the oxidation of nickel ions from Ni2+ to Ni4+ [3,5].
The only evidence specifically supporting the Ni4+ state is the sim-
ilarity in the first Ni–O bond length or Ni K-edge shift energy ob-
served though X-ray absorption studies [6–10]. The actual Ni4+
valence state has been conventionally thought to occur in the fully
oxidized c-NiOOH state because of the X-ray adsorption experi-
ments in which BaNiO3 or KNiIO6 reference materials were as-
sumed to represent an actual Ni4+ state. These materials were
originally assumed to contain nickel in the Ni4+ valence state be-
cause there was reasonable confusion between the formal and ac-
tual valence states at the time [11]. A formal oxidation state of Ni4+
includes the possibility that the actual valence state is Ni3+ because
oxidation of oxide has occurred. It has now been well-established
(even by some of the same authors, O’Grady and Mansour) from
Mössbauer, magnetic susceptibility, and X-ray photoelectron spec-
troscopy (XPS) experiments that the BaNiO3 or KNiIO6 reference
materials (as well as Li(1x)NiO2 and c-NiOOH) all have the same

http://dx.doi.org/10.1016/j.jelechem.2014.01.022
1572-6657/Published by Elsevier B.V.
178 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
D53d space group symmetry involving a Ni3+ which has been some-
what distorted due to the oxidation of some O2 [12–18]. XPS
experiments specifically indicate that nickel is Ni3+ in c-NiOOH
and not Ni4+ [19–22]. The uncertainty involved with Jahn-Teller
distortions causes the modeling of nickel oxide structures and
bond lengths alone to be insufficient for determining nickel oxida-
tion states despite the numerous attempts to do so [14,15,18].
Recent experiment and theoretical modeling of the LiXNiO2
electrode in lithium batteries has also indicated that oxygen redox
chemistry is primarily responsible for the charge storage instead of
the oxidation of nickel ions to the 4+ state [16,18,23–26]. The oxy-
gen redox chemistry occurs when electron vacancy forms on O2
instead of on Ni3+ and so the nickel ions are not being oxidized
to an actual Ni4+ valence state. Similarly, the electron vacancies
in LiXNi(1X)O also occur on oxygen species to form O plus Ni2+
when X < 0.4 and the electron vacancy only shifts to nickel to form
O2 plus Ni3+ as X approaches the 0.5 of LiNiO2 [17,27–30]. It
should therefore be expected that the aqueous ‘‘NiOOH’’ mecha-
nism involves oxygen redox activity instead of the actual Ni4+ state.
A more thorough and comprehensive review and analysis of the
‘‘NiOOH’’ mechanism is available online in the Supporting Informa-
tion section. We develop a new mechanism for the charging mech-
anism of nickel oxide materials which is consistent with both our
experimental results and the literature. This mechanism incorpo-
rates the oxygen redox activity caused by electron vacancies on
oxide species. A better understanding of oxygen redox center for-
mation and stabilization could lead to better battery active materi-
als regardless of whether it is a H+-insertion system of an aqueous
battery or a Li+-insertion systems of a Li+ ion battery.
‘‘NiOOH’’ in its most active form is expected to constitute the
optimal condition for mechanistic analysis. Minimizing the pres-
ence of side reactions will improve the resolution in the thermody-
namic analysis and minimizing the presence of inactive states will
improve the clarity of spectroscopic analysis. The oxidation of
Ni(OH)2 produced from cathodic electrodeposition from nickel ni-
trate solutions has resulted in the highest ‘‘NiOOH’’ activity (1.65
e per Ni ion) reported in the literature and will therefore be
implemented here [4]. The cathodic electrodeposition of Ni(OH)2
from nickel nitrate and other nickel solutions is likely more com-
plex than the conventional mechanism typically presupposed in
the literature because of the multiplicity of possible reaction path-
ways. The five reactions of Table 1 collectively cause the four dif-
ferent mechanisms (i–iv) below through which Ni(OH)2 can be
cathodically deposited. Ni(OH)2 electrodeposition mechanisms (i)
and (ii) have been previously proposed in the literature. It is not
likely that electrodeposition mechanisms (i) and (ii) can achieve
ideal electrodeposition current efficiency because they depend
upon the chemical precipitation of Ni(OH)2 from OH concentra-
tion gradients. The observation that the most highly active nickel
oxide materials are produced under the condition of ideal electro-
deposition current efficiency motivates the re-evaluation of the
electrodeposition mechanism and the introduction of mechanisms
Table 1
Eqs. (1)–(5): the reactions of the Ni(OH)2 deposition mechanisms [32].
NO    solutions at 1.0 mA/cm2 for 10 min. The spectra were collected
3 þ 2e þ H2 O $ NO2 þ 2OH
E0 ð1Þ ¼ 0:835  0:0592  pH þ 0:0296  logð½NO 
3 =½NO2 Þ
Ni
2þ
þ 2OH $ NiðOHÞ2
2þ
logðNi Þ ¼ 12:18  2  pH
2H2 O þ 2e $ H2 þ 2OH
E0 ð3Þ ¼ 0:000  0:0592  pH
Ni2þ þ 2e $ Ni
E0 ð4Þ ¼ 0:250 þ 0:0296  logð½Ni2þ Þ
Ni þ 2H2 O $ NiðOHÞ2 þ 2e þ 2Hþ
E0 ð5Þ ¼ 0:110  0:0592  pH
(iii) and (iv) [4]. The characterization of the cathodic electrodepos-
ition mechanism is relevant for explaining why deposits of Ni(OH)2
with optimum activity as ‘‘NiOOH’’ can only be produced in a rel-
atively narrow range of conditions as well as why the limited con-
ditions may be challenging for practical implementation in
batteries.
i. The OH gradient generated by the electrochemical reduc-
2+
tion of NO 3 through (1) chemically precipitates Ni as
Ni(OH)2 according to (2) in an 1 Ni(OH)2 per 2 e ratio
[4,31–33]. The further reduction of NO 2 to either N2 or
NH4+ can also electrochemically generate a pH gradient,
both with an additional 2 Ni(OH)2 per 3 e ratio [34].
ii. The reduction of water to form molecular hydrogen, H2,
according to reaction (3) is another mechanism which can
electrochemically an OH gradient for chemically precipitat-
ing Ni(OH)2 through reaction (2) in a 1 Ni(OH)2 per 2 e ratio
[5]. This mechanism is a likely mechanism when NO 3 is
absent from the electrodeposition solution.
iii. The reduction of Ni2+ to the metallic state Ni0 according to
(4) occurs at relevantly cathodic potentials and leads to
two more possible cathodic deposition mechanisms. Water
is not electrochemically stable on Ni at moderate potentials
and will oxidize Ni0 to Ni2+ through (4) or to Ni(OH)2
through (5), depending upon pH [34,35].
iv. The nitrate ion is such a strong oxidizing agent that it can
oxidize Ni0 even under a large, negative potential bias. The
direct coupling of nitrate reduction (1) to Ni0 oxidation (4)
is expected to cause the immediate precipitation of Ni(OH)2
before either the Ni2+ or OH products could diffuse back
into the bulk electrolyte phase and should therefore be
expected when ideal electrodeposition current efficiency is
occurring.
2. Methods
General electrochemical experiments were performed in a Tef-
lon electrochemical cell. A Teflon face plate and corrosion-resistant
silicone gasket (McMaster-Carr) exposed a circular, 1 cm diameter
(0.785 cm2) working electrode surface of highly oriented pyrolytic
graphite (HOPG, ZYH grade, Bruker AFM Probes) on the bottom of
the chamber. A 2 cm2 HOPG (ZYH grade, K-Tek Nanotechnology)
was used in conjunction with a 1.75 cm diameter Teflon face plate
for in situ Raman experiments. Used HOPG surfaces were removed
with (Scotch) tape to expose a new, clean surface. A 30 ga platinum
wire (Hauser & Miller) counter electrode was coiled around the tip
of a double junction SCE reference electrode (Radiometer Analyti-
cal) to be jointly inserted into the cell chamber from the top. The
general electrochemical experiments were performed with a Bio-
logic VSP electrochemical workstation while the in situ Raman
spectroscopy experiments were performed with a Biologic SP200.
Raman spectra were collected with Nicolet Almega XR dispersive
micro-Raman spectrometer, using 633 nm excitation length and a
water immersible Olympus LumPlanFl 60x objective for collecting
the Raman scattered radiation. The in situ Raman samples were
prepared by electrodeposition onto HOPG from 0.08 M Ni(NO3)2
(1)
for the first discharge and the second charge in 3 M KOH. The ‘‘NiO-
(2) OH’’ was initially charged at 20 C, the potential was held constant
for approximately 15 min during spectra collection, and the poten-
(3) tial was changed between spectra at 1 mV/s. Metallic nickel layers
of about 20 nm in thickness were deposited onto HOPG substrates
(4)
from Techni Nickel HT-2 solutions (Technic Inc.) at 10 mA/cm2 cur-
rent density for 5 s.
(5)
Masses of the Ni(OH)2 active material were determined by dis-
solving samples in 2 ml of 10% weight/volume nitric acid under
 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
sonication for 15 min and then analyzing the Ni through induc-
tively coupled plasma mass spectrometry (ICP–MS) (Thermo_Sci-
entific X series 2). ICP–MS samples were prepared with a 500
dilution. The diluent contains an internal standard suite in 2% nitric
acid. A fully quantitative analysis using a linear calibration curve
based on known standards was performed. The internal standard
corrects for instrument drift. NIST traceable certified reference
materials and serial dilutions were analyzed for accuracy and
duplicate samples were analyzed for precision. Scanning electron
