Professional Documents
Culture Documents
art ic l e i nf o a b s t r a c t
Article history: Molten nitrate salts (KNO3–NaNO3) are used extensively as heat transfer fluids in concentrated solar
Received 13 August 2015 power (CSP) applications. The oxidative character of nitrate salts at high temperature conditions leads to
Received in revised form challenges in the corrosion protection of CSP installations in contact with molten salts. This work reports
6 October 2015
efforts to protect low-cost ferritic–martensitic P91 (Cr9–Mo1) and austenitic stainless steel 304SS
Accepted 13 November 2015
materials against molten salt corrosion using slurry aluminizing. The gravimetric results show a sig-
nificant improvement for aluminized alloys up to 1700 h immersion in molten nitrate salt at 600 °C. The
Keywords: extent of protection in P91 becomes comparable with the most competitive resistant materials e.g.
Nitrate salt Inconel 625. Metallographic investigations and scanning electron micrographs reveal no sign of corrosion
Concentrated solar power
in the substrates beneath the aluminide coatings during the immersion test. Electron probe micro-
P91
analysis results reveal Na(Fe,Al)O2 as the protective corrosion product forming on the surface of both
Aluminizing
Slurry aluminized steels.
Solar salt & 2015 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.solmat.2015.11.024
0927-0248/& 2015 Elsevier B.V. All rights reserved.
A. Soleimani Dorcheh, M.C. Galetz / Solar Energy Materials & Solar Cells 146 (2016) 8–15 9
2Cr þ3O2 2Cr2 O3 þ 6e ð5Þ Rectangular coupons of ferritic–martensitic steel P91 and
stainless steel 304 with 25 10 4 mm3 dimensions were cut by
Feþ Cr2 O3 þO2 2FeCr2 O4 þ2e ð6Þ an electric discharge machine. Coupons were ground to 320 grit
SiC paper, sand blasted, degreased with acetone in an ultrasonic
Moreover, commercial salts contain levels of impurities which bath, and finally dried in air. The compositions of the studied
can accelerate the rate of corrosion. Thus, the mechanism of cor- alloys were obtained using optical emission spectroscopy (OES)
rosion has to be investigated and the life time of these materials (see Table 2).
has to be correctly estimated.
The performance of different materials in molten salt corrosion 2.2. Slurry aluminizing procedure
at different temperature ranges has been studied [12,14]. It has
been found that high chromium Ni-base alloys (e.g. IN625) and Two groups of specimens from both steels were used for the
high chromium austenitic steels (e.g. 304 and 316) have lower slurry coating process. To prepare the coating slurry, aluminum
corrosion rates in molten salt compared to the low chromium powder was mixed with polyvinyl alcohol and distilled water. The
steels [11–13,15–17]. The influence of chromium on molten salt slurry was air brushed onto the specimen's surface. Dried samples
oxidation behavior is controversial. It is known that high chro- were heat treated at 300 °C for 4 h (to cure the binder) and then
mium alloyed steels show higher corrosion protection via forma- heat treated at 700 °C for 10 h under an Ar atmosphere. After heat
tion of a slow growing continuous chromium oxide or chromate treatment samples were washed and weighed before being used
spinel layer [4,12]. On the other hand, depletion of chromium in in the corrosion test.
the substrates and Cr-dissolution in molten salt (however slow) Two different salt mixtures composed of NaNO3 (60 wt%) and
has been evident [13]. In addition, alloys with high chromium KNO3 (40 wt%) were used for the corrosion test. The salt mixtures
contents are usually more expensive than the low-chromium were different in the level of impurities, primarily the chloride
10 A. Soleimani Dorcheh, M.C. Galetz / Solar Energy Materials & Solar Cells 146 (2016) 8–15
content. Details of the impurity levels in the salts are listed in products and the microstructural evolution of the coating were stu-
Table 3. died using electron probe microanalysis.
Corrosion tests were conducted for non-coated and aluminized
alloys under similar conditions. The corrosion behavior of alloy-
steel P91 was studied in salt A and that of austenitic steel 304SS in 3. Results and discussion
salt B with higher chloride levels.
