Journal of Energy Storage 15 (2018) 408–414

Contents lists available at ScienceDirect

Journal of Energy Storage

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

Electrochemical measurement of corrosive impurities in molten

chlorides for thermal energy storage
Wenjin Dinga,* , Alexander Bonka , Joachim Gussoneb , Thomas Bauerc
a
Institute of Engineering Thermodynamics, German Aerospace Center (DLR), Stuttgart, Germany
b
Institute of Materials Research, German Aerospace Center (DLR), Cologne, Germany
c
Institute of Engineering Thermodynamics, German Aerospace Center (DLR), Cologne, Germany

A R T I C L E I N F O A B S T R A C T

Article history:
Received 17 July 2017 Molten chlorides are promising alternative thermal energy storage (TES) materials to be applied in
Received in revised form 27 November 2017 concentrated solar power (CSP) plants owing to their higher thermal stability (stable at >800
C) than the
Accepted 7 December 2017 commercial TES materials – nitrate salt mixtures (decomposed at 550
C). Higher operating
Available online 6 February 2018 temperatures of TES can increase efficiencies of thermal into electrical energy conversion for CSP
power plants, but cause additional challenges, particularly increase corrosiveness of metal containers
Keywords: and structural materials. Corrosion rates significantly depend on concentration of corrosive hydroxide
Corrosion control impurity in the molten chlorides. This paper presents an electrochemical method based on cyclic
Cyclic voltammetry (CV)
voltammetry (CV) to in-situ measure the concentration of the hydroxide impurity in molten MgCl2/KCl/
Concentrated solar power (CSP)
NaCl (60/20/20 mol%) salts at 500–700
C. Before each CV experiment, the concentration of the hydroxide
Magnesium hydroxyl ion (MgOH+)
impurity in the molten salts was measured via acid consumption (AC) measurements on the
simultaneously collected salt sample. The results of CV and AC experiments show that the peak current
densities obtained from CV are proportional to the concentrations of the hydroxide impurity obtained
from AC in the studied system. The slopes of peak current densities vs. concentrations of the hydroxide
impurity are determined, which compare well with literature data. This electrochemical method for
monitoring the corrosive hydroxide impurity is expected to assist the corrosion control on containers and
structural materials in the molten chlorides.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction Molten chloride salt mixtures such as MgCl2/KCl/NaCl (melting

Concentrated solar power (CSP) technology is emerging as one
of the important technologies in the future renewable energy
system. CSP plants concentrate a large area of sunlight onto a small
area by using mirrors or lenses to generate solar power. Electricity
is generated when the concentrated light is converted to heat,
which drives a heat engine connected to an electrical power
generator. Heat storage in molten salts, e.g., commercial nitrate salt
mixtures, allows CSP plants to generate dispatchable power during
the absence of sunlight and adds value to such power plants when
compared to photovoltaic panels. It is reported that global installed
CSP-capacity has increased nearly fifteen-fold from 2005 to 2015
(up to 4.8 Gigawatts) and grew at an average rate of 50 percent per
year from 2010 to 2015 [1].
* Corresponding author.
E-mail address: wenjin.ding@dlr.de (W. Ding).
temperature: 380
C) are promising thermal energy storage (TES)
materials to store the heat from sunlight in CSP plants, due to their
high thermal stability, good thermal conductivity, high heat
capacity but low price, as summarized in Table 1 together with
the properties of other common TES materials. In particular, their
high thermal stability makes them appropriate candidates to
replace the commercial TES materials in CSP – nitrate salt mixtures
(stable up to 550
C), when higher operating temperatures are
required (e.g., 700
C) for higher efficiency of thermal to electrical
energy conversion [2].
However, the application of molten chloride salts at higher
temperatures causes additional challenges, particularly increased
corrosiveness of containers and structural materials [2,3,7–10].
Table 2 summarizes the corrosion rates of commercial steels and
Ni-based alloys in various molten salts at different temperatures
[3,7]. When using the molten chloride salts (e.g., ZnCl2/KCl/NaCl) at
high temperatures, e.g., 800
C, even using expensive Ni-based
alloys with high corrosion resistance (e.g., Hastelloys C-22 and C-
276) cannot meet the requirements of corrosion resistance in

https://doi.org/10.1016/j.est.2017.12.007
2352-152X/© 2017 Elsevier Ltd. All rights reserved.
 W. Ding et al. / Journal of Energy Storage 15 (2018) 408–414 409

Table 1
Properties of commonly used molten salts as TES in CSP [3–3–6].

