-
1
Corrosion of steel alloys in eutectic NaCl+Na2CO3 at 700 °C and
Li2CO3 + K2CO3 + Na2CO3 at 450 °C for thermal energy storage
Madjid Sarvghad*, Theodore A. Steinberg, Geoffrey Will
Science and Engineering Faculty, Queensland University of
Technology (QUT), Queensland, Australia 4001
Abstract
Stainless steel 316, duplex steel 2205 and carbon steel 1008
were examined for compatibility with the eutectic mixtures of
NaCl+Na2CO3 at 700 °C and Li2CO3 + K2CO3 + Na2CO3 at 450 °C in air
for thermal energy storage. Electrochemical measurements combined
with advanced microscopy and microanalysis techniques were
employed. Oxidation was found as the primary attack in both molten
salt environments. However, the availability of O2 controlled the
degree of oxidation. Alloy 316 showed the lowest corrosion current
densities in each molten salt owing to the formation of films on
the surface. The attack morphology on the surface of all materials
was uniform corrosion with no localized degradation at 450 ºC.
Microscopy observations showed grain boundary oxidative attack as
the primary corrosion mechanism for all studied alloys at 700 °C
with depletion of alloying elements from grain boundaries in alloys
316 and 2205. The protective nature of austenite phase reduced
selective oxidation of the underlying ferrite layers of alloy 2205
in chloride carbonate at 700 °C.
Keywords
Steel; Molten Salt; Corrosion; Impedance spectroscopy;
Polarization; Microscopy
* Corresponding author; Email: [email protected]; [email protected]
-
2
Graphical abstract
-
3
1. Introduction
Recent interest in increasing plant efficiency requires improved
compatibility between structural alloys and molten salts for
Thermal Energy Storage (TES) used in Concentrated Solar Thermal
(CST) power plants [1-3]. Hot corrosion and oxidation from retained
storage media (molten salts) as Phase Change Materials (PCM) causes
significant deterioration to containment materials used in TES
systems [1-14]. Whilst steel alloys are considered as economic
candidates for containment materials in TES systems, eutectic
compositions of molten chlorides, nitrates, carbonates and
fluorides are favorite candidates as storage medium [15-20].
Oxyanions like carbonates have recently attracted more attention
not only owing to their great heat capacity, but also for their
high energy density and less corrosive behavior compared to
traditional chlorides [2, 21]. Even the mixture of carbonate and
chloride have been reported to be less corrosive than pure
chlorides [21].
The compatibility of structural alloys with molten salts depends
on the metal’s potential for oxidation, passivating nature and the
solubility of corrosion products in the salt [22]. In steel alloys,
scaling rate depends on three variables: steel chemistry,
temperature and atmosphere [23]. Based on the working temperature
low-alloyed carbon steels, Cr-Mo steels and Cr-Ni stainless steels
might be employed in combination with molten salts [10]. Groll et
al. [24] reported intergranular corrosion of steel alloys in
contact with molten chloride salts at temperatures up to 420 °C.
Other studies confirm that intergranular attack in Fe-Ni-Cr alloys
is more severe than metal loss in molten chlorides [21]. Ferritic
steel has shown to oxidize more easily than austenitic steel in hot
chloride containing atmospheres [25]. However, recent research on a
group of metal alloys also showed acceptable resistance of
stainless steel 310 to molten carbonate salts at 750 °C [2]. Steel
alloys with around 20 wt% Cr and/or high nickel content show a
greater resistance to high-temperature corrosion [23, 26].
The stability of oxide layer should also be taken into
consideration when using carbon or stainless steels. Most metals
oxidize over a wide range of conditions at elevated temperatures.
Thus, oxidation rate and morphology are of importance to determine
the material lifetime. Corrosion resistance in many
high-temperature environments is achieved by the formation of a
protective oxide film on the material surface [27]. An early study
by Azzi et al. [28] showed that the corrosion rate of iron in
molten carbonate is limited by the diffusion of oxidizing species
through the corrosion products rather than in the melt; where FeO,
Fe2O3 and Fe3O4 were determined as the general corrosion products
formed on the material surface. Hot corrosion behavior of steel
alloys in molten Li2CO3-K2CO3 showed the formation of a porous
layer composed of Fe2O3 and LiFe5O8 on the material surface [29].
However, the usually thick Fe2O3 layer has shown not to be
protective in molten salt environments [30]. Al and Si also
contribute to the development of self-healing protective oxide
films on the alloy surface acting as diffusion barriers against
further oxidation [15]. Cr and Al oxides have also proved to be
more stable than those of iron and thus Cr2O3 and Al2O3 are
thermodynamically favored [31].
The compatibility of containment material with molten salt and
its stability is a concerning issue in TES systems [8]. The
selection of appropriate and optimum structural materials as
Thermal Energy Storage vessel, subject to corrosive molten salts as
PCM and high temperatures atmospheres, is essential in developing
economic and functionally efficient systems. This study will
examine the corrosion behavior of three commercial steel alloys in
two eutectic mixtures of molten salts for the next generation of
TES applications.
2. Experimental procedure
Austenitic stainless steel 316 (SS316), ferritic carbon steel
1008 (CS1008) and ferritic/austenitic duplex steel 2205 (DS2205)
were examined in two eutectic mixtures of molten salts. Table 1
summarizes the structures and nominal compositions of the alloys
used in the current study.
-
4
Sodium carbonate anhydrous LR (CAS No. 497-19-8), sodium
chloride AR (CAS No. 7647-14-5), potassium carbonate anhydrous LR
(CAS No. 584-08-7) and lithium carbonate 99% (CAS No. 554-13-2)
were placed for 24 h in a 180 °C furnace to dry and then were
measured and mixed according to Table 2. The eutectic mixtures
melting points and test temperatures are also provided in the
table. Test temperature for each salt was selected close to its
melting point based on the assumption that the salt will be used as
a PCM in a CST plant.
Table 1 Nominal elemental composition and crystal structure of
the studied alloys (wt%).
Table 2 Chemical composition, melting point and test temperature
of salt mixtures.
Salt mixture Composition (wt%) Melting point
(°C) [32] Test temperature
(°C) Chloride carbonate 40 NaCl + 60 Na2CO3 632 700 Ternary
carbonate 33.4 Na2CO3 + 32.1 Li2CO3+ 34.5 K2CO3 397 450
2.1. Electrochemical corrosion investigation
Electrochemical experiments were conducted using a
three-electrode cell containing the molten salts in alumina
crucibles open to air at 700 °C and 450 °C in a preheated
cylindrical furnace. Test coupons of 25 mm long, 5 mm wide and 1.4
mm thick were mechanically wet ground and polished down to 0.04 m
by colloidal silica, washed with ethanol and dried in air.