microscopy (SEM) and energy-dispersive X-ray spectroscopy
(EDX) characterization were performed on a JEOL 7401-F at
1-5 keV (20 mA) in lower secondary electron image mode with a
working distance of 8 mm. X-ray diffraction (XRD) measurements
were performed on a AXS D8 ADVANCE X-ray diffractometer
(Bruker) equipped with a 1-dimensional linear Si strip detector
(LynxEye). The X-ray diffraction samples were initially prepared
by cathodic electrodeposition from 0.08 M Ni(NO3)2 solutions at
1 mA/cm2 for 30 min. The samples were scanned from 5° to 75°
2h in 0.02° steps with a 2 s counting time per step, a 0.499° diver-
gence slit, and a 0.499° antiscatter slit. The X-ray source was Ni-fil-
tered Cu radiation from a sealed tube operated at 40 kV and 40 mA.
Goniometer alignment was ensured using an Al2O3 standard
(Bruker).
The problem with using Pourbaix’s equations for aqueous elec-
trochemical equilibria of metal oxides regards the thermodynamic
treatment of hydrating H2O in the enthalpy of formation values.
Pourbaix assumes that all of the H2O hydrating metal oxides have
entropies equivalent to that of the bulk liquid phase. Pourbaix’s
treatment of H2O intercalated in hydrated metal oxides with liquid
phase entropies was likely motivated by limited data and an inter-
est in the uniform treatment of intercalated H2O across all ele-
ments and oxidation states. This treatment causes inaccuracy in
the calculated equilibria because confining water in a solid phase
lowers the H2O’s entropy. For an example at 25 °C, liquid water
(H2O(l)) has a standard entropy, DS0m , of 69.9 J/K/mol while the con-
fined movement of water in the solid phase (H2O(s)) lowers the DS0m
to 38.0 J/K/mol [36]. A comparison between NiO and NiO(H2O) free
enthalpies of formation indicates that the hydrating water can
have liquid-like DS0m values of 65.5 J/K/mol when associated with
Ni2+ ions and will be designated as H2O(l0 ) [34,37]. It is known that
water associated with smaller, more highly charged ions will have
even lower entropies [38]. It will be assumed here that the water
associated with Ni3+ has the solid phase entropy of ice and will
be designated as H2O(s). The states considered are therefore
NiO(H2O(l0 )), Ni3O4(H2O(s))(H2O(l0 )) and Ni2O3(H2O(s)). The redox
potentials of Table 3 have been adjusted from their referenced val-
ues to reflect this more accurate treatment of intercalated H2O.
3. Results and discussion
3.1. Ni(OH)2 electrodeposition characterization and optimization
The cathodic electrodeposition of Ni(OH)2 on HOPG likely in-
volves Ni0 as an intermediate state. HOPG has a large overpotential
for evolving H2 from water molecules in KCl and KNO3 (Fig. 1a).
The hydrogen evolution reaction (3) is considered the only plausi-
ble reduction reaction on HOPG in the KCl solution. The linear
sweep voltammogram of KNO3 on HOPG was approximately the
same as that of KCl. The hydrogen evolution reaction is expected
to also be the primary reduction reaction on HOPH in the KNO3
solution due to the similarities in KCl and KNO3 voltammograms
in Fig. 1a. The absence of peaks or shoulders in HOPG’s KNO3 vol-
tammogram indicates insignificant rates of NO 3 reduction (1) in
comparison with the H2 evolution reaction. The H2 evolution reac-
tion experimentally occurs on HOPG at more positive potentials
179
0
j / mA cm-2
0.16 M KCl on HOPG
0.16 M KNO3 on HOPG
0.16 M KCl on Ni/HOPG
0.16 M KNO3 on
Ni/HOPG
-5
-1.75 -1.25 -0.75 -0.25
E vs SHE / V
0
j / mA cm-2
0.08 M Ni(NO3)2
0.08 M NiCl2
0.08 M Ni(ClO4)2
-5
-1.75 -1.25 -0.75 -0.25
E vs SHE / V
Fig. 1. Linear sweep voltammograms of electrolytes and substrates relevant to the
Ni(OH)2 electrodeposition mechanism(s). (a) Linear sweeps at 10 mV/s in KCl and
KNO3 solutions on HOPG and HOPG coated with about 20 nm of metallic nickel (Ni/
HOPG). (b) Linear sweeps at 10 mV/s in Ni(NO3)2, NiCl2, and Ni(ClO4)2 electrolytes
on HOPG.
than the nitrate reduction reaction near neutral pH despite the
large difference in equilibrium potentials. The deposition of about
20 nm of metallic nickel onto the HOPG from a commercial electro-
deposition solution (Nickel HT-2) lowered the H2 evolution over-
potential by 0.4 V for the same KCl and KNO3 solutions. A low
current shoulder in the KNO3 voltammogram on the nickel metal
likely corresponds to NO 3 reduction at potentials similar to that
of H2 evolution and is consistent the literature for Ni and Pt sub-
strates [32,39]. Low current voltammetric shoulders/peaks ap-
peared for the Ni(NO3)2, NiCl2, and Ni(ClO4)2 solutions (Fig. 1b)
and likely corresponded with metallic nickel deposition (4) on
HOPG because the shoulders/peaks did not appear when Ni2+
was absent (Fig. 1a). Some Ni0 state nickel also must have depos-
ited at the lower current/potentials because the hydrogen evolu-
tion kinetics which predominate >2 mA/cm2 in the nickelous
solutions (Fig. 1b) were the same for the non-nickelous solutions
only when a thin film of Ni0 was coated on the HOPG (Fig. 1a).
Electrochemically charging the Ni(OH)2 deposits formed at
1 mA/cm2 from NiCl2 and Ni(ClO4)2 electrolytes indicated the
presence of metallic nickel, Ni0, in these deposits. Residual Ni0
was considered detected in the NiCl2 and Ni(ClO4)2 deposits be-
cause their first charges initiated near 0.4 to 0.6 V vs. SHE in
Fig. 2 chronopotentiograms, which corresponds with the oxidation
potential of metallic nickel through (5). In the absence of an ap-
plied potential, anhydrous NiO will form through the oxidation
of Ni0 by the water molecule hydrating neighboring Ni(OH)2
(which can be alternatively expressed as NiOH2O) through (5) if
the Ni0 is no longer exposed to bulk water. In contrast to the
semi-transparent green Ni(OH)2 observed for uncharged deposits
produced from Ni(NO3)2 solutions at <1.25 mA/cm2, significant
quantities of the black, anhydrous NiO was visually detected in
the uncharged NiCl2 and Ni(ClO4)2 deposits produced at compara-
ble current densities [34]. The presence of Ni0 and NiO was also
180 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
1 seems to be less consistent for ClO4- than it was for NO
Potential vs SHE / V
0.6
0.2
Ni(NO3)2
NiCl2
-0.2
Ni(ClO4)2
-0.6
0 50 100 150
Time / s
Fig. 2. Chronopotentiograms for the initial charge of freshly deposited Ni(OH)2. The
Ni(OH)2 deposits were formed from 0.08 M nickel solutions at 1 mA/cm2 for
30 min and were initially charged with 8 mA/cm2.
detected when deposits from Ni(NO3)2 solutions were generated
with current densities >1.25 mA/cm2 (data not shown), which
would be consistent with a mechanism in which Ni0 is depositing
through (4) faster than it can be oxidized by NO 3 through (1).
The role of Ni0 in the deposition mechanisms was confirmed by
the similarity of Fig. 3’s chronopotentiograms of Ni(OH)2 deposi-
tion onto HOPG from 0.08 M Ni(NO3)2 and Ni0 deposition onto
HOPG from a commercial deposition solution, Nickel HT-2. These
chronopotentiograms are characteristically unusual because of
the exponential decay of the applied potential to lower overpoten-
tials (more positive potentials) with time. An increase in the over-
potential of Fig. 3’s chronopotentiograms would be expected due
to a decrease in Ni2+ concentration with time as the interfacial
reactant diffusion layer grows yet this effect was not observed.
An increase in the overpotential of the chronopotentiograms
would also be expected if the deposition mechanisms primarily re-
lied upon building up a OH concentration gradient according to
reactions (1) or (3). Significant increases in overpotential would
have additionally been expected due to self-limiting processes if
the Ni(OH)2 was behaving like an insulator due to its band gap of
about 2.8 eV (the material becomes more electrically conductive
upon oxidation) and only the current collector was electrochemi-
cally active [4,19,31]. Note that the Ni2+ concentration of the Ni
HT-2 solution (0.48 M) appropriately accounts for the 40 mV off-
set observed at <300 s for the 0.08 M Ni(ClO4)2 deposition solution,
meaning Ni0 was likely the only material being deposited from
Ni(ClO4)2 during the first 300 s. The 0.08 M Ni(ClO4)2 deposition
solution acquired an extra 170 mV offset at >300 s to become sim-
ilar to the Ni(NO3)2 deposition solution’s potential (Fig. 3). Like the
nitrate ion, the perchlorate ion can oxidize metallic nickel even un-
der the large, negative potential bias although this mechanism
-0.5
Potential vs SHE / V
-0.7
Ni HT-2
0.08 M Ni(NO3)2
0.08 M Ni(ClO4)2
-0.9
0 600 1200
Time / s
Fig. 3. Electrodeposition chronopotentiograms. Metallic nickel and/or Ni(OH)2
were cathodically deposited onto HOPG at 1 mA/cm2 for 30 min.
3 [34].
The extra 170 mV offset was therefore likely caused by the differ-
ence between deposition onto Ni0 and Ni(OH)2. Alternatively, fac-
toring in reactant and product concentrations into the
electrochemical equilibria indicates that water will spontaneously
oxidize metallic nickel to Ni(OH)2 through reaction (5) even at a
potential bias of 0.7 V vs. SHE if the interfacial Ni2+ concentration
is depleted low enough from deposition and interfacial pH reaches
pH P 11 due to hydrogen evolution [34]. Bubbles indicative of H2
evolution were observed during deposition from Ni(ClO4)2 and
NiCl2 solutions but not from Ni(NO3)2, indicating that the oxidation
200
of Ni0 by NO 3 is favorable if nitrate is present.