Molten salts composition (wt.%) Melting point Stability limit Viscosity (mPa Density Heat capacity Cost
(
C) (
C) s) (g cm3) (kJ kg1 K1) ($/kg)
Solar Salt NaNO3/KNO3 (60/40) 220 [3] 600 [3] 3.26 [3] 1.84 [4] 1.5 [4] (560
C) 0.5 [3]
(300
C)



Hitec NaNO3/KNO3/NaNO2 (7/53/40) 142 [3] 535 [3] 3.16 [3] (300 C) 1.9 [4] 1.56 [3] (300 C) 0.93 [3]
LiNaK carbonates Li2CO3/Na2CO3/K2CO3 (32.1/33.4/ 400 [3] 800–850 [3] 4.3 [3] (800
C) 1.85 [4] 1.4–1.5 [3] 1.2–1.3
34.5) [3]
LiNaK fluorides KF/LiF/NaF 454 [4] >700 [4] N/A 1.89 [4] 1.89 [4] (700
C) >2a
(59/29/12)
ZnNaK chlorides ZnCl2/NaCl/KCl 204 [3] 850 [3] 4 [3] 2 [5] 0.81 [3] (300–600
C) <1 [3]
(68.6/7.5/23.9) (600–800
C)


MgNaK chlorides MgCl2/NaCl/KCl (68.2/14.0/17.8) 380 [6] >800 4 [5] 1.65 [5] 1.0 [5] (500–800 C) <0.5a
(500–800
C)
a
The prices of LiNaK fluorides and MgNaK chlorides are estimated from the data on www.alibaba.com.

Table 2
Corrosion in commonly used molten salts as TES [3,7].

Molten Salts Alloy Temperature (
C) Corrosion rate in inert gas (mm/year) Refs.
Solar Salt SS 316 600 15.9 [3]
Ha230 600 47 [3]
Hitec SS 321 570 2 [3]
LiNaK carbonates (32.1/33.4/34.5) In600 900 <1000 [3]
ZnNaK chlorides (68.6/7.5/23.9) SS 304 800 110–200 [7]
SS 304 500 >300 (in air) [7]
Ha C-22 800 <50 [7]
Ha C-276 800 <50 [7]