Measurements were implemented by means of a VMP3-based BioLogic
instrument controlled by EC-Lab® software. The three-electrode cell
was implemented with the polished sample as the working electrode
and two same sized platinum sheets (25×5×1 mm) as pseudo reference
and counter electrodes [33-36]. Samples were subjected to open
circuit potential (OCP), electrochemical impedance spectroscopy
(EIS) and potentiodynamic polarization (PDP) measurements in this
order to avoid sample deterioration. Equilibrations of potentials
(OCP) were carried out for 1 h immediately after immersion. EIS
measurements were then obtained using a frequency range of 100 kHz-
100 mHz with the amplitude of ±10 mV. Finally, PDP was conducted at
the potential scan rate of 10 mV/min and potential range of -400 to
+500 mV with respect to the open circuit potential. ZFit analysis
of EC-Lab software was used to fit successive impedance cycles.
2.2. Static corrosion
Fresh metal coupons were cut to around 25×7×1.4 mm for static
corrosion tests while the front sides were mechanically wet
polished down to 1 µm in colloidal silica using standard grinding
and polishing procedures. A schematic representation of the test
condition and alignment of samples in the furnace is shown in Fig.
1. Cylindrical alumina crucibles were used as salt vessels and the
furnace temperature was set to 700 ± 10 °C for chloride carbonate
and 450 ± 10 °C for ternary carbonate.
The salt containing vessels were placed into the furnace at room
temperature and then gradually heated up to the test temperature.
Once the salt melted and the chamber conditions stabilized, the
metal coupons were immersed so that the top half was exposed to the
air and bottom half submerged into the molten salt. Such a
configuration enabled us to make a comparison not only between the
impact of the salt and that of the oxygen from the environment, but
also the decomposition gases of the salt which are expected to
increase the corrosive impact in this atmosphere. Metal coupons
were removed after 120 h of exposure. All coupons were then mounted
exposing the side indicated in Fig. 1 into a conductive resin,
Alloy Structure Fe Ni Cr C Mo Mn S P SS316 fcc Bal 12 17.5 0.05
3 … … …
DS2205 fcc/bcc Bal 6.5 23 0.03 3.5 … … … CS1008 bcc Bal … … 0.14
… 0.5 0.04 0.04
-
5
ground and polished down to 0.04 µm from the side of the sample
in colloidal silica using standard procedures, washed with ethanol
and finally dried in air. Microscopy points in Fig. 1 provide us
with a view of the material close to the polished side. It is
worthy to note that although 120 h does not seem long enough to
study corrosion rate and related phenomena, aggressive nature of
molten salts makes studying short-term impacts like attack
morphology and corrosion mechanisms possible.
Fig. 1 Schematic of the corrosion vessel and a sample inside the
furnace.
2.3. Metallography samples
Small coupons of each alloy were selected and mechanically wet
ground and polished down to 0.04 μm by colloidal silica, washed
with ethanol and finally dried in air. These coupons were used to
study the corroded metal morphology and short-term impacts of the
molten salts on the microstructure. The coupons were submerged into
the molten salts for only 3 min and then were ultrasonically
cleaned for 15 min in demineralized water for corrosion product
residues to be removed from the surface. The samples were then
studied under an optic microscope from the surface.
2.4. Macro and micro-structural investigations 2.4.1. Optical
microscopy analysis
Optic microscopes model Leica DMi8A (magnification 1.25x-50x)
and Leica M125 (magnification 0.8x-10x) both equipped with Leica
Application Suite software were used to take macro and micro-images
for microstructural and corrosion observations.
2.4.2. SEM, EDS and EBSD analysis
Complimentary techniques like scanning electron microscopy
(SEM), energy-dispersive X-ray spectroscopy (EDS) and electron
back-scatter diffraction (EBSD) were employed for further
microstructural investigations using a field emission SEM (model:
JEOL 7001F, with automated feature detection equipped with
secondary electron, EDS analysis system, OXFORD EBSD pattern
analyzer and Channel 5 analysis software).
3. Results and discussion 3.1. Electrochemistry
OCP graphs of the alloy specimens in the studied molten salts
are represented in Fig. 2a. CS1008 shows stable OCP values over the
1 h of exposure in both salts after an initial rapid drop in
ternary carbonate. DS2205 and SS316 show more noble potential
values over time in chloride carbonate which could be
-
6
attributed to the gradual development of a film on the surface.
These alloys also show steady OCP values in ternary carbonate.
Fig. 2 (a) Open circuit potential, and (b) potentiodynamic
polarization curves of samples in molten chloride carbonate at 700
°C and ternary carbonate at 450 °C.
Ecorr values in PDP plots in Fig. 2b reflect OCP values in Fig.
2a. Data extracted from PDP curves are summarized in Tables 3 and 4
depicting the corrosion current density values in Fig. 3. SS316
represents the best performance and shows the lowest current
densities in each salt. Graphs show that DS2205 and SS316 perform
similarly in each salt with similar potential and comparable Icorr
values. Higher Icorr value of DS2205 compared to that of SS316 in
ternary carbonate could be attributed to the faster rates of
cathodic and anodic reactions according to Fig. 2b and Table 4. On
the other hand, maximum Icorr values belong to CS1008 in both
environments. Ecorr for CS1008 is comparable however, the
significant change in Icorr could indicate a change in mechanism
between the two molten salts.
Table 3 Extracted data from polarization plots in Fig. 2b for
the alloys in chloride carbonate at 700 ºC. Alloy Ecorr (mV)
Icorr (µA.cm-²)
Βa (mV/decade)
|Βc| (mV/decade)
SS316 -24.0 247.2 220.9 312.9DS2205 -35.6 293.5 239.4 360.6
CS1008 -920.5 1686.5 163.1 409.0
Table 4 Extracted data from polarization plots in Fig. 2b for
the alloys in ternary carbonate at 450 ºC. Alloy Ecorr (mV)
Icorr (µA.cm-²)
Βa (mV/decade)
|Βc| (mV/decade)
SS316 -538.5 5.7 55.9 142.6 DS2205 -547.1 29.7 100.3 299.8CS1008
-902.5 85.2 153.6 264.6
-
7
Fig. 3 Column charts of corrosion current density values in
chloride carbonate at 700 °C (a) and ternary carbonate at 450 °C
(b).
