Engineering Failure Analysis 28 (2013) 6977

Contents lists available at SciVerse ScienceDirect

Engineering Failure Analysis

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

Caustic corrosion in a boiler waterside tube: Root cause and mechanism

Farhad Daneshvar-Fatah, Amir Mostafaei , Reza Hosseinzadeh-Taghani, Farzad Nasirpouri
Laboratory of Failure Analysis, Research Centre for Advanced Materials and Minerals Processing, Department of Materials Engineering, Sahand University
of Technology, Tabriz, Iran

a r t i c l e i n f o a b s t r a c t

Article history: We report a failure occurred in a boiler tube of low alloy steel, SA-210 Grade A-1. A severe
Received 4 July 2012 thickness reduction was detected in a boiler tube during periodic investigations. Two iden-
Received in revised form 6 September 2012 tical parallel grooves along the horizontal tube were observed. The inner surface of the tube
Accepted 12 September 2012
consisted of three distinct parts: (1) dark gray scale under the grooves, (2) hemi-elliptical
Available online 13 October 2012
grooves, and (3) brown porous scale above the grooves. Microscopic investigations and
microhardness tests revealed that the microstructure and properties of the base metal is
Keywords:
intact along the tube. Moreover, some needle shape crystals which represented Na2FeO2
Failure analysis
Metallurgical examination
and NaFeO2 compounds were observed in SEM images taken from the groove scale. An
Boiler tube excessive amount of Na and Cu was detected by EDS in the groove. The XRD results con-
Caustic corrosion rmed the presence NaFeO2 which is the main product of caustic corrosion. Therefore, it
Low alloy steel was concluded that the evaporation at the waterline of the partially lled tube leaded to
locally caustic concentration which dissolves the protective magnetite layer and subse-
quently attacked the bare metal.
2012 Elsevier Ltd. All rights reserved.

1. Introduction

Boiler failure is a major concern in power plants and other industrial units [13] and the tube corrosion is known to be the
main cause. The total cost of boiler tube failures in power plants is estimated to be about $5 billion a year [4]. For ensuring
long term and convenient operation of the boiler, water hardness and contaminant control is vital.
The water treatment process begins with initial step of clarication and ltration in which suspended solids are separated
from the water. Then the water passes through a resin bed (mostly zeolite) and hardness ions (Ca2+ and Mg2+). In the next
step by heating the feedwater the solubility of the dissolved gas is reduced so the water is deaerated. Thermal deaeration can
reduce feedwater oxygen concentration to less than 7 ppb. The remained oxygen is removed by adding oxygen scavengers.
The most commonly used oxygen scavenger for boilers are sodium sulte and hydrazine.
The most common internal water treatment programs applied to the boiler are: (1) phosphate treatment, (2) caustic
treatment, and (3) amine or polyamine treatment. To decrease the susceptibility of corrosion, the pH of the boiler water
should be maintained in the range of 911. Addition of alkaline agents such as sodium hydroxide (12 ppm) and sodium
phosphate (Na3PO4) to the makeup water is a common treatment for providing the immune alkaline environment. Moreover,
this alkaline environment contributes to formation of nonadherent sludge. It should be considered that above 93 C, NaOH
can act as a severe corrodent to steel [5].
Scale deposits in boiler system are a main concern since they considerably increase the fuel consumption. In order to con-
trol the scale formation, phosphates are used to form controllable participates (phosphate treatment). Phosphate reacts with
calcium and magnesium to form a soft particle which can be suspended in the boiler water and removed through blowdown.
Chelant and dispersant programs are other common methods to control the scale formation [6].

Corresponding author. Tel.: +98 9126216801.

E-mail addresses: [email protected], [email protected] (A. Mostafaei).

