Corrosion Science 154 (2019) 254–267

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Study on corrosion kinetics of 310H in different simulated MSW combustion T

environment. The influence of SO2 and H2O on NaCl assisted corrosion
⁎
Terrence Wengaa, Guanyi Chena,b, Wenchao Maa, , Beibei Yana
a
School of Environmental Science and Engineering, Tianjin Key Lab of Biomass-wastes Utilization, Tianjin University, Tianjin, 300072, China
b
School of Science, Tibet University, Lhasa, 850012, China

A R T I C LE I N FO A B S T R A C T

Keywords: In this study, the corrosion kinetics of alloy 310H with and without each of NaCl, H2O-vapor, and SO2 under an
Alloy oxidizing environment at 500 °C was investigated and the corrosion characteristics in each exposure were
High temperature corrosion compared. The contribution of NaCl, H2O, and SO2 to the corrosion rate was calculated and the corrosion
NaCl-induced corrosion mechanisms in each condition were characterized by means of ex-situ and in-situ thermogravimetric mass-gain
APCS-MLR model
measurements, SEM/EDS, XRD, XPS, and theoretical calculations. The contribution was in the order of
NaCl > H2O > SO2. Corrosion was linear in the absence of SO2 while in the absence of NaCl, corrosion dis-
played a multi-stage behavior.

1. Introduction these studies revealed that different metals/alloys have different sus-
ceptibility to the environment containing NaCl. Pure Cr, Fe-Cr alloys
Due to the increasing population sizes, rapid economic growth, and and austenitic stainless steels exhibit accelerated corrosion behavior
uncontrolled urbanization, the quantities of municipal solid waste [10,11,18]. In the exposure condition containing NaCl and water vapor,
(MSW) generated are rapidly increasing. Thermal treatment of MSW in the synergistic effect of NaCl and H2O vapor converts the protective
waste-to-energy (WtE) plants has gained worldwide attention and ac- oxide scale of chromium oxide into non-protective oxide scales con-
ceptance as the preferred method for sustainable management and safe sisting of loose oxides and sodium salts (Na2FeO4 and Na2CrO4)
disposal of wastes [1]. WtE operating and maintenance costs, however, [19,20]. Furthermore, gaseous chlorides including HCl and MClx are
increases due to superheater corrosion induced by corrosive compo- formed where MClx are speculated to evaporate and diffuse out of the
nents released from MSW during combustion [2–6]. This problem oxide scale where they are oxidized; nonetheless, the actual gaseous
forces WtE plants to operate at low steam parameters (400 °C and MClx have not been experimentally attested.
4 MPa) [7] which reduces the boiler electricity generation efficiency to Extensive research concerning how alkali chlorides accelerate the
˜15-25% [8] compared to fossil-fuelled power plants with an efficiency corrosion rate of different metal alloys have been conducted
of ˜47% [9]. Because combustion of MSW releases low amounts of SO2 [12,13,21–23]. Generally, alkali chloride in simplified environments
and large quantities of alkali chlorides and water vapor, deposits [21,24] cause corrosion in chromia-forming steels by reacting with the
formed on superheater are generally rich in alkali chlorides and thus oxide scale forming alkali chromate (VI). The reactions are rapid at
corrosion is mainly induced by alkali chlorides. Unlike pure biomass temperatures above 600 °C and deplete the protective oxide in chromia,
with large amounts of KCl, MSW is rich in NaCl mainly originating from converting it into poorly protective iron oxide [12,18,22]. This results
kitchen wastes [2,7]; hence this study focuses on NaCl, water vapor and in a sudden increase in the corrosion rate (breakaway corrosion).
SO2. Understanding the corrosion behavior of superheater material in Chlorine then diffuses through the porous oxide scale and propagates
corrosive environments with different composition of these components the corrosion through the oxidation-chloridation cycle [13,23,25].
is crucial for the industry to mitigate the corrosion problem. Shinata and coworkers [20,24] showed that large amounts of NaC1 at
The corrosion behavior of pure Cr [10], Fe-Cr alloys with different 450 °C results in the formation of Na2CrO4 while at 580 °C eutectic
Cr content [11–13], 304 l steel [14], K38 G [15], 1Cr-11Ni-2W-2Mo-V mixture of Na2CrO4/NaC1 is formed which accelerate the corrosion
steel [16], and Ti-6Al-4 V [17] under NaCl and wet environment have rate. Small amounts of NaCl accelerated oxidation of chromium due to
been investigated at temperatures between 450–700 °C. Results from the formation of CrCl3 which become oxidized to Cr2O3 releasing Cl2

⁎
Corresponding author.
E-mail address: [email protected] (W. Ma).

https://doi.org/10.1016/j.corsci.2019.04.013
Received 2 November 2018; Received in revised form 6 April 2019; Accepted 11 April 2019
Available online 23 April 2019
0010-938X/ © 2019 Elsevier Ltd. All rights reserved.
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

[20]. these species in corrosion rate has not been established.

Addition of SO2 in the flue gas changes the corrosion phenomenon. This paper studied and elaborated the corrosion behavior of auste-
Some researchers believe that alkali chloride in the deposits accelerate nitic stainless steel 310H which is a good candidate currently used in
the corrosion by reacting with SO2 releasing HCl [26] which dissociate most parts of the waste incinerator, in an environment consisting of the
on the gas-deposit surface according to the deacon reaction forming presence or absence of each of NaCl, SO2, and H2O in an oxidizing
chlorine that diffuse through cracks and pores to the metal/oxide in- environment at 500 °C. This temperature was chosen as the potential at
terface where it reacts with transition metals forming volatile metal which future MSW boiler aim to achieve in order to improve electricity
chlorides thus initiating the corrosion and the chlorine cycle [27]. In generation efficiency. The objectives are 1) to observe and examine
contrast, other researchers put forward that sulfur in the gas participate variations in the corrosion rate of 310H alloy with temporal resolution
in the corrosion mechanism by changing the deposit composition con- in the environments containing different combinations of NaCl, H2O
verting alkali chlorides to less corrosive sulfates that do not react with vapor, and SO2; 2) determine the percentage contribution of NaCl, H2O,
the protective chromia-rich oxide to form alkali chromate [28,29], and SO2 to the corrosion rate of 310 H; 3) to obtain a better under-
formation of sulfates on the metal that prevents the volatilization of standing of the corrosion mechanisms responsible for changes in cor-
CrO2(OH)2 and decreases the rate of oxygen reduction on the oxide rosion behavior in different exposure atmospheres. The results of the
surface [30,31] and the formation of sulfates that prevents the surface study provide a basis for guiding further strategies aimed at preventing
reaction of Cl gas with the metal surface [31]. high temperature corrosion in waste incinerators.
In addition, the influence of H2O vapor on high-temperature cor-
rosion of different steels has been extensively studied in the absence of 2. Experimental
deposits [31–34]. The corrosion findings revealed that scale breakdown
is triggered by the vaporization of CrO2(HO)2 depleting the substrate in 2.1. Material preparation and exposure
Cr leading to the formation of oxide islands on the alloy grain centers
consisting of an outward growing hematite layer and an inward Austenitic stainless steel 310H (wt. %: 26 Cr, 20 Ni, 2 Mn, 0.9 Si,
growing FeCrNi spinel. Moreover, the effect of H2O vapor in simplified 0.08 C, balance Fe) was used in the present study. The metal samples
exposure atmospheres and synthetic deposits has also been conducted with the geometric surface area of 400 mm2 (20 mm × 20 mm) and the
[12,16,35–37]. Nevertheless, it is still an open question regarding the thickness of 3 mm were cut from the virgin steel used as the super-
influence of H2O vapor as some studies showed accelerating heater. The samples were ground with a 320 and a 1000-grit silicon
[16,34,37,38] while others suppressing [25,35,36] effect of H2O vapor carbide paper and then polished with a 1 μm diamond paste until a
on high temperature corrosion of the tested materials. Hence, a strong mirror-like surface appeared. Cleaning and degreasing of the samples
necessity for the further understanding of the influence of water vapor were achieved using ethyl ethanol in an ultrasonic bath, rinsed with
on high temperature corrosion by systematically changing the corrosive distilled water and then dried to a constant mass. Water was added to a
conditions is important as well as on the influence of alkali chloride beaker containing NaCl with particle sizes ranging from 110 to 250 μm
reaction with SO2. and then ethanol (ratio of 1: 9 by volume of added water) was added to
Although extensive researches have been conducted on the influ- the mixture to form slurry which was sprayed on to the metal surface to
ence of NaCl, SO2, and water vapor on material oxidation, there is still form a deposit 250 ± 5 mg/cm2. The samples with salt deposits were
lack of a comprehensive investigation and contrasting of the corrosion air dried and the sample masses were measured before and after salt
kinetics of superheater material under exposure environment with application using an electronic mass balance with four decimal place.
different composition of NaCl, H2O, and SO2 that may occur in MSW Corrosion tests were carried out in an electrically heated furnace
boiler. The environment containing varying components of NaCl, H2O, which was preheated at 500 °C for 2 h to obtain a stable temperature.
and SO2 differently affect the corrosion behavior of boiler tube metals. Fig. 1(a) shows a schematic diagram of the experimental setup of the
Grabke et al. [13] studied the effect of NaCl deposits in O2 and HCl on furnace used in the present study. The device consists of a two-zone
the corrosion kinetics of 2 1 Cr-1Mo, 12CrMoV, 20Cr-12N1, and 25Cr- electrically heated oven, in which 1 m cylindrical silica tubes were
2
29Ni at different temperatures (500, 600, and 700 °C). The corrosion mounted horizontally and each tube has a water-cooled flange at one
process was observed to start with nearly no incubation time after the end to cool the hot gases which were then adsorbed into de-ionized
deposition of NaCl on the scale. Shu et al. [16] investigated the influ- water, NaOH solution and dried by silica gel drier before emitted to the
ence of NaCl and water vapor on the corrosion kinetics of 1Cr-11Ni-2W- air. The reactor also has a water preheating chamber where water was
2Mo-V steel at temperatures between 500–700 °C. The steel showed a preheated and evaporated steam was injected into the furnace by
similar trend of corrosion at all the temperatures used with the ac- controlling the opening valve to the required volume percentage. Inside
celerated corrosion rate at higher temperatures. Yang et al. [39] ex- each cylindrical silica tube, a removable semi-circle quartz-made
amined the influence of molten NaCl on corrosion behavior of Fe, Cr, sample holder containing metal samples was placed. The samples, six at
Ni-base alloys at 850 °C and all the alloys obeyed the linear rate law. a time, were mounted on a sample holder and positioned parallel to the
The reactions of chromium and NaCl were suggested as the cause for gas flow direction.
the accelerated corrosion rate of all the alloys with the characteristic of All the corrosion exposure conditions are shown in Table 1 and the
corroding along the grain boundaries which provide channels for the flow rate of each gas (O2 and SO2) was kept at 100 ml/min using DO7
infiltration of molten NaCl into alloy matrix. Cao et al. [12] compared mass flow controllers. The exposures were isothermal with ex-situ re-
NaCl solution spray and solid NaCl deposit in water vapor on the cor- cording of the weight changes using an electronic mass balance after 1,
rosion behavior of Fe-20Cr alloy at 600 °C. The initial corrosion rate 24, 72 and 168 h. The method used here for the corrosion testing was a
was increased significantly but approaches a parabolic rate law with backward approach where the first exposure consisted of all the cor-
time. It was suggested that due to the formation of a thick oxide scale rosive species studied (taken as the control) representing a simplified
on Fe-20Cr alloy samples, the diffusion of chlorine and MClx were real plant. In the subsequent exposures, one species was removed at a
slowed down and thus corrosion rate slowed down. While these studies time, to determine how the alloy 310H responds to such an environ-
investigated the corrosion kinetics of different chromia-forming steels ment. Prior to each test, the furnace was evacuated and purged with
under NaCl and/or with water, they focused much on the influence of nitrogen for 2 min to avoid the furnace gas history effects then followed
temperature, and the effect of H2O vapor on NaCl assisted corrosion, yet by the gas mixture without each of the parameters.
the influence of exposure environment with varying component of The flow apparatus combined with thermogravimetric analysis
NaCl, SO2, and H2O vapor on corrosion kinetics of metals with temporal (TGA) measurements were also performed with the same gas flow rate
resolution has been largely neglected and the relative importance of to study the corrosion by NaCl deposits. Fig. 1(b) shows the schematic