It is known from literature that molten carbonate corrosion
becomes less severe at lower temperatures and the corrosion
mechanism can vary widely based on temperature and salt composition
[21]. Regardless of the salt composition and test temperature,
early studies on the molten salt corrosion of metals express the
general corrosion reaction as a combination of metal dissolution,
partial anodic reaction and reduction of oxidants, and partial
cathodic reaction [37]. Liquid salts are ionic and their
interaction with metals is therefore electrochemical [38-40].
Consequently, the partial anodic reaction could be considered as
the oxidation of the metal [37, 39]:
M→Mn ne- Equation 1 2Mn nO2-→2MOn Equation 2 where M is the
transition metal like Fe, Cr and Ni. Partial cathodic reaction
follows the equation below [37]:
Ox ne-→R Equation 3 with n pointing to the number of electrons
while Ox and R refer to oxidant and reductant, respectively.
Corrosion mechanisms of metals in molten carbonate salts have
been earlier discussed in detail [41]. It has been also reported
that the solubility of a range of oxides in molten salts highly
depends on the melt basicity and so does the type of dissolution
mechanism [29]. Therefore, corrosion will be dependent upon the
solubility of the formed metal oxides on the material surface. Azzi
et al. [28] reported that oxygen can be dissolved in molten
Na2CO3-K2CO3 at 800 °C. Fe2+ ions were then reported to be formed
in the melt due to the dissolution of oxidizing species and the
simultaneous oxidizing of iron according to reaction 1. The
presence of oxygen causes the formation of peroxide ions in the
melt leading to higher corrosion rate. Oxygen participates
indirectly in the corrosion of iron by O peroxide formed via
equilibrium 4 and reduced to oxygen anions according to reaction 5
[28]:
12 O2 O2-↔O22- Equation 4 O22- 2e-→2O2- Equation 5 Early studies
[37, 42, 43] suggest that highly reactive peroxide ions O and
superoxide ions O2- could be considered as the basic component of
the solvent in molten salts, which oxidize the metal when reduced
on the metal surface. In O2 atmosphere, reaction 5 participates in
the corrosion process and
-
8
leads to the formation of iron oxide products like FeO, Fe2O3
and Fe3O4 [28, 29]. Therefore, high concentration of oxide ions in
the melt leads to the assumption that the carbonate molten salts
used in this study act as a basic melt implying that the corrosion
of the metal is under anodic control [37].
Kruizenga et al. [44], on the other hand, reported that stable
passivated oxide scales are not likely to form on structural alloys
in molten chloride salts. It is already known from literature that,
thermodynamically, alkali chlorides are more favored than
transition metal chlorides [15]. Consequently, as the common
alloying elements are not expected to reduce the salt, corrosion is
expected to be driven by oxidants like H2O, O2, H+ and other
impurities in molten chlorides [15, 37, 44]. Therefore, any
moisture content in the system could lead to the formation of HCl
and chlorine gas [15, 40, 45].
Despite of the presence of HCl in the system, Cl2 has been found
as the main corrosive medium because of its ability to penetrate
protective oxide scales and react with transition metals [40, 46].
Therefore, the production of chlorine gas and its reaction with
metal oxides on the surface could be defined as the dominant
corrosion mechanism in molten chlorides [40].
For further analyses, EIS was measured in the molten salts;
Nyquist and Bode-Phase plots are shown in Fig.4. Simulated
equivalent circuit (EC) is included in Fig. 4a and the data
extracted from the model is summarized in Table 5. Here RS, Rct and
Rox point to solution resistance, charge transfer (polarization)
resistance and transfer resistance of ions in the oxide film,
respectively. Qdl is the double layer Constant Phase Element (CPE),
Qox is the film CPE, n represents the CPE parameter and W
corresponds to Warburg resistance. As reported previously [41], CPE
points to the non-ideal capacitance of the surface and is used to
calculate the surface charge transfer capacitance during corrosion
[47-49].
In chloride carbonate at 700 ºC, the presence of CPE (Fig. 4a)
confirms the formation of scale layers and subsequent microscopic
roughness on the metal’s surfaces where the transfer of ions in the
scale is rate limiting [50, 51]. This is also in agreement with OCP
plots in Fig. 2a. However, the appearance of Warburg mass transfer
impedance indicates corrosion due to the non protective film on the
surface [51]. Warburg coefficient is inversely proportional to the
square root of diffusion coefficient [52]. Therefore, the highest W
coefficient for DS2205 means lowest diffusion through the surface
scale for this alloy compared to that of the SS316 with the highest
diffusion coefficient. Table 5 and Bode plots in Fig. 4b show the
highest impedance values for DS2205 at low frequencies. Therefore,
probably due to the high Cr content in the alloy composition, a
semi-protective film with very low diffusivity on the surface of
DS2205 should have been developed [29, 53]. In addition, as will be
shown later, alloy CS1008 shows the development of a very thick
oxide layer on the surface which does not provide protection
against further corrosion. According to Table 5, the diffusivity of
this layer should be quite low.
In ternary carbonate at 450 ºC, the EC model confirms the
formation of scale layers on the surface of the alloys (Fig. 4c).
Again, the appearance of Warburg mass transfer impedance casts a
shadow of doubt over the protective nature of the film and points
to high corrosion rates for all studied alloys. Once more, DS2205
shows the highest impedance values, according to Fig. 4d and Table
5, which could be a result of the formation of a semi-protective
film on the surface due to high Cr content in the alloy. In
agreement with its highest corrosion current density value, CS1008
shows the lowest impedance values while SS316 displays moderate
values. Contrary to the chloride carbonate salt, W coefficient
values here point to the lowest diffusivity through the surface
layer on SS316 versus the highest in DS2205. However, these values
are comparable in this ternary carbonate salt.
-
9
Fig. 4 (a) Nyquist, and (b) Bode-Phase plots resulted from
impedance measurements of the alloys in chloride carbonate at 700
°C, (c) Nyquist, and (d) Bode-Phase plots resulted from impedance
measurements of the alloys in ternary carbonate at 450 °C. Larger
Nyquist plots of CS1008, calculated equivalent circuit and fit data
points are also included in (a) and (c).
Table 5 EIS data extracted from equivalent circuit models in
Fig. 4.