The most active ‘‘NiOOH’’ materials could only be generated at
low concentrations (0.08 M) of Ni(NO3)2 and relatively high cur-
rent densities, where the most efficient deposition current efficien-
cies and deposit activities occurred around 1 mA/cm2 (Fig. 4). The
quantity of deposited nickel and the corresponding ideal electrode-
position current efficiency was determined by dissolving thor-
oughly-washed deposits in 10% nitric acid and then determining
the Ni2+ concentration with ICP–MS. The ideal current efficiency
for the cathodic electrodeposition of Ni(OH)2 was sustainable for
over 10 min (Fig. 5).
3.2. Ni(OH)2 electrodeposition mechanisms
Ni(OH)2 electrodeposition mechanisms (i) and (ii) both rely
upon the chemical precipitation of Ni(OH)2 due to the formation
of a OH concentration gradient through (2). In this commonly pre-
sumed process, OH generated at the interface can diffuse back
into the bulk phase much faster than Ni2+ can diffuse to the inter-
face because OH’s diffusion coefficient of 5.27  105 cm2/s is al-
most an order of magnitude faster than ½Ni2+’s diffusion
coefficient of 0.66  105 cm2/s [37]. The application of a constant
current electrodeposition technique to this system will cause a
non-ideal electrodeposition current efficiency which will further
decrease with time because the OH generated at the interface will
diffuse back into the bulk solution phase faster than Ni2+ can dif-
fuse from the bulk phase to the interface. Mechanisms (i) and (ii)
are therefore incompatible with the electrodeposition condition
of 0.08 M Ni(NO3)2 at 1 mA/cm2 because the electrodeposition
current efficiency remained constant at the ideal 1 Ni(OH)2 per 2
e with respect to the duration of electrodeposition for over
10 min (Fig. 5). Mechanisms (i) and (ii) are only incompatible with
cathodic Ni(OH)2 electrodeposition when the well-known ideal
electrodeposition current efficiency of 1 Ni(OH)2 per 2 e is
0.5
e- discharge / e- deposit
0.4
0.3
0.2
0.1
0
-1.5 -1.2 -0.9 -0.6 -0.3 0
Deposition current density / mA cm-2
Fig. 4. Optimum deposition current density. The electrodeposition current density
1800 for a 0.08 M Ni(NO3)2 solution was varied while the quantity of e (1.44 C) used for
electrodeposition was held constant. The Ni(OH)2 deposits were charged and then
discharged with 2 mA/cm2. A greater ratio of the number of electrons discharged
from the Ni(OH)2 deposit per the number of electron used to form the deposit
reflect greater electrodeposition current efficiencies and/or deposit activities.
 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
occurring [40,41]. Mechanisms (i) and (ii) are otherwise likely
mechanisms involved when conditions result in non-ideal electro-
deposition current efficiency.
Mechanism (iii) is likely only present when the black NiO prod-
uct is formed at low overpotentials. The oxidation of Ni0 by water
through (4) is negative of the hydrogen evolution equilibrium po-
tential of (3) between pH 5–8, depending on the interfacial Ni2+
concentration, and so the coupling of the two reactions can be
spontaneous in the absence of a sufficiently negative potential bias.
This coupling of H2 evolution with metal corrosion has been more
widely recognized in the case of iron [34]. NiO can also be formed
by depositing Ni0 at relatively negative potentials and then shifting
the potential to a more neutral value for oxidation through (4) or
(5) with cyclic voltammetry or pulse techniques [42,43]. It is
known that Ni(OH)2 can be converted to NiO during electrodepos-
ition through electrochemically driven H2 evolution at large over-
potentials and current densities (>50 mA/cm2) however this was
likely not the case observed here for the deposits formed with
1 mA/cm2 from the 0.08 M NiCl2 and Ni(ClO4)2 solutions [44].
Mechanism (iv) is most consistent with the observed results for
the Ni(NO3)2 electrolytes when the highest activities are occurring
because they correlate with ideal Ni(OH)2 electrodeposition cur-
rent efficiency. The direct coupling of Ni0 oxidation through (4)
with the conjugate reduction of NO 3 through (1) to form Ni(OH)2
prevents the diffusion of the nitrate reduction’s product OH back
into the solution’s bulk phase to produce ideal electrodeposition
current efficiencies. The nitrate ion is such a strong oxidizing agent
according to (1) that it could oxidize Ni0 at potential biases more
negative than 1.0 V vs. SHE. The characterization and comparison
of electrolyte compositions in Section 3.1 using ion substitutions
and omissions applied in this work indicated that both a metallic
Ni intermediate state and NO 3 are involved with achieving ideal
electrodeposition current efficiency for Ni(OH)2 as well as the ab-
sence of Ni and NiO in the final product. Mechanism (iv)’s con-
straint of matching the rate of Ni0 electrodeposition with the rate
of its oxidation by NO
3 would also support the dramatic depen-
dence of the ideal electrodeposition current efficiency (or greatest
activity) upon both Ni(NO3)2 concentration and deposition current
density (Fig. 4). Similarly, the rate dependence of upon both reac-
tions (4) and (1) would also explain why neither ideal electrode-
position current efficiency nor high activity could be achieved
using galvanostatic pulse or potentiostatic techniques for Ni(OH)2
electrodeposition.
3.3. Coulometric analysis
The Ni(OH)2 deposited at 1.0 mA/cm2 became fully charged
when oxidized by 1.84 ± 0.02 e per Ni ion and could subsequently
deliver 1.58 ± 0.02 e per Ni ion. The ability to both charge and
160
experimental
Mass Ni in deposit / µg
120 1 Ni per 2 e-
80
40
0 Rate of initial charge / log (C rate)
0 0.1 0.2 0.3 0.4
Deposition charge / C
Fig. 5. Electrodeposition current efficiency. The mass of Ni deposited as Ni(OH)2 per
coulomb of deposition charge with a 1.0 mA/cm2 deposition current density from
0.08 M Ni(NO3)2. The slope of 304 lg/C corresponds to an ideal deposition current
efficiency of 1 Ni deposited per 2 electrons.
181
discharge >1 e per deposited Ni2+ ion of Ni(OH)2 requires the ‘‘NiO-
OH’’ mechanism to either charge nickel to oxidation states greater
than Ni3+ or the occurrence of oxygen redox chemistry if Ni3+ is
the highest, stable nickel oxidation state. The discharge occurred
in a narrow voltage window of <0.2 V and yielded values of >8000
F/g. The application of this capacitor-appropriate metric to battery
materials is however misleading because the material can only de-
liver a maximum of 1640 C/g, regardless of the potential range uti-
lized. The maximum charge was experimentally limited by oxygen
evolution from water and was similar to the reported value of 1.65
e per Ni ion [4]. The ability to discharge 1.58 e per Ni represents a
high activity for a material where reports of activities >1 e per Ni
are rare and 0.4 e per Ni are more typical [45,46].
The freshly deposited Ni(OH)2 required exceptionally fast rates
of initial charge to achieve the highest activities. Subsequent
charging and discharging was performed at more conventional
rates such as 2C. The ability to discharge >1.4 e per Ni could only
be achieved if the freshly electrodeposited Ni(OH)2 material was
charged under extremely fast rates of >17 C (charged under
200 s) (Fig. 6). The activity of the freshly electrodeposited Ni(OH)2
material decreased if the initial charge rate was slower. The freshly
deposited Ni(OH)2 oxidized by >1.7 e per Ni for all initial charging
rates. The greater discharge capacities caused by the higher initial
charging rates indicates that the higher initial charging rates in-
creased the reversibility of the charge storage system. Charging
the freshly electrodeposited Ni(OH)2 materials at rates >17 C
caused overpotentials of 200–300 mV and may have affected the
initial oxidation reaction pathways or structure of the Ni(OH)2.
The initial discharge rate also had to be 62 C in order to obtain
the highest activities, where slower discharge rates did not release
more charge. Assuming a molecular weight of 92.7 g/mol and a
density of 4.15 g/cm3, a measured deposit mass of 60.15 mg of
Ni(OH)2 would correspond with a thickness of 60.22 lm.
3.4. Raman spectroscopy
The features of the (dis)charge cycle illustrated in Fig. 7 will be
supported here with in situ Raman spectroscopy and with addi-
tional support from the thermodynamic analysis of Section 3.5. A
greater resolution in spectral potential-dependence (25 mV) than
previous characterizations enabled the identification and charac-
terization of less prominent and/or previously unreported spectral
features and behaviors [47,48]. The cathodically deposited Ni(OH)2
was operated in 3 M KOH electrolyte and was initially charged at
20 C. The spectra illustrated in Fig. 8A–D represent the first
1.6
e- discharged / Ni atom
1.4
1.2
1
0.8
0.3 0.7 1.1 1.5
0.5
Fig. 6. Ni(OH)2 activity as a function of initial charge rate. The activity of 0.25 mg/
cm2 deposits of Ni(OH)2 produced under ideal electrodeposition current efficiency
was characterized as a function of initial charge rate. The rate of initial charge is
described in terms of the log of the C rate, where a 1.0 C rate means the electrode
was charged in 1.0 h. The number of e discharged per Ni atom in the deposit
represents the material’s activity.
182 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188