255
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 1. Schematic diagram of (a) experimental tube furnace (b) TGA experiment.

Table 1 it could dissolve some corrosion products like metal chlorides. The
Exposure conditions for corrosion testing of 310H. cross-section was covered with Au to prevent charging. For detailed
Run NaCl SO2 H2O O2 N2 Temperature Flue gas velocity microstructure analysis, the microscope was operated at a 15KV ac-
mg/cm2 ppm vol.% vol.% vol.% °C ml/min celerating voltage, a working distance of 8 mm which is sufficient for
detection of elements with high atomic numbers like Fe, and a spot size
1 250 300 20 8 Bal. 500 100
of 5.5 with secondary electron imaging. The quantification of chemical
2 250 0 20 8 Bal. 500 100
3 250 300 0 8 Bal. 500 100
elements was achieved using a HITACHI X-ACT energy dispersive X-ray
4 0 300 20 8 Bal. 500 100 spectroscopy analyzer connected to the SEM. Accelerating voltage 5-
15KV was used.

set-up used for the TGA. The samples were mounted on the quartz-made
sample holder and introduced to the tube furnace in the Setaram 2.3. XRD analysis
system. The flow apparatus delivered the gases N2, O2, and SO2 by
capillary flow meters into the hot zone at 500 °C. The desired amount of The crystalline phases of the corrosion products on the plan view
water vapor was then injected into the hot zone of the thermobalance surface of the steel after salt residues were removed and on the corro-
where O2, SO2, and N2 were injected, so that reaction take place at high sion scale were detected using X-ray diffraction D8 Advance from
temperature in order to avoid the condensation of sulfiric acid. The Bruker AXS operated in grazing incidence configuration. The incident
thermobalance itself was flushed and protected by a flow of N2. The X-ray beams were focused using polycap optics system to converge the
data were plotted from the time at which the isothermal exposure beam. The diffractometer was operated with Cu Kα radiation
temperature was reached, which corresponded to 8 min. (λ = 1.5418Ǻ) at 40 KV operation voltage and 40 mA current, an X-ray
incidence angle of 5°, giving a constant penetration depth less than
2.2. Scanning electron microscopy and energy dispersive spectroscopy 1.3 μm under the assumption of a completely flat surface. The detector
analysis (LYNXEYE) collected data between 10° < 2θ < 100° with the step size
of 0.05°. The penetration depth in the surface oxide scale could not be
The morphology of the corrosion products on the top surface of the determined accurately due to the void spaces in the oxide with un-
deposit and the plan view surface underneath the salt deposit were known absorption coefficient. The peak identification was achieved by
analyzed using scanning electron microscopy (SEM) (su3500 HITACHI). the search–match function in the X`Pert HighScore Plus software.
Furthermore, the metal samples were cut at the middle; the cross sec-
tion was polished and cleaned using kerosene. No water was allowed as

256
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

2.4. XPS analysis gain up to 168 h (b) in-situ mass gain measurements by TGA up to 80 h.
The exposure conditions from now on will be named as; run 1 for re-
The composition of the corrosion products were analyzed using X- ference condition (NaCl + 20 vol. % H2O + 300 ppm SO2 + 8 vol. %
ray Photoelectron Spectroscopy (XPS). The samples for the analysis O2), run 2 for (NaCl + 20 vol. % H2O + 8 vol. % O2), run 3 for
were prepared by scraping the corrosion products from the surface of (NaCl + 300 ppm SO2 + 8 vol. % O2) and run 4 for (300 ppm SO2 +
the samples using a razor blade and the metallography of the alloy 20 vol. % H2O + 8 vol. % O2). As shown in Fig. 2(a), in run 1, corrosion
samples were subsequently checked to confirm the corrosion products steadily increased between 1 h and 24 h and plateaued thereafter
were removed completely from the surface. All specimens were ana- reaching a maximum mass gain of 0.03 g after 168 h of exposure. In run
lyzed in a Thermo Scientific Escalab 250Xi X-ray Photoelectron 2, mass abruptly changed in the first 24 h with a steep gradient reaching
Spectrometer using Al Kα X-ray (1486.6 eV) source at 150 W (15 kV, about 400% increase in mass gain relative to the control environment
10 mA) to avoid X-ray damage on the samples. An ultrahigh vacuum (run 1) after 168 h. By removing H2O vapor and introducing SO2, that is
condition with base pressure of 10−8 mbar was used. Each analysis was changing the exposure environment to run 3, the mass slightly in-
conducted over a 1400 eV binding energy range with 1.0 eV step size creased in the first 1 h then levelled off. The mass gained was almost
and a dwell time of 100 ms. To obtain a narrow high resolution of in- 83% less than the mass gained in the reference exposure condition after
dividual elements spectrum, 0.05 eV steps and a dwell time of 250 ms 168 h. In run 4, the mass gain was very low as compared to run 1 and 2
were adopted. XPSPEAK version 4.1 software was subsequently used for but was almost equal to run 3. The error bars for the exposure condi-
peak fitting and a linear baseline was used for background subtraction. tions in run 4 and 3 overlapped indicating there was no significant
The binding energies were calibrated with respect to the C 1 s peak at difference in mass gain between these two exposure environments. As
285 eV. displayed in Fig. 2(a), it was evident that the mass gain of the samples
in the absence of NaCl and H2O that is in run 4 and 3 respectively
2.5. Thermodynamic calculations remained constantly low during the testing time, while the mass gain of
the samples in the other two exposure conditions where NaCl and H2O
Thermodynamic equilibrium calculations were conducted using were present that is in run 1 and 2 increased considerably with run 2
HSC Chemistry 6.0 software for all possible interactions between salt strongly accelerated.
deposit, gas species, and the metal surface in order to propose the The comparison of corrosion kinetics by in situ mass gain mea-
corrosion mechanisms. The temperature was set at 500 °C in all the surement conducted for 80 h in TGA are shown in Fig. 2b. At the be-
calculations. ginning in reference exposure (run 1), corrosion started slowly that is
incubation phase then turned into an accelerated phase for a short
2.6. Ranking of the corrosive species period of time and then levelled off (stagnation phase). In run 2, no
initial slow stage was observed, corrosion started with a linear increase
In order to determine the contribution of the studied species in which lasted for an extended period of time then slowed down fol-
corrosion rate, a multivariate statistical approach was employed. The lowing a law that is almost parabolic. This was attributed to the si-
correlation coefficient was used to assess multi-collinearity issues multaneous activity of NaCl and water vapor attacking the elements of
among corrosive species on the corrosion rate. Absolute principal the metal substrate and the protective oxide scale, thereby drastically
component score-multiple linear regression (APCS-MLR) model can be reducing the initiation or incubation period resulting in the rapid cor-
used to get robust contributions of species to the dependent variable, rosion attack. This result is in accordance with the previous literature
therefore, APCS-MLR model was applied to calculate the contributions [12,16]. In run 3 and 4, no significant changes in the corrosion kinetics
(%) of corrosive species on the corrosion rate of the 310H alloy. The were seen. The kinetics of the corrosion process of the in-situ mass gain
model procedure is defined in Eq. (1) measurement by TGA can be further elucidated by presenting the mass
n
gain data on a ln(m) versus ln(t) plot (Fig. 3) with the slope 1/ α of the
CR = ro + ∑k =1 rk ×APCSk (1) kinetic Eq. (2):
mα ∝ t (2)
where CR is the mass gain (g/hour), r0 is the constant, the rk is the
regression coefficient of the species k and APCSk is the absolute prin- where, if the exponent α = 1 shows that there was a linear increase in
ciple component score of species k. The term rk × APCSk represents the mass gain, α = 2 there was a parabolic behavior and α < 1 indicate
contribution of species (k) to the corrosion rate of the metal. the rapid corrosion attack (breakaway). In situ mass gain was converted
to natural logarithm to clearly trace and compare the corrosion stages
3. Results at different times in different exposures environment. The kinetics in all
corrosive environments displayed multistage corrosion behaviors. In
3.1. Corrosion kinetics the reference exposure (run 1) environment, corrosion started slowly
with α = 2.3, then turned into nearly linear with α = 1.4 and stabilized
The influence of different exposure conditions on the mass gain of thereafter almost 26 h of exposure. Unlike the reference condition, run
the alloy 310H against exposure time is shown in Fig. 2(a) ex-situ mass 4 showed many corrosion stages occurring within a short period after