3.2. Microscopy and microanalysis
Optical and SEM images of samples submerged for 3 min in molten
chloride carbonate at 700 ºC are presented in Fig. 5. An adhesive
layer was formed on DS2205 and CS1008 while the deposit on SS316
was physically detached during handling and washing. The film on
CS1008 could be removed after 15 min ultrasonic cleaning with the
film on DS2205 remaining attached to the surface, Figs. 5c and
f.
Molten Salt Alloy
Rs Rct Rox Qdl ndl
Qox nox
W (Ω.cm2) (Ω.cm2) (Ω.cm2) (F.sn-1.cm-2) (F.sn-1.cm-2)
(Ω.s-1/2)
Chloride carbonate
SS316 1.03 1.60 9.72 0.19 0.47 0.05 1.00 8.96
DS2205 0.62 32.1 40.1 0.05 0.64 0.05 0.68 1.1e+6
CS1008 0.97 0.32 10.95 0.11 0.57 0.43 0.49 6.9e+5
Ternary carbonate
SS316 1.61 3.10 204.38 1.24e-03 0.76 3.12e-03 0.79 69.01
DS2205 1.23 543.85 5.76 3.68e-03 0.77 8.15e-04 0.86 31.02
CS1008 1.21 3.08 2.86 3.50e-03 0.50 5.78e-03 0.30 44.27
-
10
Grain boundary (GB) attack seems to be the dominant corrosion
mechanism for all studied alloys at 700 °C. Even on DS2205, surface
contamination seems to follow grain boundaries, Fig. 5f. This is
more obvious in the SEM image in Fig. 5e where salt residues follow
GBs on alloy CS1008. However, as will be shown later, the oxidation
is so severe on CS1008 that after long exposure times the
morphology looks like uniform rather than GB attack. Further
detailed microstructural observations after long-term exposure to
each eutectic salt are provided later.
Fig. 5 Optical (a to c) and SEM (d to f) images of as-polished
samples after 3 min exposure to the molten chloride carbonate at
700 °C. Inserted EBSD phase color map of as-received material and
parallel features in (c) point to the alternate layers of
ferrite/austenite in the texture of the duplex steel 2205 corroded
with different rates.
3.2.1. SS316
An optical image of a coupon of SS316 subjected to chloride
carbonate at 700 °C for 120 h is shown in Fig. 6a. Hanging the
sample over the salt bath resulted in the top half being exposed to
air, salt vapor and potentially salt creep and the bottom half to
the molten salt. Figs. 6b and c compare and contrast top and bottom
halves of the sample, respectively. Images confirm GB attack in
both environments. In the absence of a reference surface and
ignoring the material already removed through corrosion, attack
above the salt penetrates around 54.4 µm towards the bulk material
(Fig. 6b) versus 35.1 µm at the bottom part under the salt level
(Fig. 6c). No adherent film is detectable on the top part while a
semi-protective deposit seems to be formed under the salt level.
However, no film could be observed at the bottom side of the sample
under the salt where the penetration depth of 57.5 µm is close to
that of the top part exposed to air. Hence, lower GB degradation
depth under the salt level could be attributed to the formation of
a semi-protective film on the surface as previously predicted by
electrochemistry.
-
11
Fig. 6 (a) Macro-graph of a SS316 sample hung over the molten
chloride carbonate at 700 °C for 120 h, (b) optical image of the
alloy cross-section from sidewall above the molten salt, (c)
optical image of the alloy cross-section from sidewall under the
salt level.
Further SEM-EDS analyses in Figs. 7 and 8 clarify the behavior.
54.4 µm penetration depth above the molten salt accompanies Fe and
Cr depletion from GBs and the following GB oxidation. The lower
penetration depth under the salt level also follows the formation
of GB oxides, Fig. 8. Therefore, GB degradation due to the
de-alloying of Fe and Cr from GBs and formation of Fe-Cr oxides
could be considered as the main corrosion mechanism which, in the
presence of the molten salt, is hindered owing to the development
of a semi-protective film on the surface. This is in agreement with
electrochemical measurements reported previously. The reduced
amount of oxygen in the salt, compared to the above atmosphere,
could have also contributed. However, as previously shown in Figs.
5a and d and according to the high diffusivity previously observed
through the surface layer, the oxide film does not seem adherent as
it might have been detached from the bottom side in the salt and
also from the top side exposed to air due to gravity or during
handling.
-
12
Fig. 7 (a) SEM image, and (b) to (f) its corresponding EDS map
analysis of an area on the top corner of SS316 above the chloride
carbonate salt level and exposed to air for 120 h at 700 °C.
Fig. 8 (a) SEM image, and (b) to (f) its corresponding EDS map
analysis of an area on the bottom corner of SS316 submerged into
the chloride carbonate salt for 120 h at 700 °C.
Fig. 9 shows the formation of a continuous and adherent
Fe-Ni-Cr-K oxide on the material surface in ternary carbonate at
450 °C. No localized de-alloying and oxidation/corrosion attack is
detectable on the sample without any noticeable discrepancy between
the attack depth above and below the salt level. Even the possible
formation of lithium oxide, according to the fact that Li is hard
to be detected by EDS [54], does not seem to have contributed to
corrosion under the salt level. Thus, it could be concluded that
the oxide layer formed on the material surface acts as a solid
barrier against further corrosion in contact with the molten salt
at 450 °C.
-
13
Fig. 9 (a) Optical image of the top corner of SS316 above the
ternary carbonate salt and exposed to air at 450 °C for 120 h, (b)
SEM image of the red rectangular area in (a), (c) its corresponding
EDS map analysis, (d) optical image of the bottom corner of SS316
under the ternary salt at 450 °C for 120 h, (e) SEM image of the
red rectangular area in (d), and (f) its corresponding EDS map
analysis.
3.2.2. CS1008
Degradation of the ferritic carbon steel CS1008 above the
chloride carbonate salt at 700 °C for 120 h accompanies the
formation of a thick layer of iron oxide on the surface, Figs. 10a
and b. The deposit seems to consist of two separate oxide layers
with the total thickness of 1.16 mm on the sample surface which was
exposed above the salt level. Although a similar oxide seems to
have been developed on the material surface under the salt level,
its total thickness of 0.80 mm is 31% less than that of the above;
Fig. 10c. On the other hand, 0.80 mm thickness of the metal
remained intact after 120 h exposure is 21% higher than that above
the salt (0.63 mm). EDS analysis confirms the penetration of
chlorine through the film and subsequent chlorine attack as
discussed previously. Therefore, oxidation seems as the main
corrosion behavior for CS1008 in contact with chloride carbonate
salt at 700 °C while the molten salt here seems to have played a
protective role against further oxidation probably due to the lower
O2 in the salt.
