energies

Article
Identifying the Active Species in Li-Na Dual-Ion “Saltwater
Battery” Based on Spinel Lithium Manganese Oxide, Sodium
Titanium Phosphate and Aqueous Electrolyte
Jonathan Schubert 1, *, Lukas Grossmann 2 , Stefan Seidlmayer 2 , Karl-Heinz Pettinger 1 , Ralph Gilles 2
and Michael A. Danzer 3

1 Technology Center for Energy, University of Applied Science Landshut, Wiesenweg 1,

94099 Ruhstorf a.d. Rott, Germany; [email protected]
2 Heinz-Mayer Leibnitz Zentrum (MLZ), Technical University Munich (TUM), Lichtenbergstraße 1,
85748 Garching, Germany; [email protected] (L.G.); [email protected] (S.S.);
[email protected] (R.G.)
3 Chair of Electrical Energy Systems, University of Bayreuth, Universitätsstraße 30, 95447 Bayreuth, Germany;
[email protected]
* Correspondence: [email protected]

Abstract: The dual-ion “Saltwater Battery” based on aqueous electrolyte containing sodium ions and
lithium ions is believed to be one of the safest and environmentally friendliest battery technologies.
The anode consists of sodium titanium phosphate, whereas the cathode is spinel lithium manganese
oxide. It has been reported that both materials can intercalate sodium and lithium ions depending
on their availability in the electrolyte. This study aims to identify the dominant active species in
“Saltwater Batteries” with dual-ion electrolyte. Therefore, cyclic voltammetry of single electrodes
as well as full cell measurements are performed with electrolyte containing lithium sulfate, sodium
sulfate or a mixture. Moreover, the study is complemented by ion analysis of the electrolyte, as
well as X-ray diffraction of fresh and cycled electrodes at different states of charge. The results
Citation: Schubert, J.; Grossmann, L.; show that the cathode only (de)intercalates lithium ions, revealing that Li is the dominant active
Seidlmayer, S.; Pettinger, K.-H.; cathode species. The anode can react with both ions and undergoes a formation reaction accompanied
Gilles, R.; Danzer, M.A. Identifying by partial dissolution of sodium titanium phosphate. Nevertheless, cyclic voltammetry and full
the Active Species in Li-Na Dual-Ion cell measurements indicate that lithium is also the dominant active species on the anode side. In
“Saltwater Battery” Based on Spinel conclusion, the dual-ion battery is dominated by lithium and shows a superior performance when
Lithium Manganese Oxide, Sodium removing sodium from electrolyte.
Titanium Phosphate and Aqueous
Electrolyte. Energies 2023, 16, 4485. Keywords: Li-Na dual-ion battery; aqueous electrolyte; lithium manganese oxide; sodium titanium
https://doi.org/10.3390/en16114485
phosphate; post-lithium-ion battery materials; sustainable developments
Academic Editor: Carlos
Miguel Costa

Received: 8 May 2023

1. Introduction
Revised: 25 May 2023
Accepted: 27 May 2023 The rapid increase in energy prices as well as climate change have led to a boom
Published: 1 June 2023 in the use of energy storage systems for household applications. The increased self-
sufficiency and independence of power grids are strong arguments for investments in home
storage systems.
Today, lithium-ion batteries (LIBs) are the most available and commercialized home
Copyright: © 2023 by the authors. storage solution. Remaining safety issues and critical material supply chains require further
Licensee MDPI, Basel, Switzerland. improvements in energy storage systems. Currently, LIBs are irreplaceable for mobile
This article is an open access article applications due to their unrivaled high energy and power density. In contrast, stationary
distributed under the terms and
storage systems have other requirements, consequently enabling different technologies to
conditions of the Creative Commons
enter the market. Size, volume and weight, and thus specific energy and power density, are
Attribution (CC BY) license (https://
less important. Safety issues and environmental impact receive more attention as well as
creativecommons.org/licenses/by/
the material availability (resource self-sufficiency). Additionally, redox-flow batteries [1],
4.0/).

Energies 2023, 16, 4485. https://doi.org/10.3390/en16114485 https://www.mdpi.com/journal/energies

Energies 2023, 16, 4485 2 of 18

potassium-ion batteries [2,3], bimetallic-sulfide batteries [4], sodium-ion batteries [3,5]

and dual-ion batteries [6] have drawn more attention to fulfil safety, environmental and
material requirements with a sufficient energy density. Research includes dual-ion systems
based on cations/anions, such as hybrid cathodes for LIBs [7], sodium bromide [8] or
sodium fluoride [9], as well as dual-cation technologies, such as lithium-sodium dual-ion
batteries [10]. The aim of the latter battery type is to combine the benefits of lithium
regarding its capacity with the advantages of sodium regarding its material abundance.
Research has been performed on organic-based Li-Na dual-ion batteries [11] as well as
aqueous-based systems [10,12]. The development of aqueous-based systems includes
intrinsic advantages regarding toxicity, costs, and flammability.
In the last decade, research has led to the development of the commercially avail-
able “Saltwater Battery” for stationary applications as an alternative to LIBs [10,12]. The
technology is based on aqueous electrolyte, a lithium manganese spinel cathode and a
sodium titanium phosphate anode [10,13]. Thus, it is a dual-ion system containing lithium
(cathode) and sodium (anode) ions. Both ions are also part of the electrolyte. In contrast to
the organic solvents used in most LIBs, water functions as a solvent and the basis for ion
transport. Sodium sulfate and lithium sulfate act as conducting salts in order to deliver
both species for the dual-ion battery type [10,13]. Notably, the ionic transfer in aqueous
solutions can exceed that of organic electrolytes significantly [14]. The aqueous electrolyte
also leads to a wide range of possible separators, such as synthetic cotton, due to fewer
requirements according to the chemical stability [10,14].
The electrodes are multicomponent systems, as in conventional LIBs. Two additives,
graphite and carbon black, with different particle sizes and thus various lengths of conduc-
tive connections between the grains of active materials are used. Due to the thickness of the
electrodes being above 1 mm, these conductive agents play a major role in the performance
of the cells [10]. For the “Saltwater Battery”, polytetrafluoroethylene works as a binder. It
performs well within this dual-ion system [15]. Since aqueous electrolyte can become acidic
or alkaline during cell operation, at least locally for a short time, stainless steel collectors of
types 1.4301 or 1.4404 [10,15] are the most prominent in “Saltwater Technology”.
The active materials of the cathode and anode are spinel lithium manganese oxide
LiMn2 O4 (LMO) and sodium titanium phosphate NaTi2 (PO4 )3 (NTP). Both phases have
been reported to work successfully as the host structure for reversible ion intercalation in
aqueous-based electrolyte containing lithium and sodium ions with a beneficial ecological
impact [10,16]. Other commonly known active materials, such as lithium iron phosphate or
cobalt nickel manganese oxides cannot be used in this dual-ion battery type due to their
sensitivity to water. Exemplary SEM pictures of the used LMO and NTP can be found in the
supplementary material (Figure S9 and Figure S10). Due to benefits regarding its lifetime,
electrical conductivity, and surface structure, NTP is coated with carbon (NaTiPO), as
reported by several groups [15,17–20]. Additionally, the anode contains a fifth compound:
activated carbon [10,13]. Previous studies [10,13] have shown that activated carbon with
a high surface area is beneficial for the reaction environment, resulting in enhanced cycle
stability. The resulting electrode formulations in our work are summarized in Table 1.
The working principle of the active materials in the used electrodes is described
as an intercalation process [10,21,22]. There are different studies in which a manganese
oxide structure has worked as a host for sodium ions [5,9,12,15,17,18,20,21,23–27] or for
lithium ions [8,12,14,16,19,22,28,29], depending on the selected electrolyte and thus the
available ion. Whitacre et al. [10] assumed that the cathode is interacting with both ions
during operation. They mentioned that lithium containing manganese spinel changes to a
λ-manganese phase with empty space for sodium ions. Nevertheless, lithium intercalation
is likely to be preferred due to its smaller ionic radius [10,25].
Energies 2023, 16, 4485 3 of 18

Table 1. Electrode compositions of the commercially available “Saltwater Battery” according to

weight percentage [10].

Component Function Weight Percentage in Anode/% Weight Percentage in Cathode/%

NaTiPO Anode intercalation material 70 0
LMO Cathode intercalation material 0 80
Carbon black Electrical conduction 7 2
Graphite Electrical conduction 3 10
Polytetrafluoroethylene Binder material 5 8
Activated Carbon Anode additive for cycle stability 15 0

Similarly, the charging mechanism of the NaTiPO anode is based on intercalation.

Comparably to the cathode material, some groups have described lithium titanium phos-
phate in a lithium environment intercalation process (Equation (1)) [10,12,15,20,28], whereas
others have described the intercalation process based on a sodium environment for the
isostructural sodium titanium phosphate (Equation (2)) [10,12,15,18,20,24].