Fig. 2. Mass changes of alloy 310H in different exposures at 500 °C (a) ex-situ mass gain up to 168 h (b) in-situ TGA mass gain measurements up to 80 h.

257
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 3. ln(mass gain) versus ln(time) of in-situ TGA for 310H in different exposure environment at 500 °C.

the exposure began. It started with a very rapid attack (breakaway) this dark area could be composed of iron oxide, Na2CrO4, and Na2SO4
α = 0.2 within the first 2 h, slowed down to almost parabolic behavior (35.56 atomic % O2, 15.43 atomic % Na, 14.67 atomic % S, 8.87 atomic
α = 2.1 and increased for another short period of time then after 11 h % Cr, 23.17 atomic % Fe and 2.3 atomic % Ni at 5 KV). XRD analysis
the mass gain became almost flat α increased to 7.2 (Fig. 3). The results shown later in Fig. 10 was conducted on metal/scale surface
plausible mechanism responsible for the multistage corrosion behavior after salt residues were removed, showed the presence of Na2SO4,
in this environment was discussed in Section 4.2. Under the case of run Fe2O3 and sodium salts (Na2Fe2O4 and Na2CrO4). Close examination of
2, corrosion started with a nearly linear increase α = 1.5, then turned the dark area, small oxide features 1–10 μm irregularly shaped corro-
into accelerated α = 0.9 for a long period of time ˜20 h and then slowed sion product particles on top of the surface were seen. EDX point ana-
α = 2.3. The longer period of the accelerated phase caused this ex- lysis showed chlorine, Fe and Cr but no sodium in these particles,
posure condition to have a very high mass gain compared to the other suggesting that they consist of chlorides formed by the alloying ele-
exposures as shown in Fig. 2. In run 3, the corrosion started slowly and ments (Fe, Cr,). After 24 h, the surface was fully covered by a moder-
turned into a very rapid stage within 1 h of exposure. The rapid increase ately thick oxide layer (Fig. 4b) corresponding to Na2SO4 and Fe2O3
was attributed to the reaction of NaCl with alloy metals before it was (30.01 atomic % O2, 18.87 atomic % Na, 8.75 atomic % S, 37.6 atomic
sulfated. Interestingly, the corrosion seemed to have ceased α=-10.4 % Cl and 4.76 atomic % Fe at 5 KV). This indicated that Na2SO4 has
for a long period of time resulting in a decrease in mass. After 65 h of formed on top of a Fe2O3 layer. At this time no Cr and Ni were detected
exposure, the corrosion process resumed but very slowly. Similar be- in this oxide scale. No unreacted NaCl could be seen, the NaCl particles
havior was observed by Grabke et al. [13]. had been replaced by irregularly shaped agglomerates pointed with an
arrow in Fig. 4b. The detection of Cl in the scale indicated incomplete
3.2. Morphological changes sulfation of NaCl. This is because the sulfation rate of the solid NaCl is
very slow at temperatures around 500 °C [40].
The secondary electron SEM images of the surface characteristics of Removal of one parameter at a time resulted in the formation of
the samples exposed to the reference condition run 1 are shown in different surface morphologies. Fig. 5 shows the surface morphology of
Fig. 4(a) metal sample exposed for 1 h (b) metal sample exposed for the corrosion products when SO2 was removed from the flue gas system
24 h. After 1 h of exposure, some unreacted NaCl particles (5–50 μm in in run 2. Unlike all other exposures where large salt coverage/amounts
size) were observed on the surface. The whole surface was covered by a were used, two cases were investigated in this condition that is small
thin dark oxide scale. The EDX analysis (3 mm × 3 mm) revealed that amounts (˜0.1 mg/cm2) of NaCl and large amounts (250 mg/cm2) of
NaCl with approximately 1 mm thick from the metal surface. After 1 h
of exposure with small amounts of NaCl, residuals of the salt particle
were seen on the surface. Between two salt particles, two distinct re-
gions, the light and the dark areas were clearly observed. In the direct
vicinity of the former salt particle, a dark region corresponding to thick
iron oxide (72.27 atomic % O2 and 27.73 atomic % Fe at 5KV) was
growing outside from the position of NaCl. The light area corresponds
to a thin oxide scale (23.49 atomic % O2, 17.15 atomic % Na, 43.00
atomic % Cl, 1.03 atomic % Cr and 16.33 atomic % Fe at 5KV). This
composition suggests the presence of Na2CrO4, NaCl, and Fe2O3 as
Fig. 4. Secondary electron SEM images of 310 samples exposed to run 1 for (a)
1 h (b) 24 h. supported by XRD analysis. The thickness of the oxide layer can be

258
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 5. SEM images of 310 samples exposed to run 2 (a) sample with small coverage NaCl for 1 h, (b) insert of marked up area in panel (a), and (c) 24 h.