Corrosion of CS1008 in ternary carbonate at 450 °C for 120 h is
shown in Fig. 11. According to Fig. 11a, a 50.5 µm thick and
adherent film of iron oxide has been formed on the alloy surface
for the portion of the sample above the molten salt. The bottom
half of the metal submerged into the molten salt shows an adherent
film which appears thinner; Figs. 11c and d. EDS analysis shows the
similar film composition of iron oxide formed under the salt
compared to the above. Therefore, oxidation could be defined as the
main corrosion attack in the ternary carbonate at 450 °C with the
salt decelerating the development of the oxide layer.
-
14
Fig. 10 (a) SEM image of an area on the top half of CS1008 above
the chloride carbonate salt exposed to air for 120 h at 700 °C, (b)
SEM-EDS map analysis corresponding to the red rectangular area in
(a), (c) SEM image of an area on the bottom half of CS1008 under
the chloride carbonate salt for 120 h at 700 °C, (d) SEM-EDS map
analysis corresponding to the red rectangular area in (c).
Fig. 11 (a) Optical image of an area on the top corner of CS1008
above the ternary carbonate salt at 450 °C for 120 h, (b) SEM-EDS
map analysis corresponding to the red rectangular area in (a), (c)
optical image of an area on the bottom corner of CS1008 submerged
into the ternary carbonate salt at 450 °C for 120 h, (d) SEM-EDS
map analysis corresponding to the red rectangular area in (c).
-
15
3.2.3. DS2205
Although the corrosion current density values of alloy DS2205 is
closer to that of the SS316, this dual phase ferritic-austenitic
steel inherits the combined behavior of SS316 and CS1008 regarding
the corroded metal morphology in the studied molten salt
environments. Fig. 12 shows oxidation attack of ferrite above the
chloride carbonate salt at 700 °C for 120 h. Despite the outcomes
of electrochemical measurements, no surface deposit is detectable
on the alloy surface above and under the salt level, Figs. 12a and
d. In fact, as shown in Figs. 5c and f, a surface film is likely to
form on the alloy surface in contact with the salt. However, the
film could have been later dissolved as a result of the its
solubility in the salt (or its vapors) due to the acidic nature of
the molten chloride as discussed earlier [40].
EBSD phase and elemental analysis in Fig. 12c show selective
oxidation of ferrite, rich in Ni-Cr-Mo, while the alternate
austenite phase remained relatively intact. Optical and EBSD images
in Figs. 12d to f show the same behavior on the sample under the
salt level.
Fig. 12 (a) Optical image from top corner of DS2205 above the
chloride carbonate salt and exposed to air for 120 h at 700 °C, (b)
EBSD band contrast image of the red rectangular area in (a), (c)
EBSD phase and elemental analysis maps corresponding to (b), (d)
optical image from bottom corner of DS2205 under the chloride
carbonate salt for 120 h at 700 °C, (e) EBSD band contrast image of
the red rectangular area in (d), (f) EBSD phase and elemental
analysis maps corresponding to (e).
Again, ignoring the material already removed through corrosion,
the oxidation depth from the top corner above the salt is around
23.2 µm in the direction normal to alternate austenite/ferrite
phases (rolling direction) while it extends roughly twice that
distance (up to 43.1 µm) on the surface parallel to the rolling
direction; Figs. 13a and b. In fact, oxidation proceeds faster
along ferrite while austenite acts as a barrier against further
oxidation and protects the underlying ferrite layer. However, the
presence of austenite as a more noble phase at the vicinity of
ferrite (ref. Fig. 2a) might have contributed to faster oxidation
due to the galvanic corrosion of ferrite which is less noble than
austenite [25, 41, 55, 56]. This is similar to that part of the
metal under the salt level, Figs. 13c and d. The 20.5 µm
penetration depth from the edge is only 11.6% lower than that of
the above. However, compared to above the salt level (Fig. 13b), it
seems that the limited solubility of oxygen in the salt has reduced
the oxidation by more than 50% parallel to the rolling direction
below the salt.
-
16
Fig. 13 (a) and (b) SEM images from the side edge and top edge,
respectively, on the top part of DS2205 above the chloride
carbonate salt exposed to air for 120 h at 700 °C, (c) and (d) SEM
images from the side edge and bottom edge, respectively, on the
bottom part of DS2205 under the chloride carbonate salt for 120 h
at 700 °C.
Fig. 14 shows the top part of alloy 2205 above the ternary
carbonate salt after 120 h at 450 °C suffered from oxidation on the
surface. The degradation depth is around 22 µm. On the other hand
and having in mind that the sample surface was polished before
exposure, a sharp edge is detectable in Fig. 15 for the front side
of the sample in direct contact with the molten salt. That
sharpness indicates that the metal remained intact under the salt.
Comparing Fig. 15 to the previous observations, it seems that 120 h
was not long enough for considerable corrosion attack in the
ternary carbonate salt at 450 °C. This also supports the salt’s
protection role because of the reduced availability of O2 under the
salt.
Fig. 14 (a) SEM image from the top part of DS2205 above the
ternary carbonate salt exposed to air at 450 °C for 120 h, (b) EDS
map analysis corresponding to (a).
-
17
Fig. 15 SEM image from the bottom part of DS2205 below the
ternary carbonate salt at 450 °C for 120 h.
Conclusion
Corrosion behavior of three commercial alloys including
stainless steel 316, duplex steel 2205 and carbon steel 1008 in two
eutectic mixtures of NaCl+Na2CO3 at 700 °C and Li2CO3 + K2CO3 +
Na2CO3 at 450 °C in air were studied as candidates for containment
materials. A combination of optical microscopy, electrochemical
measurements, SEM, EDS and EBSD techniques were employed to
characterize the degradation mechanisms. Results are summarized as
below.
Electrochemical measurements showed the most anodic potential
values for alloy 1008 in both molten salt environments. Alloys 316
and 2205 showed more noble potential values with the same
susceptibility to each molten salt. Impedance spectroscopy
suggested the formation of films on the surface of the studied
alloys in both molten salts which was not confirmed for alloy 2205
in chloride carbonate. And alloy 316 showed the lowest corrosion
current densities in each molten salt.