industrial applications (corrosion rate <10 mm/year). In addition,
the concentration of corrosive impurities in the molten chloride
salts should be controlled.
It is well known that corrosion rates significantly depend on the
concentration of hydroxide/O2 impurities in the molten chloride
salts [2,8,9]. The corrosive impurities are mainly produced from
water and O2 existing in the chloride salts, as well as moisture and
O2 in the atmosphere dissolving in the chloride salts [2,8–11]. The
presence of water in the chloride salts from hydrated phases, e.g.,
strong hydrophilic chloride salt – MgCl2, induces its hydrolysis
during heating following the simplified equations Eqs. (1) and (2)
to produce the corrosive HCl and MgOHCl (more details shown in
Eqs. (6) and (7)). HCl and MgOHCl have corrosion reactions with
the metal alloys of containers and structural materials according to
the simplified equations Eqs. (3) and (4), while oxygen containing
species existing in the chloride salts (e.g. dissolved O2) corrode the
containers and structural materials by oxidizing the alloying
components as shown in the simplified Eq. (5) [11,12]. Table 2
shows that when the molten ZnCl2/NaCl/KCl salt is exposed in air,
which leads to a high concentration of dissolved O2 and hydroxide
species in melt, the corrosion rate of steel 304 at 500
C is much
higher than that at 800
C when sweeping the molten salt with an
inert gas.
MgCl2 + H2O ! MgO + 2 HCl,
MgCl2 + H2O ! MgOHCl + HCl,
x HCl + M ! MClx + (x/2) H2,
x MgOHCl + M ! x MgO + MClx + (x/2) H2,
x/2O2 + y M ! MyOx,
M = the alloying components in the metal alloys.
Considerable effort has been made to control the hydrolysis
during heating to reduce the corrosion effect of the molten
chloride salts [10–18]. For instance, it is reported in [10,11] that
when the heating of the salts with strong hydrophilic MgCl2 in the
dehydration process is carefully performed, the irreversible
decomposition side reactions producing MgOHCl and MgO (see
Eqs. (6) and (7)) can be suppressed substantially. Only a small
amount of oxide/hydroxide impurities remains in the salts, i.e., a
small amount of corrosive HCl and MgOHCl is produced.
MgCl2H2O ! MgOHCl + HCl, (6)
MgCl2H2O ! MgO + 2 HCl. (7)
MgOHCl dissolved in molten salts can be decomposed further to
MgO and corrosive HCl at high temperatures. Therefore, to reduce
the corrosive impurity ion MgOH+ in molten salts further to an
acceptable concentration level regarding to the corrosion rate (the
experimental study on the acceptable concentration level in
MgCl2/KCl/NaCl is in progress in our work), some techniques [12–
18] to purify the salts have been developed. In literature,
purification was accomplished by a chlorinating process (CCl4,
HCl, other chlorinating compounds [12–14]), by reduction of the
(1)
salt by using active metals such as Mg [15], or saturation of a salt
with a liquid metal (i.e., LiCl with Li metal [16–18]), to reduce the
redox potential of the melt.
(2) For an efficient control of corrosive impurity concentrations in
molten salts in industrial applications, reliable, fast but affordable
concentration measurements of the corrosive impurities are
(3) required. At the start of this work, hydroxide containing ions in
molten chlorides were examined with ex-situ analytical methods
[19]. However, compared to these ex-situ methods, generally via
(4) collecting and analyzing a salt sample out of the melt, in-situ
measurements of hydroxide containing ions with an electrochem-
ical analytic system based on cyclic voltammetry (CV) provide
(5) unique possibility to monitor impurity concentrations and thus
410 W. Ding et al. / Journal of Energy Storage 15 (2018) 408–414
control the technological processes [19,20]. Some electrochemical
investigations to examine hydroxide species in chloride melts such
as in MgCl2/NaCl (730–850
C) [19], LiCl/KCl (350–500
C) [22] and
(430–500
C) [23] have been performed. In many publications [19–
24], it was found that in melts such as MgCl2/NaCl [19], LiCl/KCl
[22], CaCl2/KCl [25], the height of the reduction peak of hydroxide
species in CV was directly proportional to the concentration of
hydroxide species. Accordingly, in-situ CV appears to be the most
appropriate method for online quantitative and qualitative
analysis of corrosive hydroxide impurities in molten chlorides.
To our knowledge, no published papers are available which
investigate the correlation between the reduction peak of
hydroxide species with its concentration in molten MgCl2/KCl/
NaCl at 500–700
C, a promising TES material in CSP and other
power plants in this temperature range. In this work, we focus on
understanding the electrochemical behavior of MgCl2/KCl/NaCl
with the hydroxide impurities via the CV experiments, in order to
develop an electrochemical technique for monitoring the hydrox-
ide impurities in molten chlorides to assist the corrosion control.
2. Experimental
2.1. Chemicals
KCl(99%) and NaCl(99%) were purchased from Alfa Aesar, while
anhydrous MgCl2(99%) came from Magnesia, a global distribution
company of Magnesium based compounds in Germany. They were
used to synthesize the eutectic salt mixture of MgCl2/KCl/NaCl (in
this work with the molar composition of 60/20/20, respectively).
Referring to our previous work [11], the heating process of the salts
is performed as follow: after vacuuming for 30 min (30 mbar),
the salt mixture (150 g) was heated under an argon atmosphere
(purity 99.9999%, H2O 0.5 ppm, 20 l/h (STP), pressure above
atmospheric pressure is about 0.1 bar) to 200
C with the heating
rate of 5
C/min, then kept at 200
C for 1 h, and after that was
heated to the studied temperature (500–700
C) of the electro-
chemical experiments. The titration measurements, which are
Fig. 1. Schematic of experimental set-up for cyclic voltammetry experiments on molten salts MgCl2/KCl/NaCl, adopted from our previous work [26]. The glassy carbon
crucible is open on the top.
described in Section 2.3, show that the concentrations of oxygen
species in the salts after the heating process were about 8000,
1500, 800 ppm O at 500, 600 and 700
C, respectively (ppm O
represents mass parts per million of oxygen species in the salt, e.g.,
ppm of all oxygen species are expressed as O2/MgOH+, see
Eq. (10)). In order to study the correlation between the peak
current densities and concentrations of hydroxide species and
oxide-ion, de-ionized water was added in the salts before heating
or NaOH (pellet, 99%) purchased from Alfa Aesar was added in the
molten salts, in order to vary the concentration of hydroxide
species in the molten salts, while MgO (powder, 99.95%) from VWR
Chemicals was added in the salts before heating to vary the
concentrations of oxide-ions.
2.2. Experimental set-up
Fig.1 shows a schematic of the experimental set-up used for CV
experiments, which has been used for CV experiments at 500
C in
our previous work [26]. A high temperature resistance glassy
carbon crucible purchased from HTW Germany (Sigradur1 G) was
used to avoid the reaction of the molten salts with the materials of
the crucible. During the experiments, the temperature of the
molten salts under an argon atmosphere (same parameters as
mentioned in Section 2.1) was controlled by the programmable
furnace and the thermocouple close to the surface of the molten
salt.
The pure tungsten wire purchased from Alfa Aesar (99.95%,
diameter: 1 mm) were applied in this work as the working, counter
and reference electrodes for the concentration measurements on
hydroxide species, while glassy carbon electrodes from HTW
Germany (diameter: 1 mm) were used as the working electrodes for
oxide species. Moreover, two tungsten wires were immersed in the
molten salts as shown in Fig. 1, which have larger contact areas than
the working electrode. They are served as the counter and reference
electrode, respectively. In the CV experiments, the immersion depth
of the working electrode was fixed to 5 mm by using an ohmmeter
(the contact area of the tungsten and glassy carbon working
 W. Ding et al. / Journal of Energy Storage 15 (2018) 408–414 411