determined from the oxygen content in the layer. At 5KV EDX analysis, reacted with chromium or chromium oxide and oxygen. This process
the beam penetration depth is 0.5 μm. In the case of the light oxide released chlorine at the deposit/metal interface which reacted with
layer, oxygen was only 23.49 atomic % which means the beam only alloy substrate such as Cr forming CrCl2. The eutectic compound NaCl-
penetrate a part of the oxide layer. In the case of the dark oxide, oxygen CrCl2 with melting point 437 °C [41] could be the most probable can-
detected was 72.27 atomic % meaning the oxide layer at least ap- didate for the molten phase. This eutectic compound spreads on the
proached the maximum penetration depth. The salt remains were partly metal surface and also participate in accelerating the corrosion. The
converted to Fe oxide from beneath the salt particle (22.31 atomic % CrCl2 in the melt, due to high volatility, evaporates, diffusing out of the
O2, 7.25 atomic % Fe, 12.61 atomic % Na and 57.83 atomic % Cl at metal/oxide interface to the deposit/gas interface and oxidizes in a
5KV). The marked area in Fig. 5(a) was magnified as shown in Fig. 5(b); place with the appropriate partial pressure of oxygen to form oxide and
some nodules like features were seen all over the surface and they have chlorine while the remaining NaCl becomes solid. The released chlorine
been previously characterized as Na2CrO4 [12,18]. After 24 h, no salt is returned back to react with either Fe or Cr again forming metal
particles and no sign of the surface nodule-like features of Na2CrO4 chlorides which can form a eutectic compound with NaCl. Hence,
particles on top of a thick oxide were observed as shown in Fig. 5c. The during the chloridation/oxidation reaction process, chlorine acts as a
surface was completely covered by a thick, loose and porous dark oxide catalyst to accelerate corrosion. The removal of salt residue layer to
scale corresponding to iron oxide (70.70 atomic % O2 and 29.30 atomic study the plan view of the metal/oxide layer enabled the identification
% Fe at 5KV). Inside the salt particles residuals, Fe2O3 were detected of important molten phase that could not be detected by examination of
and outside NaCl were observed. deposit surface morphology or cross-section only.
The surface morphology of the corrosion products when large In run 3 exposure, the surface morphology of the corrosion product
amounts of NaCl salt was used in run 2 is shown in Fig. 6. On top of the as shown in Fig. 7 appeared completely different from run 1 and 2.
salt deposit (Fig. 6(a)), some large NaCl particles and small particles After 1 h, the surface developed into a compact scale all over the sur-
growing from beneath were observed. Oxide sticks (20.73 atomic % O2, face. EDX analysis result showed that the scale layer was composed of
7.04 atomic % Fe, 16.82 atomic % Na, and 55.42 atomic % Cl at 5KV) 29.8 atomic % O2, 8.3 atomic % Na, 4.1 atomic % S, and 51.2 atomic %
also developed on the surface. Fig. 6(b) shows the structure of the Fe at 5 KV. XRD analysis result showed the presence of salt residues,
corrosion product (deposit/metal) after the greater part of salt-corro- Fe2O3, Na2SO4, and Na2CrO4. After 24 h of exposure, the compact
sion products layer was removed. Some molten phase appeared. Since grains particles (5.31 atomic % O2, 25.36 atomic % Na, 13.28 atomic %
pure NaCl has a melting point of 801 °C, a question was raised how S, and 55. 85 atomic % Cl at 5KV) became larger. Fe and other alloy
could a molten phase be formed under the present condition. Na2CrO4 elements were not detected by EDX analysis. This observation was
was formed by the oxychloridation process, in which NaCl could have supported by the XRD analysis results in Fig. 10 which showed Na2SO4

Fig. 6. Images of samples exposed to run 2 with large coverage of NaCl salt after 24 h (a) on top of salt deposit (b) after salt deposit was removed.

259
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 7. Images of samples exposed to run 3 at 500 °C for (a) 1 h (b) 24 h.

with strong signal and Fe2O3 with weak diffraction signal. This implies Table 2
that corrosion was minimal and also support the decrease in mass gain Comparison of EDX result for the marked position (red circle) on Figs. 4(b), 5
shown in Fig. 2. By comparing the surface morphologies of the products (c), 7 (b), and 8 (b) on sample exposed for 24 h at 500 °C.
obtained in run 3 with run 1 and 2 (Fig. 7(b) with 4 (b), and 6(a)), it Element O Na Cl S Cr Fe
was clear that the scale formed in the former was more compact and
denser than the others. This compact and dense layer somewhat, pre- Run 1 30.01 18.87 37.60 8.75 – 4.76
Run 2 70.70 – – – – 29.30
vented the entry of Cl2 to react with the metal at the metal/scale in-
Run 3 42.35 25.45 – 32.21 – –
terface and therefore made the corrosion process slower with time. Run 4 57.07 – – 2.34 11.17 29.42
When the metal sample was exposed to run 4 exposure, some small
oxide features occurred sporadically after 1 h as shown in Fig. 8. It
showed that a very thin corrosion scale (15.25 atomic % O2, 12.25 observed in this study are displayed in Fig. 9(a) run 2 and (b) run 4. As
atomic % Cr and 72.50 atomic % Fe at 5KV) was formed on the surface seen in Fig. 9(a), there was a distinct border between the oxides and the
of the sample, in which some scratches produced by SiC paper that metal with a significant number of visible pits and crevices distributed
could not be totally removed during the preparation stage of virgin along the border. The substrate (1#) is intact; it did not suffer any
samples could be seen. After 24 h of exposure, some patch-like features corrosion attack, as no alkalis or chlorides were detected inside.
(oxide islands) were observed. These oxide islands were present in the Nevertheless, position (2#) experienced a high level of corrosion attack
center of most of the grains. If the Cr content decreases below a critical and the elemental composition was too complicated, consisting of 2.7
value (usually 12 wt.%) needed for passivation or when the Cr supply atomic % Cl, 5.0 atomic % Na, 28.4 atomic % O2, 4.0 atomic % Cr, and
rate could not meet with the evaporation rate, a “short circuit” is 59.9 atomic % Fe at 5 KV. This proved that chlorides and chromates
produced where grain boundaries act as the anode and grains act as the accumulated in this area by the reactions of NaCl with substrate ele-
cathode resulting in rapid dissolution of the grain boundaries while ments. In position (3#), chlorine content was higher than that in the
grains remaining passive [42]. The thin oxide film present between the position (2#). The elemental composition of (3#) consisted of 4.3
islands contains mainly iron and chromium (35.25 atomic % O2, 1.87 atomic % Cl, 36.8 atomic % O2 and 58.9 atomic % Fe at 5KV. The large
atomic % S, 20.24 atomic % Cr, and 44.51 atomic % Fe at 5 KV). The amounts of chlorine at (3#) indicated that large amounts of chlorine
outer half of the oxide islands was made up of almost pure iron oxide, resulted from diffusing metal chloride out of the metal/oxide interface
corresponding to the hematite seen by XRD, whereas when the oxide to the oxide scale/gas surface. Finally, position (4#) consisted of 5.9
was scraped, the bottom layer of the oxide islands comprised of chro- atomic % Na, 38.0 atomic % O2, 10.7 atomic % Cl, and 45.4 atomic %
mium and iron corresponding to the spinel layer detected by XRD Fe at 5 KV. At this position, the amount of chlorine was ascribed the
shown in Fig. 10. residual salt remains. From these results, it can be concluded that the
The EDS results shown in Table 2 for the marked position in red metal chlorides formed at the metal/scale interface migrates out of the
cycles in Fig. 4(b) (run 1), 5(c) (run 2), 7(b) (run 3), and 8(b) (run 4) metal due to its high volatility to areas of a higher oxygen partial
indicated that the corrosion products were rich in elements O, Cl, Na, S, pressure near the deposit/gas interface and converted to oxide and
and Fe for run 1 (Fig. 4(b)), while in Fig. 5(c) it was rich in O and Fe for chlorine. This behavior explains the voids seen at the corrosion front in
run 2, in Fig. 7(b) it was rich in Na, S, and O for run 3, and in Fig. 8(b) it the metal/deposit interface. The sample exposed to run 4 (Fig. 9(b)) did
was rich in O, Cr, Fe and very small amount of S. It was clear that the not experience much noticeable corrosion attack as compared to run 2
chemical compositions of all corrosion products were significantly dif- shown in Fig. 9(a). The absence of a thick oxide layer in this run is in
ferent from each other. conformity with the mass gain observed in Fig. 2. Only some voids
The cross-section analyses of the metal samples exposed to run 2 could be seen sporadically near the metal/gas interface.
with large NaCl salt coverage and run 4 which are the extreme cases

Fig. 8. SEM surface morphology of the 310 sample in run 4 at 500 °C (a) 1 h (b) 24 h.

260
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 9. Cross-section analysis of samples exposed for 24 h in (a) run 2 and (b) run 4.