LiTi2 (PO4 )3 → Li1+x Ti2 (PO4 )3 (0 ≤ x ≤ 2) (1)

NaTi2 (PO4 )3 → Na1+x Ti2 (PO4 )3 (0 ≤ x ≤ 2) (2)

Sun et al. [19] presented the benefits of using sodium titanium phosphate in a lithium
environment showing lower material costs and more space for lithium ions. In this environ-
ment, the material can host both ions during the two-step intercalation process. Wang et al. [8]
described the chemical process as the mixed uptake of sodium and lithium (Equation (3)):

NaTi2 (PO4 )3 → Na1−x Liy Ti2 (PO4 )3 (0 ≤ x ≤ 1; 0 ≤ y ≤ 3) (3)

Chen et al. [30] showed that total or partial ion exchange happens when NaTiPO or
LiTiPO is cycled in electrolytes containing the other species.
To summarize, several groups have described the ability of LMO as well as NaTiPO to
work as a host for Li ions as well as Na ions, depending on the used electrolyte. However,
the preferred active ion for a battery system using Li-Na dual-ion electrolyte, such as the
analyzed system, remains unclear. Therefore, this study aimed to analyze this commercially
available “Saltwater Battery” working principle based on the given electrode and electrolyte
formulation to identify the active ions for each electrode. The results should be used in
future work on target-oriented development to increase the capacity, e.g., via material
selection or variation of the electrode formulation. Herein, we focused on single electrodes
to investigate the active species by cyclic voltammetry (CV) as well as X-ray diffraction
(XRD). Additionally, we performed full cell measurements with electrolytes containing
only one ion type to analyze the performance by exclusion. Finally, an ion analysis of
the electrolyte regarding sodium and lithium ions at different states of charge (SoC) was
performed to give further insight into the behavior of the active ions.

2. Experimental
In the following, the necessary electrode as well as the electrolyte preparation are
described. Afterwards, the experimental setup and execution of the above-mentioned
measurements are presented.

2.1. Electrode Preparation

For CV, a few mg (<50 mg) of material was used to prepare electrode samples. Material
details are summarized in Table 2. The electrode mixture was prepared solvent-free with
a mortar and pestle for both the anode and cathode. By mixing for 10 min, an optically
homogenous electrode mixture was produced.
Energies 2023, 16, 4485 4 of 18

Table 2. Materials used for electrode preparation according to the weight percentages stated in Table 1.

Type Electrode Product Name Supplier

LMO Cathode HLM-Y01 Eachem, Hunan, China
NaTiPO Anode Customized synthesis
according to [10]
Activated carbon Anode PAK C-1000C CarboTech, Essen, Germany
Carbon black Anode and cathode Super P Imerys, Willebroek, Belgium
Graphite Anode and cathode KS6 Imerys, Bodio, Switzerland
Polytetrafluoroethylene Anode Algoflon L203 Solvay, Bollate, Italy
Polytetrafluoroethylene Cathode Dyneon TF 2021Z 3M, Burgkirchen, Germany

The mixtures were pressed with a manual hydraulic garage press into a stainless steel
1.4404 wire cloth with a 60 × 60 mesh type and a thickness of 50 µm. The electrode was
0.5 × 0.5 cm2 in size and was pressed with 8 metric tons into the mesh. The final electrode
weight was 25 mg.
For full cells, from which the samples for XRD were also extracted, the same materials
with the same composition were used for electrodes. Here, we used two-stage mixing with
a highly intensive mixer (El1, Eirich, Hardheim, Germany). At first, active materials and
conductive additives were mixed for 7.5 min. Secondly, a binder was added, and mixing
continued for another 8 min. The procedure remained the same for the anode and cathode.
Afterwards, the electrode powder mixture was laminated onto the current collector, which
was a primer-coated stainless steel 1.4301 sheet (50 µm thick, MK Metallfolien, Hagen,
Germany). Lamination was performed with a roller press laminator (KFK-L 600, Maschi-
nenfabrik Herbert Meyer, Rötz, Germany) under conditions of 190 ◦ C and 40 kN/cm2 . The
final electrodes had a size of 4.5 × 5.5 cm2 with thicknesses of 1.3 mm and 1.4 mm for the
cathode and anode, respectively. The mass ratio of the cathode to the anode was 1.4:1.0,
resulting in a capacity balance of 1.2:1.0.

2.2. Electrolyte Preparation

The electrolyte required for the measurements was mixed with a magnetic stirrer and
agitator. The solvent was deionized water, and the salts were added during mixing. Stirring
for 15 min dissolved the salt completely under ambient conditions. Depending on the
measurement, 10 wt.% sodium sulfate (Purity ≥ 99.9%, CALC, Brenntag, Wien, Austria),
10 wt.% lithium sulfate (Purity ≥ 99.0%, Anhydrous Technical, Interchim, Wörgl, Austria)
or 10 wt.% of both salts (resulting in 20 wt.% total salt) were added.

2.3. Cyclic Voltammetry

The CV measurements were performed with a potentiostat (VMP3, Biologic, Seyssinet-
Pariset, France). Therefore, the above-described electrodes were inserted into a glass cell
with a 1 × 1 cm2 platinum counter electrode. As a reference, electrode mercury/mercury
sulfate was used. Both the anode and cathode were measured with the prepared sodium
electrolyte (ENa) and lithium electrolyte (ELi) separately. The scan rate was set to 0.1 mV/s.
For cathodes, the voltage sweep was set between 0 V and 0.9 V against Hg/HgSO4 . The
anode voltage sweep was between −0.6 V and −1.65 V.

2.4. Full Cell Measurement

Full cell cycling was executed with a battery tester (BCS 815, Biologic) in a voltage
range of 1.1 to 1.81 V. We used two formation cycles at a rate of C/18, followed by cycling
at a rate of C/10. Finally, 5 cycles were conducted at a rate of C/6. The C-rate was based on
the theoretical capacity of the limiting anode with 90 mAh/g. The electrodes were prepared
as described above. Full cells were assembled by stacking the anode, separator (FS2226,
Energies 2023, 16, 4485 5 of 18

Freudenberg, Weinheim, Germany) and cathode to form a single-layer cell, which was
then placed between two stainless steel metal plates with thicknesses of 0.5 cm. Insulation
against these plates was ensured with two additional layers of separator. A force of 340 N
was applied to the cell. Measurements were performed with all three electrolyte types:
solely sodium sulfate, solely lithium sulfate and both salts (EStd). The cells were opened
after electrochemical testing (results see Section 3.2) and used for ion analysis (results
Section 3.3) as well as for XRD (results Section 3.4).

2.5. Electrolyte Ion Analysis

Three electrolyte samples were investigated with ion exchange chromatography
(Sykam S155 ion chromatograph with conductivity detector) by the company “Clean
Controlling, Emmingen-Liptingen, Germany”, which is accredited by DAkkS in accordance
with [31]. The samples were the following: (1) electrolyte containing both salts taken
directly after mixing and before filling, (2) electrolyte taken from a full cell after formation
at a state of charge of 0%, and (3) electrolyte taken from a full cell after formation at a 100%
state of charge. In this work, 0% SoC corresponds to the battery at a discharged state of
1.1 V (@C/10), whereas 100% SoC corresponds to the charged state of 1.81 V (@C/10).

2.6. X-ray Diffraction (XRD)

XRD measurements were performed on an Empyrean multipurpose diffractometer
from Malvern Panalytical. For the experiments, we extracted the electrode coatings from
full cells containing Li2 SO4 and Na2 SO4 in electrolyte after the given cycling procedure.
Therefore, anode and cathode coatings were mechanically scratched from the current
collector and used as samples without further cleaning or drying to retain the electrode
state. Four anodes as well as four cathodes were analyzed: two fresh anode and cathode
electrodes after two formation cycles at 0% and 100% SoC as well as two aged anode
and cathode electrodes after two formation cycles and an additional 100 cycles, also at
these SoCs.
Weighed amounts of extracted anode and cathode powder materials were placed
between two Kapton foils and measured in transmission geometry with Mo-Ka radiation.
The sample details as well as the detailed analysis procedure (e.g., used software, analysis
method, initial settings, etc.) are presented in Appendix A.

3. Results and Discussion

In the following, the results of single electrode measurements in CV as well as com-
bined electrodes in full cells for different electrolyte variations are presented and discussed.
Additionally, the results of the ion analysis as well as XRD are shown and discussed for
different SoCs.