electrode with the melt is 16.5 mm2), while the counter and
reference electrodes had a much larger contact area of about
50 mm2, as shown in Fig. 1. The CV experiments were conducted
using a ZENNIUM electrochemical workstation from Zahner GmbH
(Germany). Since it is shown that the plots of the peak current
densities vs. v1/2 at 500, 600 and 700
C are linear and pass through
the origin when the potential sweep rate v is below 500 mV/s. Thus,
a sweep rate of 200 mV/s was used in all CV experiments in this
work. Moreover, in order to remove the noise in the cyclic
voltammograms and reduce the impurity level, 25 min pre-
electrolysis (PE) was performed to pre-treat the melts as in our
previous work [26].
Using the sample collector (Fig. 1), salt samples were collected
directly before the CV measurements for analysis. More details
about the sample collector can be found in our previous work [26].
In our experiments in this work, about 2 g salt was collected during
each extraction step for successive sample analysis.

2.3. Acid consumption for oxide/hydroxide concentration

measurements

An acid consumption (AC) method based titration, which was

also introduced in our previous work [26], was used to determine
the total amount of oxide/hydroxide in a salt sample ex-situ. The
sample (msample = 0.5–1 g) was dissolved in a known amount of HCl
(10 ml 0.1 M HCl solution). The dissolution reactions are:
O2 + 2HCl = 2Cl + H2O, (8)

MgOH+ + HCl = Mg2+ + Cl + H2O. (9)

The excess HCl is titrated with NaOH solution (0.1 M, VNaOH).
The titration indicator is 1% phenolphthalein solution. References
[19,27] show that the solubility of MgO in eutectic MgCl2/NaCl
(41.5/58.5 mol%) and MgCl2/NaCl (60/40 mol%) at 730
C was only
88 and 75 ppm O, respectively. Compared to the amount of MgOH+ Fig. 2. Cyclic voltammogram in the eutectic MgCl2/KCl/NaCl at 500
C obtained by a
in the chloride salts (400–8000 ppm O) in this work, the amount of working electrode: a) Tungsten (adopted from our previous work [26]); b) Glassy
carbon. Sweep rate: 200 mV/s. Tungsten reference electrode. ip(B): peak current
oxide-ion was neglected in the titration measurements in this density for reaction B.
work due to so low solubility of MgO at lower temperatures (this
work 500–700
C). This was also proved by our CV experiments
(see Section 3.1.) that all the amount of the reacted HCl was due to
the dissolution reaction with MgOH+. Thus, the concentration of Reaction A’: Mg (s) = Mg2+ + 2e (12)
MgOH+ in the salt sample can be calculated as follow:
C(MgOH+) [ppm O] = (CHCl * VHCl  CNaOH * VNaOH) * MO/msam- Reaction B: MgOH+ + e = MgO (s) + 1/2H2 (g) (13)
ple * 1000, (10)

where ppm O represents mass parts per million of oxygen in

MgOH+ in the salt sample, CHCl and CNaOH are 0.1 mol/L, VHCl is the Reaction C: Cl = 1/2 Cl2 (g) + e (14)
volume of HCl, 10 ml, VNaOH is the titration volume of NaOH in ml; Skar [19] showed that the peak current density of the peak B
MO is the molar mass of oxygen 16 g/mol, msample is the mass of the (ip(B)) was proportional to the bulk concentration of MgOH+
sample in g. (C1(MgOH+))) in molten NaCl/MgCl2 salts:

ip ðBÞ ¼ kðT; nÞðBÞ  C 1 ðMgOHþ Þ; ð15Þ

3. Results and discussion
where ip represents the peak current density in mA/cm ; k(T, n) is a
2

constant, i.e., slope of peak current densities vs. concentrations of

3.1. Cyclic voltammograms
the reacting species in (ppm O)/(mA/cm2), which depends on the
temperature T and n potential sweep rate; C1 is the bulk
Typical cyclic voltammograms by using the tungsten and glassy
concentrations of the reacting species in ppm O. Thus, in this
carbon working electrode are shown in Fig. 2. As references [19,28]
case, the peak current densities of the peak B obtained via CV can
conclude that the most stable H-containing positive ion is
simply be used to in-situ monitor the concentration of the
Magnesium hydroxyl (MgOH+) via experiments, the peaks A, A’,
hydroxide impurity in the chloride molten salts [19].
B and C in Fig. 2a) are considered to represent the following
When glassy carbon electrodes were used as working electro-
electrochemical reactions, respectively:
des, Mohamedi et al. [20] observed two obvious peaks corre-
Reaction A: Mg2+ + 2e = Mg (s) (11) sponding to the electrochemical oxidation of O2 with carbon of
412 W. Ding et al. / Journal of Energy Storage 15 (2018) 408–414
the glassy carbon electrodes in the anodic potential range of the
cyclic voltammogram for molten CaCl2 containing CaO at 850
C
([CaO] = 3.3  105 mol cm3, i.e., 260 ppm O):
O2(diss) + xC = CxO(ads) + 2e (+0.7 V vs. Pt reference)
O2(diss) + CxO(ads) = CO2(g) + (x  1)C + 2e (+1.2 V vs. Pt refer-
ence) (17)
When molten MgCl2 or NaCl/MgCl2 was added with MgOHCl or
saturated with MgO at 730–850
C, a peak (+0.7 V vs. Pt reference)
corresponding to the electrochemical oxidation of O2 with carbon
of glassy carbon electrodes was observed in the anodic potential
range of the cyclic voltammogram [19]. However, not as two peaks
for CaO in molten CaCl2 [20], this peak was difficult to detect, as the
solubility of MgO in molten MgCl2 or NaCl/MgCl2 was very low. The
measured average saturation concentration of MgO in eutectic
MgCl2/NaCl (41.5/58.5 mol%) was only 80 ppm O at 730–845
C,
which could cause a peak with the peak current density of only
4 mA/cm2 [19,27].
In the cyclic voltammograms obtained by glassy carbon
electrodes in this work (see Fig. 2b)), such peaks corresponding
to the electrochemical oxidation of O2 with carbon of the glassy
carbon electrodes were not observed in the anodic potential range
of the cyclic voltammogram, when the molten MgCl2/KCl/NaCl salt
was also saturated with MgO. This may be explained with low
solubility of MgO in molten chlorides as reported in [19,27]. The
peaks corresponding to the electrochemical oxidation of O2 with
carbon of glassy carbon electrodes might be too weak and hence
hide in the noise caused by other impurities.
3.2. Correlation between current density and impurity concentration
In order to study the correlation between the current density of
peak B in CV and the concentration of MgOH+ in molten MgCl2/KCl/
NaCl salts at 500–700
C, the direct current pre-electrolysis (PE) was
performed to remove the impurities as well as to reduce the
concentration of MgOH+. More details about the PE and the obtained
CVs after PE at 500
C can be found in our previous work [26]. Fig. 3
shows the cyclic voltammograms of chloride molten salts before and
after the PE at 600
C. In the PE experiments, the voltage on the
working and counter electrodes was kept at 2 V.5, 25 and 65 min PE
were performed to purify the chloride molten salts. It can be found
from Fig. 3 that the noise in the CV was removed after 5 min PE.
Moreover, the height of peak B decreased with the increasing PE time
Fig. 3. Cyclic voltammograms of chloride molten salts before direct current pre-
electrolysis (PE), after 5, 25 and 65 min PE at 600
C. The working and counter/
reference electrodes were tungsten. The voltage on the working and counter
electrodes was kept at 2 V. Sweep rate: 200 mV/s.
significantly, in particular in the first 25 min. However, PE seemed to
be less effective to remove MgOH+ with the increasing PE time, in
particular after 25 min. After PEs, some solids (e.g., MgO) were
observed at the surface of the working electrode. Thus, in order to
(16)
keep the effectiveness of PE for removal of MgOH+, these solids are
suggested to be removed during the PE experiments.
After PE, about 1.5 g NaOH (4286 ppm O) was added to 140 g
molten salt, in order to increase the concentration of MgOH+. The
NaOH pellets are dissolved into OH and Na+ ions in the molten
salts. As shown in Eq. (18), the dissolved OH ions from NaOH
combine with Mg2+ ions to form the most stable hydroxide species
MgOH+ in the molten MgCl2/KCl/NaCl. The MgOH+ ions can
decompose to MgO and H+ ions according to Eq. (19).
OH + Mg2+ = MgOH+ (18)
MgOH+ = MgO (s) + H+ (19)
H+ + Cl = HCl (g) (20)
+
The dissolved H ions can corrode the containers and structural
materials. Moreover, as shown in Eq. (20), they can combine with
Cl ions to form HCl, most of which is swept out in the form of gas
due to it low solubility in molten MgCl2/KCl/NaCl [19].
Fig. 4 shows selected cyclic voltammograms at different times
after adding NaOH at 500
C, while Fig. 5 summarizes the
concentrations of MgOH+ obtained with acid consumption method
(grey points and curve), as well as the current densities of the peak
relating the concentration of MgOH+ obtained by CV (black points
and curve), as a function of time after adding NaOH. The result
similar as literature [19] was observed that the peak current
density (as well as concentration of MgOH+) increases initially.
After fifteen minutes, it starts to decrease. This implies that the
dissolution of the added NaOH (Eq. (18)) and the decomposition of
MgOH+ (Eq. (19)) to MgO occur simultaneously. In literature [19],
MgOHCl pellets were added to the MgCl2/NaCl melt at 730–850
C
for changing the concentration of MgOH+ in the melt. The
concentration of MgOH+ starts to decrease only after an hour
[19]. It may be explained that MgOHCl pellets dissolves more
slowly than NaOH pellets used in this work.
Fig. 6a) and b) summarize all peak current densities and
concentrations of MgOH+ in molten MgCl2/KCl/NaCl obtained from
CV and AC experiments in the PE and after adding NaOH pellets at
500–700
C. It can be found that the peak current densities
according to the electrochemical reduction reaction of MgOH+ are
proportional to the concentrations of MgOH+. Via the least-square
method, the slope of peak current densities vs. concentrations of
MgOH+ (k(T, n)(B) in Eq. (15)) is determined to be 38.2 7.6,
13.7 2.7 and 9.4 1.9 (ppm O)/(mA/cm2) at 500, 600 and 700
C,
respectively. It can be found that the slope has a significant
decrease from 500
C to 600
C, while the decrease of the slope is
slight from 600
C to 700
C. This decrease could be explained with
increasing diffusivity of MgOH+ ions in molten MgCl2/KCl/NaCl
with increasing temperature.
Table 3 shows comparison of the slopes of peak current
densities vs. concentrations of MgOH+ in molten chlorides (k(T, n)
(B)) determined in this work and reported in literature [19]. The
same finding was shown in literature [19] that the slope of peak
current densities vs. concentrations of MgOH+ in molten MgCl2/
NaCl salts had a small change at high temperatures, i.e., at 730–
850
C. Moreover, literature [19] shows the effect of the salt
composition on the slope of peak current densities vs. concen-
trations of MgOH+ that the slope decreased when going from 100 to
80 mol% MgCl2, while only a very small difference was observed
 W. Ding et al. / Journal of Energy Storage 15 (2018) 408–414 413