The corrosion test was conducted according to ISO/NP 17245 – 3.1. Microstructure of the aluminized steels
Corrosion of Metals and Alloys – Test method for high-temperature cor-
rosion testing of metallic materials by immersing in molten salt or other Former studies on aluminized low-chromium ferritic steels
inorganic liquids. The immersion test was conducted on 5 coupons of (e.g. P91 and P92) illustrated their promising oxidation perfor-
each alloy immersed in molten salt at 600 °C under flowing laboratory mance in steam environments [18,34]. Aluminum enrichment of
air for up to 2500 h. Only one sample was removed after each time the sub-surface region in steel substrates results in the formation
of Al-rich intermetallic phases which are prone to the formation of
interval (100, 500, and 1000, 1700, and 2500 h) and the weight change
a protective alumina scale [18,22].
kinetics of oxidation was investigated. The chemistry of corrosion
After the heat treatment process, an aluminide coating with a
thickness of 80 mm formed on the P91 ferritic substrate. The
structure of the aluminide coating is already known on P91 sub-
Table 2
Chemical composition of the studied steels. strates [22,26,39]. It is composed of an outer Fe2Al5 layer with Cr-
rich precipitates and underneath a much thinner FeAl layer as
Alloy P91 304SS shown in Fig. 2a. Vertical cracks in the aluminide coating (Fig. 2a)
are typical features in such coatings on ferritic steels [22,26]. These
EN number 1.4903 1.4301
C 0.102 0.037 cracks form during aluminizing due to the mismatch between the
Si 0.356 0.649 coefficient of thermal expansion of the ferritic substrate and the
Mn 0.412 1.21 phases in the aluminide coatings [26].
P 0.014 0.013
A thinner aluminide coating of 52–65 mm thickness was
S 0.01 0.021
Cr 8.75 18.37 formed on 304SS. A larger thickness of the aluminide coating on
Ni 0.17 8.69 steel P91 can be explained by a higher diffusivity of Al in bcc ferrite
Mo 0.944 0.32 compared to the fcc austenitic structure of 304. The inward dif-
Fe 88.7 69.92
fusion of Al leaves pores in the coatings [26]. The amounts of these
Al 0.001 0.002
Ti 0.002 0.005 pores depend on the diffusion rate of Al during the heat treatment
V 0.203 0.065 process. As the Al diffusion rate in the ferrite structure is higher
Co o 0.002 0.058 than in austenite, a higher porosity level in the steel P91 diffusion
W 0.008 0.02
Cu 0.119 0.493
zone should not be surprising (see Fig. 2).
Nb 0.08 0.013 The coating on austenitic steel 304 has a different morphology.
The SEM micrograph (see Fig. 2b) illustrates three distinct layers
composed of a single phase layer with a dark gray level, a two
Table 3 phase layer with a lighter gray scale and underneath a thin single
Levels of impurities in nitrate salt mixtures. phase layer. EDS results revealed that the outer layer resembles
Impurity Salt 1 (wt%) Salt 2 (wt%)
the Fe2Al5 phase containing similar Al and Fe contents. The dif-
ference in their gray scale was attributed to the higher Ni content
Nitrite (NO2 ) o0.01 o 0.02 (2 at%) in the inner Fe2Al5 layer. A higher phase fraction of the Cr-
Chloride (Cl ) o0.02 o 0.25 rich precipitates due to the higher Cr content is found in alloy 304
Magnesium (Mg2 þ ) o0.0007 o 0.05
o0.1 o 0.1
compared to that of P91. EDS results revealed the formation of an
Carbonate CO23
o0.01 o 0.1 FeAl layer under the Fe2Al5 layer in both steels. The aluminide
Sulfate SO24
Insolubles o0.01 o 0.05 coating on the 304SS substrate did not reveal vertical crack for-
Moisture o0.2 o 0.1 mation. This can be due to the smaller mismatch between the
coating and the substrate [26,36].
Fig. 2. Back scattered SEM micrographs of (a) as-coated P91 and (b) as-coated 304.
A. Soleimani Dorcheh, M.C. Galetz / Solar Energy Materials & Solar Cells 146 (2016) 8–15 11
Fig. 3. (a, b) Gravimetric results for uncoated and aluminized P91 and 304SS immersed in molten nitrate salt at 600 °C and (c) corresponding macrographs of uncoated and
aluminized substrates at t ¼1000 h.