3.3. XRD Fig. 12(a) shows the survey spectrum for the corrosion products. The C
1 s high resolution shown in Fig. 12(b) divulged two peaks similar to
The XRD analysis results for corrosion products obtained in different those of run 2 (Fig. 11(b)). The peaks were ascribed to hydrocarbon
exposure environments at 500 °C after corrosion for 1 h and 24 h are adsorbed to the corrosion products (BE 284.51 eV) and to pure chro-
summarized in Fig. 10. The results showed that Fe2O3 and Cr2O3 were mium carbonate. Fig. 12(c) shows the Cl 2p3/2 high resolution peak
the main crystalline corrosion product formed in run 4. Although SiO2 spectra of the corrosion products. The spectra showed two peaks at BE
was detected, it was intentionally neglected since the preparation stage 199.4 eV corresponding to CrCl3 and the peak at 198.0 eV which is close
used silicon carbide paper. The strong signal of the Fe-Cr alloy and the to 198.3 eV was attributed to NaCl. Fig. 12(d) shows O 1 s high re-
low intensity of oxides indicate that the corrosion scale was very thin, solution peak spectra and a peak at BE 531.5 eV corresponding to
which support the SEM images and the mass gain results. Some in- NaCrO2 was observed. Fig. 12(e) shows S 2p3/2 high resolution of the
formation on corrosion product mixtures in the exposure containing corrosion products. Two peaks at BE 167.5 eV corresponding to NaSO4
NaCl + H2O + O2, such as Fe2O3, Cr2O3, Na2CrO4, Na2Fe2O4, and re- and a small peak at BE 162.3 eV which was attributed to FeS2 were
sidual NaCl which is in accordance with previous results [12,16] are observed. An examination of Fe 2p3/2 peak spectra shown in Fig. 12(f)
also detected. For run 1 and 3, weak diffractions of Na2SO4 were de- revealed that FeCl2 (peak at BE 710.6 eV) was present in the corrosion
tected only after 1 h of exposure. However, after 24 h diffraction peaks products. Fe 3 s peak spectra in Fig. 12(g) showed the peak at BE
with strong intensity for Na2SO4 and low intensity for Fe2O3 were de- 92.8 eV corresponding to Fe2O3 was also present in the corrosion pro-
tected. Additionally, in both cases of run 1 and 3, some compounds duct layer and is in agreement with XRD result in Fig. 10.
including Na2Fe2O4, NaCrO4 and residual NaCl were identified. The XPS result for the corrosion product from 310H exposed to run
4 for 24 h at 500 °C is shown in Fig. 13 with the survey spectrum for the
3.4. XPS corrosion product shown in Fig. 13(a). This spectrum confirms the
absence of Cl as no peak of Cl was found since NaCl was removed from
XPS analysis for the chemical composition was used to further the system. Fig. 13(b) shows C 1 s high resolution peak spectra. Two
confirm the existence of some elements in the corrosion products peaks at BE 184.6 eV corresponding to carbon adsorbed on the metal
formed on 310H under different corrosive environment (i.e, run 2, 3, and a small peak at BE 188.0 eV corresponding to pure iron carbonate
and 4 which are the experimental conditions of special interest than the were observed. Fig. 13(c) displays S 2p3/2 peak spectra. The main peak
control condition run 1). Fig. 11 shows the XPS result for the corrosion occurred at BE 168.6 eV corresponding to iron sulfates, possibly due to
products on 310H samples exposed to run 2 at 500 °C for 24 h. Elements the adsorption of sulfate on metal surface. An additional small peak
of special interest were Cl, O, Fe, and Cr. Fig. 11(a) shows survey occurs at BE 171.9 eV which is close to 172.2 eV and could be
spectrum of the corrosion product. As expected, no signs of sulfur were (CH3)2SO. Fig. 13(d) shows O 1 s XPS peak spectra and only one peak at
detected on the samples exposed to run 2 since SO2 was removed from BE 532.2 eV corresponding to iron sulfates was observed. An analysis of
the system. The absence of S from the XPS results confirms both the Fe 2p3/2 peak spectra in Fig. 13(e) indicate that Fe2O3 occurred in the
EDX (Table 2) and XRD (Fig. 10) results for run 2. Fig. 11(b–g) shows corrosion product of 310H exposed to run 4.
the typical peak deconvolution into their constituent peaks. The C 1 s
high-resolution XPS scans in Fig. 11(b) reveal two peaks at the binding 3.5. Species contribution to the corrosion rate
energy (BE) 287.6 (chromium carbonate) and 284.51 eV (hydro-
carbon). Fig. 11(c) shows the O 1 s high-resolution XPS peak spectra of Pearson's correlation coefficients for the corrosive species are dis-
the corrosion product layer. The spectra have two peaks at BEs 535.68 played in Table 3. The correlation coefficients indicated that the species
and 531.3 eV. The peak at BE 531.3 eV correspond to Fe(OH)3, while (NaCl, H2O, and SO2) were not correlated to each other; however, all
the peak at BE 535.68, however, was not defined precisely due to the the species were somehow correlated to the mass gain of 310H alloy.
presence of the complex oxides. Fig. 11(d) shows Cl 2p3/2 XPS peak NaCl and H2O were positively correlated with the mass gain while SO2
spectra which also have two peaks at BE 199.7 eV and 198.8 eV. The was strongly negatively correlated with the mass gain (P < 0.01). For
peak at BE 198.8 eV corresponds to FeCl2. An analysis of Fe 2p3/2 in the PCA analysis, two principal components (PC1) and (PC2) were
Fig. 11(e) revealed the presence of FeCl2 and Fe2O3 in the corrosion extracted. PC1 and PC2 explain about 91% of the total variance with
product as identified at peaks BE 710.6 eV and 724.17 eV respectively. PC1 57.94% and PC2 33.33%. PC1 was dominated by NaCl and H2O
The presence of Fe2O3 was also confirmed by an examination of Fe 3 s while PC2 was dominated by SO2. On the basis of the results above,
peak spectra shown in Fig. 11(f). Similar to Fe 2p3/2 peak spectra APCS-MLR model was applied to apportion the contributions of the
analysis, another peak spectra examination was conducted for Cr 2p3/2 species studied in this work to the corrosion of the alloy 310H. The R2
(Fig. 11(g)). The Cr 2p3/2 peak spectra showed the presence of Cr2O3 at value of APCS-MLR model was 0.999, showing that the allotment re-
peak BE 576.8 eV in the corrosion products. sults were credible. The percentage contributions of NaCl, H2O, and SO2
The XPS results for the corrosion products on 310H exposed to run 3 to the mass gain of the alloy studied were 50.36, 49.63, and 0.00026%
for 24 h at 500 °C are shown in Fig. 12. Since the condition was dif- respectively. The results from the statistical analysis are in excellent
ferent, the elements of interest in this case were Cl, O, S, and Fe. agreement with the result from Fig. 2. NaCl and H2O have a strong

261
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 10. X-ray diffraction for corrosion products obtained in different corrosive environments at 500 °C.

impact on the corrosion rate while the amount of SO2 used in the alloys, 304 l and 1Cr-11Ni-2W-2Mo-V steel. However, the differentia-
present study does not contribute to the mass gain of the alloy tested. tion of the corrosion kinetics of the alloys between these exposure en-
vironments was not given much attention as well as the contribution of
each species to the corrosion rate. The aim of this study was to elabo-
4. Discussion rate and contrast the effect of SO2 or H2O on NaCl-induced corrosion
kinetics of alloy 310H at 500 °C. The corrosion mechanisms were
The corrosion kinetics, composition and the structure of the corro- clarified based on the above experimental results and theoretical cal-
sion product formed in NaCl containing either SO2 or H2O in O2 were culations. From the data analysis, it was divulged that the corrosion
very different from each other and were important to the reveal of the attack was a result of the direct reaction of the salt deposit with the
responsible corrosion mechanisms. From the previous studies alloy elements or preformed protective oxide scale, sulfation of the
[10–12,16] it was already shown that NaCl accelerates the corrosion initial deposit, and direct attack of the flue gas on the alloy.
rate of different materials including pure Cr, Fe-20Cr, various Fe-Cr

262
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 11. XPS peak fitting spectra obtained from corrosion product layer at 500 °C for 24 h in run 2 exposure condition: (a) survey spectrum of corrosion product layer,
(b) C 1 s characteristic peak spectra, (c) O 1 s characteristic peak spectra, (d) Cl 2p3/2 characteristic peak spectra, (e) Fe 2p3/2 characteristic peak spectra, (f) Fe 3 s
characteristic peak spectra, and (g) Cr 2p3/2 characteristic peak spectra.

4.1. Influence of H2O on NaCl-induced corrosion of 310H at 500 °C Cr + 2NaCl + 3/2O2(g) + H2O(g) = Na2CrO4 + 2HCl(g) (1)

2NaCl + 2Fe + 3/2O2(g) + H2O(g) = 2NaFeO2 + 2HCl(g) (2)

Exposing the alloy 310H samples to the environment consisting of
NaCl + 20 vol. % H2O + 8 vol. % O2 resulted in a catastrophic cor- 2NaCl + Fe2O3 + H2O(g) = 2NaFeO2 + 2HCl(g) (3)
rosion rate. Prior to the investigation, it was hypothesized that H2O
vapor could react with NaCl on the deposit surface giving gaseous HCl Cr2O3 +4NaCl +2H2O +3/2O2 = 2Na2CrO4 + 4HCl(g) (4)
which could be lost into the flowing gas thereby decreasing the corro- The standards Gibbs free energy for the reactions at 500 °C calcu-
sive nature of the exposure condition. However, the hypothesis was lated using HSC chemistry are shown in Table 4. NaCl disrupts the
falsified as the exposure condition produced the highest mass gain of formation of Cr2O3 on the surface of the metal samples. It reacts with
310H alloy compared to the exposure environment without water vapor Cr2O3 preliminarily formed just after the exposure (reaction (4)). Al-
or NaCl (Fig. 2). Under this condition, the alloy 310H exhibited an though HCl produced in reaction (1)-(4) can be able to dissociate ac-
initial rapid attack which can be called the breakaway stage and then cording to reaction (5) giving chlorine gas, the equilibrium will be
with time the corrosion behavior turned into a parabolic rate law. This pushed to the left in this condition due to the continuous supply of H2O
behavior has also been reported for Fe-20Cr alloy [12]. The simulta- vapor and consumption of O2 in the oxidation of alloy elements:
neous presence of water vapor and NaCl reacted with substrate ele-
ments forming either sodium chromate (IV) or sodium ferrate (III) ac- 2HCl(g) + 1/2O2(g) = Cl2(g) + H2O(g) (5)
cording to reactions (1)–(4):
HCl generated diffuses to the metal/scale interface and reacts with

263
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 12. XPS peak fitting spectra obtained from corrosion product layer at 500 °C for 24 h in run 3 exposure condition: (a) survey spectrum of corrosion product layer,
(b) C 1 s characteristic peak spectra, (c) Cl 2p3/2 characteristic peak spectra, (d) O 1 s characteristic peak spectra, (e) S 2p3/2 characteristic peak spectra, (f) Fe 2p3/2
characteristic peak spectra, and (g) Fe 3 s characteristic peak spectra.