3.1. Cyclic Voltammetry

As a first step, we executed CV measurements to identify whether Na+ and/or Li+
can act as active ions for single electrodes. Therefore, cathodes and anodes were analyzed
separately with electrolyte containing solely lithium sulfate or sodium sulfate. For the
cathode, two peaks and a reversible reaction were expected to be seen for lithium electrolyte,
comparable to measurements in [19,28]. According to [10,22], the cathode should also work
with sodium electrolyte and was thus expected to show a similar curve shape.
Figure 1 shows the cyclic voltammograms of the cathode using electrolytes with
dissolved lithium sulfate (ELi) and sodium sulfate (ENa). By employing lithium sulfate
as the conductive salt, two oxidation and reduction peaks were observable, expressing a
reversible two-stage reaction.
ER REVIEW 7 of 18
Energies 2023, 16, 4485 6 of 18

Figure 1. Exemplary cyclic voltammograms (10 cycles) of the spinel lithium manganese oxide cathode
with aqueous lithium sulfate electrolyte (ELi) or sodium sulfate electrolyte (ENa) versus a platinum
counter electrode and Hg/HgSO4 reference at a scan rate of 0.1 mV/s.
Figure 2. Cyclic voltammograms (10 cycles) of the sodium titanium phosphate anode with aqueous
For the
lithium sulfate electrolyte (ELi) or case of using
sodium sodium
sulfate sulfate as the
electrolyte conductive
(ENa) salt the
against (ENa), no intensecounter
platinum peak was
identified, even over 10 cycles. Only a weak and broad reduction peak was observed, yet
electrode and Hg/HgSO4 reference at a scan rate of 0.1 mV/s.
an oxidation peak was missing completely.
The results with ELi were expected, due to the similar shape compared to results in the
literature [19,28]. The reversible reaction indicates that lithium ions can be the active species
200 for the cathode electrode. On the contrary, the results with ENa indicate an irreversible
Relative specific capacity

reaction. As described in [10], a slow exchange of lithium with sodium in the cathode
normalised to EStd in %

material was expected in the case of lithium absence in the electrolyte. This indicates a
similar curve shape to that shown in Figure 1 with ELi with small shifts in the peak intensity
150 and voltage level of the peaks. Even when cycling for 10 cycles, no increase or change in
shape occurred. More than that, a slight decrease in the little existing peak was observed
after cycling. The cathode seemed to be electrochemically inactive under these conditions.
The amount of lithium in the pristine LMO was too low to undergo a measurable reversible
100 reaction in the measurement set up. This could be due to the small electrode size and thus
the low amount of pristine lithium ions compared to the excess amount of electrolyte. The
CV data led to the assumption that sodium ions are not possible active species for LMO
50 cathodes under dual-ion battery conditions. Therefore, the given CV results reveal that
lithium ions are active species for cathode electrodes.
For the anode, one reversible peak was expected in combination with electrolyte
containing only sodium sulfate as well as that containing only lithium sulfate, as shown
0 in [10,13,17,18,20,28,30]. Notably, the intercalation process is considered to be a two-phase
reaction for both electrolyte types. The lithium (ELi) and sodium (ENa) systems are shown
in Figure 2. @C/10 @C/6

ENa EStd ELi

The measurement with ENa electrolyte showed one oxidation and reduction peak,
indicating a reversible reaction, which is in excellent agreement with the literature
[10,13,15,18,20]. In contrast, the measurement with ELi indicated a more complex
 in [10,13,17,18,20,28,30]. Notably, the intercalation process is considered to be a two-phase
reaction for both electrolyte types. The lithium (ELi) and sodium (ENa) systems are shown
Energies 2023, 16, 4485 7 of 18
in Figure 2.

Figure 2. Cyclic voltammograms (10 cycles) of the sodium titanium phosphate anode with aqueous
lithium sulfate electrolyte (ELi) or sodium sulfate electrolyte (ENa) against the platinum counter
Figure 2. Cyclic voltammograms (10 cycles) of the sodium titanium phosphate anode with aqueous
electrode (ELi)
lithium sulfate electrolyte and Hg/HgSO
or sodium4 reference
sulfate at a scan rate(ENa)
electrolyte of 0.1 mV/s.
against the platinum counter
electrode and Hg/HgSO4 reference at a scan rate of 0.1 mV/s.
The measurement with ENa electrolyte showed one oxidation and reduction peak, indi-
cating a reversible reaction, which is in excellent agreement with the literature [10,13,15,18,20].
The measurement withthe
In contrast, ENa electrolyte with
measurement showed one oxidation
ELi indicated a more and reduction
complex peak,Here, the
behavior.
indicating a reversible reaction, which is in excellent agreement with the literature
signal was smeared out compared with that of the ENa system, and additionally, the shape
[10,13,15,18,20]. changed
In contrast, the first
over the measurement
three cycles. with
From ELi indicated
cycles a more
one to three, complex
the peaks became more
pronounced
behavior. Here, the signal wasandsmeared
shifted to out
higher voltages by
compared with−1.1 V vs.
that of Hg/HgSO 4 . After the
the ENa system, andthird cycle,
additionally, the shape changed over the first three cycles. From cycles one to three, Conclusively,
the system stabilized, and further sweeps showed almost identical behavior. the
peaks became more in this case, we also
pronounced observed
and shifteda reversible
to higher reaction.
voltages by −1.1 V vs. Hg/HgSO4.
The changes in shape and peak position led us to assume that there is a side reaction in
After the third cycle, the system stabilized, and further sweeps showed almost identical
the first two cycles, such as the formation of another species, before a stable reaction takes
behavior. Conclusively, in this case, we also observed a reversible reaction.
place. Nevertheless, the anode showed reversible reactions with both ions in accordance
The changeswith
in shape and peak
the literature andposition
in contrastled
to us
theto assume that
investigated LMO.there is aboth
Thus, sideions
reaction
can be the dom-
in the first two cycles, such as the formation of another species, before a stable reaction
inant active species for anode electrodes in dual-ion batteries. However, after stabilization
takes place. Nevertheless,
with ELi, thethepeakanode
positionshowed
differed reversible
by ≈80 mV to reactions with both
higher voltages (meaningionsless
in negative)
accordance with compared to theand
the literature ENa insystem.
contrastConsequently, lithium LMO.
to the investigated intercalation
Thus, isboth
slightly
ionspreferred
in cases
can be the dominant where
active both ions
species for are available,
anode indicating
electrodes that a lithium-dominated
in dual-ion batteries. However, reaction also
occurs for anode electrodes.
after stabilization with ELi, the peak position differed by ≈80 mV to higher voltages
(meaning less negative)
3.2. Full compared to the ENa system. Consequently, lithium intercalation
Cell Measurement
is slightly preferred in cases where both ions are available, indicating that a lithium-
To further validate the results found by CV, we performed full cell measurements with
dominated reaction also occurs
electrolytes for anode
containing electrodes.
different conductive salts. From previous findings, we expected
that the full cell setup with electrolyte containing only sodium sulfate (ENa) would not
3.2. Full Cell Measurement
really work due to the cathode behavior, whereas the lithium-based electrolyte (ELi) should
perform at
To further validate least
the similarly
results foundto standard
by CV, electrolyte (EStd) full
we performed containing both ions, as used in the
cell measurements
“Saltwater Battery”. The results are shown in Figure 3.
with electrolytes containing different conductive salts. From previous findings, we
expected that the full cell setup with electrolyte containing only sodium sulfate (ENa)
would not really work due to the cathode behavior, whereas the lithium-based electrolyte
(ELi) should perform at least similarly to standard electrolyte (EStd) containing both ions,
as used in the “Saltwater Battery”. The results are shown in Figure 3.
Figure 2. Cyclic voltammograms (10 cycles) of the sodium titanium phosphate anode with aqueous
lithium sulfate electrolyte (ELi) or sodium sulfate electrolyte (ENa) against the platinum counter
Energies 2023, 16, 4485 8 of 18
electrode and Hg/HgSO4 reference at a scan rate of 0.1 mV/s.