Fig. 4. Cyclic voltammograms after adding NaOH in the molten salts: a) 0 min (solid) and 15 min (dotted); b) 15 min (dotted) and 60 min (solid). T = 500
C. Working
electrode: tungsten. Sweep rate: 200 mV/s. 0 min means the measurement directly after adding NaOH.

16000

ConcentraƟon of MgOH+ [ppm O]

a) y = 38.2x
±20 %
12000
500°C

8000 y = 13.7x

4000
±20 %
600°C
0
0 100 200 300 400 500
Peak current density ip(B) [mA/cm2]
Fig. 5. Concentration of MgOH+ (black points and curve: CV data; Grey points:
titration data) as a function of time after adding 1.5 g NaOH (4286 ppm O). Error bars
4000
ConcentraƟon of MgOH + [ppm O]

represent the standard deviations in the CV (three measurements) and AC

measurements (three measurements). b) y = 9.4x
3000
700°C
between 80 and 20 mol% MgCl2. From 20 to 10 mol% MgCl2, the 2000
slope decreased again. As shown in Table 3, the slope of peak
±20 %
current densities vs. concentrations of MgOH+ in molten MgCl2/ 1000
NaCl salts (80/20 mol%) is about 22 4 (ppm O)/(mA/cm2) at 730–
848
C. The slopes of peak current densities vs. concentrations of
0
MgOH+ in molten MgCl2/KCl/NaCl (60/20/20 mol%) determined in
0 100 200 300
this work compare well with these reported in literature [19].
Peak current density ip(B) [mA/cm2]
4. Conclusions Fig. 6. Peak current densities vs. concentrations of corrosive MgOH+ in molten
MgCl2/KCl/NaCl (60/20/20 mol%) at a) 500, 600
C and b) 700
C. Error bars
An electrochemical method based on cyclic voltammetry (CV) represent the standard deviations in the CV (three measurements) and AC
has been developed to in-situ measure the concentration of measurements (three measurements).

corrosive impurity MgOH+ in molten chloride salts at temperatures

of 500–700
C. Pre-electrolysis was performed to purify the molten

Table 3
 
Comparison of slopes of peak current densities vs. concentrations of MgOH+ (kðT; nÞðBÞ) or O2 (kðT; nÞ O2 ) in molten chlorides determined in this work and reported
in literature [19].
 