discharge and second charge. The spectra did not change signifi- Table 2
cantly for the subsequent charge cycle. All of the in situ Raman Raman peak assignments.

spectral features have been previously reported except for the Peak shift(s) Species State(s) Ref(s).
peaks in the 800–1500 cm1 range. The Raman peak assignments (cm1)
are in Table 2 and made in accordance with the literature. 445 A1 symmetric Ni–OH Ni(OH)2 [4,10,45,46]
The in situ Raman spectra indicated that a lower potential (0.1– 470–490 Ni(2+)–O stretching b-Ni(OH)2 or NiO(H2O) [45–47]
0.5 V vs. SHE) and a higher potential (0.5–0.65 V vs. SHE) phase 470–490a ? Ni3O4(H2O)2 [47]
470–490b Eg Ni(3+) = O or Ni2O3(H2O) or [10,46,50]
change occur during cycling between charged and discharged Ni(3+)–O bending c-NiOOH
states. The two phase changes have been designated according to 547 – 490c A1g Ni(3+)–O Ni2O3(H2O) or [10,46,50]
their relative potential regions because distinctly different spectral stretching c-NiOOH
changes occur with respect to potential within the two potential 900–1200 Active oxygen O0
1332 –OH Intercalated H2O [46,47]
regions. Fig. 9 presents the potential dependent shifting of the
spectral peaks and Fig. 10 presents the potential dependent change a
With the absence of 547–555 cm1 peak.
b
in relative peak intensity to better distinguish the different spectral With presence of 547–555 cm1 peak.
c
With presence of 470–490 cm1 peak.
changes which occur between the lower and upper potential
ranges. Two phase changes occur for both the charge and the dis-
charge mechanisms because both mechanisms have an intermedi-
to a state referred to as c-NiOOH. A new peak emerges around
ate state (Fig. 7).
547 cm1 and rapidly shifts to 555 cm1 as the 490 cm1 peak
The lower potential phase change during charging was associ-
shifts to 470 cm1 and more rapidly increases in intensity. The
ated with the oxidation of b-Ni(OH)2 to Ni3O4(H2O)2 due to the
similarity of the shifts and relative 470/555 cm1 peak intensities
changes of the 445 cm1 and 490 cm1 peaks upon charging from
are indicative of the c-NiOOH state, which is similar to the Raman
0.250 V to 0.500 V (Figs. 9 and 10) [47,49–51]. Chemically precip-
spectra of discharged LiNiO2 [18,47,51,57,58]. The nomenclature of
itated Ni(OH)2 is typically a more or less ordered b-Ni(OH)2 while
the c-NiOOH state may be somewhat misleading as the lack of any
the more active a-Ni(OH)2 produced from cathodic electrodeposit-
other visible or Raman peaks corresponding to an OH group indi-
ion is typically characterized as the more disordered and hydrated
cate that the charged ‘‘NiOOH’’ is both dehydrated and has approx-
a-Ni(OH)2 [31,32,52–54]. Cathodically deposited a-Ni(OH)2 has
imately no (<0.2 H+ per Ni) protons for the most active battery
been shown to adopt b-Ni(OH)2 character upon charge/discharge
materials [4,5,47,57,59]. Similarly, the remaining 1/3 of the inter-
and this was consistent with our results [5,55]. The shift of the
calated water is lost upon the completion of charge as indicated
490 cm1 peak towards 470 cm1 as the potential surpasses
by the disappearance of the 1332 cm1 peak at the highest poten-
0.450 V while the 540–560 cm1 peak remains absent is indicative
tials. The absence of OH or intercalated H2O species as well as the
of Ni3O4(H2O)2 as the lower potential phase product during charge
similarity with discharged LiNiO2 Raman spectra supports both the
(Fig. 9) [51]. The fully discharged state has spectral characteristics
absence of protons and the presence of activated oxygen, respec-
of both b-Ni(OH)2 and Ni3O4(H2O)2 as expected because coulomet-
tively, in Fig. 7’s higher potential product state.
ric and spectroscopic results from both this work and the literature
The higher potential phase stopped changing with respect to
indicate that the ‘‘NiOOH’’ never fully discharges back to a Ni(OH)2
potential during charging at about 0.600 V (Figs. 9 and 10), which
state [51,55,56]. The lower potential phase change also includes
coincided with the onset of the oxygen evolution reaction. The Ra-
the loss of 2/3 of the 1332 cm1 peak’s intensity up to 0.500 V
man spectra did not change noticeably even at 0.650 V even
and the emergence of the broad, low intensity peak ranging from
though the oxygen evolution reaction was proceeding at about
900 to 1200 cm1. The 1332 cm1 peak has been attributed to
1 mA/cm2. The results suggest that the characteristics of the 470,
intercalated H2O, which can appear in the 1000–1400 cm1 range
555, and/or 900–1200 cm1 peaks correspond with the presence
instead of contaminating NO 3 because of its consistent disappear-
of oxygen atoms, O0 (½O2), in ‘‘NiOOH’’ for the following three rea-
ance/reappearance with charging/discharging (NO 3 redox reac- sons: (1) these peak intensities and positions only emerge above
tions are irreversible in this voltage range) [50,51]. The loss of 2/
the oxygen evolution equilibrium potential of 0.400 V vs. SHE at
3 of the 1332 cm1 peak’s intensity up to 0.500 V indicates that
pH 14, (2) these peak intensities and positions become constant
the less hydrated Ni3O4(H2O) is the lower phase change’s predom-
as oxygen evolution dominates the charging current, and (3) there
inant product instead of the more hydrated Ni3O4(H2O)2 state
are no other features or phase changes which could be correlated
(Fig. 7).
to an oxygen evolution intermediate state. The broad peak of
The upper potential phase change during charging initiated
900–1200 cm1 may correspond to the O0, because it is much more
near 0.500 V and was associated with the oxidation of Ni3O4(H2O)
prominent in c-NiOOH than b-NiOOH and because it is red-shifted
from the intercalated water peak [47]. A more definitive

Table 3
Eqs. (6)–(10): known nickel and oxygen redox reactionsa [32].