200
Relative specific capacity
normalised to EStd in %

150

100

50

0
@C/10 @C/6

ENa EStd ELi

Figure 3. Capacity of the full cell measurement at C/10 (5th overall cycle, 3rd with C/10) and C/6
(23rd overall cycle, 1st with C/6) with solely lithium sulfate (ELi) with respect to sodium sulfate
The measurement (ENa)
withelectrolyte
ENa electrolyte showed one oxidation and reduction peak,
normalized to the capacity of standard full cell and electrolyte containing both
indicating a reversibleions
reaction,
(EStd). which is in excellent agreement with the literature
[10,13,15,18,20]. In contrast, the measurement with ELi indicated a more complex
At a rate of C/10, the ENa system reached 60% of the capacity compared to standard
behavior. Here, the signal was containing
electrolyte smeared out
both compared
ions, while the ELi with
systemthat of120%.
reached the At
ENa C/6,system,
the trend andand
discrepancy between the systems were even more prominent:
additionally, the shape changed over the first three cycles. From cycles one to three, the the ENa system was nearly
inert, reaching only 0.1% of the original capacity, whereas the ELi system performed 90%
peaks became more pronounced
better than theand shifted to higher voltages by −1.1 V vs. Hg/HgSO4.
standard.
After the third cycle, the system
The sodium stabilized,
system @C/10 and further
showed sweeps
a relatively high showed almost
capacity of 60%, whichidentical
was not
expected due to the CV results of the cathode with ENa showing no significant peak. On
behavior. Conclusively,theincontrary,
this case, we alsoat observed
the behavior a reversible
@C/6 was expected. reaction.
The anode works well with sodium
The changes in shape and to
according peak
CV, soposition
the cathodesled us to assume
presumably cause thethat
poorthere is a side
performance of thereaction
full cell
with ENa. The fact that the system performed quite well at C/10 leads to two possible
in the first two cycles, such as the formation of another species, before a stable reaction
explanations that are both based on the small current. On one hand, sodium ions with
takes place. Nevertheless, the ionic
their larger anode radiusshowed
(116 pm forreversible
Na+ vs. 90 pmreactions withintercalate
for Li+ [32]) might both ions in the in
manganeseand
accordance with the literature oxidein spinel when theto
contrast discharge period is long enough
the investigated LMO.(10Thus,h at C-rate
both C/10).
ions
The intercalation of sodium should work at the same C-rate according to [10]. A possible
reason why our CV results showed no interaction with sodium could be the shorter time
used for one complete charge–discharge cycle during the voltage sweep due to the higher
specific current (2.5 h compared to 20 h, respectively 160 mA/g vs. 6 mA/g). On the
other hand, the lithium which is extracted from the cathode during charging may again
reintercalate during discharge. This process might also be invisible in CV due to the lower
specific current and/or the lower amount of electrolyte per electrode mass in the full cell
(4000 mL/g vs. 75 mL/g). This results in both a longer charge–discharge cycle and shorter
diffusion pathways in the full cell. Considering the first case, the results indicate that 60%
of the original lithium content in LiMn2 O4 can be replaced with sodium during a 10 h
discharge period. If our second interpretation holds true, it shows that 60% of the lithium
ions can be reintercalated within this 10 h window. Since there is no evidence for significant
Energies 2023, 16, 4485 9 of 18

Na+ intercalation into the LMO structure in our XRD results (see below), we conclude that
the second case is more plausible, yet further research is required to validate this finding.
The capacity reached with ELi compared to EStd shows the ability of both electrodes
to work with lithium ions. The significantly higher capacity compared to the standard
system substantiates the idea that lithium is the preferred species in both electrodes. This is
in accordance with the shift in our CV measurements of the anode. Finally, the superior
performance of ELi at a rate of C/6 compared to C/10 could have been caused by the
sodium ions in the electrolyte hindering and/or slowing the electrode reaction.
The analyzed system works best with electrolyte containing only lithium sulfate as
the conductive salt. The addition of sodium sulfate to electrolyte deteriorates the capacity
retention of a working battery.

3.3. Ion Analysis of Electrolyte

To further investigate the role of sodium in our full cells, we performed an ion analysis
of electrolyte. Here, two different aspects were of major importance. First, if the working
principle described in the literature [5,8–10,12,14–29] is valid in this system, the charged
battery (SoC 100%) should be composed of a sodium-saturated anode and an Li-depleted
cathode. Thus, the Li+ content in electrolyte should be increased compared to fresh elec-
trolyte, whereas the opposite should be the case for the Na+ content. In the discharged
battery (SoC 0%), however, it would be the opposite: a higher sodium and a lower lithium
content in electrolyte. If both electrodes only react with lithium, as concluded from the CV
and full cell results, the lithium content in electrolyte should remain unchanged. The same
holds true for the sodium content. Second, by comparing the ion content of electrolyte
after the activation cycles with fresh electrolyte, we drew conclusions about the formation
reactions. If there is no formation of an interphase, such as the solid electrolyte interphase
(SEI) in standard LIB or other side reactions, the ion amount should remain similar before
and after the activation cycles. If a formation reaction takes place, a change in the electrolyte
concentration should be detectable. The results are shown in Table 3.

Table 3. Results of the electrolyte ion analysis regarding sodium and lithium contents with ion
exchange chromatography referring to the full cells described in Section 3.2.

Fresh Electrolyte before Filling Sample 0% SoC Sample 100% SoC

Lithium content in g/L 11.1 13.1 14.7
Sodium content in g/L 58.4 69.2 70.4

The contents of both ions were 18% higher after activation compared to those in fresh
electrolyte, indicating a strong link between the ion reactions. Furthermore, this change
clearly indicates a formation reaction and can be associated with the change in the anode
CV curve in a lithium-containing electrolyte during the first two cycles (Figure 2). The
activation step for the full cell, and thus the basis for the ion analysis, also comprises
two full cycles. The irreversible capacity loss measured in the previously mentioned full
cell measurements over activation cycles was approximately 20%, which is close to the
changes in electrolyte concentrations. This indicates a strong link between the full cell
activation losses in the first two cycles and the increase in the ion content in electrolyte,
along with a decrease in the ion content in the electrodes. All three measurements led to
the conclusion that formation is happening during cell activation. A formation mechanism
for sodium titanium phosphate anodes in aqueous solution is presented in [26]. The
authors identified the formation of titanium phosphates as well as titanium sulfates during
activation. Additionally, they measured an increase in the sodium content in electrolyte due
to this reaction, which is in accordance with our data. This hints also to a partial dissolution
of the initial phase NaTi2 (PO4 )3 .
Energies 2023, 16, 4485 10 of 18

Furthermore, the ion analysis revealed an increase in lithium ions in electrolyte after
charging. This goes along with the literature [8,12,19,22,25,28], showing the expected dein-
tercalation of lithium ions during charging and the associated release of lithium ions into
the electrolyte. Interestingly, the sodium content also increased slightly during charging.
This might be another hint of partial dissolution of the initial NaTi2 (PO4 )3 phase, which
is the only viable source of additional sodium in the system. According to the working
principle described in the literature [12,18,20,24], the sodium content in electrolyte should
be reduced in the charged battery state. This trend is decoupled from the formation process,
since it was virtually finished after two cycles according to the CV results (Figure 2) and,
consequently, did not influence the ion content during cycling. The system seems to be quite
stable, as the change in sodium content of 2% is small compared to the change in the lithium
concentration of 12%. Thus, these findings cannot be described with a solely lithium-based
(de)intercalation working principle on the anode and cathode sides, as expected from the
CV and full cell results, or with a sodium-dominated (de)intercalation process on the anode
side, as described in the literature [10,12,18]. We conclude that sodium is either inactive,
as indicated by the full cell measurements and CV, or undergoes a reversible reaction
during charging and discharging without affecting the actual electrolyte concentration.
This contradicts the sodium-dominated working principle on the anode side assumed in
the literature [10,12,18,20,24] and lets us assume that another reaction has to take place.
However, during cycling, the sodium as well as the lithium content are equally increased
on a molar basis compared to fresh electrolyte, indicating a strong link between sodium
and lithium reactions.

3.4. X-ray Diffraction

So far, our present results partially contradict the reported working principle of the
“Saltwater Battery” in its hybrid form, in which Li+ is (de)intercalated on the cathode
side and Na+ is (de)intercalated on the anode side. Consequently, we performed an XRD
analysis to clarify which crystalline phases are present for different SoCs as well as different
cycle numbers. For normal intercalation processes, we expected that the utilized spinel-
type LMO would show a cubic phase with varying lattice parameters (e.g., [33,34]). Since
lithium is a very light element with a low electron density, its (de)intercalation will not
lead to significant intensity changes in the reflections. However, shifts of reflections by 2θ
resulting from changes in lattice parameters should occur. Electrochemical delithiation of
Lix Mn2 O4 is known to involve only one single spinel phase and not a coexistence of two
phases [35]. Alternatively, for the intercalation of sodium, we would expect significant
intensity changes, while new phases might also form, as indicated by unassigned peaks in
the diffraction pattern [36].
The intercalation process on the anode side was reported to be a two-phase process,
where no transition compound Na1+x Ti2 (PO4 )3 was detected. Hence, during intercalation, a
new phase Na3 Ti2 (PO4 )3 is formed, which would result in the appearance of new reflections
in XRD and a coexistence of the two phases [37,38]. An intercalation of solely lithium or
the formation of another phase would lead to peak shifts as well as additional or missing
peaks. Another alternative reaction might be the partial intercalation of lithium and the
formation of a mixed structure (Na1−x Lix )3 Ti2 (PO4 )3 . Such mixed compounds would also
result in changed lattice parameters and additional or markedly shifted reflections.
The XRD measurement data are shown in Figure 4, whereas the detailed refinement
results, including all refined parameters, phases as well as the instrument parameters can
be found in the Supplementary Materials.
 Energies 2023, 16, 4485 11 of 18
Energies 2023, 16, x FOR PEER REVIEW 11 of 18