T (
C) kðT; nÞ O2 kðT; nÞðBÞ molten chlorides Refs.
(ppm O)/(mA/cm2)
(ppm O)/(mA/cm2)
500 not detected 38.2 7.6 MgCl2/KCl/NaCl(60/20/20 mol%) this work
600 not detected 13.7 2.7 MgCl2/KCl/NaCl(60/20/20 mol%) this work
700 not detected 9.4 1.9 MgCl2/KCl/NaCl(60/20/20 mol%) this work
728–848 24 6 37 8 MgCl2 [19]
741–842 21 4 28 6 MgCl2/NaCl(90/10 mol%) [19]
730–848 17 5 22 4 MgCl2/NaCl(80/20 mol%) [19]
730–845 9 2 21 6 MgCl2/NaCl(41.5/58.5 mol%) [19]
730–850 n.a. 22 3 MgCl2/NaCl(30/70 mol%) [19]
801 n.a. 23 2 MgCl2/NaCl(20/80 mol%) [19]
798–850 n.a. 10 5 MgCl2/NaCl(10/90 mol%) [19]
414 W. Ding et al. / Journal of Energy Storage 15 (2018) 408–414
salt, i.e., remove the impurities and reduce the concentration of
MgOH+. After that, by adding NaOH pellets into the molten salts,
the peak corresponding to the electrochemical reduction of MgOH+
was determined in the cathodic potential range of the cyclic
voltammogram. However, the peak corresponding to the electro-
chemical oxidation of O2 with carbon of glass carbon electrodes
was not observed in the anodic potential range in the studied
temperature range, when enough MgO was added to ensure the
saturation of O2 in the molten salts. This may be explained that
the concentration of O2 in the studied molten salts was very low
at studied temperatures due to low solubility of MgO. The results of
CV and AC experiments show that the peak current densities
obtained from CV are proportional to the concentrations of MgOH+
obtained from AC in the studied temperature range. The slopes of
peak current densities vs. concentrations of MgOH+ are deter-
mined, which compare well with literature data. These slopes will
be used for monitoring the corrosive MgOH+ impurities in our
future work.
This electrochemical method can assist impurity monitoring
and hence the corrosion control on containers and structural
materials in the molten chloride salts. Overall, this work presented
the successful commissioning of an experimental setup to study
molten chloride melts in-situ (by cyclic voltammetry) and ex-situ
(post-analysis of salt samples by titration). Correlation between
both methods could be established. This allows for further studies
of the molten salt chloride chemistry and development of the
process technology for thermal energy storage systems up to
700
C in the future.
Acknowledgments
This research has been performed within the DLR-DAAD
fellowship programme, which is funded by German Academic
Exchange Service (DAAD) and German Aerospace Center (DLR). The
authors would like to thank Yanlei Xiu, a Master student from
Karlsruhe Institute of Technology (KIT), Markus Braun, Jonas
Tombrink, Marc-Stephan Glumm and Andreas Kohzer at the DLR-
Institute of Engineering Thermodynamics for their support.
References
[1] REN21, Renewables 2016: Global Status Report, (2016) (ISBN 978-3-9818107-
0-7).
[2] A.M. Kruizenga, Corrosion Mechanisms in Chloride and Carbonate Salts,
SANDIA, 2012 Report SAND 2012-7594.
[3] K. Vignarooban, X. Xu, A. Arvay, K. Hsu, A.M. Kannan, Heat transfer fluids for
concentrating solar power systems—a review, Appl. Energy 146 (2015) 383–
396, doi:http://dx.doi.org/10.1016/j.apenergy.2015.01.125.
[4] T. Bauer, N. Pfleger, D. Laing, S. Kaesche, High-temperature molten salts for
solar power application, Chapter 20 in Book: Molten Salts Chemistry, 415
(2013) (ISBN: 978-0-12-398538-5).
[5] Y. Li, X. Xu, X. Wang, P. Li, Q. Hao, B. Xiao, Survey and evaluation of equations for
thermophysical properties of binary/ternary eutectic salts from NaCl, KCl,
MgCl2, CaCl2, ZnCl2 for heat transfer and thermal storage fluids in CSP, Sol.
Energy 152 (2017) 57–79, doi:http://dx.doi.org/10.1016/j.solener.2017.03.019.
[6] Y. Tian, C.Y. Zhao, A review of solar collectors and thermal energy storage in
solar thermal applications, Appl. Energy 104 (2013) 538–553, doi:http://dx.
doi.org/10.1016/j.apenergy.2012.11.051.
[7] K. Vignarooban, X. Xu, K. Wang, E.E. Molina, P. Li, D. Gervasio, A.M. Kannan,
Vapor pressure and corrosivity of ternary metal-chloride molten-salt based
heat transfer fluids for use in concentrating solar power systems, Appl. Energy