metal elements in the substrate according to reaction (6) and (7). to cause accelerated corrosion of Fe by the formation of iron hydroxide
which was evidently detected by XPS analysis in Fig. 11(c), according
Fe+2HCl(g) =FeCl2 +H2(g) (6)
to reaction (10):
Cr+3HCl(g) =CrCl3(g) +3/2H2(g) (7)
2Fe + 3H2O(g) + 3/2O2(g) = 2Fe(OH)3 (10)
CrCl3 and FeCl2 formed diffused outwards and react with O2 and
Fe(OH)3 that is formed decomposes rapidly into Fe2O3:
H2O vapor which diffused inwards
2Fe(OH)3 = Fe2O3 + 3H2O(g) (11)
2FeCl2 + 1/2O2(g) +2H2O(g) =Fe2O3 +4HCl(g) (8)
Cao et al. [12] reported that a continuous supply of moisture leads
2CrCl3 + 3H2O(g) = Cr2O3 + 6HCl(g) (9)
to enhanced degradation of iron due to the formation of iron hydroxide.
Cr2O3 formed in reaction (9) continue reacting with the remaining Similarly, due to the continuous supply of moisture in the present si-
unreacted NaCl on the surface according to reaction (4). The reactions tuation, the penetration of the H2O to the metal/scale interface to form
of NaCl with Cr2O3 and Cr break the protective Cr2O3 layer, forming Fe(OH)3 presumably accelerated corrosion of the alloy for a longer time
some pits and crevices as shown in Fig. 9(a) that exposed poor pro- as depicted in Fig. 2(b). Once the decomposition of iron hydroxide
tective Fe-rich oxide that was underneath. H2O and HCl rapidly diffused occurred, hematite was formed on the metal/scale surface. This cor-
to the region depleted in Cr at metal/scale interface. Because NaCl was rosion product was clearly confirmed by XRD results shown in Fig. 10.
consumed in the reactions with substrate elements, a large amount of With increasing exposure time, the formation of hematite caused the
H2O remained after NaCl reacted, and H2O entered the region rich in Fe thickness of the corrosion products to increase resulting in slowed

264
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Fig. 13. XPS peak fitting spectra obtained from corrosion product layer at 500 °C for 24 h in run 4 exposure condition: (a) survey spectrum of corrosion product layer,
(b) C 1 s characteristic peak spectra, (c) S 2p3/2 characteristic peak spectra, (d) O 1 s characteristic peak spectra, (e) Fe 2p3/2 characteristic peak spectra.

Table 3 Table 4
Pearson's correlations matrix for the corrosive components. Reactions proposed for the mechanisms and their corresponding calculated
values of standard Gibbs free-energy changes (ΔGo) (Calculated using HSC
NaCl SO2 H2O Mass gain
chemistry 6.0).
NaCl −0.33 −0.33 0.525** Chemical reactions ΔG°500 (kJ/mol)
°C
SO2 −0.33 −0.33 −0.98**
H2O −0.33 −0.33 0.515** NaCl + H2O + O2 (run 2) environment
Mass gain 0.525 −0.98 0.515 Cr + 2NaCl + 3/2O2(g) + H2O(g) = Na2CrO4 + 2HCl(g) −378.195
2NaCl + 2Fe + 3/2O2(g) + H2O(g) = 2NaFeO2 + 2HCl(g) −403.203
** Correlation is significant at the 0.01 level (1-tailed). 2NaCl + Fe2O3 + H2O(g) = 2NaFeO2 + 2HCl(g) +212.708
Fe+2HCl(g) =FeCl2 +H2(g) −44.843
inward diffusion of H2O, O2, and HCl and outward diffusion of CrCl3, Cr+3HCl(g) =CrCl3(g) +3/2H2(g) −27.001
2FeCl2 +1/2O2 (g) + 2H2O(g) =Fe2O3 + 4HCl(g) −232.397
and FeCl2. Eventually, this situation reduced the total amounts of O2,
2CrCl3 (g) + 3H2O(g) = Cr2O3 +6HCl(g) −259.632
HCl, and H2O in the subscale, and thus corrosion rate gradually slowed Cr2O3 +4NaCl +2H2O +3/2O2 = 2Na2CrO4 + 4HCl(g) +101.891
down after about 22.1 h and proceed slowly after 67.4 h as shown in 2Fe + 3H2O(g) + 3/2O2(g) = 2Fe(OH)3 −403.631
Fig. 3. The mechanism is supported by our observation that Kp1 > Kp5 2Fe(OH)3 = Fe2O3 + 3H2O(g) −212.281
Reference condition (run 1)
in NaCl + H2O + O2 (run 2) as displayed in Table 5. The parabolic rate
Cr + 2NaCl + 2O2(g) = Na2CrO4 + Cl2(g) −384.598
constants were calculated using plots of (m vs t1/2) for each exposure at 2Fe + 2NaCl + 2O2(g) = 2NaFeO2 + Cl2(g) −276.874
different time intervals. The minimum corrosion rate is the steady-state 2NaCl + SO2(g) + H2O(g) + 1/2O2(g) = Na2SO4 + 2HCl −95.341
growth of the scale. (g)
Fe + Cl2(g) = FeCl2 −243.452
Cr + Cl2(g) = CrCl2 −298.285
4.2. Influence of SO2 on the corrosion kinetics of 310H 2FeCl2 + 3/2O2(g) = Fe2O3 + 2Cl2(g) −129.006
2CrCl2(g) + 3/2O2(g) = Cr2O3 + 2Cl2(g) −593.438
NaCl + SO2 + O2 environment (run 3)
The corrosion behavior of the alloy 310H in the reference exposure 2NaCl + SO2(g) + O2(g) = Na2SO4 + Cl2(g) −101.745
(NaCl + 300 ppmSO2 + 20 vol.H2O + 8 vol.O2) (run 1) showed three
different stages. The first stage was slow, turning into accelerated phase
and then plateaued off (Fig. (2b)). Corrosion started with the reaction of shown in Fig. 10. These reactions could have occurred during the in-
NaCl with the metal at the substrate/salt interface leading to the for- cubation phase.
mation of Na2CrO4 or Na2Fe2O4 which were detected by XRD analysis

265
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

Table 5
Parabolic rate constant for 310H corroded in different corrosive environment at 500 °C at different time.
Exposure condition Kp1 Kp2 Kp3 Kp4 Kp5

−3 −3 −4 −4
NaCl + SO2+H2O + O2 (run 1) 1.17× 10 (0.74-3.5 h) 1.36× 10 (4.5-12.7 h) 2.86× 10 (26.1-29.2 h) 1.70× 10 (34.8-59.1 h) 1.40× 10−4 (67.3-78.0 h)
NaCl+H2O + O2 (run 2) 7.81× 10−3 (1.6-3.1 h) 7.45× 10−3 (4.0-10 h) 2.99× 10−3 (20.9-29.9 h) 2.01× 10−3 (35.2-66.8 h) 2.53× 10−3 (69.7-78.7 h)
NaCl + SO2+O2 (run 3) 7.96× 10−3 (0.2-0.9 h) −2.91× 10 (1.3-6.4 h)
-4
−6.6× 10 (21.0-30.2 h)
-5
6.01× 10−5 (34.8-65.2 h) 3.83× 10−6 (70.3-79.5 h)
SO2+H2O + O2 (run 4) 1.14× 10−3 (1.4-2.0 h) 2.07× 10−4 (5.4-10.2 h) 2.09× 10−5 (11.3-28.3 h) 5.21× 10−5 (28.3-47.8 h)