Oxidation was found as the primary attack to the alloys in both
molten salt environments. However, as the availability of O2
controls the degree of oxidation, both molten salts proved to slow
the oxidation down when the materials were submerged.
The corroded metal morphology was grain boundary oxidation for
all studied alloys in chloride carbonate at 700 °C. Corrosion in
ternary carbonate at 450 °C was of uniform morphology on the
surface of the alloys with no localized degradation.
Degradation of alloy 1008 in both environments involved the
formation of a thick layer of iron oxide on the surface. The grain
boundary oxidative attack was so severe on this alloy at 700 °C
that it looked like uniform corrosion after long exposure
times.
Depletion of alloying elements from grain boundaries at 700 °C
contributed to intergranular attack in alloys 316 and 2205.
Formation of a semi-protective oxide on the surface of alloy 316
provided more protection against chloride carbonate at 700 °C. A
continuous and adherent oxide layer was also observed on the
surface of this alloy at 450 °C. In alloy 2205, selective oxidation
of ferrite was observed at 700 °C with no adherent film detected on
the surface. The austenite phase in this alloy appeared to help
protect the underlying ferrite layers at 700 °C.
Acknowledgement
This work was funded by the Australian Solar Thermal Research
Initiative (ASTRI), which is supported by the Australian Government
via the Australian Renewable Energy Agency (ARENA). The authors
would also like to thank AINSE Ltd for providing financial
assistance (Award-PGRA-2016) to enable
-
18
work on the reported topic. The data reported in the paper were
obtained at the Central Analytical Research Facility (CARF)
operated by the Institute for Future Environments at Queensland
University of Technology (QUT). Access to CARF was supported by
generous funding from the Science and Engineering Faculty, QUT.
References
[1] M. Liu, N.H.S. Tay, S.
Bell, M. Belusko, R. Jacob, G.
Will, W. Saman, F. Bruno,
Review
on concentrating solar power plants and new developments in high temperature thermal energy storage technologies, Renew Sust Energ Rev, 53 (2016) 1411‐1432. [2] J.C. Gomez‐Vidal, J. Noel, J. Weber, Corrosion evaluation of alloys and MCrAlX coatings in molten carbonates for thermal solar applications, Solar Energy Materials and Solar Cells, 157 (2016) 517‐525. [3] J.C. Gomez‐Vidal, E. Morton, Castable cements to prevent corrosion of metals in molten salts, Solar Energy Materials and Solar Cells, 153 (2016) 44‐51. [4] M. Sarvghad, S. Bell, R. Raud, T.A. Steinberg, G. Will, Stress assisted oxidative failure of Inconel 601 for thermal energy storage, Solar Energy Materials and Solar Cells, 159 (2017) 510‐517. [5] W. Guo, Y. Wu, J. Zhang, S. Hong, L. Chen, Y. Qin, A Comparative Study of Cyclic Oxidation and Sulfates‐Induced Hot Corrosion Behavior of Arc‐Sprayed Ni‐Cr‐Ti Coatings at Moderate Temperatures, Journal of Thermal Spray Technology, 24 (2015) 789‐797. [6]
J. Gomez, High‐Temperature Phase
Change Materials (PCM) Candidates for
Thermal
Energy Storage (TES) Applications, in, National Renewable Energy Laboratory (NREL), Golden, CO., 2011. [7] M.M. Kenisarin, High‐temperature phase change materials for thermal energy storage, Renew Sust Energ Rev, 14 (2010) 955‐970. [8]
S. Kuravi, J. Trahan, D.Y.
Goswami, M.M. Rahman, E.K. Stefanakos,
Thermal energy
storage technologies and systems for concentrating solar power plants, Progress in Energy and Combustion Science, 39 (2013) 285‐319. [9]
L.F. Cabeza, A. Gutierrez, C. Barreneche, S. Ushak, A.G. Fernandez, A.I. Fernadez, M. Grageda, Lithium in thermal energy storage: A state‐of‐the‐art review, Renew Sust Energ Rev, 42 (2015) 1106‐1112. [10] T. Bauer, N. Pfleger, D. Laing, W.‐D. Steinmann, M. Eck, S. Kaesche, 20 ‐ High‐Temperature Molten Salts for Solar Power Application, in: F.L. Groult (Ed.) Molten Salts Chemistry, Elsevier, Oxford, 2013, pp. 415‐438. [11] K. Lovegrove, W.S. Csiro, 1 ‐ Introduction to concentrating solar power (CSP) technology, in: K. Lovegrove, W. Stein (Eds.) Concentrating Solar Power Technology, Woodhead Publishing, 2012, pp. 3‐15. [12] K. Lovegrove, J. Pye, 2 ‐ Fundamental principles of concentrating solar power (CSP) systems, in: K. Lovegrove, W. Stein (Eds.) Concentrating Solar Power Technology, Woodhead Publishing, 2012, pp. 16‐67. [13] K. Federsel, J. Wortmann, M. Ladenberger, High‐temperature and corrosion behavior of nitrate nitrite molten
salt mixtures regarding
their application in concentrating
solar power plants, Enrgy Proced, 69 (2015) 618‐625. [14] B.A.T. Mehrabadi, J.W. Weidner, B. Garcia‐Diaz, M. Martinez‐Rodriguez, L. Olson, S. Shimpalee, Multidimensional Modeling of Nickel Alloy Corrosion inside High Temperature Molten Salt Systems, Journal of The Electrochemical Society, 163 (2016) C830‐C838. [15] K. Sridharan, T.R. Allen, 12 ‐ Corrosion in Molten Salts, in: F.L. Groult (Ed.) Molten Salts Chemistry, Elsevier, Oxford, 2013, pp. 241‐267. [16] L.C. Olson, Materials corrosion
in molten LiF‐NaF‐KF eutectic salt,
in, University of Wisconsin‐‐Madison, 2009. [17] L.C. Olson, J.W. Ambrosek, K. Sridharan, M.H. Anderson, T.R. Allen, Materials corrosion in molten LiF‐NaF‐KF salt, Journal of Fluorine Chemistry, 130 (2009) 67‐73.