Ni3 O4 ðH2 OÞ2 þ 2Hþ þ 2e $ 3NiOðH2 OÞ (6)

E0 ð6Þ ¼ 0:837  0:0592  pH
Ni2 O3 ðH2 OÞ þ 2Hþ þ 2e $ 2NiOðH2 OÞ (7)
E0 ð7Þ ¼ 0:912  0:0592  pH
3Ni2 O3 ðH2 OÞ þ 2Hþ þ 2e $ 2Ni3 O4 ðH2 OÞ2 (8)
E0 ð8Þ ¼ 1:184  0:0592  pH
2NiO2 ðH2 OÞ2 þ 2Hþ þ 2e $ Ni2 O3 ðH2 OÞ þ 4H2 O (9)
E0 ð9Þ ¼ 1:434  0:0592  pH
1=2O2 þ 2Hþ þ e $ H2 O (10)
E0 ð10Þ ¼ 1:288  0:0592  pH
a
See Methods section for details of how these potentials were adjusted from
Fig. 7. The ideal ‘‘NiOOH’’ active cycle. original source.
 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188 183

Intensity / Arb. Units

Intensity / Arb. Units

350 550 750 800 1100 1400

Raman shift / cm-1 Raman shift / cm-1
Intensity / Arb. units

Intensity / Arb. units

350 550 750 800 1100 1400

Raman shift / cm-1 Raman shift / cm-1

Fig. 8. The low (A) and high (B) in situ Raman shifts during charge and the low (C) and high (D) shifts during discharge. The lines indicate the positions for the 445 cm1 (red),
470 M 490 cm1 (blue), 447 M 545 cm1 (green), and 1332 cm1 (orange) peaks. The transitions between lower and higher potential phase changes occurs at approximately
0.500 V during charge and 0.350 V during discharge. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

20 1.0

0.8
Relative intensity
Shift in peak / cm-1

10
0.6

0.4
0
0.2

-10 0.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 0.1 0.2 0.3 0.4 0.5 0.6
Potential vs SHE / V
Potential vs SHE / V
Fig. 10. The change in relative Raman peak intensity with respect to potential.
Fig. 9. The change in Raman peak cm1 with respect to potential.

assignment of this peak as the active oxygen species is difficult be-

being oxidized and not the permanently anionic O2 responsible
cause the 900–1200 cm1 range is commonly ignored in in situ Ra-
for structurally holding the nickel cations together. The oxidation
man studies of relevant metal oxides. Thermodynamic evidence
of only hydrating/intercalated H2O and not the H2O of bulk water
presented in the next section indicates that only the O2 belonging
is also consistent with quartz crystal microbalance analysis [60].
specifically to the hydrating/intercalated H2O at 1330 cm1 are
184 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188

The ‘‘NiOOH’’ discharge mechanism was different than the Table 4

charge mechanism. The complete emergence of the intercalated Eqs. (11)–(18): relevant charge and discharge mechanisms.

H2O (1332 cm1) peak and the complete disappearance of the Ni3 O4 O0 ðH2 Oðl0 Þ Þ þ 4Hþ þ 4e $ 3NiOðH2 Oðl0 Þ Þ (11)
broad 900–1200 cm1 peak directly coincided with the decrease E0 ð11Þ ¼ 2:125  0:1183  pH
of the c-NiOOH (470 and 555, cm1) peaks during the discharge’s NiOðH2 Oðl0 Þ Þ þ Ni2 O3 ðH2 OðsÞ Þ $ Ni3 O4 ðH2 OðsÞ ÞðH2 Oðl0 Þ Þ (12)
higher potential phase down to 0.350 V (Fig. 10). The shifting of E0 ð12Þ ¼ 0:347
the 470 and 555 cm1 peaks back to their original discharge posi- Ni2 O3 O0 þ 4Hþ þ 4e $ 2NiOðH2 Oðl0 Þ Þ (13)

tions of 490 and 547 cm1, respectively, plus the reappearance of E0 ð13Þ ¼ 2:200  0:1183  pH
3Ni2 O3 O0 þ 4Hþ þ 4e $ 2Ni3 O4 O0 ðH2 Oðl0 Þ Þ (14)
the intercalated water peak at 1332 cm1 indicates the formation
of NiO(H2O) at potentials 60.450 V [47,50]. This indicates that E0 ð14Þ ¼ 2:472  0:1183  pH
2NiOO0 $ Ni2 O3 O0 (15)
the Ni2O3 state in combination with the active oxygen character
0 (16)
of the broad 900–1200 cm1 peak was being maintained through- NiOO þ 2Hþ þ 2e $ NiOðH2 OðsÞ Þ
E0 ð16Þ ¼ 1:288  0:0592  pH
out the charge delivery potentials and discharged to the NiO(H2O)
Ni3 O3 F2 ðH2 OðsÞ ÞðH2 Oðl0 Þ Þ þ H2 O þ 2e $ 3NiOðH2 Oðl0 Þ Þ þ 2F (17)
state instead of Ni(OH)2. The 445 cm1 peak indicating the proton-
Ni3 O4 O0 FðH2 Oðl0 Þ Þ þ 4Hþ þ 3e þ F $ Ni3 O3 F2 ðH2 OðsÞ ÞðH2 Oðl0 Þ Þ (18)
ated Ni–OH species of Ni(OH)2 did not reappear during discharge
until 0.150 V. The 470–490 cm1 peak was always accompanied
by the 547–555 cm1 peak during discharge, which indicates that
a hydrated Ni3O4 state did not form during discharge. Charging 0.01 M KOH
through a state with Ni3O4 character and discharging to a state 0.75 0.03 M KOH
with NiO(H2O) character is consistent with the well-known hyster- 0.1 M KOH
0.3 M KOH
esis of the ‘‘NiOOH’’ and underscores the importance of reporting

E vs SHE / V
0.65 1 M KOH
whether potential-dependent in situ spectra have been obtained 3 M KOH
during charge or discharge. 0.55

3.5. Thermodynamic analysis

0.45

An active oxygen mechanism of ‘‘NiOOH’’ consistent with spec-

0.35
troscopic and voltammetric characterizations (Fig. 7) can be con- 0 1 2 3 4 5 6
structed from the thermodynamics of known nickel and oxygen time / ks
redox reactions. The original source of the electrochemical poten-
tials of reaction (6)–(10) in Table 3 considered the hydrated nickel Fig. 11. ‘‘NiOOH’’ (dis)charge chronopotentiograms. The potential dependence of
oxide’s intercalated H2O to have the entropy of bulk phase liquid the 2nd charge/discharge curves with respect to KOH concentration for 0.25 mg/
cm2 loadings deposited from 0.08 M Ni(NO3)2 at 1 mA/cm2. The markers indicate
water. The nickel and oxygen redox potentials in Table 3 have been
the inflection points evaluated in Fig. 12.
adjusted because the water intercalated in the nickel oxides do not
have the entropy of bulk phase liquid water. Water associated with
Ni2+ ions have been designated as H2O(l0 ) because they have an en- 1.0
tropy similar to that of liquid water and the water associated with
Ni3+ ions have been designated as H2O(s) because they have the en-
E vs SHE / V