Cr + 2NaCl + 2O2(g) = Na2CrO4 + Cl2(g) (12) corrosion attack was very minimal as compared to the other conditions
(run 1 and 2). The observation of a very low corrosion attack in this
2Fe + 2NaCl + 2O2(g) = 2NaFeO2 + Cl2(g) (13)
condition is in agreement with [31]. The corrosion kinetics in this ex-
On top or within the salt deposit, sulfation of NaCl took place ac- posure environment showed a multi-stage corrosion process (Fig. 3).
cording to the Hargreaves mechanism reaction (14) [43] converting The corrosion attack started rapidly within the first 2 h and was at-
NaCl to Na2SO4. tributed to the vaporization of CrO2(OH)2. The loss of chromium, de-
2NaCl + SO2(g) + H2O(g) + 1/2O2(g) = Na2SO4 + 2HCl(g)(14) pleted the oxide and the underlying alloy in chromium. The scale
Because H2O vapor was participating in the sulfation reaction of formed, which was non-protective consisted of hematite layer. After 2 h
NaCl, there could be little water vapor at the metal/scale interface. As a the corrosion attack slowed down. The only components present in this
result, reactions (12) and (13) released chlorine as a gas at the deposit/ condition were SO2, O2 and H2O. The oxidation of SO2 to SO3 in the
metal interface and reacted with the exposed metals forming metal absence of a catalyst is a very slow reaction [31] and the formation of
chlorides which evaporated to the deposit/gas surface where they be- sulfuric acid under the present condition was prevented by maintaining
come oxidized in regions with appropriate oxygen partial pressure. This the temperature at 500 °C and was not thermodynamically favored as
process resulted in the accelerated mass gain during the period between shown in reaction (15).
9.9–26.9 h after the exposure test begins (Fig. 3). The combination of
SO3(g) + H2O(g) = H2SO4(g) ΔG500°C = 6.66KJ/mol (15)
the resulting sulfate layer and Fe2O3 somewhat imposed some degree of
resistance to the inward transport of chlorine to the corrosion front. The Therefore, since the surface consists of metal oxide, SO2 and O2
corrosive species (NaCl) which were not replenished become consumed reacted with the oxide scale forming adsorbed metal-sulfur trioxide
in the corrosion process with time. As a result, the corrosion rate slowed which was detected by XPS in Fig. 13(c and d).
down with time due to the formation of less corrosive Na2SO4, deple-
3SO2(g) + 3/2O2(g) + M2O3(s) = M2(SO4)3(s) (16)
tion of chlorine at the metal/deposit interface and the continuous flow
of flue gas components. This observation is in line with previous lit- Fe: ΔG500°C = -230.19KJ/molCr: ΔG500°C = -219.83KJ/mol
erature [26,44].
When H2O vapor was removed from the system (run 3), the alloy 3SO2(g) + 3/2O2(g) + M2O3(s) = 2 M(SO4)3 (ads) (17)
310H displayed different corrosion behavior from the reference ex- Where M = Fe, Cr
posure as well as to the kinetics displayed for the condition in run 2. Under the present condition of 300 ppm SO2 and 8 vol. % O2, the
Corrosion was very rapid within the first hour but decreased until after formation of solid metal sulfate was not thermodynamically possible
about 65 h when it was nearly equal to the rate without deposits since the minimum SO2 concentration needed to form Fe2(SO4)3 and
(Fig. 3). This observation was in agreement with Grabke et al. [13] who Cr2(SO4)3 are 1500 ppm and 500 ppm respectively. However, adsorbed
used low alloyed steels under NaCl in HCl and SO2. The main product sulfate on the metal surface can still be formed and reduced the oxi-
within the first hour was Fe2O3 as detected by XRD (Fig. 10) which dation of the alloy metal, slowing down the corrosion attack and thus a
indicate that NaCl reacted with the alloy elements (Fe and Cr) forming very low mass gain was observed as shown in Fig. 2(a) and (b). The
sodium ferrate and chromate releasing chlorine at the deposit/scale small atomic % (2.34 at.%) of S detected by EDX after 24 h of exposure
interface. Chlorine then formed volatile FeCl2 which becomes oxidized (Table 2) is in agreement with the inability of the metal alloy elements
at the deposit/gas interface forming Fe2O3, thus rapid corrosion attack to form solid sulfate and hence only correspond to the adsorbed metal
was observed in the first hour. However, the reason for the decrease in sulfate. Panas et al. [45] explained using the first principle approach for
corrosion rate after 1 h was not clear. Grabke et al. [13] pointed that in the conversion of Cr2O3 to CrO2(OH)2. Cr2O3 readily forms tetra-
the presence of SO2 and HCl on NaCl-induced corrosion of low alloyed hedrally coordinated Cr(VI) with oxygen molecules. When it reacts with
steel, sulfation and chlorination come to an equilibrium resulting in no water molecules, hydrolysis of the oxygen bonds connecting Cr(VI) to
change in mass. Nevertheless, this explanation cannot apply in the the oxide occurs, forming CrO2(OH)2(g). S(VI) have the potential to
present situation in which HCl was not expected to form since H2O bind to a Cr2O3 surface in the same way as Cr(VI) since chromate and
vapor was removed. Likewise, the decrease in mass was assumed to be sulfate have similar chemical properties [31]. This suggests that there
connected to the formation of sulfates which were evidently detected by might be a competition between the formation of chromate and sulfate
XPS (Fig. 12(e) and XRD (Fig. 10) analyses. Chlorine was generated by on the Cr2O3 surface and thus the vaporization of CrO2(OH)2(g) may be
the sulfation reactions within or on top of the deposit, not at the metal/ suppressed in environments that produce surface sulfate. Nevertheless,
deposit interface, and, therefore, most of it diffused to the atmosphere this valuable effect of sulfate does not entirely stop rapid oxidation of
and get lost into the flowing gas. The Na2SO4 formed has a dense and the metal samples as seen in Fig. 3 that between 5.3–6.5 h of exposure
compact layer as seen in Fig. 7. It prevented the ingress of chlorine to some increase in corrosion attack was noticed.
the metal/scale interface, as well as prevented the surface reactions As shown in this study, a slight change in the exposure environment
between chlorine and metal elements. Sodium sulfate could not react has a significant impact on the corrosion behavior of alloy 310H which
with the metal elements [18]. Thus, chlorination and active oxidation is mostly used as superheater material. The presence of SO2 either in the
were increasingly suppressed by the sulfation reaction. This mechanism absence or presence of NaCl salt or H2O reduced the corrosion attack
could possibly be the reason why corrosion rate slowed down in the and the competition for the formation of chromate and sulfate maybe
period of 1–65 h of exposure shown in Fig. 3. another way of elucidating the beneficial effect of SO2. When water
When NaCl was removed in the system that is in the presence of vapor was removed, the corrosion attack was minimal and it was
300 ppm SO2 + 20 vol. % H2O + 8 vol.% O2 (run 4), the overall comparable to the environment without NaCl. Therefore, based on the