-
19
[18] R.B. Rebak, 7 ‐ Stress corrosion cracking (SCC) of nickel‐based alloys, in: V.S. Raja, T. Shoji (Eds.) Stress Corrosion Cracking, Woodhead Publishing, 2011, pp. 273‐306. [19] R.B. Rebak, Environmentally
assisted cracking of nickel
alloys —a review, in:
S.A.S.H.J.M.O.B. Rebak (Ed.) Environment‐Induced Cracking of Materials, Elsevier, Amsterdam, 2008, pp. 435‐446. [20] F.J. Ruiz‐Cabañas, C. Prieto, R. Osuna, V. Madina, A.I. Fernández, L.F. Cabeza, Corrosion testing device
for in‐situ corrosion characterization
in operational molten salts storage
tanks: A516 Gr70 carbon steel performance under molten salts exposure, Solar Energy Materials and Solar Cells, 157 (2016) 383‐392. [21] Molten Salt Corrosion, in: G.Y. Lai (Ed.) High‐temperature corrosion and materials applications, ASM International, 2007, pp. 409–421. [22]
M.S. Sohal, M.A. Ebner, P.
Sabharwall, P. Sharpe, Engineering
database of liquid
salt thermophysical and thermochemical properties, in, Idaho National Laboratory, Idaho Falls, 2010. [23] J. Young, Chapter 1 The Nature of High Temperature Oxidation, in: Y. David John (Ed.) Corrosion Series, Elsevier Science, 2008, pp. 1‐27. [24] M. Groll, O. Brost, D. Heine, Corrosion of steels in contact with salt eutectics as latent heat storage materials: Influence of water and other impurities, Heat Recovery Systems and CHP, 10 (1990) 567‐572. [25] M.A. Uusitalo, P.M.J. Vuoristo, T.A. Mäntylä, High temperature corrosion of coatings and boiler steels in reducing chlorine‐containing atmosphere, Surface and Coatings Technology, 161 (2002) 275‐285. [26] A. Schütz, M. Günthner, G. Motz, O. Greißl, U. Glatzel, High temperature
(salt melt) corrosion tests with ceramic‐coated steel, Materials Chemistry and Physics, 159 (2015) 10‐18. [27] A.S. Khanna, Introduction to high temperature oxidation and corrosion, ASM international, 2002. [28] M. Azzi, J.J. Rameau, Corrosion
in molten Na2CO3‐K2CO3 at 800° C—I. Effect of oxygen partial pressure on iron corrosion, Corrosion Science, 24 (1984) 935‐944. [29] T. Tzvetkoff, A. Girginov, M. Bojinov, Corrosion of nickel, iron, cobalt and their alloys in molten salt electrolytes, Journal of Materials Science, 30 (1995) 5561‐5575. [30] D. Bankiewicz, P. Yrjas, M. Hupa, High‐Temperature Corrosion of Superheater Tube Materials Exposed to Zinc Salts†, Energy & Fuels, 23 (2009) 3469‐3474. [31] J. Young, Chapter 5 Oxidation of Alloys
I: Single Phase Scales,
in: Y. David John (Ed.) Corrosion Series, Elsevier Science, 2008, pp. 185‐246. [32]
Collection of Phase Diagrams, in,
FTsalt ‐ FACT Salt Phase
Diagrams, http://www.crct.polymtl.ca/fact/documentation/FTsalt/FTsalt_Figs.htm, 2016. [33] A.I. Bhatt, G.A. Snook, Reference Electrodes for
Ionic Liquids and Molten Salts,
in: G.
Inzelt, A. Lewenstam, F. Scholz (Eds.) Handbook of Reference Electrodes, Springer Berlin Heidelberg, 2013, pp. 189‐227. [34] G. Inzelt, Pseudo‐reference Electrodes, in: G. Inzelt, A. Lewenstam, F. Scholz (Eds.) Handbook of Reference Electrodes, Springer Berlin Heidelberg, 2013, pp. 331‐332. [35]
G.Z. Chen, D.J. Fray, T.W.
Farthing, Direct electrochemical reduction
of titanium dioxide
to titanium in molten calcium chloride, Nature, 407 (2000) 361‐364. [36] T. Nohira, K.
Yasuda, Y. Ito, Pinpoint and bulk
electrochemical reduction of insulating
silicon dioxide to silicon, Nature Materials, 2 (2003) 397‐401. [37] A. Nishikata, H. Numata, T. Tsuru, Electrochemistry of Molten‐Salt Corrosion, Mat Sci Eng a‐Struct, 146 (1991) 15‐31. [38] R.V. Carter, D.L. Douglass, F. Gesmundo, Kinetics and mechanism of
the
sulfidation of Fe‐Mo alloys, Oxidation of Metals, 31 (1989) 341‐367. [39] J. Young, Chapter 8 Corrosion by Sulfur, in: Y. David John (Ed.) Corrosion Series, Elsevier Science, 2008, pp. 361‐396. [40] S.N. Liu, Z.D. Liu, Y.T. Wang, J. Tang, A comparative study on the high temperature corrosion of TP347H stainless steel, C22 alloy and
laser‐cladding C22 coating
in molten chloride salts, Corrosion Science, 83 (2014) 396‐408.