Figure 4. XRD patterns of cathode (a) as well as anode (b) samples for fresh cells directly after
Figure 4. XRD patterns of cathode (a) as well as anode (b) samples for fresh cells directly after
formation and cycled ones after 100 cycles. (c) shows the experimental lattice parameter as a function
formation and cycled ones after 100 cycles. (c) shows the experimental lattice parameter as a function
of x in LixMn2O4. Thereby, x can be extracted from the charge flow measured by the potentiostat,
of x in Lixthe
assuming Mn(de)intercalation
2 O4 . Thereby, x can
of 1 be
Li+extracted fromrefinement.
/e− or Rietveld the charge flowHere,measured
our data areby the potentiostat,
compared with
assuming the[35,39]
(de)intercalation + −
the literature where onlyofLi1 Li
+ was/e or Rietveld
intercalated. (d)refinement.
Comparison Here,
of xour
in Lidata
x(Na are compared
x)Mn 2O4 withwiththe
the literature
charge [35,39] where
flow extracted with aonly Li+ was intercalated.
potentiostat and Rietveld (d) Comparison
refinement. The of x inline
black Lix (Na
withx )Mn 2 O4 with
a slope of 1
shows the idealized
the charge case with
flow extracted in which 1 ion is (de)intercalated
a potentiostat per measured
and Rietveld refinement. e−, while
The black in reality,
line with a slope
additional
of 1 showselectrons are needed
the idealized case intowhich
account for is
1 ion losses, cell resistance,
(de)intercalated per side e− , etc.
reactions,
measured whileTheindotted
reality,
lines are fitted
additional to the data.
electrons Fit equations:
are needed (bluefor
to account dotted line)
losses, xMn2O4 -> f(x) = 0.8487x + 0.1999; (orange
cellLiresistance, side reactions, etc. The dotted
dotted line)
lines are NaxMn
fitted 2O4 -> f(x) = 0.1815x + 0.0429). (e) Evolution of the NaTi2(PO4)3 amount in the
to the data. Fit equations: (blue dotted line) Lix Mn2 O4 -> f(x) = 0.8487x + 0.1999;
measured anode samples. Here, the weight fraction of NaTi2(PO4)3 is divided by the weight fractions
(orange dotted line) Nax Mn2 O4 -> f(x) = 0.1815x + 0.0429). (e) Evolution of the NaTi2 (PO4 )3 amount
of TiO2 and graphite, respectively. The result is then plotted relative to the “fresh Soc 0%” sample.
in the measured anode samples. Here, the weight fraction of NaTi2 (PO4 )3 is divided by the weight
fractions of TiO2 and graphite, respectively. The result is then plotted relative to the “fresh Soc
XRD measurements of the cathode electrode showed expected reflections of the cubic
0%” sample.
spinel-type LixMn2O4 phase for both the discharged and charged states. The reflections
shifted
XRDdue to extension/contraction
measurements of the cathodeofelectrode
the lattice during
showed the insertion/extraction
expected of
reflections of the cubic
lithium ions. Furthermore, we identified reflections from hexagonal graphite, which
spinel-type Lix Mn2 O4 phase for both the discharged and charged states. The reflections is
part of the cathode mixture. Finally, we found three Li and/or Na containing sulfates
shifted due to extension/contraction of the lattice during the insertion/extraction of lithium
(hydrated and nonhydrated)
ions. Furthermore, in tiny
we identified amounts.
reflections fromWe assume that
hexagonal thesewhich
graphite, salts precipitated
is part of the
from electrolyte that was not removed from the sample material before measurement.
cathode mixture. Finally, we found three Li and/or Na containing sulfates (hydrated and In
summary, there were no prominent new phases that played major roles during cycling.
Energies 2023, 16, 4485 12 of 18

nonhydrated) in tiny amounts. We assume that these salts precipitated from electrolyte that
was not removed from the sample material before measurement. In summary, there were
no prominent new phases that played major roles during cycling. The refined progress of
the lattice parameters (see Figure 4c) is in perfect accordance with the literature, in which
only Li+ was used for (de)intercalation [33]. Since the bigger Na+ ion requires more space
in the structure, increased lattice expansion would be expected. Equation (4) gives the
calculation for the number of electrons with respect to the x-value of lithium in the formula
Lix Mn2 O4 based on our measured charge flow. Furthermore, the Li occupancy obtained
from Rietveld refinement of the structure fits well with electrochemical measurements
assuming the (de)intercalation of 1 ion per electron (Figure 4d). From XRD, we obtained a
transfer of ~0.85 Li+ /e− ; the deviation from unity could have been caused by uncertainties,
since only four data points were used to determine our linear fit. The underestimation of
lithium transfer can be explained either by electron losses due to cell resistance and side
reactions or by partial Na intercalation (<5%). On the other hand, when assuming pure
Nax Mn2 O4 chemistry in the Rietveld refinement, the Na occupancy would explain the
measured charge flow, only if merely 0.18 ions were (de)intercalated per electron, which is
far from an expected behavior of such materials.

Qcharge (i ) − Qdischarge (i )
x LMO = 1 − ∑ (4)
i
e·ntot ( Li )

Here, XLMO is the x value in Lix Mn2 O4 , Qcharge (i) and Qdischarge (i) are the measured
capacities of the ith charge and discharge cycle (unit Coulomb), e is the elementary charge
(unit Coulomb), and ntot (Li) is the total amount of Li atoms in the pristine LMO material of
the cathode electrode.
The XRD measurements of the anode electrode are summarized in Table 4 and showed
the NASICON-type phase NaTi2 (PO4 )3 . Surprisingly, this phase showed no significant
changes during charging or during cycling. While there were slight variations in lattice
parameters a and c, we are not able to make a clear statement about its origin, yet we
cannot rule out that it might originate from slight Li uptake by the material. However, as
the literature indicates, reduced NASICONs A3 Ti2 (PO4 )3 (A = Li/Na) tends to be sensitive
to air or unstable vs. dissolved oxygen in aqueous solutions [40–42] undergoing a self-
discharge reaction. This self-discharge could occur during sample contact with air during
transportation or during the long lasting XRD measurements, which might explain why
we did not see much change between the SoC 0% and SoC 100% anode samples in XRD.
Consequently, we could not identify the active intercalation compound on the anode side.
Nevertheless, we observed graphite as well as some sulfates in tiny amounts in the anode
samples. Furthermore, we identified reflections of the anatase (TiO2 ) phase that originated
from NASICON synthesis, as similarly reported in [10]. To our astonishment, the only
indication of a chemical reaction was found in the charged state (SoC 100%) in form of a
Li3 PO4 phase, in both fresh cells as well as in cells that were cycled 100 times. This phase
was missing in the discharged state and, therefore, seems to be formed via charging and
remains stable in an oxygen atmosphere. The Li3 PO4 phase could also be a decomposition
product of mixed sodium–lithium–titanium–phosphate, which would decompose due to
self-discharge on oxygen contact into Li3 PO4 . However, the resulting compositions of the
anode in weight percents are summarized in Table 5. Since the only compound in the
cell that contains phosphate is the anode active material, the precipitated Li3 PO4 has to
originate from here. A partial dissolution of NaTiPO is also described in [8,19], resulting in
findings of Li3 PO4 [8]. The Li3 PO4 has non-negligible solubility in aqueous media, such
as the used electrolyte in [43] with 450 mg/L; consequently, the XRD observed quantity
of crystalline Li3 PO4 was the amount above its saturation point. In addition, by using
inactive compounds as the internal reference in our XRD measurements, we were able to
derive the degree of dissolution of NaTiPO (Figure 4e). This result has to be interpreted
with care, since graphite often shows a preferred orientation and, thus, is not ideal as an
internal reference. On the other hand, for TiO2 , we had to presume that it was inactive
Energies 2023, 16, 4485 13 of 18

during electrochemical cycling, which is plausible, but we cannot completely rule out any
electrochemical interference just by the experiments performed. However, both lines (in
Figure 4e) show the same trend, and thus, we deduce that a measurable dissolution of
NaTiPO took place.

Table 4. Results of the cathode XRD measurement with refined sample compositions in
weight percentages for the four different test samples. The numbers in brackets show the ±1σ
standard deviation.

Sample Lix Mn2 O4 in wt.% Graphite in wt.% LiNaSO4 in wt.% Li2 SO4 (H2 O) in wt.% Na3 Li(SO4 )2 (H2 O)6 in wt.%
Fresh SoC 0% 86.9(2) 7.8(2) 1.95(9) 4.5(8) 2.9(2)
Fresh SoC 100% 87.4(2) 8.6(2) 1.3(1) 2.7(2) 0.000
Cycled SoC 0% 86.8(2) 9.5(2) 3.2(1) 0.000 0.5(1)
Cycled SoC 100% 87.4(2) 8.4(2) 1.4(2) 2.8(2) 0.000

Table 5. Results of the anode XRD measurement with the refined sample compositions in weight per-
centages for the four different test samples. The numbers in brackets show the ±1σ standard deviation.