To study the corrosion behavior, the aluminized and un-coated both aluminized alloys reveals that up to 100 h, an initial mass-
ferritic steel P91 and the austenitic 304SS were immersed in salt A gain occurs followed by a slow rate mass-loss after 500 h (Fig. 3a).
and salt B at 600 °C, respectively. The initial weight gain was not observed in Carling's gravimetric
curves [38]. The difference can be attributed to the FeAl aluminide
3.2. Corrosion kinetics formation in the pack cemented P91 of Carling's work and Fe2Al5
aluminide formation in slurry aluminized P91 of this work. The
3.2.1. Un-coated alloys higher Al activity in Fe2Al5 aluminide is more prone to form an
Fig. 3 illustrates the gravimetric weight change result for coated aluminum rich scale during the course of corrosion. Thus, it is
and un-coated alloys immersed in the molten salt at 600 °C. expected to observe increased protection in this work. Partial
Results represent the net weight of coupons after rinsing with dissolution of aluminum and iron-oxide to sodium-aluminate and
water and drying. In accordance with previous works [40], it was sodium-ferrite (NaAlO2 and NaFeO2) is expected for the alumi-
shown that P91 does not act protective in molten nitrate salts as nized steels under long term exposure in molten nitrate salts [38]
its weight gain increases massively during the exposure. Moreover, which can describe the weight loss in the gravimetric curves. The
the oxide scales formed on the un-coated P91 substrates are prone corresponding macrographs of specimens did not show signs of
to buckling and blistering as shown in Fig. 3b. local spallation, buckling, or blistering (see Fig. 3b).
The austenitic steel 304, however, shows notably higher cor-
rosion resistance with parabolic behavior (see Fig. 3). Higher 3.3. Corrosion products
chromium content results in the formation of a protective chro-
mium oxide layer which accounts for the lower mass gain in 304 The contrast in corrosion behavior of coated and un-coated
[11,15]. alloys can be reflected by comparing their metallographic cross
sections after 1000 h immersion in molten salts (Fig. 4).
3.2.2. Aluminized alloys The massive weight gain of the un-coated P91 gravimetric
Slurry aluminide coatings provide an alumina forming layer on results is reflected as a thick scale composed of multi-layers of iron
the surface of alloys which significantly increases the corrosion oxides (Fe2O3, Fe3O4) on top and inner layers of iron chromium
behavior of low chromium alloys in oxidizing environments. The spinel FeCr2O4 (see Fig. 4) [38]. The major oxidation product in
weight change curves of the un-coated and aluminized steels in 304SS is FeCr-spinel and sodium ferrite as a continuous protective
salts A and B reveal the beneficial influence of the coating on layer that prevents molten salt access to the substrate. This
corrosion resistance of both steels. In the case of P91, the protec- behavior is typically observed in other stainless steels with high
tive behavior of the coating is much more pronounced to an extent chromium contents (e.g. 316SS and 347SS) [11,13]. Despite the
that is comparable with the protective behavior of the high- acceptable corrosion resistance of the un-coated 304SS, further
chromium stainless steel 304 (see Fig. 3a). The coated 304 steel improvements are achieved by aluminizing 304SS.
shows about 2.5 times less mass gain compared to un-coated 304 Fig. 5 shows scanning electron micrographs of the aluminized
in salt B. Higher molten salt corrosion protection in aluminized steels after 100 and 1000 h immersion. Thin layers of corrosion
steels is in agreement with the results of Carling et al. on pack products on top of the coatings on both steels are evident. On both
cemented aluminide coatings on low and high-chromium steels aluminized steels the corrosion front could not penetrate the
[38]. However, detailed observation of the weight change curves of coating and the substrate remained protected. The aluminized P91
12 A. Soleimani Dorcheh, M.C. Galetz / Solar Energy Materials & Solar Cells 146 (2016) 8–15
Fig. 4. Metallographic cross section of (a) uncoated P91 (b) aluminized P91, (c) uncoated 304SS, and (d) aluminized 304SS after 1000 h immersion in molten nitrate salts at
600 °C.
and 304SS show higher porosities in the aluminide layers adjacent thermodynamic calculations of Tortorelli et al. for Al–K–O–N–Na
to the salts compared to their as-coated samples. Continuous system at 507 and 677 °C (Fig. 7) revealing that NaAlO2 is the only
inward diffusion of aluminum from the high activity region during stable phase at this temperature range [13]. Since the temperature
the course of exposure leaves these pores at the outer parts of the used in this work lies in the range used for the calculations, it is
aluminide layer. Further diffusion process during exposure results safe to say that no Al2O3 phase can form under the conditions
in a thicker coating. In addition, the reaction between the inward
used. The microstructure of the coating also underwent an evo-
diffusing oxygen and Al fills pores and vertical cracks with internal
lution during the course of exposure. Due to ongoing inter-
oxides (Fig. 5). The austenitic structure of 304SS provides lower
diffusion in the aluminide layer, after 1000 h, Fe2Al5 thinned
CTE mismatch between the Fe2Al5 aluminide layer and the sub-
strate preventing vertical crack formation (Figs. 4 and 5) [41,42]. while the underlying FeAl layer thickened compared to their
The evolution of coating microstructure during the corrosion test thicknesses after 100 h immersion.