266
T. Wenga, et al. Corrosion Science 154 (2019) 254–267

results of this study, it is recommended to dry the waste prior to [12] M. Cao, L. Liu, Z. Yu, L. Fan, L. Ying, F. Wang, Studies on the corrosion behavior of Fe-
combustion as it may prolong the service lifetime of the boiler tube 20Cr alloy in NaCl solution spray at 600 °C, Corros. Sci. 133 (2018) 165–177.
[13] H.J. Grabke, E. Reese, M. Spiegel, The effects of chlorides, hydrogen chloride, and sulfur
materials and reduce the maintenance cost. dioxide in the oxidation of steels below deposits, Corros. Sci. 37 (1995) 1023–1043.
[14] C. Wang, F. Jiang, F. Wang, Corrosion inhibition of 304 stainless steel by nano-sized Ti/
silicone coatings in an environment containing NaCl and water vapor at 400–600 °C,
5. Conclusion Oxid. Met. 62 (2004) 1–13.
[15] F. Wang, S. Geng, S. Zhu, Corrosion behavior of a sputtered K38G nanocrystalline coating
This study investigated the corrosion kinetics and associated cor- with a solid NaCl deposit in wet oxygen at 600 to 700 °C, Oxid. Met. 58 (2002) 185–195.
[16] Y. Shu, F. Wang, W. Wu, Synergistic effect of nacl and water vapor on the corrosion of
rosion mechanisms when the stainless steel 310H samples were exposed 1Cr-11Ni-2W-2Mo-V Steel at 500-700 °C, Oxid. Met. 51 (1999) 97–110.
to corrosive environments containing different compositions of NaCl, [17] C. Ciszak, I. Popa, J.-M. Brossard, D. Monceau, S. Chevalier, NaCl induced corrosion of Ti-
6Al-4V alloy at high temperature, Corros. Sci. 110 (2016) 91–104.
H2O, and SO2. To the author's best knowledge, the present work is the
[18] S. Karlsson, J. Pettersson, L.G. Johansson, J.E. Svensson, Alkali induced high temperature
first study to investigate and contrast the influence of different exposure corrosion of stainless steel: the influence of NaCl, KCl and CaCl2, Oxid. Met. 78 (2012)
environment on the corrosion kinetics of 310H which normally used as 83–102.
[19] H.P. Nielsen, F.J. Frandsen, K. Dam-Johansen, L.L. Baxter, The implications of chlorine-
superheater material and to determine the contribution of the corrosive associated corrosion on the operation of biomass-fired boilers, Prog. Energy Combust. Sci.
species to the corrosion rate. The results from weight gain kinetics and 26 (2000) 283–298.
corrosion product characterization using scanning electron microscopy [20] Y. Shinata, F. Takahashi, K. Hashiura, NaCl-induced hot corrosion of stainless steels,
Mater. Sci. Eng. 87 (1987) 399–405.
(SEM), energy dispersive spectroscopy (EDS), x-ray diffraction (XRD), [21] N. Hiramatsu, Y. Uematsu, T. Tanaka, M. Kinugasa, Effects of alloying elements on NaCl-
x-ray photoelectron spectroscopy (XPS), and theoretical calculations induced hot corrosion of stainless steels, Mater. Sci. Eng. A 120-121 (1989) 319–328.
[22] C.-C. Tsaur, J.C. Rock, C.-J. Wang, Y.-H. Su, The hot corrosion of 310 stainless steel with
were used to divulge the underlying corrosion mechanisms. The ex- pre-coated NaCl/Na2SO4 mixtures at 750 °C, Mater. Chem. Phys. 89 (2005) 445–453.
posure environment containing NaCl + 20 vol.% H2O + 8 vol.% O2 [23] T. Jonsson, N. Folkeson, M. Halvarsson, J.E. Svensson, L.G. Johansson, Microstructural
showed a linear corrosion attack for prolonged exposure time, then investigation of the HCl-induced corrosion of the austenitic alloy 310S (52Fe26Cr19Ni) at
500 °C, Oxid. Met. 81 (2014) 575–596.
turned into a steady-state parabolic rate law. This condition is the most [24] Y. Shinata, Y. Nishi, NaCl-induced accelerated oxidation of chromium, Oxid. Met. 26
aggressive environment as it resulted in approximately 400% mass gain (1986) 201–212.
[25] S.C. Okoro, M. Montgomery, F.J. Frandsen, K. Pantleon, Effect of water vapor on high-
increase than the condition with NaCl + 20 vol.% H2O + 300 ppm SO2
temperature corrosion under conditions mimicking biomass firing, Energy Fuels 29
+ 8 vol.% O2. The environment containing NaCl + 300 ppm SO2 + (2015) 5802–5815.
8 vol.% O2 result in breakaway corrosion attack within a very short [26] D.W. McKee, D.A. Shores, K.L. Luthra, The Effect of SO2and NaCl on High Temperature
Hot Corrosion, J. Electrochem. Soc. 125 (1978) 411–419.
period of time (1 h), and a decrease in mass gain due to suppression of [27] A. Zahs, M. Spiegel, H.J. Grabke, Chloridation and oxidation of iron, chromium, nickel
chloridation by sulfation of NaCl. In the absence of NaCl salt, the for- and their alloys in chloridizing and oxidizing atmospheres at 400–700 °C, Corros. Sci. 42
mation of adsorbed surface sulfate inhibited the vaporization of (2000) 1093–1122.
[28] N. Folkeson, J. Pettersson, C. Pettersson, L.G. Johansson, E. Skog, B.Å. Andersson,
chromic acid. The contribution of the corrosive species to the corrosion S. Enestam, J. Tuiremo, A. Jonasson, B. Heikne, J.E. Svensson, fireside corrosion of
were in the order of NaCl > H2O vapor > SO2. stainless and low alloyed steels in a waste-fired CFB boiler; The effect of adding sulfur to
the fuel, Mater. Sci. Forum 595-598 (2008) 289–297.
[29] K. Iisa, Y. Lu, K. Salmenoja, Sulfation of potassium chloride at combustion conditions,
Data availability Energy Fuels 13 (1999) 1184–1190.
[30] S. Karlsson, J. Pettersson, J.E. Svensson, L.G. Johansson, KCl-induced high temperature
corrosion of the austenitic stainless steel 304L – the influence of SO2, Mater. Sci. Forum
All the necessary raw/processed data are included in this paper 696 (2011) 224–229.
[31] A. Järdnäs, J.-E. Svensson, L.-G. Johansson, Influence of SO2 on the oxidation of 304L
Acknowledgements Steel in O2 + 40%H2O at 600 °C, Oxid. Met. 69 (2008) 249–263.
[32] H. Asteman, J.E. Svensson, M. Norell, L.G. Johansson, Influence of water vapor and flow
rate on the high-temperature oxidation of 304L; effect of chromium oxide hydroxide
The Supports from National Key Technologies R&D Program of evaporation, Oxid. Met. 54 (2000) 11–26.
[33] M. Halvarsson, J.E. Tang, H. Asteman, J.E. Svensson, L.G. Johansson, Microstructural
China (No. 2018YFC190135), NSFC (No.51776139), Tianjin Science investigation of the breakdown of the protective oxide scale on a 304 steel in the presence
and Technology Project (No.18ZXSZSF00120) are gratefully acknowl- of oxygen and water vapour at 600 °C, Corros. Sci. 48 (2006) 2014–2035.
edged. [34] T. Jonsson, B. Pujilaksono, H. Heidari, F. Liu, J.E. Svensson, M. Halvarsson,
L.G. Johansson, Oxidation of Fe–10Cr in O2 and in O2+H2O environment at 600 °C: a
microstructural investigation, Corros. Sci. 75 (2013) 326–336.
References [35] J. Lehmusto, B.J. Skrifvars, P. Yrjas, M. Hupa, Comparison of potassium chloride and
potassium carbonate with respect to their tendency to cause high temperature corrosion
of stainless 304L steel, Fuel Process. Technol. 105 (2013) 98–105.
[1] J. Song, Y. Sun, L. Jin, PESTEL analysis of the development of the waste-to-energy in- [36] N. Israelsson, K. Hellström, J.E. Svensson, L.G. Johansson, KCl-Induced corrosion of the
cineration industry in China, Renew. Sustain. Energy Rev. 80 (2017) 276–289. FeCrAl Alloy Kanthal®AF at 600 °C and the effect of H2O, Oxid. Met. 83 (2015) 1–27.
[2] A.J. Pedersen, S.C. van Lith, F.J. Frandsen, S.D. Steinsen, L.B. Holgersen, Release to the [37] J. Pettersson, J.E. Svensson, L.G. Johansson, KCl-Induced Corrosion of a 304-type aus-
gas phase of metals, S and Cl during combustion of dedicated waste fractions, Fuel tenitic stainless steel in O2 and in O2 + H2O environment: the influence of temperature,
Process. Technol. 91 (2010) 1062–1072. Oxid. Met. 72 (2009) 159–177.
[3] A. Phongphiphat, C. Ryu, Y.B. Yang, K.N. Finney, A. Leyland, V.N. Sharifi, [38] T. Jonsson, B. Pujilaksono, S. Hallström, J. Ågren, J.E. Svensson, L.G. Johansson,
J. Swithenbank, Investigation into high-temperature corrosion in a large-scale municipal M. Halvarsson, An ESEM in situ investigation of the influence of H2O on iron oxidation at
waste-to-energy plant, Corros. Sci. 52 (2010) 3861–3874. 500 °C, Corros. Sci. 51 (2009) 1914–1924.
[4] W. Ma, T. Wenga, N. Zhang, G. Chen, B. Yan, Z. Zhou, X. Wu, Full-scale experimental [39] L. Yang, C. Bo, W. Junwei, W. Zhiping, L. Wensheng, Corrosion behavior of Cr, Fe and Ni
investigation of deposition and corrosion of pre-protector and 3rd superheater in a waste based superalloy in molten NaCl, Rare Met. Mater. Eng. 43 (2014) 17–23.
incineration plant, Sci. Rep. 7 (2017) 17549-11750. [40] L. Boonsongsup, K. Iisa, W.J. Frederick, Kinetics of the sulfation of NaCl at combustion
[5] G. Chen, N. Zhang, W. Ma, V.S. Rotter, Y. Wang, Investigation of chloride deposit for- conditions, Ind. Eng. Chem. Res. 36 (1997) 4212–4216.
mation in a 24MWe waste to energy plant, Fuel 140 (2015) 317–327. [41] Y.S. Li, M. Spiegel, S. Shimada, Corrosion behaviour of various model alloys with
[6] D.O. Albina, K. Millrath, N.J. Themelis, Effects of feed composition on boiler corrosion in NaCl–KCl coating, Mater. Chem. Phys. 93 (2005) 217–223.
waste-to-energy plants, 12th North American Waste-to-Energy Conference (2004) 1–11. [42] S. Liu, Z. Liu, Y. Wang, J. Tang, A comparative study on the high temperature corrosion of
[7] W. Ma, G. Hoffmann, M. Schirmer, G. Chen, V.S. Rotter, Chlorine characterization and TP347H stainless steel, C22 alloy and laser-cladding C22 coating in molten chloride salts,
thermal behavior in MSW and RDF, J. Hazard. Mater. 178 (2010) 489–498. Corros. Sci. 83 (2014) 396–408.
[8] F. Frandsen, Next Generation of High-Efficient Waste Incinerators, in, Technical [43] S. Uchida, H. Kamo, H. Kubota, K. Kanaya, Reaction kinetics of formatton of HCI in
University of Denmark, Department of Chemical Engineering, Final Report, FORSKEL- municipal refuse incinerators, Ind. Eng. Chem. Process Des. Dev. 22 (1983) 144–149.
10487, (2010). [44] W. Ma, Study on Identification Method and Thermal Behavior of Chlorine in Municipal
[9] F.J. Frandsen, Ash Formation, Deposition and Corrosion When Utilizing Straw for Heat Solid Waste Incineration, Environmental Science and Engineering, Tianjin University,
and Power Production, Department of Chemical Engineering, Technical University of China, 2010.
Denmark, 2010. [45] I. Panas, J.-E. Svensson, H. Asteman, T.J.R. Johnson, L.-G. Johansson, Chromic acid
[10] Y. Shu, F. Wang, W. Wu, Corrosion behavior of pure Cr with a solid NaCl deposit in O2 evaporation upon exposure of Cr2O3(s) to H2O(g) and O2(g) – mechanism from first
plus water vapor, Oxid. Met. 54 (2000) 457–471. principles, Chem. Phys. Lett. 383 (2004) 549–554.
[11] F. Wang, Y. Shu, Influence of Cr content on the corrosion of Fe–Cr alloys: the synergistic
effect of NaCl and water vapor, Oxid. Met. 59 (2003) 201–214.

267