tropy of ice (see Methods section). The ‘‘NiOOH’’ reaction mecha-
nisms of Table 4 have been constructed using the potentials of
Table 3 and will be supported by experimentally derived thermo-
dynamic values.
The thermodynamics of ‘‘NiOOH’’ active material produced un-
der ideal electrodeposition current efficiency were galvanostatical-
ly (dis)charged with respect to variable electrolyte KOH
concentrations (Fig. 11). The pH and potential dependence of the
(dis)charge curve inflection points are illustrated in Fig. 12. The
(dis)charge curve inflection points should correspond to rational
equilibria in an approximately ideal system and correlate with
the pH and potential dependence of the mechanisms in Table 4.
The slight deviations of the experimentally determined inflection
points in the chronopotentiometric curves from the calculated
equilibria at the highest KOH concentrations (Fig. 12) was likely
due to the significant deviation of the solvent water’s activity (or
osmotic) coefficient from the ideal conditions assumed in the ther-
modynamic calculations [34,61]. Note in Fig. 12 that the 120 mV/
pH dependence of the charge curve inflection points and the
60 mV/pH dependence of the discharge curve inflection points
at steady state conditions further indicated that the hysteresis oc-
curs due to a permanent thermodynamic effect and that the charge
and discharge mechanisms were not simply the reverse of each
other [62].
The oxidation states derived from the thermodynamic analysis
corroborate the interpretation of the in situ Raman experiments.
The inflection points of the galvanostatic (dis)charge curves match
both the potential and pH dependence of mechanism (11). It is not
0.8
0.6
0.4
11.5 12.5 13.5 14.5
pH
Fig. 12. Correlation of experimental ‘‘NiOOH’’ thermodynamic values with calcu-
lated equilibria. The experimental equilibrium values derived from inflection points
of Fig. 11’s chronopotentiograms are compared with the equilibria calculated for
mechanisms described in Table 4.
clear from thermodynamics alone why an active oxygen mecha-
nism would occur instead of the more energetically favorable oxi-
dation of just the nickel ions yet equilibrium thermodynamics
alone yield no insight into the active material’s coordination chem-
istry. It is apparent from in situ X-ray adsorption experiments that
the coordination of the nickel ions remains constant and only dis-
tortion in the (a)symmetry occurs during (dis)charge [6–10,54,63].
The differences between the experimentally determined inflection
points and the calculated equilibria in Fig. 12 indicate that the acti-
vation energies for the active oxygen mechanism are small. The
activation energies for changing the material’s coordination chem-
istry to accommodate only nickel redox chemistry must therefore
 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
be relatively large for the observed selectivity for the active oxygen
mechanism to occur. Thermodynamics indicate that O0 is stabi-
lized in the H2O(s) position so that the product is Ni3O4O0(H2O(l0 ))
(Fig. 7). The preference for specifically the Ni3O4O0(H2O(l0 )) state
near 0.500 V in 3 M KOH was consistent with the Raman spectra
in that the Ni3O4 state appears during charge, a second water-cor-
related peak never actually appears for H2O(s), and that 1/3 of the
water species present in the discharged state notably persisted
up to 0.500 V (Fig. 10).
The ‘‘NiOOH’’ oxidizes from the Ni3O4O0(H2O(l0 )) state towards
the Ni2O3O0 state while charging through the upper potential
phase change (Fig. 7). The potential during charge surpasses
the disproportionation reaction of (12) so that Ni3O4(H2O(s))(H2-
O(l0 )) can spontaneously transition into separate NiO(H2O(l0 )) and
Ni2O3(H2O(s)) states. The degeneracy of (12) provides a means
for the Ni2+ ion in Ni3O4O0(H2O(l0 )) to reach the Ni2O3O0 state
as the latter becomes stable through (13). The ‘‘NiOOH’’ can
therefore oxidize through (14) at the potential of (13) in a solid
state model. Oxygen atoms, O0, seem to be able to combine to
form molecular oxygen, O2(g), in the Ni2O3O0 state and so the
experimental charging curve inflection point caused by oxygen
evolution (Fig. 12) corresponds with the potential of mechanism
(13) as well as the lack of further potential-dependent evolution
in the Raman spectra above 0.6 V vs. SHE (Figs. 8A–D). The pro-
posed fully charged state, which is composed of an equilibrium
between the Ni3O4O0(H2O(l0 )) and Ni2O3O0 states due to the con-
straints of O2(g) evolution, is approximately absent of protonated
species and in accordance with Corrigan’s and our inability to
detect OH interactions with Ni ions in the fully charged condi-
tion. The fully charged state is also in accordance with a pro-
posed mechanism in which optimal O2(g) evolution catalytic
efficiency occurs when the M3O4 and M2O3 oxidation states
interact in predominantly dehydrated first row transition metal
oxide compounds [44]. The intermediate state of Ni3O4O0(H2O(l0 ))
has released 1.67 e per Ni ion, the ideal charged state of Ni2O3-
O0 has released 2.00 e per Ni ion, and the equilibrium (average)
condition between the two states corresponded to the experi-
mental limit of oxygen evolution where the ‘‘NiOOH’’ was oxi-
dized by 1.84 ± 0.02 e per Ni ion.
The proposed charging process indicates that the fully charged
condition was composed of predominantly Ni2O3O0 plus a fraction
of Ni3+ whose O0 is absent as a result of O2 evolution catalysis. The
Ni3+ without O0 constitutes about 20–22% of the nickel in the fully
charged condition and was inert during the discharge process be-
cause the ‘‘NiOOH’’ finished discharging before approaching the
potential of (7). While the discharge potential of Ni2O3 to Ni3O4
was similar to the experimental discharge asymptote according
to the conventional potential of (8), the adjustment of the redox
potential of (8) to account for the lower entropies of the hydrating
H2O mean that the ‘‘NiOOH’’ is fully discharged at potentials 50 mV
more positive than the unadjusted value. The absence of Ni3O4 Ra-
man spectroscopic character during discharge means that the dis-
charge of the Ni3+ unassociated with O0 to Ni2.67+ cannot be a
predominant mechanism if it is active at all. Similarly, ‘‘NiOOH’’
finished discharging at potentials 0.2–0.3 V positive of the unad-
justed equilibria of (6) and (7) and 0.3–0.4 V positive of the ad-
justed equilibria of (6) and (7) (see Fig. 11) and so these nickel
reduction reactions by themselves could not have been significant
pathways during discharge. This means that all of the Ni3+ unasso-
ciated with the active oxygen species in the fully charged state did
not discharge to lower oxidation states during discharge and re-
mained as Ni3+ in the fully discharged state. This may indicate that
changes in coordination chemistry have occurred for the inert Ni3+.
Charge and discharge through other reactions, which could include
active peroxides instead of oxygen, may have developed after sig-
nificant deactivation of the ‘‘NiOOH’’ through repetitive cycling.
185
The potential-dependent selectivity of viable nickel redox reac-
tions during discharge was also supported by coulometric analysis.
The 7–9% charge loss due to the inability to completely charge the
‘‘NiOOH’’ because of oxygen evolution plus a 4 – 5% charge loss due
to instability with cycling causes the ‘‘NiOOH’’ to only be able to
discharge by 1.58 ± 0.02 e per Ni ion to an average Ni valence of
+2.26 ± 0.02. The discharge potentials and inflection points
matched that expected for the discharge of NiOO0 to NiO(H2O(s))
through (16) (Fig. 12). This means that both the thermodynamics
and coulometrics indicate that Ni2O3O0 relaxes into 2 NiOO0
according to (15) before discharging to 2 NiO(H2O(s)) as H+ are in-
serted into the local environment. Note that only the conversion of
Ni2O3O0 to 2 NiO(H2O(s)) was detected in the in situ Raman exper-
iments while only the conversion of NiOO0 to NiO(H2O(s)) was de-
tected through thermodynamics and so the latter conversion must
have proceeded immediately after the former conversion (Fig. 7).
The proposed discharge mechanism is in agreement with other
cation insertion electrode materials such as LiNiO2 as expected due
to the similarities of between H+ and Li+ (de-)intercalation. Struc-
turally, the biggest difference between LiNiO2 and ‘‘NiOOH’’, which
can be alternatively written as HNiO2, is that LiNO2 is convention-
ally well-ordered in Li-ion electrode materials while HNiO2 is con-
ventionally highly defective structurally in electrochemically
deposited ‘‘NiOOH’’ materials. Both experimental and modeling
techniques report that at defect sites of LiNiO2, it is thermodynam-
ically favorable for oxygen species to oxidize so that Ni3+ can re-
duce to Ni2+ as the Ni2+ moves into Li+ ion vacancies [15,16,18].
This discharge mechanism is consistent with the report that hys-
teresis is thermodynamically governed [62]. The experimental re-
sults here indicate that the ‘‘NiOOH’’ material was so defective
that 95–96% of the Ni3+ moved into H+ vacancies as Ni2+ to form
NiOO0 during discharge and would explain why the residual Ni3+
does not discharge to lower oxidation states until the evolution
of aging effects. The degree of structural and bonding changes
which must be occurring at defects for the active oxygen mecha-
nism unfortunately may indicate that instability in aqueous elec-
trolytes may be inherent for such disordered and highly active
materials.
3.6. Fluoride partial inhibition
The H2O(l0 ) and H2O(s) was experimentally distinguishable be-
cause of the exchange of H2O(l0 ) with the aqueous solvent. Water’s
self-diffusion is exceptionally fast because it involves the dissocia-
tion into H+ + OH in an intermediate state [64,65]. This solvent ex-
change mechanism can be inhibited in a condition when an F ion
(133 pm) can kinetically and thermodynamically outcompete OH
(133 pm). Solution phase calculations supported by experimental
results indicated that the electronegative F is competitive with
OH for Ni2+ in a 1 M KF:0.01 M KOH solution [66]. The inhibition
by F caused oxidation by 0.67 e per Ni ion to the Ni3O3F2(H2-
O(s))(H2O(l0 )) according to (17) and the coulometric charge indi-
cated in Fig. 13. Elemental analysis with EDX after the first
charge step at the point marked by h in Fig. 13 indicated that there
was 0.59 of the 0.67 F per Ni predicted in the product of (17). The
subsequent oxidation of Ni3O3F2(H2O(s))(H2O(l0 )) to Ni3O4O0F(H2-
O(l0 )) by another e per Ni ion through (18) was limited by oxygen
evolution, where 0.31 of the expected 0.33 F per Ni measured at
the point marked by d in Fig. 13. The potential-dependence dem-
onstrated by the effects of KF in the electrolyte suggests that the
number of hydrating H2O which can exchange with the solution
phase could change with activity, cycling, or exchange with KOH
and may explain the potential shifts which occur with aging [3].
The potential shifts correlated with aging could alternatively be
due to the oxygen redox chemistry switching to a peroxide product
state instead of the O0 state.
186 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
1.2 3 M KF + 0.001 M KOH
1 M KOH
E vs SCE / V
0.8
0.6
0.4
0 1 2 3 4
time / ks
Fig. 13. Partial inhibition of ‘‘NiOOH’’ with F. 0.25 mg/cm2 loadings deposited
from 0.08 M Ni(NO3)2 at 1 mA/cm2 were galvanostatically characterized in both
1 M KOH and 3 M KF + 0.001 M KOH to determine if F ions could selectively inhibit
the active oxygen mechanism of H2O(l0 ) but not H2O(s). The symbols h and d
indicate where the ‘‘NiOOH’’ charged in KF was sampled for elemental analysis.
Fig. 14A. SEM image of Ni(OH)2 deposited under ideal electrodeposition current
efficiency. The cathodic deposition of Ni(OH)2 from 0.08 M Ni(NO3)2 at 1 mA/cm2
would have produced a 0.22 lm thick film if the deposit was uniformly flat and
fully dense at 4.15 g/cm3.
3.7. Prospects for practical implementation
The morphology of the Ni(OH)2 deposited onto HOPG was not
ideal for practical implementation. The narrow range of optimum