1350-6307/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engfailanal.2012.09.010
70 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977

Moreover, phosphate addition to the makeup water prevents the formation of concentrated free caustic. However, as a
result of improper maintenance, phosphate hideout and subsequently free caustic formation may occur. Due to evaporation,
local concentration of caustics can reach to 10100 mg/l which leads to caustic corrosion or caustic embrittlement.
In general, in the case of caustic treatment, use of excessive alkalines such as NaOH may lead to corrosion attack (caustic
gouging and stress corrosion cracking) [5,7,8]. Caustic embrittlement which also referred to as caustic cracking is a form of
stress corrosion cracking that takes place in highly concentrated alkaline environments. This brittle failure is common at
welded joints in which there is excessive residual stress [912]. Caustic corrosion, also known as caustic attack or caustic
gouging, is one of the most common causes of boiler tube failures [4]. In boiler tubes, generally due to evaporation, caustics
compounds such as sodium or potassium hydroxides concentrate locally. Once the pH exceeds 12.9, highly concentrated
caustic solution dissolves the protective magnetite lm and attacks the bare steel.
Khajavi et al. [13] recommended that in order to eliminate formation of corrosion products, the molar ratio of free caustic
sodium to phosphate should not exceed 2.6. Additionally, they also reported that if the ratio goes under 2.2, phosphate cor-
rosion will occur which is apparently similar to caustic corrosion.
In this current study, failure analysis of a boiler tube was investigated by visual inspection, X-ray diffraction (XRD),
microhardness testing and microstructure studies using optical microscopy and scanning electron microscopy (SEM)
equipped with energy dispersive X-ray (EDS). Based on the results and discussions, a possible mechanism for under deposit
corrosion in this study was proposed.

2. Experimental procedure

2.1. On-site inspection

The boiler was out of service for a few months and the tubes were washed by caustic compounds for scale removal. When
the boiler was idle, wet lay-up method was applied to prevent the corrosion. In this approach, all the tubes were completely
lled with demineralized and deaerated water. The water was also treated with sodium hydroxide to establish a pH of at
least 11 and oxygen scavenger (sodium sulte).
After repairs and several water ush-outs, the boiler was restarted and worked for 4 months. During operation, boiler
water was under phosphate treatment. For providing the proper alkaline environment, sodium hydroxide was used. Sodium
sulte was added to the storage space for deaerating heater. It was reported that at the beginning of the operation, water
analysis showed unusual uctuations in pH and conductivity. Operators carried out several blowdowns to control the water
chemistry.
During periodic investigations, a severe thickness reduction was detected in a horizontal boiler tube by non-destructive
testing (NDT). It was reported that there was a uniform oxide layer on the outer surface of the tube. Moreover, two parallel
longitudinal trenches were observed on the internal surface. In order to examine the internal surface of the failed tube, a part
of it was cut lengthwise and sent to the laboratory.

2.2. Characterization methods

Three samples from different parts of the as-received specimen were selected for quantometric, metallographic, electron
microscopy, X-ray diffractometery and hardness examinations. Furthermore, the oxide layer composition and corrosion
products were analyzed. Spectroscopic methods were utilized as follows:

(i) Chemical composition. To specify the chemical composition of the tube, quantometric analysis was applied.
(ii) Optical microstructure studies. For microstructural examination, specimens were cut cross-sectional from the adjacent
and apart from the groove. The samples were then mounted and polished. After preparation, the samples were etched
by nital 2% solution for 15 s to study the microstructure morphology of the sample.
(iii) Microhardness measurement. Vickers microhardness test was performed by Akashi (M400 G1 Leco) under indentation
load of 100 g for 15 s.
(iv) Electron microscopy analysis. The scanning electron micrographs of samples were analyzed using SEM (MV2300 Cam
Scan), equipped with an energy dispersive X-ray (EDS-Oxford instrument).
(v) XRD analysis. The X-ray diffraction 0
(XRD) patterns were recorded on a D8 Advance-Bruckers AXS diffractometer with
Cu Ka radiation source (k = 1.54 A ) operated at 40 kV and 40 mA in the 2h range 10110 at the scan rate of 0.05 /s.

3. Results and discussions

3.1. Visual inspection

An image of the as-received tube is shown in Fig. 1 which presents three perspective of the sample: (a) inner side, (b)
outer side and (c) cross section images of the damaged tube. Based on the appearance and the color of the scale around
the groove, the tube could be separated to three distinct parts (Fig. 1a). As it can be seen in Fig. 1a, on the inner side of
 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977 71

Fig. 1. The as-received failed superheater tube: (a) inner side view, (b) outer side view and (c) cross sectional view.

the tube, there was an adherent gray scale (part A) and on the other side of the groove, there was porous brown scale (part B)
which can be attributed to the formation of magnetite (Fe3O4) and hematite (Fe2O3), respectively. On the outer surface of the
tube (Fig. 1b), an adhesive, uniform, and relatively thick oxide layer formed and there was not any sign of seam or weldment
on the tube. From Fig. 1c, it is clear that the maximum thickness reduction is approximately 40% and the trench is rather
elliptical.