NaTi2 (PO4 )3 in Na3 Li(SO4 )2 (H2 O)6 Na2 SO4

Sample Graphite in wt.% Li3 PO4 in wt.% TiO2 in wt.%
wt.% in wt.% in wt.%
Fresh SoC 0% 86.4(2) 6.5(2) 4.4(2) 0.000 2.7(1) 0.000
Fresh SoC 100% 76.5(3) 9.5(2) 0.000 11.0(3) 3.0(1) 0.000
Cycled SoC 0% 86.8(2) 9.7(2) 0.000 0.000 3.5(1) 0.000
Cycled SoC 100% 73.5(2) 9.2(2) 0.000 6.8(2) 3.1(1) 7.4(1)

In summary, our findings lead us to conclude that the lithium intercalation reaction
in LMO takes place on the cathode side, while we identified the dissolution of NaTiPO
during operation. Due to the self-discharge mechanism in the presence of oxygen, similar
reflections occur in the charged and discharged states, hindering conclusions about the
active ion intercalation mechanism for anodes. However, our results show that a partial
dissolution of NaTi2 (PO4 )3 is a significant part of the cycling process. The dissolved
(PO4 )3− ions are the only source to form the precipitated Li3 PO4 , while the sodium and
titanium ions remaining in the solution during charging are hidden from XRD in an
amorphous state. According to [26], dissolved titanium ions can undergo a reaction,
changing the oxidation state reversibly from 4+ to 3+ and, thus, might contribute slightly
to the electron transfer during operation. However, during discharge, solid Li3 PO4 on the
anode side disappears again, indicating the removal of this salt from electrolyte. This could
be due to lithium intercalation taking place on the cathode side. Additionally, a formation
mechanism of Li3 PO4 only based on the oxygen self-discharge could explain its appearance
in the charged state. Nevertheless, the dissolution of NaTiPO in a lithium-containing
environment as well as the assumed reversible formation of Li3 PO4 can explain the peak
shift during formation and the stabilized peaks found afterwards in CV. More than that,
the irreversible capacity loss during two activation cycles is explained by the previously
described formation phenomenon. Furthermore, the increased sodium in electrolyte after
formation can be explained by the NaTiPO dissolution. The dissolution measured by XRD
is ≈12–15 wt.%, which is close to the measured sodium increase of ≈18 wt.% in electrolyte.
Three possibilities remain as the anode reaction mechanism: (1) the intercalation of sodium;
(2) solely the intercalation of lithium; and (3) the intercalation of lithium leading to an ion
mixture in NaTiPO. Even though the specific reacting ion species on the anode side remains
unclear, the third option seems to be the most likely, and XRD measurements still revealed
a formation process accompanied by NaTiPO dissolution.
Energies 2023, 16, 4485 14 of 18

4. Conclusions
This research focused on identifying the active species on the anode as well as the
cathode side in a commercially available dual-ion battery system containing lithium and
sodium on an aqueous basis. According to the results from previous literature, manganese
oxide structures, as the active cathode material, as well as NASICON-type sodium titanium
phosphate, as the active anode material, can intercalate both types of ions, depending on
the ion availability in electrolyte [5,8–10,12,14–29].
The findings of this work support the assumption that the anode material can interact
with lithium and sodium in a reversible reaction. In contrast to this, the CV measurements
indicate that the activity of the spinel (lithium manganese oxide) is limited to lithium
and does not intercalate sodium in a significant amount. The phase analysis of XRD
patterns supports this conclusion and substantiates the described intercalation process on
the cathode side. The active species on the cathode side was unequivocally identified as
lithium. More than that, sodium seems to hinder the electrochemical process, as found in
electrochemical full cell measurements.
The results regarding the anode contradicted expectations. A formation process based
on the partial dissolution of sodium titanium phosphate during the battery activation cycles
was revealed, which seems to lead to a reversible formation of Li3 PO4 during charging.
However, intercalation was not detectable in XRD on the anode side for Li+ or Na+ due
to the dominant self-discharge reaction in the presence of oxygen during transportation.
Although the XRD measurement did not help to identify the active ion on the anode side,
the CV as well as the full cell results led to the conclusion that lithium is the dominant ion
in the hybrid system, supporting the reaction described by Wang et al. [8]. To conclude, for
the analyzed Li-Na dual-ion system, Li ions were identified as the preferred active ion for
both electrodes.
Thus, these results indicate that the dual-ion “Saltwater Battery”, as available on the
market, is only a single-ion aqueous lithium system or at least a battery type dominated
by lithium. Therefore, the system and materials can be improved with a new focus on the
lithium reaction. The sodium sulfate dissolved in electrolyte can be eliminated, saving a
minor part of the electrolyte costs. However, more than that, it leads to superior perfor-
mance of the battery, as shown in the full cell measurements. This behavior needs to be
further investigated in full cell measurements as well as with a prompt gamma activation
analysis and will open new pathways for improving the battery’s performance, such as the
development of new electrode formulations regarding the composition and used materials
for beneficial lithium (de)intercalation characteristics.

Supplementary Materials: The following supporting information can be downloaded at: https:
//www.mdpi.com/article/10.3390/en16114485/s1, Figure S1: Rietveld fits of the fresh cathode at 0%
SoC. Figure S2: Rietveld fits of the fresh anode at 0% SoC. Figure S3: Rietveld fits of the fresh cathode
at 100% SoC. Figure S4: Rietveld fits of the fresh anode at 100% SoC. Figure S5: Rietveld fits of the
cycled cathode at 0% SoC. Figure S6: Rietveld fits of the cycled anode at 0% SoC. Figure S7: Rietveld
fits of the cycled cathode at 100% SoC. Figure S8: Rietveld fits of the cycled anode at 100% SoC.
Figure S9: SEM picture of sodium titanium phosphate. Figure S10: SEM picture of lithium manganese
spinel. Table S1: Listed results of the refined parameters extracted from Rietveld refinement of the
anode samples. Table S2: Parameters of the utilized instrument resolution function for Rietveld
refinement. Table S3: Listed results of the refined parameters extracted from Rietveld refinement of
the cathode samples.
Author Contributions: Conceptualization, J.S., K.-H.P., S.S. and M.A.D.; methodology, J.S., K.-H.P.,
L.G., S.S., R.G. and M.A.D.; validation, J.S., K.-H.P., L.G., R.G. and S.S.; formal analysis, J.S., L.G.,
R.G. and S.S.; investigation, J.S., L.G. and S.S.; resources, J.S. and L.G.; data curation, J.S. and L.G.;
writing—original draft preparation, J.S., L.G. and S.S.; writing—review and editing, J.S., K.-H.P., L.G.,
S.S., R.G. and M.A.D.; visualization, J.S. and L.G.; supervision, J.S., K.-H.P., R.G. and M.A.D.; project
administration, J.S.; funding acquisition, J.S., R.G. and K.-H.P. All authors have read and agreed to
the published version of the manuscript.
Energies 2023, 16, 4485 15 of 18

Funding: This research was funded by BMWK, grant number 03EI3046F and BMBF (Federal Ministry
of Education and Research, Germany) “ExZellTUM III”, project number 03XP0255.
Data Availability Statement: Data is unavailable due to privacy restrictions.
Acknowledgments: We thank the Physics Lab of the Heinz Maier-Leibnitz Zentrum for measuring
time at the diffractometer and Alexander Book for adjustment of the instrument.
Conflicts of Interest: The authors declare no conflict of interest.