electrodeposition current densities (Fig. 4) could be difficult to
extrapolate from flat to porous electrodes. The 300 mC charge
passed during deposition at 1 mA/cm2 was ideally expected to
form a uniform film 0.22 lm thick, assuming the density of
4.15 g/cm3, because the electrodeposition current efficiency was
100%. The Ni(OH)2 instead formed clumps with heights of about
1.5 lm (Fig. 14A). The morphology of Fig. 14A suggests that the
density of the Ni(OH)2 may be approximately half of the expected
4.15 g/cm3 due to porosity. A low density and high surface area de-
posit from cathodic electrodeposition was not unexpected due to
the resulting material’s proficiency for supercapacitor applications
[67–70]. Switching to a 0.3 s pulse at 1.0 mA/cm2 and 2.1 s
pauses resulted in a more evenly distributed film with a signifi-
cantly lower electrodeposition current efficiency and/or activity
(Fig. 14B).
The highly active nickel material was unstable throughout re-
peated charge and discharge cycles (Fig. 15). Charge storage capa-
bility degenerated with a 5% loss with each successive charge or
discharge. The film thickness, charge and discharge rates, depth of
discharge, overcharge, electrolyte KOH concentration, or the addi-
tion of LiOH to the electrolyte did not improve the stability signif-
icantly enough (data not shown) to be useful in practical
rechargeable battery applications. Despite the new insight into
Fig. 14B. SEM image of Ni(OH)2 deposited with pulsed current. The Ni(OH)2 was
deposited from 0.08 M Ni(NO3)2 at 1 mA/cm2 in 0.3 s pulses and 2.1 s pauses.
100
charge retention / %
80
60
40
20
0
0 30 60 90 120
cycle #
Fig. 15. Cycle stability of ‘‘NiOOH’’ at optimum activity. A 0.25 mg/cm2 deposit of
‘‘NiOOH’’ at optimum activity was prepared by electrodeposition at ideal current
efficiency followed by an initial charge at 20 C in 1 M KOH. The ‘‘NiOOH’’ was then
cycled between 0.4 and 0.6 V vs. SCE with a 2 C rate.
the charge/discharge mechanism, it is not yet clear whether highly
active ‘‘NiOOH’’ materials can have competitive stabilities or if the
deactivation by changes in the hydrated states are unavoidable.
4. Conclusion
An optimal combination of Ni(NO3)2 concentration and current
density resulted in an ideal cathodic electrodeposition current effi-
ciency and the highest deposit activities. The cathodic electrode-
position of nickel hydroxide involved the deposition of Ni2+ into
the metallic (Ni0) state. The Ni0 is oxidized back to Ni2+ by the
reduction of NO 
3 to NO2 and the coupling of these reactions and
their products causes Ni(OH)2 precipitation with an ideal deposi-
tion current efficiency of 1 Ni2+ per 2 e. Ni0 can alternatively cat-
alyze NO 
3 reduction or hydrogen evolution to generate the OH
product gradient which can chemically precipitate Ni(OH)2 from
aqueous Ni2+ under conditions of non-ideal electrodeposition cur-
rent efficiencies.
Highly active ‘‘NiOOH’’ deposits could be initially oxidized to
store up to 1.84 e per Ni ion and could subsequently deliver a
charge of 1.58 e per Ni ion at optimal performance. The approxi-
mately ideal charge capacity of the ‘‘NiOOH’’ facilitated a thermo-
dynamic analysis through which the specific charge and discharge
mechanisms could be resolved from the enthalpies of formation for
known states. Significant changes in the galvanostatic charge and
discharge curves occurred as the ‘‘NiOOH’’ aged with cycling. A
quantitative thermodynamic or in situ Raman spectroscopic
 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188
analysis of the aged material to determine whether active oxygen
species switched from O0 to O1 should be the subject of future
work. Poor electrodeposition constraints, morphology and stability
with cycling currently prevent practical application of this highly
active material in secondary batteries.
The results of both the thermodynamic analysis and the in situ
Raman spectroscopy characterization of the ‘‘NiOOH’’ charge and
discharge pathways at optimal performance were consistent with
an active oxygen mechanism. Successfully modeling the thermo-
dynamics of the ‘‘NiOOH’’ (dis)charge cycle represents a distinct
achievement which cannot be obtained through other proposed
mechanisms (see Supplementary Information). Because it is now
known that (1) bond lengths alone cannot distinguish between
the Ni4+ state vs. the Ni3+ state plus activated oxygen, and (2) that
the BaNiO3 and KNiIO6 X-ray adsorption references actually corre-
spond to the Ni3+ state plus activated oxygen, there is no longer
any significant evidence substantiating the Ni4+ state in the ‘‘NiO-
OH’’ mechanism. Constructing the ‘‘NiOOH’’ mechanism from
known redox reactions required adjustment of the nickel oxide
reduction potentials because Pourbaix considered all hydrating
H2O species to have an entropy equivalent to the liquid phase,
which is not an accurate assumption. Modifying the entropies of
H2O(l0 ) associated with Ni2+ to a known value slightly smaller than
that of the liquid phase and the entropy of the H2O(s) associated
with Ni3+ to an ice-like entropy resulted in an excellent correlation
with the experimentally derived equilibria. The selectivity for only
H2O(s)’s in the oxygen redox activity was consistent with the
Li(1X)NiO2 system in which oxygen electron vacancies only occur
in the presence of Ni3+ upon charging. The ‘‘NiOOH’’ discharge
mechanism through the formation of oxygen electron vacancies
associated with Ni2+ was similarly consistent with the discharge
of a Li(1X)NiO2 system into a LiXNi(1X)O system at defect sites. To-
gether, these (dis)charge mechanisms are also consistent with a
thermodynamically governed hysteresis as well as proton interca-
lation. The elucidation of these relatively subtle mechanistic
behaviors was facilitated in part by generating Raman spectra with
a greater potential-dependent resolution to observe a secondary,
less prominent phase change and distinguishing between the
charge and discharge processes.
Acknowledgements
This work was performed under the auspices of the U.S. Depart-
ment of Energy by Lawrence Livermore National Laboratory under
Contract DE-AC52-07NA27344. Project 11-LW-037. The authors
would also like to acknowledge Rachel E. Lindvall’s contribution
for the ICP–MS analysis and Jonathan Lee’s discussions regarding
X-ray absorption spectroscopic techniques.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jelechem.20
14.01.022.
References
[1] J. Phillips, S. Mohanta, M. Geng, J. Barton, B. McKinney, J. Wu, ECS Trans. 16
(2009) 11–17.
[2] T. Reddy, D. Linden, Linden’s Handbook of Batteries, McGraw-Hill, 2010.
[3] A.H. Zimmerman, Nickel–Hydrogen Batteries: Principles and Practice,
Aerospace Press, 2010.
[4] D.A. Corrigan, S.L. Knight, J. Electrochem. Soc. 136 (1989) 613–619.
[5] J. Desilvestro, D.A. Corrigan, M.J. Weaver, J. Electrochem. Soc. 135 (1988) 885–
892.
[6] T.W. Capehart, D.A. Corrigan, R.S. Conell, K.I. Pandya, R.W. Hoffman, Appl. Phys.
Lett. 58 (1991) 865–867.
187
[7] N.R.S. Farley, S.J. Gurman, A.R. Hillman, Electrochim. Acta 46 (2001) 3119–
3127.
[8] A.N. Mansour, R.A. Brizzolara, C.A. Melendres, M. Pankuch, J. Electrochem. Soc.
141 (6) (1994) L69–L72 (United States).
[9] W.E. Ogrady, K.I. Pandya, K.E. Swider, D.A. Corrigan, J. Electrochem. Soc. 143
(1996) 1613–1616.
[10] Y.N. Hu, I.T. Bae, Y.B. Mo, M.R. Antonio, D.A. Scherson, Can. J. Chem. – Rev. Can.
De Chim. 75 (1997) 1721–1729.
[11] X. Qian, H. Sambe, D.E. Ramaker, K.I. Pandya, W.E. O’Grady, J. Phys. Chem. B
101 (1997) 9441–9446.
[12] R. Gottschall, R. Schöllhorn, M. Muhler, N. Jansen, D. Walcher, P. Gütlich, Inorg.
Chem. 37 (1998) 1513–1518.
[13] H. Sambe, T.M. Nabi, D.E. Ramaker, A.N. Mansour, W.E. O’grady, The Oxidation
State of Ni in the Nickel Oxide Electrode and Related Nickel Oxide Compounds:
I. Spectroscopic Evidence, Defense Technical Information Center, 1997.
[14] H. Sambe, T.M. Nabi, D.E. Ramaker, The Oxidation State of Ni in the Nickel
Oxide Electrode and Related Nickel Oxide Compounds: II. Geometric Evidence,
Defense Technical Information Center, 1997.
[15] A. Rougier, C. Delmas, A.V. Chadwick, Solid State Commun. 94 (1995) 123–127.
[16] M.K. Aydinol, A.F. Kohan, G. Ceder, K. Cho, J. Joannopoulos, Phys. Rev. B 56
(1997) 1354–1365.
[17] J. van Elp, H. Eskes, P. Kuiper, G.A. Sawatzky, Phys. Rev. B 45 (1992) 1612–
1622.
[18] P. Kalyani, N. Kalaiselvi, Sci. Technol. Adv. Mater. 6 (2005) 689–703.
[19] M.C. Biesinger, B.P. Payne, L.W.M. Lau, A. Gerson, R.S.C. Smart, Surf. Interface
Anal. 41 (2009) 324–332.
[20] A.P. Grosvenor, M.C. Biesinger, R.S.C. Smart, N.S. McIntyre, Surf. Sci. 600 (2006)
1771–1779.
[21] I.G. Casella, M.R. Guascito, M.G. Sannazzaro, J. Electroanal. Chem. 462 (1999)
202–210.
[22] T. Dickinson, A.F. Povey, P.M.A. Sherwood, J. Chem. Soc., Faraday Trans. 1: Phys.
Chem. Condens. Phases 73 (1977) 327–343.
[23] Y. Koyama, T. Mizoguchi, H. Ikeno, I. Tanaka, J. Phys. Chem. B 109 (2005)
10749–10755.
[24] Y. Koyama, Y.-S. Kim, Isao Tanaka, H. Adachi, Jpn. J. Appl. Phys. 4 (38) (1999)
2024.
[25] Y. Uchimoto, H. Sawada, T. Yao, J. Power Sources 97–98 (2001) 326–327.
[26] J. Graetz, C.C. Ahn, H. Ouyang, P. Rez, B. Fultz, Phys. Rev. B 69 (2004) 235103.
[27] P. Kuiper, G. Kruizinga, J. Ghijsen, G.A. Sawatzky, H. Verweij, Phys. Rev. Lett. 62
(1989) 221–224.
[28] F. Reynaud, D. Mertz, F. Celestini, J.M. Debierre, A.M. Ghorayeb, P. Simon, A.
Stepanov, J. Voiron, C. Delmas, Phys. Rev. Lett. 86 (2001) 3638–3641.
[29] A.W. Moses, H.G.G. Flores, J.-G. Kim, M.A. Langell, Appl. Surf. Sci. 253 (2007)
4782–4791.
[30] H. Chen, C.L. Freeman, J.H. Harding, Phys. Rev. B 84 (2011) 085108.
[31] M. Wohlfahrt-Mehrens, R. Oesten, P. Wilde, R.A. Huggins, Solid State Ionics
86–88, Part 2 (1996) 841–847.
[32] R.S. Jayashree, P.V. Kamath, J. Power Sources 93 (2001) 273–278.
[33] C.C. Streinz, A.P. Hartman, S. Motupally, J.W. Weidner, J. Electrochem. Soc. 142
(1995) 1084–1089.
[34] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, National
Association of Corrosion Engineers, 1974.
[35] C.D. Taylor, S.A. Wasileski, J.-S. Filhol, M. Neurock, Phys. Rev. B 73 (2006)
165402.
[36] P. Atkins, J.D. Paula, Physical Chemistry, W. H. Freeman and Co., 2009.
[37] W.M. Haynes, CRC Handbook of Chemistry and Physics: A Ready-Reference
Book of Chemical and Physical Data, CRC Press, Boca Raton, FL, 2011.
[38] Y. Marcus, Chem. Rev. 109 (2009) 1346–1370.
[39] K. Bouzek, M. Paidar, A. Sadílková, H. Bergmann, J. Appl. Electrochem. 31
(2001) 1185–1193.
[40] Zhao, Zhao, Li, Chem. Mater. 19 (2007) 3882–3891.
[41] T. Subbaiah, S.C. Mallick, K.G. Mishra, K. Sanjay, R.P. Das, J. Power Sources 112
(2002) 562–569.
[42] A.C. Sonavane, A.I. Inamdar, P.S. Shinde, H.P. Deshmukh, R.S. Patil, P.S. Patil, J.
Alloy. Compd. 489 (2010) 667–673.
[43] K.-W. Nam, E.-S. Lee, J.-H. Kim, Y.-H. Lee, K.-B. Kim, J. Electrochem. Soc. 152
(2005) A2123–A2129.
[44] M.D. Merrill, R.C. Dougherty, J. Phys. Chem. C 112 (2008) 3655–3666.
[45] H.G. Zhang, X.D. Yu, P.V. Braun, Nat. Nanotechnol. 6 (2011) 277–281.
[46] H. Wang, H.S. Casalongue, Y. Liang, H. Dai, J. Am. Chem. Soc. 132 (2010) 7472–
7477.
[47] Y.L. Lo, B.J. Hwang, Langmuir 14 (1998) 944–950.
[48] D. Gosztola, M.J. Weaver, Langmuir 5 (1989) 776–782.
[49] S. Deabate, F. Fourgeot, F. Henn, J. Power Sources 87 (2000) 125–136.
[50] C.A. Melendres, S. Xu, J. Electrochem. Soc. 131 (1984) 2239–2243.
[51] C. Johnston, P.R. Graves, Appl. Spectrosc. 44 (1990) 105–115.
[52] P. Oliva, J. Leonardi, J.F. Laurent, C. Delmas, J.J. Braconnier, M. Figlarz, F. Fievet,
A.D. Guibert, J. Power Sources 8 (1982) 229–255.
[53] G. Nazri, D.A. Corrigan, S.P. Maheswari, Langmuir 5 (1989) 17–22.
[54] K.I. Pandya, W.E. O’Grady, D.A. Corrigan, J. McBreen, R.W. Hoffman, J. Phys.
Chem. 94 (1990) 21–26.
[55] R. Kostecki, F. McLarnon, J. Electrochem. Soc. 144 (1997) 485–493.
[56] B.C. Cornilsen, P.J. Karjala, P.L. Loyselle, J. Power Sources 22 (1988) 351–
357.
[57] S. Deabate, F. Fourgeot, F. Henn, Electrochim. Acta 51 (2006) 5430–5437.
[58] B.S. Yeo, A.T. Bell, J. Phys. Chem. C 116 (2012) 8394–8400.
188 M. Merrill et al. / Journal of Electroanalytical Chemistry 717–718 (2014) 177–188