3.2. Chemical composition analysis

The quantometric analysis result is given in Table 1. With regard to the chemical composition, the tube was made of low
alloy steel, SA-210 Grade A-1.

3.3. Metallographical examination

The microstructure and morphology of the tube at the damaged region was examined by optical microscopy and the
images are shown in Fig. 2. Metallographical images after polishing and etching procedure showed that the morphology
of the tube at the center of the sample consisted of pearlite islands in ferritic matrix (Fig. 2a). Fig. 2b and c represents the
area adjacent to the groove; no microcrack was observed, the scale layer is smooth and free from pits. Finally, Fig. 2d illus-
trated an optical micrograph taken from the groove region. By comparing Fig. 2ad, one can deduce that there were no
changes in the microstructure and it is uniform along the tube, and the grain boundaries remained intact.

3.4. Microhardness test

The microhardness tests were carried out on the cross-section of the sample in different distances from the groove.
Microhardness prole of the specimen is represented in Fig. 3. As it can be observed, there is not any signicant difference
in the hardness values. The brief results are given in Table 2.

3.5. SEM and EDS examination

SEM was used to study morphology of the oxide layer and corrosion products. Fig. 4 shows the cross-section micrograph
of the tube at the groove region. As it can be seen, the deposits on the surface of the base alloy consist of two different layers;

Table 1
Chemical analysis of the failed boiler tube.

Component C Si P Mn S Cr Ni Mo Fe
Wt% 0.15 0.19 0.015 0.52 0.004 0.03 0.02 0.02 99.051
72 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977

Fig. 2. Metallographic structures of the tube taken from (a) the center of the sample, (b) the area adjacent to the groove at dark gray scale, (c) the area
adjacent to the groove at brown scale and (d) the groove region. All micrographs are in 500 magnication.

Fig. 3. (a) The corresponding image and line scan on which the hardness was measured and (b) the hardness prole of the tube.
 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977 73

Table 2
Microhardness test results in different distances from the groove.

Distance from the groove (mm) 12 10 8 6 4 2 0 2 4 6 8 10 12

Hardness (HV) 160 159 160 158 159 158 159 159 160 159 159 158 159

a relatively thick layer adjacent to the base metal which probably contains products of a severe corrosion attack (A), and a
relatively thin uniform layer (B). Additionally, EDS results showed that in dark gray region, weight percentage of some ele-
ments such as Cu, Ca, Zn and Na is higher than brown area which can be attributed to presence of them in boiler circulating
water.
Surface of the tube at three regions (as determined in Fig. 1a) were studied using SEM. No microcrack was observed on the
surface of the specimen in different areas. Figs. 57 show SEM micrographs taken from groove groove region, brown scales
and dark gray scales, respectively. As it can be observed, there were some needle shape crystals on the groove scales (Fig. 5)
which did not exist on the hematite layer Fe2O3, (Fig. 6) and magnetite layer, Fe3O4 (Fig. 7).
EDS was utilized to identify the corrosion products on the inner side of the tube. The EDS samples were chosen from the
groove, brown and dark gray scales. Three points was analyzed in each sample and the average is reported in Table 3. Ele-
ments such as Al, Zn, P, Cu and Ca were detected in an abnormal amount. A signicant amount of Na also was detected in the
groove scale which conrmed presence of the caustic compounds in treatment water.
Si, Mg, Ca are common water contaminants which remain in the feedwater or makeup water even after ltration. Cu is
sometimes added to the makeup water intentionally due to its great conductive properties. Al presence can be attributed to
copper base alloys and zeloite that is used for boiler water treatment. Moreover, oxygen scavengers which are used for water
deaeration may contain sulfur. Finally, as mentioned before, the source of Na and P is chemicals that are used for boiler water
treatment [13].