Appendix A
For XRD, a flat plate transmission geometry was used with a reflection/transmission
spinner stage. Samples consisting of weighed amounts of anode material (373 mg) and
cathode material (144 mg) were placed between two Kapton foils and measured on a
rotating stage. This sample amount was calculated to result in less than ≈70% transmission
of the primary beam to give a good signal/noise ratio in XRD. The sample was set up with
a transmission geometry. The fixed beam irradiated the plate at a normal incidence. The
sample plate was in the center of the goniometer with a radius of 240 mm. The detector
scanned the goniometer circle at around 2θ. Mo radiation with a focusing mirror was
applied to focus the beam onto the detector and serve as a filter, resulting in nearly pure
Kα1/2 radiation (λ(Kα1 ) = 0.7093187 Å; λ(Kα2 ) = 0.7136090 Å). We used 1/4◦ Ta divergence
and antiscatter slits on the incident beam to achieve a high resolution and low noise. Soller
slits with 0.02 rad placed on the incident and diffracted beam sides further enhanced the
resolution. The Malvern Panalytical GaliPIX3D detector was equipped with a transmission
cap, reducing the background level caused by air scattering. We collected diffraction data
in the range of 2θ = 7◦ –49◦ . The measurement procedure stepped through the range with a
step size of 0.007◦ and step time of 130.26 s, giving a total duration of ≈30 min for a single
scan. Eight scans were successively gathered and summed up for each sample, resulting in
a total measurement time of ≈4 h/sample. To enhance the particle statistics, the sample
stage was rotated at 0.25 Hz.
The diffraction patterns were analyzed with the software package GSAS-II v.5468 [44].
We applied the Rietveld method [45] using a combination of the following phases as initial
settings (cif-files provided either by the Crystallographic Open Database (COD) or the
Inorganic Crystal Structure Database (ICSD)):
• LiMn2 O4 (LMO): COD #1514006
• Graphite: ICSD #76767
• LiNaSO4 : COD #2106021
• Li2 SO4 (H2 O): COD #1008190
• Na3 Li(SO4 )2 (H2 0)6 : COD #2243890
• NaTi2 (PO4 )3 (NaTiPO): ICSD #19995
• Li3 PO4 : ICSD #10257
• TiO2 (Anatase): COD #9008214
• Na2 SO4 : COD #9004092
The instrument resolution function was obtained by Rietveld refinement of powder
data from a NIST-660c LaB6 sample using ~20 mg of LaB6 powder sandwiched between
two Kapton foils measured in the same sample holder and with identical instrumental
settings as described above. For consistency, we refined the same set of parameters for all
anode and cathode samples, respectively. For the anode samples we refined
• Six coefficients of the Chebyshev-1 background polynomial model;
• Two separate background peaks at 2θ = 8.29◦ and 8.17◦ that originated from the binder;
• Lattice parameters and phase fractions of all phases;
• One isotropic microstrain parameter for each phase except for NaTiPO;
• Three anisotropic microstrain parameters for NaTiPO using the generalized model;
• One isotropic atomic displacement factor U for all atoms of the NaTiPO and Li3 PO4
phases, respectively;
Energies 2023, 16, 4485 16 of 18

• All possible atomic positions of the NaTiPO and Li3 PO4 phases.
For the cathode samples we refined
• Six coefficients of the Chebyshev-1 background polynomial model;
• One separate background peak at 2θ = 8.29◦ that originated from the binder;
• Lattice parameters and phase fractions of all phases;
• One isotropic microstrain parameter for each phase except for LMO and Na3 Li(SO4 )2 (H2 O)6 ;
• Two anisotropic microstrain parameters for LMO using the generalized model;
• No broadening model was used for Na3 Li(SO4 )2 (H2 O)6 because its phase fraction was
too low;
• One isotropic atomic displacement factor U for all atoms of the LMO phase;
• The x-position of the oxygen atom in the LMO phase;
• The fraction of Li in the LMO phase was set to a value extracted from the summed
transferred charge measured by the potentiostat. We assume that the same amount of
Li+ was removed from LMO by assuming a transfer ratio of 1 Li+ /e− .
All isotropic size parameters were set to the maximum possible value of 10 µm, since
the broadening did not improve the fit. All other isotropic atomic displacement factors
were set to 0.01, because it lies in an acceptable range and could not be refined for these
atoms due to the low intensity contribution of the respective phases. For LiNaSO4 and
Na3 Li(SO4 )2 (H2 O)6 , we set fixed anisotropic atomic displacement factors, as provided by
the used cif-file.

References
1. Zugschwert, C.; Dundálek, J.; Leyer, S.; Hadji-Minaglou, J.-R.; Kosek, J.; Pettinger, K.-H. The Effect of Input Parameter Variation
on the Accuracy of a Vanadium Redox Flow Battery Simulation Model. Batteries 2021, 7, 7. [CrossRef]
2. Oh, H.G.; Park, S.-K. Co-MOF Derived MoSe2@CoSe2/N-Doped Carbon Nanorods as High-Performance Anode Materials for
Potassium Ion Batteries. Int. J. Energy Res. 2022, 46, 10677–10688. [CrossRef]
3. Deng, Q.; Wang, M.; Liu, X.; Fan, H.; Zhang, Y.; Yang, H.Y. Ultrathin Cobalt Nickel Selenides (Co0.5 Ni0.5 Se2 ) Nanosheet Arrays
Anchoring on Ti3 C2 MXene for High-Performance Na+/K+ Batteries. J. Colloid Interface Sci. 2022, 626, 700–709. [CrossRef]
4. Li, X.; Liang, H.; Qin, B.; Wang, M.; Zhang, Y.; Fan, H. Rational Design of Heterostructured Bimetallic Sulfides (CoS2 /NC@VS4)
with VS4 Nanodots Decorated on CoS2 Dodecahedron for High-Performance Sodium and Potassium Ion Batteries. J. Colloid
Interface Sci. 2022, 625, 41–49. [CrossRef] [PubMed]
5. Bin, D.; Wang, F.; Tamirat, A.G.; Suo, L.; Wang, Y.; Wang, C.; Xia, Y. Progress in Aqueous Rechargeable Sodium-Ion Batteries. Adv.
Energy Mater. 2018, 8, 1703008. [CrossRef]
6. Li, W.-H.; Wu, X.-L. Advanced Cathode Materials in Dual-Ion Batteries: Progress and Prospect. Electrochem. Sci. Adv. 2022,
2, e2100127. [CrossRef]
7. Wang, X.-T.; Yang, Y.; Guo, J.-Z.; Gu, Z.-Y.; Ang, E.H.; Sun, Z.-H.; Li, W.-H.; Liang, H.-J.; Wu, X.-L. An Advanced Cathode
Composite for Co-Utilization of Cations and Anions in Lithium Batteries. J. Mater. Sci. Technol. 2022, 102, 72–79. [CrossRef]
8. Wang, H.; Wang, R.; Song, Z.; Zhang, H.; Zhang, H.; Wang, Y.; Li, X. A Novel Aqueous Li+ (or Na+ )/Br− Hybrid-Ion Battery with
Super High Areal Capacity and Energy Density. J. Mater. Chem. A 2019, 7, 13050–13059. [CrossRef]
9. Zhang, Z.; Hu, X.; Zhou, Y.; Wang, S.; Yao, L.; Pan, H.; Su, C.-Y.; Chen, F.; Hou, X. Aqueous Rechargeable Dual-Ion Battery Based
on Fluoride Ion and Sodium Ion Electrochemistry. J. Mater. Chem. A 2018, 6, 8244–8250. [CrossRef]
10. Whitacre, J.F.; Shanbhag, S.; Mohamed, A.; Polonsky, A.; Carlisle, K.; Gulakowski, J.; Wu, W.; Smith, C.; Cooney, L.; Blackwood,
D.; et al. A Polyionic, Large-Format Energy Storage Device Using an Aqueous Electrolyte and Thick-Format Composite NaTi 2
(PO 4 ) 3 /Activated Carbon Negative Electrodes. Energy Technol. 2015, 3, 20–31. [CrossRef]
11. Kalapsazova, M.; Rasheev, H.; Zhecheva, E.; Tadjer, A.; Stoyanova, R. Insights into the Function of Electrode and Electrolyte
Materials in a Hybrid Lithium–Sodium Ion Cell. J. Phys. Chem. C 2019, 123, 11508–11521. [CrossRef]
12. Kim, H.; Hong, J.; Park, K.-Y.; Kim, H.; Kim, S.-W.; Kang, K. Aqueous Rechargeable Li and Na Ion Batteries. Chem. Rev. 2014,
114, 11788–11827. [CrossRef] [PubMed]
13. Mohamed, A.I.; Whitacre, J.F. Capacity Fade of NaTi2(PO4)3 in Aqueous Electrolyte Solutions: Relating PH Increases to Long
Term Stability. Electrochim. Acta 2017, 235, 730–739. [CrossRef]
14. Whitacre, J.F.; Wiley, T.; Shanbhag, S.; Wenzhuo, Y.; Mohamed, A.; Chun, S.E.; Weber, E.; Blackwood, D.; Lynch-Bell, E.;
Gulakowski, J.; et al. An Aqueous Electrolyte, Sodium Ion Functional, Large Format Energy Storage Device for Stationary
Applications. J. Power Sources 2012, 213, 255–264. [CrossRef]
Energies 2023, 16, 4485 17 of 18