in both steels is much the same. They are composed of a homo-
geneous Fe2Al5 layer with more porosities and Cr-rich precipitates
at longer immersion times. The precipitation-free zone of the as- 3.3.2. Aluminized 304SS
coated 304SS (see Fig. 2b) is altered to a single zone with dis- Fig. 8 shows the element distribution maps and line scan pro-
persed Cr-rich precipitates after 1000 h exposure. files of the aluminized 304SS after 100 and 1000 h immersion in
salt B. Similar to the behavior of aluminized P91, the 304SS forms a
3.3.1. Aluminized P91 continuous corrosion product with a thickness similar to that of
Element distribution maps and quantitative EPMA line scans of aluminized P91. A comparison of the Na/O signal ratio after 100 h
the aluminized P91 after 100 and 1000 h immersion in molten salt and 1000 h reveals that similar to the observations for the alu-
at 600 °C (Fig. 6) reveal details of the compositional evolution of minized P91, the initially formed sodium-rich phase tends to form
corrosion products and coatings. After 100 h, a complex layer with oxide as the course of exposure continues. This is illustrated as a
relatively similar O and Na contents is observed. After 1000 h,
significant reduction in the Na/O ratio in the line scan profiles.
however, the oxygen content dominates forming a more protective
Several differences can be noted, however, by comparing to the
layer of 5 μm thickness. Mutual solubility of elements in the
results of the aluminized P91 steel. EPMA results reveal higher
oxide and ferrite/aluminate structure makes it difficult to conclude
oxygen/sodium ratios in the oxidation products and notable Fe
the stoichiometric composition of this layer using EPMA results. X-
ray diffraction analysis (not shown) could also not resolve the depletion within the Fe2Al5 in aluminized 304SS after 1000 h (see
different phases as the existing peaks corresponding to the can- Figs. 6 and 8). Moreover, co-diffusion of chromium and Ni in the
didates (Al2O3, sodium aluminate) were too weak and overlapped. aluminide layer is evident in the distribution profiles. The effect of
It may be attributed to a complex aluminate (Na(Fe,Al)O2) based substrate Cr content does not seem to be significant on the alu-
on distribution profiles (see Fig. 6b). This is in agreement with minide corrosion performance.
A. Soleimani Dorcheh, M.C. Galetz / Solar Energy Materials & Solar Cells 146 (2016) 8–15 13
Fig. 5. Back-scattered micrographs of the aluminized P91 and 304SS steels after (a, c) 100 h and (b, d) 1000 h immersion in molten salts at 600 °C. Note the evolution of
coating during the course of exposure.
Fig. 7. Predominance diagram for Al–Na–K–N–O system for the temperature range
507–677 °C (after Tortorelli et al. [13]).
4. Conclusions
[38] R.W. Carling, R.W. Bradshaw, R.W. Mar, Protective coatings for alloys in con- [41] B.A. Pint, Y. Zhang, P.F. Tortorelli, J.A. Haynes, I.G. Wright, Evaluation of iron-
tact with molten drawsalt (NaNO3–KNO3), JMES 4 (1983) 229–233. aluminide CVD coatings for high temperature corrosion protection, Mater.
[39] A. Agüero, R. Muelas, B. Scarlin, Coatings for steam power plants under High Temp. 18 (2001) 185–192.
advanced conditions, Schriften Des Forschungszentrum Jü lich Reihe Energie- [42] A. Naji, M.C. Galetz, M. Schütze, Design model for diffusion coatings formed
technik 21 (2002) 1143–1157. via pack cementation, Mater. Corros. 65 (2014) 312–318.
[40] R.W. Carling, C.M. Kramer, R.W. Bradshaw, D.A. Nissen, S.H. Goods, R.W. Mar, J.
W. Munford, M.M. Karnowsky, R.N. Biefeld, N.J. Norem, Molten Nitrate Salt
Technology Development, NASA STI/Recon Technical Report N, 81, 1981, p.
25540.