3.6. XRD examination

Qualitative analysis of deposits at the groove region was carried out by XRD. For sample preparation, the damaged surface
was scraped and the prepared oxide powder was used. The XRD result is depicted in Fig. 8. It is clear that the groove scale
mostly consist of magnetite and hematite. Moreover, presence of Cu, Cu2O and NaFeO2 was conrmed by XRD.

3.7. Root cause and mechanism of the boiler failure

Generally, it was reported that there are three under deposit corrosion namely: (1) hydrogen damage, (2) acid phosphate
corrosion and (3) caustic gouging. All of them need heavy deposits and a concentration mechanism within those deposits
[8,13,14]. For hydrogen damage, the concentrating medium is usually chloride and contaminant levels are sufcient to cor-
rode the protective oxide layer (magnetite and hematite). Moreover, hydrogen atoms can easily penetrate in carbon steel

Fig. 4. SEM image of the cross-section of the groove region, where a two layered deposit is formed on base metal: porous corrosion products (part A) and
uniform scale (part B).
74 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977

Fig. 5. SEM micrographs of the groove region at two different magnications.

Fig. 6. SEM micrographs of the brown region (hematite layer) at two different magnications.

Fig. 7. SEM micrographs of the dark gray region (magnetite layer) at two different magnications.
 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977 75

Table 3
EDS elemental analysis of the inner surface of the tube from the different part of the specimen.

C O Zn Cu Mn Si S P Ca Al Na Ni Fe
Groove scale 2.75 21.21 1.09 8.06 2.16 0.80 0.83 1.06 3.90 4.16 2.70 0.53 50.75
Brown scale 1.84 18.98 0.11 1.27 0.64 1.31 0.24 0.10 0.61 4.62 0.22 0.24 69.82
Gray scale 2.04 24.90 0.39 3.93 1.49 1.00 0.49 2.16 2.02 4.50 0.65 0.39 56.04

Fig. 8. X-ray diffraction pattern of the corrosion products powder at groove region.

matrix. Then, hydrogen atoms tend to combine and form H2 molecules which cause distortion in structure and increases
stress in the metal matrix.
In decient phosphate treatment, the sodium to phosphate molar ratios (Na:PO4) may goes below 2.2 which leads to acid
phosphate corrosion. Khajavi et al. [13] reported that when a solution of phosphate compound such as trisodium phosphate
is concentrated at high temperature, precipitation of disodium phosphate crystals appears and the liquid would be rich in
NaOH. Also, due to presence of phosphate anions in the water, monitoring and control of pH is complicated and the situation
is prepared for acid phosphate corrosion. In the following equation, production of NaOH is presented:

Na3 PO4 H2 O $ Na2 HPO4 NaOH 1

When hydrogen damage and acid phosphate corrosion are common failure mechanisms in boilers, caustic treatment is
the best solution to prevent them. However, if the monitoring and control of the boiler water pH are not undertaken prop-
erly, excessive amount of NaOH may enter the boiler water and caustic gouging will be inevitable [8,15]. On the other hand,
insufcient NaOH dosing may leads to phosphate attack and formation of undesired scales.
There are two main factors in occurrence of caustic corrosion: (1) presence of caustics in boiler water and (2) concentra-
tion mechanism. With respect to the water treatment program applied, caustic soda was added to the water. Moreover, after
restarting the boiler, several load changes was recorded. Load change is one of the main causes of phosphate hideout in
which phosphate precipitates and consequently free caustic may form [13,16]. Therefore, as a result of load changes and bulk
phosphate hideout, boiler water analysis had shown unusual uctuations. By blowdowns the operators tried to control the
water chemistry, but the operating pressure was not changed. This situation leaded to water/steam stratication and forma-
tion of waterline in horizontal tubes.
Above the waterline a high heat ux area formed and the water in contact with the tube wall at the surface evaporated
rapidly leaving behind contaminants. Under these conditions, dissolved solids such as NaOH or suspended solids such as
copper, began to concentrate. Simultaneously, phosphates accumulate in the concentrated liquid and precipitated or reacted
with surface scales. Then the locally concentrated NaOH at the waterline attacked the protective oxide layer and bare metal.
The EDS results show a signicant amount of Na in the trench scale. The iron oxide which is formed on the inner surface of
the tube is an amphoter compound and it is susceptible to be attacked by sodium hydroxide. At temperatures above 90 C,
the concentrated NaOH attacks into the protective magnetite lm by the following reaction:

Fe3 O4 4NaOH ! 2NaFeO2 Na2 FeO2 2H2 O 2

In some literatures, the following reaction for dissolving the magnetite layer in caustic solution is also recommended
[17,18]:

3NaOH Fe3 O4 ! 3NaFeO2 H H2 O 3

Gradually, the highly concentrated NaOH dissolves the protective magnetite layer based on reactions 2 and 3, and the
bare iron reacts with NaOH according to the following reactions:
76 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977

Fig. 9. Scheme of caustic gouging along a longitudinal waterline where the tube was not lled completely.

NaOH H2 O Fe ! NaFeO2 3H 4

2NaOH Fe ! Na2 FeO2 H2 5

As a result of Eqs. (3)(5), Corrosion of base metal is taken place by gouging. According to these equations, although gen-
eration of hydrogen can cause hydrogen damage, it has not been conrmed in practice [18]. In caustic solution, hydrogen
produced by dissolving the bare metal move to the solution instead of penetrating to the base metal which is common in
acidic electrolytes.
Meanwhile, the base iron tries to restore the magnetite lm and consequently a porous non-protective magnetite layer is
formed. After elimination of the protective lm, the caustic solution enters the oxide layer. The trapped solution evaporates
and consequently, pH beneath the oxide layer increases. Fig. 4 clearly represents the diffusion of caustic solution inside the
magnetite layer which causes formation of a two layered scale on the base metal. Furthermore, it is reported that in caustic
solutions, Na2FeO2 and NaFeO2 crystals are needle shape which can be observed on the remained trench scale (Fig. 5).
From the appearance of the oxide layers in Fig. 1a, one can deduce that the brown and dark gray oxide layers mostly con-
sist of hematite and magnetite, respectively. It should be considered that magnetite often forms in aqueous environments.
With respect to the position of dark scale and higher amount of elements within it, one can conclude that the tube has been
partially lled with water through the boiler operation.
By studying the morphology of the deposits (Figs. 57), it is clear that uniform layer was formed on the inner side of the
tube. Although grain boundaries are highly susceptible to corrosion attacks such as stress corrosion cracking or intergranular
corrosion, there was no evidence of microcrack. Metallographical studies (Fig. 2) showed that the grain boundaries adjacent
to the trench are intact and no microcracking or hydrogen damage evident has appeared. The microhardness test results
which are presented in Table 2 showed that the properties of the region adjacent to the trench are similar to the parts rel-
atively far from it. Thus, results indicated that pitting corrosion, hydrogen damage, stress related corrosions, and oxygen at-
tack did not occur.
Fig. 9 presented the scheme of caustic gouging along a longitudinal waterline where the tube was not lled completely.
In steam generating systems such as boilers, copper is extensively added to the makeup water for water heating, steam
condensing and process temperature control [13]. When copper is dissolved or present as oxide particulate form in boiler
water, usually deposits in metallic form. For instance, cuprite (Cu2O) and the cupric oxide (CuO) reduce by the following
reactions and intensify the corrosion [18]:

CuO 3Fe ! 4Cu Fe3 O4 6

4CuO 3Fe ! 8Cu Fe3 O4 7

Therefore, copper is deposited either by direct exchange with iron or by reduction of copper oxide by hydrogen evolved
during corrosion. It is common to see large reddish stains of elemental copper intermixed with corrosion products such as
magnetite and hematite near caustic corrosion sites because of the hydrogen generation associated with the attack [16]. In a
partially lled tube, rapid evaporation of the water at the surface, accelerate deposition of mineral precipitates and partic-
ulates such as calcium and copper. Other parameters such as pH, temperature, ow rate, amine type, amine concentration,
and dissolved oxygen concentration impact copper concentration in deposits [13].