15. Wu, W.; Yan, J.; Wise, A.; Rutt, A.; Whitacre, J.F. Using Intimate Carbon to Enhance the Performance of NaTi2(PO4)3 Anode
Materials: Carbon Nanotubes vs Graphite. J. Electrochem. Soc. 2014, 161, 561–567. [CrossRef]
16. Wang, H.; Zhang, H.; Cheng, Y.; Feng, K.; Li, X.; Zhang, H. Rational Design and Synthesis of LiTi 2 (PO 4 ) 3−x F x Anode Materials
for High-Performance Aqueous Lithium Ion Batteries. J. Mater. Chem. A 2017, 5, 593–599. [CrossRef]
17. Wu, W.; Mohamed, A.; Whitacre, J.F. Microwave Synthesized NaTi2 (PO4 )3 as an Aqueous Sodium-Ion Negative Electrode.
J. Electrochem. Soc. 2013, 160, 497–504. [CrossRef]
18. Hou, Z.; Li, X.; Liang, J.; Zhu, Y.; Qian, Y. An Aqueous Rechargeable Sodium Ion Battery Based on a NaMnO2 –NaTi2 (PO4 )3
Hybrid System for Stationary Energy Storage. J. Mater. Chem. A 2015, 3, 1400–1404. [CrossRef]
19. Sun, D.; Jin, G.; Tang, Y.; Zhang, R.; Xue, X.; Huang, X.; Chu, H.; Wang, H. NaTi2 (PO4 )3 Nanoparticles Embedded in Carbon
Matrix as Long-Lived Anode for Aqueous Lithium Ion Battery. J. Electrochem. Soc. 2016, 163, A1388–A1393. [CrossRef]
20. Park, S.I.; Gocheva, I.; Okada, S.; Yamaki, J.-I. Electrochemical Properties of NaTi2 (PO4 )3 Anode for Rechargeable Aqueous
Sodium-Ion Batteries. J. Electrochem. Soc. 2011, 158, 1067–1070. [CrossRef]
21. Whitacre, J.F.; Tevar, A.; Sharma, S. Na4 Mn9 O18 as a Positive Electrode Material for an Aqueous Electrolyte Sodium-Ion Energy
Storage Device. Electrochem. Commun. 2010, 12, 463–466. [CrossRef]
22. Wang, Y.; Chen, L.; Wang, Y.; Xia, Y. Cycling Stability of Spinel LiMn2 O4 with Different Particle Sizes in Aqueous Electrolyte.
Electrochim. Acta 2015, 173, 178–183. [CrossRef]
23. Zhu, Z.; Peelaers, H.; Van de Walle, C.G. Hydrogen-Induced Degradation of NaMnO2 . Chem. Mater. 2019, 31, 5224–5228.
[CrossRef]
24. Li, Z.; Young, D.; Xiang, K.; Carter, W.C.; Chiang, Y.-M. Towards High Power High Energy Aqueous Sodium-Ion Batteries: The
NaTi2 (PO4 )3 /Na0.44 MnO2 System. Adv. Energy Mater. 2013, 3, 290–294. [CrossRef]
25. He, X.; Wang, J.; Qiu, B.; Paillard, E.; Ma, C.; Cao, X.; Liu, H.; Stan, M.C.; Liu, H.; Gallash, T.; et al. Durable High-Rate Capability
Na0.44MnO2 Cathode Material for Sodium-Ion Batteries. Nano Energy 2016, 27, 602–610. [CrossRef]
26. Zhan, X.; Shirpour, M. Evolution of Solid/Aqueous Interface in Aqueous Sodium-Ion Batteries. Chem. Commun. 2017, 53, 204–207.
[CrossRef]
27. Sauvage, F.; Laffont, L.; Tarascon, J.-M.; Baudrin, E. Study of the Insertion/Deinsertion Mechanism of Sodium into Na0.44 MnO2 .
Inorg. Chem. 2007, 46, 3289–3294. [CrossRef]
28. Luo, J.-Y.; Xia, Y.-Y. Aqueous Lithium-Ion Battery LiTi2 (PO4 )3 /LiMn2 O4 with High Power and Energy Densities as Well as
Superior Cycling Stability**. Adv. Funct. Mater. 2007, 17, 3877–3884. [CrossRef]
29. Tang, W.; Hou, Y.; Wang, F.; Liu, L.; Wu, Y.; Zhu, K. LiMn 2 O 4 Nanotube as Cathode Material of Second-Level Charge Capability
for Aqueous Rechargeable Batteries. Nano Lett. 2013, 13, 2036–2040. [CrossRef]
30. Chen, L.; Liu, J.; Guo, Z.; Wang, Y.; Wang, C.; Xia, Y. Electrochemical Profile of LiTi2 (PO4 )3 and NaTi2 (PO4 )3 in Lithium, Sodium
or Mixed Ion Aqueous Solutions. J. Electrochem. Soc. 2016, 163, A904. [CrossRef]
31. DIN EN ISO/IEC 17025:2018-03; Allgemeine Anforderungen an Die Kompetenz von Prüf- Und Kalibrierlaboratorien. Deutsche
Beuth Verlag GmbH: Berlin, Germany, 2018.
32. Binnewies, M.; Finze, M.; Jäckel, M.; Schmidt, P.; Willner, H.; Rayner-Canham, G. Allgemeine und Anorganische Chemie; Springer:
Berlin/Heidelberg, Germany, 2016; ISBN 978-3-662-45066-6.
33. Luo, F.; Wei, C.; Zhang, C.; Gao, H.; Niu, J.; Ma, W.; Peng, Z.; Bai, Y.; Zhang, Z. Operando X-Ray Diffraction Analysis of
the Degradation Mechanisms of a Spinel LiMn2O4 Cathode in Different Voltage Windows. J. Energy Chem. 2020, 44, 138–146.
[CrossRef]
34. Marchini, F.; Rubi, D.; del Pozo, M.; Williams, F.J.; Calvo, E.J. Surface Chemistry and Lithium-Ion Exchange in LiMn 2 O 4 for the
Electrochemical Selective Extraction of LiCl from Natural Salt Lake Brines. J. Phys. Chem. C 2016, 120, 15875–15883. [CrossRef]
35. Berg, H. Neutron Diffraction Study of Electrochemically Delithiated LiMn2O4 Spinel. Solid State Ion. 1999, 126, 227–234. [CrossRef]
36. Yabuuchi, N.; Yano, M.; Kuze, S.; Komaba, S. Electrochemical Behavior and Structural Change of Spinel-Type Li[LixMn2−x]O4
(X = 0 and 0.2) in Sodium Cells. Electrochim. Acta 2012, 82, 296–301. [CrossRef]
37. Yang, J.; Wang, H.; Hu, P.; Qi, J.; Guo, L.; Wang, L. A High-Rate and Ultralong-Life Sodium-Ion Battery Based on NaTi2 (PO4 )3
Nanocubes with Synergistic Coating of Carbon and Rutile TiO2 . Small 2015, 11, 3744–3749. [CrossRef] [PubMed]
38. Kabbour, H.; Coillot, D.; Colmont, M.; Masquelier, C.; Mentré, O. α-Na3 M2 (PO4 )3 (M = Ti, Fe): Absolute Cationic Ordering in
NASICON-Type Phases. J. Am. Chem. Soc. 2011, 133, 11900–11903. [CrossRef]
39. Kanamura, K.; Naito, H.; Yao, T.; Takehara, Z. Structural Change of the LiMn2O4 Spinel Structure Induced by Extraction of
Lithium. J. Mater. Chem. 1996, 6, 33. [CrossRef]
40. Jazouli, A.E.; Nadiri, A.; Dance, J.M.; Delmas, C.; Flem, L. Relationships between Structure and Magnetic Properties of Titanium
(III) NASICON-Type Phosphates. J. Phys. Chem. Solids 1988, 7, 779–783. [CrossRef]
41. Luo, J.-Y.; Cui, W.-J.; He, P.; Xia, Y.-Y. Raising the Cycling Stability of Aqueous Lithium-Ion Batteries by Eliminating Oxygen in
the Electrolyte. Nat. Chem. 2010, 2, 760–765. [CrossRef]
42. Chen, L.; Cao, L.; Ji, X.; Hou, S.; Li, Q.; Chen, J.; Yang, C.; Eidson, N.; Wang, C. Enabling Safe Aqueous Lithium Ion Open Batteries
by Suppressing Oxygen Reduction Reaction. Nat. Commun. 2020, 11, 2638. [CrossRef]
Energies 2023, 16, 4485 18 of 18

43. Song, Y.-J. Recovery of Lithium as Li3PO4 from Waste Water in a LIB Recycling Process. Korean J. Met. Mater. 2018, 56, 755–762.
[CrossRef]
44. Toby, B.H.; Von Dreele, R.B. It GSAS-II: The Genesis of a Modern Open-Source All Purpose Crystallography Software Package.
J. Appl. Crystallogr. 2013, 46, 544–549. [CrossRef]
45. Rietveld, H.M. A Profile Refinement Method for Nuclear and Magnetic Structures. J. Appl. Crystallogr. 1969, 2, 65–71. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.