-
20
[41] M. Sarvghad, T. Chenu, G. Will, Comparative interaction of cold‐worked versus annealed inconel 601 with molten carbonate salt at 450°C, Corrosion Science, 116 (2017) 88‐97. [42] M.L. Orfield, D.A. Shores, Solubility of NiO in Molten Li2 CO 3, Na2 CO 3, K 2 CO 3, and Rb2 CO 3 at 910°C, Journal of The Electrochemical Society, 135 (1988) 1662‐1668. [43] A. Nishikata,
S. Haruyama, Electrochemical Studies
of the Corrosion of Ni and
Fe
in Molten Carbonate J. Japan Inst. Metals,, 48 (1984) 720‐725. [44]
A.M. Kruizenga, Corrosion mechanisms
in chloride and carbonate salts,
in:
Sandia National Laboratories, Livermore, CA Report No. SAND2012‐7594, 2012. [45] H.J. Grabke, E. Reese, M. Spiegel, The effects of chlorides, hydrogen chloride, and sulfur dioxide in the oxidation of steels below deposits, Corrosion Science, 37 (1995) 1023‐1043. [46] J.M. Abels, H.H. Strehblow, A surface analytical approach to the high temperature chlorination behaviour of Inconel 600 at 700 degrees C, Corrosion Science, 39 (1997) 115‐132. [47] K. Surekha, B.S. Murty, K.P. Rao, Effect of processing parameters on the corrosion behaviour of friction stir processed AA 2219 aluminum alloy, Solid State Sciences, 11 (2009) 907‐917. [48] W.F. Xu, J.H. Liu, H.Q. Zhu, Pitting corrosion of friction stir welded aluminum alloy thick plate in alkaline chloride solution, Electrochimica Acta, 55 (2010) 2918‐2923. [49] C. Shen, J. Zhang, J. Ge, Microstructures and electrochemical behaviors of the friction stir welding dissimilar weld, J Environ Sci (China), 23 Suppl (2011) S32‐35. [50] X.X. Sheng, Y.P. Ting, S.A. Pehkonen, The influence of sulphate‐reducing bacteria biofilm on the corrosion of stainless steel AISI 316, Corrosion Science, 49 (2007) 2159‐2176. [51] C.L. Zeng, W. Wang, W.T. Wu,
Electrochemical impedance models
for molten salt
corrosion, Corrosion Science, 43 (2001) 787‐801. [52] F. Ziebert, D. Lacoste, A Planar Lipid Bilayer in an Electric Field, in, 2011, pp. 63‐95. [53] Y.F. Yan, X.Q. Xu, D.Q. Zhou, H. Wang, Y. Wu, X.J. Liu, Z.P. Lu, Hot corrosion behaviour and
its mechanism of a new alumina‐forming austenitic stainless steel in molten sodium sulphate, Corrosion Science, 77 (2013) 202‐209. [54] P. Hovington, V. Timoshevskii, S. Burgess, H. Demers, P. Statham, R. Gauvin, K. Zaghib, Can we detect Li K X‐ray
in
lithium compounds using energy dispersive spectroscopy?, Scanning, 38
(2016) 571‐578. [55] A. Davoodi,
Z. Esfahani, M.
Sarvghad, Microstructure and corrosion
characterization of
the interfacial region in dissimilar friction stir welded AA5083 to AA7023, Corrosion Science, 107 (2016) 133‐144. [56] M. Sarvghad‐Moghaddam, R. Parvizi, A. Davoodi, M. Haddad‐Sabzevar, A. Imani, Establishing a correlation between interfacial microstructures and corrosion initiation sites in Al/Cu joints by SEM–EDS and AFM–SKPFM, Corrosion Science, 79 (2014) 148‐158.
Table of figures
Fig. 1 Schematic of the corrosion vessel and a sample inside the
furnace. Fig. 2 (a) Open circuit potential, and (b)
potentiodynamic polarization curves of samples in molten chloride
carbonate at 700 °C and ternary carbonate at 450 °C. Fig. 3
Column charts of corrosion current density values in chloride
carbonate at 700 °C (a) and ternary carbonate at 450 °C
(b). Fig. 4 (a) Nyquist, and (b) Bode-Phase plots resulted
from impedance measurements of the alloys in chloride carbonate at
700 °C, (c) Nyquist, and (d) Bode-Phase plots resulted from
impedance measurements of the alloys in ternary carbonate at 450
°C. Larger Nyquist plots of CS1008, calculated equivalent circuit
and fit data points are also included in (a) and (c). Fig. 5
Optical (a to c) and SEM (d to f) images of as-polished samples
after 3 min exposure to the molten chloride carbonate at 700 °C.
Inserted EBSD phase color map of as-received material and parallel
features in (c) point to the alternate layers of ferrite/austenite
in the texture of the duplex steel 2205 corroded with different
rates.
-
21
Fig. 6 (a) Macro-graph of a SS316 sample hung over the molten
chloride carbonate at 700 °C for 120 h, (b) optical image of the
alloy cross-section from sidewall above the molten salt, (c)
optical image of the alloy cross-section from sidewall under the
salt level. Fig. 7 (a) SEM image, and (b) to (f) its
corresponding EDS map analysis of an area on the top corner of
SS316 above the chloride carbonate salt level and exposed to air
for 120 h at 700 °C. Fig. 8 (a) SEM image, and (b) to (f) its
corresponding EDS map analysis of an area on the bottom corner of
SS316 submerged into the chloride carbonate salt for 120 h at 700
°C. Fig. 9 (a) Optical image of the top corner of SS316 above
the ternary carbonate salt and exposed to air at 450 °C for 120 h,
(b) SEM image of the red rectangular area in (a), (c) its
corresponding EDS map analysis, (d) optical image of the bottom
corner of SS316 under the ternary salt at 450 °C for 120 h, (e) SEM
image of the red rectangular area in (d), and (f) its corresponding
EDS map analysis. Fig. 10 (a) SEM image of an area on the top
half of CS1008 above the chloride carbonate salt exposed to air for
120 h at 700 °C, (b) SEM-EDS map analysis corresponding to the red
rectangular area in (a), (c) SEM image of an area on the bottom
half of CS1008 under the chloride carbonate salt for 120 h at 700
°C, (d) SEM-EDS map analysis corresponding to the red rectangular
area in (c). Fig. 11 (a) Optical image of an area on the top
corner of CS1008 above the ternary carbonate salt at 450 °C for 120
h, (b) SEM-EDS map analysis corresponding to the red rectangular
area in (a), (c) optical image of an area on the bottom corner of
CS1008 submerged into the ternary carbonate salt at 450 °C for 120
h, (d) SEM-EDS map analysis corresponding to the red rectangular
area in (c). Fig. 12 (a) Optical image from top corner of
DS2205 above the chloride carbonate salt and exposed to air for 120
h at 700 °C, (b) EBSD band contrast image of the red rectangular
area in (a), (c) EBSD phase and elemental analysis maps
corresponding to (b), (d) optical image from bottom corner of
DS2205 under the chloride carbonate salt for 120 h at 700 °C, (e)
EBSD band contrast image of the red rectangular area in (d), (f)
EBSD phase and elemental analysis maps corresponding to
(e). Fig. 13 (a) and (b) SEM images from the side edge and top
edge, respectively, on the top part of DS2205 above the chloride
carbonate salt exposed to air for 120 h at 700 °C, (c) and (d) SEM
images from the side edge and bottom edge, respectively, on the
bottom part of DS2205 under the chloride carbonate salt for 120 h
at 700 °C. Fig. 14 (a) SEM image from the top part of DS2205
above the ternary carbonate salt exposed to air at 450 °C for 120
h, (b) EDS map analysis corresponding to (a). Fig. 15 SEM
image from the bottom part of DS2205 below the ternary carbonate
salt at 450 °C for 120 h.