4. Conclusion and recommendations

In this study, the tube was lled in half of its content. Moreover, High heat ux causes increased deposition in which any
liquid water contacting the hot-side tube wall is vaporized. Therefore, concentration of the caustic compound and other min-
erals, which are present in water, increases in deposits. Again, water diffuses into these scales and evaporates due to high
heat ux, and consequently, the amount of caustic compound, which are extremely corrosive, increases in the waterline area
and nally leads to gouge. Finally, two identical parallel trenches were observed in horizontal tube of a boiler unit. Visual
inspections, metallurgical and morphology examinations, and composition analysis of scales showed that the failure is
due to evaporation of water inside the partially lled tube and concentrating of caustics at the water surface that leads to
caustic gouging along the waterline. However, metallographical and SEM results did not reveal any crack in microstructure
of the tube in different regions.
 F. Daneshvar-Fatah et al. / Engineering Failure Analysis 28 (2013) 6977 77

In order to prevent or postpone this problem, following suggestions are presented:

1. Precise control of water treatment and coordinated phosphate programs and caustic treatments which can prevent for-
mation of the free caustic. Also, unintentional release of caustic regeneration chemicals from makeup water demineral-
izers should be prevented.
2. By controlling the boiler load, it can be possible to prevent formation of waterline at horizontal tubes.
3. General inspection of the tube for nding any thickness reduction in wall is necessary.
4. Prevent the creation of waterlines in tubes which may be created by excessive blowdown rates, low water levels, and
excessive load reduction when pressure remains constant.

References

[1] Ameri M, Shamshirgaran SR. A case study: the effects of the design factors on the thermal prole of Shahid Rajaiee boiler. App Therm Eng
2008;28:95561.
[2] Noori S, Price A, John WH. A risk approach to the management of boiler tube thinning. Nucl Eng Des 2006;236:40514.
[3] Jonas O. Effective cycle chemistry control. In: ESAA power station chemistry conference, Rockhampton, Australia; 2000.
[4] Rahman MM, Purbolaksono J, Ahmad J. Root cause failure analysis of a division wall superheater tube of a coal-red power station. Eng Fail Anal
2010;17:14904.
[5] Damage mechanisms affecting xed equipment in the rening industry. API 571;2003:4.35 and 4.95.
[6] Amjad Z, editor. Science and technology of industrial water treatment. Boca Raton: CRC Press; 2010. p. 297342.
[7] Dooley RB, Bursik A. Hydrogen damage. PPChem 101 boiler and HRSG tube failures 2010;12(2):1227.
[8] Dooley RB, Bursik A. Caustic gouging. PPChem 101 boiler and HRSG tube failures 2010;12(3):18892.
[9] Abouswa K, Elshawesh F, Abuargoub A. Stress corrosion cracking (caustic embrittlement) of super heater tubes. Desalination 2008;222:6828.
[10] Parnian N. Failure analysis of austenitic stainless steel tubes in a gas red steam heater. Mater Des 2012;36:78895.
[11] Ranjbar K. Failure analysis of boiler cold and hot reheater tubes. Eng Fail Anal 2007;14:6205.
[12] Srikanth S, Gopalakrishna K, Das SK, Ravikumar B. Phosphate induced stress corrosion cracking in a waterwall tube from a coal red boiler. Eng Fail
Anal 2003;10:491501.
[13] Khajavi MR, Abdolmaleki AR, Adibi N, Mirfendereski S. Failure analysis of bank front boiler tubes. Eng Fail Anal 2007;14:7318.
[14] Dooley RB, Anderson B. Assessments of HRSGs trends in cycle chemistry and thermal transient performance. PPChem 101 boiler and HRSG tube
failures 2009;11(3):138.
[15] Daucik K, Dooley RB. Phosphate and NaOH treatments for the steam-water circuits of drum boilers of fossil and combined cycle/HRSG power plants. In:
The International Association for the Properties of Water and Steam. Plzen (Czech Republic): International Association for the Properties of Water and
Steam; 2011.
[16] Port RD, Herro HM. The Nalco guide to boiler failure analysis. Nalco Chemical Company, McGraw-Hill; 1991.
[17] Labuda EM, Cline DA. Waterwall tubes: waterside and reside corrosion: case studies, corrosion 2005. Houston (TX): NACE International; 2005.
[18] Howell AG, X. Energy, under-deposit corrosion mechanisms in boilers, corrosion NACExpo 2006. p. 06458.