[59] M.K. Carpenter, D.A. Corrigan, J. Electrochem. Soc. 136 (1989) 1022–1026.
[60] G.T. Cheek, W.E. O’Grady, J. Electroanal. Chem. 421 (1997) 173–177.
[61] W.J. Hamer, Y.-C. Wu, J. Phys. Chem. Ref. Data 1 (1972) 1047–1100.
[62] K.P. Ta, J. Newman, J. Electrochem. Soc. 146 (1999) 2769–2779.
[63] J. McBreen, W.E. O’Grady, G. Tourillon, E. Dartyge, A. Fontaine, K.I. Pandya, J.
Phys. Chem. 93 (1989) 6308–6311.
[64] M. Eigen, Discuss. Faraday Soc. 17 (1954).
[65] T. Dippel, K.D. Kreuer, Solid State Ionics 46 (1991) 3–9.
[66] D.o.L.a.W.R. Engineering, Visual MINTEQ 2.54, in: K.R.I.o. Technology (Ed)
<http://www.lwr.kth.se/English/OurSoftware/vminteq/>, Stockholme, 2010.
[67] E.E. Kalu, T.T. Nwoga, V. Srinivasan, J.W. Weidner, J. Power Sources 92 (2001)
163–167.
[68] D.-D. Zhao, S.-J. Bao, W.-J. Zhou, H.-L. Li, Electrochem. Commun. 9 (2007) 869–874.
[69] K.-W. Nam, K.-B. Kim, J. Electrochem. Soc. 149 (2002) A346–A354.
[70] X. Zhang, W. Shi, J. Zhu, W. Zhao, J. Ma, S. Mhaisalkar, T. Maria, Y. Yang, H.
Zhang, H. Hng, Q. Yan, Nano Res. 3 (2010) 643–652.