159 (2015) 206–213, doi:http://dx.doi.org/10.1016/j.apenergy.2015.08.131.
[8] G.Y. Lai, High-temperature corrosion and materials applications, Chapter 15 in
Book: Molten Salt Corrosion, ASM International, 2007.
[9] H.R. Copson, Corrosion of heating electrodes in molten chloride baths, J.
Electrochem. Soc. 100 (6) (1953) 257–264, doi:http://dx.doi.org/10.1149/
1.2781115.
[10] G.J. Kipouros, D.R. Sadoway, A thermochemical analysis of the production of
anhydrous MgCl2, J. Light Met. 1 (2) (2001) 111–117, doi:http://dx.doi.org/
10.1016/S1471-5317(01)00004-9.
[11] L. Maksoud, T. Bauer, Experimental investigation of chloride molten salts for
thermal energy storage applications, 10th International Conference on Molten
Salt Chemistry and Technology, Shenyang, China, 2015, pp. 1–12.
[12] D.F. Williams, Assessment of Candidate Molten Salt Coolants for the NGNP/NHI
Heat-transfer Loop, Oak Ridge National Laboratory, 2006.
[13] V.L. Cherginets, T.P. Rebrova, Studies of some acid-base equilibria in the molten
eutectic mixture KCl-LiCl at 700 C, Electrochim. Acta 45 (3) (1999) 469–476,,
doi:http://dx.doi.org/10.1016/S0013-4686(99)00274-1.
[14] D.L. Maricle, D.N. Hume, A new method for preparing hydroxide-free alkali
chloride melts, J. Electrochem. Soc. 107 (4) (1960) 354–356, doi:http://dx.doi.
org/10.1149/1.2427694.
[15] J. Ambrosek, Doctoral Thesis: Molten Chloride Salts for Heat Transfer in
Nuclear Systems in Nuclear Engineering and Engineering Physics, University of
Wisconsin-Madison, 2011.
[16] B. Mishra, D.L. Olson, Corrosion of refractory alloys in molten lithium and
lithium chloride, Min. Proc. Ext. Met. Rev. 22 (2001) 369–388, doi:http://dx.
doi.org/10.1080/08827500208547421 (4–6 SCPEC. ISS).
[17] J.E. Indacochea, J.L. Smith, K.R. Litko, E.J. Karell, A.G. Rarez, High-temperature
oxidation and corrosion of structural materials in molten chlorides, Oxid. Met.
55 (1–2) (2001) 1–16, doi:http://dx.doi.org/10.1023/A.1010333407304.
[18] J.E. Indacochea, J.L. Smith, K.R. Litko, E.J. Karell, Corrosion performance of
ferrous and refractory metals in molten salts under reducing conditions, J.
Mater. Res. 14 (5) (1999) 1990–1995, doi:http://dx.doi.org/10.1557/
JMR.1999.0268.
[19] R.A. Skar, Doctoral Thesis: Chemical and Electrochemical Characterisation of
Oxide/hydroxide Impurities in the Electrolyte for Magnesium Production,
Norwegian University of Science and Technology (NTNU), 2001.
[20] M. Mohamedi, B. Borresen, G.M. Haarberg, R. Tunold, Anodic behavior of
carbon electrodes in CaO-CaCl2 meltsat 1123 K, J. Electrochem. Soc. 146 (4)
(1999) 1472–1477, doi:http://dx.doi.org/10.1149/1.1391789.
[21] S.A. Kuznetsov, Electrochemical techniques, in: M. Gaune-Escard (Ed.), Molten
Salts: from Fundamentals to Applications, Springer, Netherlands, 2001, pp. 283
(no. 52 in NATO Science Series).
[22] S.H. White, Ionic Liquids, Plenum Press, New York, 1981 (Chapter 12).
[23] J. Gussone, Doctoral thesis: Schmelzflusselektroytische Abscheidung von Titan
auf Vertärkungsfasern zur Herstellung von Titanmatrixverbundwerkstoffen,
RWTH Aachen University, Shaker Verlag Aachen, 2012.
[24] G.J. Janz, C.B. Allen, N.P. Bansal, R.M. Murphy, R.P.T. Tomkins, Physical
Properties Data Compilations Relevant to Energy Storage. II. Molten Salts: Data
on Single and Multi-component Salt Systems, National Bureau of Standards,
United States: N, 1979.
[25] H.C. Brookes, Voltammetric investigations of CaCI2:KCI melts at 700
C, J.
Electrochem. Soc. 135 (2) (1988) 373–377, doi:http://dx.doi.org/10.1149/
1.2095618.
[26] W. Ding, A. Bonk, J. Gussone, T. Bauer, Cyclic voltammetry for monitoring
corrosive impurities in molten chlorides for thermal energy storage, Energy
Procedia 135 (2017) 82–91, doi:http://dx.doi.org/10.1016/j.
egypro.2017.09.489.
[27] S. Boghosian, A. Godø, H. Mediaas, T. Østvold, Oxide complexes in alkali-
alkaline-earth chloride melts, Acta Chem. Scand. 45 (1991) 145–157, doi:
http://dx.doi.org/10.3891/acta.chem.scand.45-0145.
[28] A. Bjørgum, U. Eriksen, R. Tunold, Voltametrik Karaktersering av Hydrogen- og
Oksygen-holdige Homponenter i Magnesiumkloridsmelter, Tech. Rep., Lab. of
Ind. Electrochem., NTH, Trondheim, Norway, Norwegian, 1986.