High-temperature corrosion behavior of thermal spray coatings
Transcript of High-temperature corrosion behavior of thermal spray coatings
Lehigh UniversityLehigh Preserve
Theses and Dissertations
1992
High-temperature corrosion behavior of thermalspray coatingsScott T. BluniLehigh University
Follow this and additional works at: http://preserve.lehigh.edu/etd
This Thesis is brought to you for free and open access by Lehigh Preserve. It has been accepted for inclusion in Theses and Dissertations by anauthorized administrator of Lehigh Preserve. For more information, please contact [email protected].
Recommended CitationBluni, Scott T., "High-temperature corrosion behavior of thermal spray coatings" (1992). Theses and Dissertations. Paper 57.
U OR: Bluni, Scott m
ITlE:High...Temperature C rr si n
Behavior of Thermal Spray
Coatings
TE: May 31,1992
High-Temperature Corrosion Behaviorof Thermal Spray Coatings
by
Scott T. Bluni
... - - ·AThesii;--
Presented to the Graduate Committee
of Lehigh University
in Candidacy for the Degree of
Master of Science
In
Materials Science and Engineering
February 1992
Acknowledgements
The author would like to thank his advisor, Professor A.R. Marder, for his guidance
and assistance during this research program and in the preparation of this document.
The numerous metallographic consultations with Mr. Arlan Benscoter during the past
two years are also greatly appreciated. The author is also indebted to many fellow
graduate students, Brian Smith in, particular, who provided assistance in becoming
acquainted with laboratory procedures at Lehigh. In addition, the financial support of
the Pennsylvania Electric Research Council (PERC), Potomac Electric Power
Company (PEPCO), Public Service Electric and Gas (PSE&G), Ohio Edison, and
Virginia Power is gratefully acknowledged. Finally, the author would like to thank his
wife, Maria, for her patience and encouragement during this effort.
1ll
Table of Contents
Page
ABSTRACT 1
I. INTRODUCTION 3
ll. BACKGROUND 5
II.A Corrosion 5
II.A.l Oxidation 5
II.A.2 Sulfidation 6
II.A.3 The use of thermal spray coatingsfor corrosion protection 11
ILB Thermal Spray Techniques 13
II.B.l Flame spraying 14
II.B.2 Electric arc spraying 15
II.B.3 D-gun spraying 16
II.BA Plasma spraying 17
.ILC The Effect of Coating Structure onCorrosion Behavior 18
III. EXPERIMENTAL PROCEDURE 21
lILA Corrosion Testing 21
lILA. 1 Oxidation testing 21
IILA.2 Sulfidation testing 22
III.A.3 Cyclic oxidation testing 22
III.B Cyclic Thennal Testing 22
iv
Ill.C Metallographic Sample Preparation 24
IILC.l Sectioning 24
IlLC.2 Mounting 24
IILC.3 Metallography 26
Ill.D Sample Analysis Techniques 26
IV. RESULTS' 32
IV.A As-Sprayed Microstructural Characterization 32
IV.A.I LOM cross-sectional observations 32
IV.A.2 SEM/EDS observations 33
IV.B Oxidation Test Results 37
IV.B.I LOM cross-sectional observations 37
IV.B.2 SEM/EDS surface observations 48
IV.C Sulfidation Test Results 54
IV.C.I LOM cross-sectional observations 54
IV.C.2 SEM/EDS surface observations 60
IV.C.3 SEM/EPMA cross-sectional observations 67
IV.D Cyclic Oxidation Test Results 67
IV.E Cyclic Thennal Test Results 67
V. DISCUSSION 72
V.A Oxidation and Sulfidation 72
V.A.I Thennodynamic considerations 72
V.A.2 Surface corrosion 73
v
V.AJ Intercoating corrosion
V.A.4 Corrosion scale formtion at thecoating/substrate interface
V.A.4.i Kinetics of scale formation
V.A.4.ii Interfacial scale thicknessas a function of coatingmicrostructure
V.B The Effect of Cycling Oxidation Testing
V.C The Effect of Cyclic Thermal Testing
V.D The Effect of Coating Composition
VI. CONCLUSIONS
VII. REFERENCES
VITA
VI
74
77
77
81
84
85
86
91
95
98
List of Tables
T~~R~ 5Equilibrium oxygen partial pressures required for the oxidationof various metals at 600°C.
Table 11.11: 13Corrosion resistant plasma spray coatings.
Table IV.I: 33As-sprayed microstructural characterization results for thecoatings examined.
Table IV.II: 37Chemical composition and oxide constitution for all coatings.
Table IV.III: 69Outgrowth size data for cyclic oxidation test results.
Table V.I: 78Coated substrate corrosion rate type and corresponding rateconstants for all coatings and both test environments.
Table V.II: 78Uncoated substrate corrosion rate type and corresponding rateconstants for all substrates and both test environments.
Table V.III: 80Substrate compositions as determined by wet chemical analysis.
Table V.IV: 87Coating reactivity based on surface scale formation andintercoating oxide content. (O=oxidation test results;S=sulfidation test results).
Vll
List of Figures
Section II: BACKGROUND
FigureIT.I: 9Stability diagram for the Fe-S-O system at 900°F.
Figure IT.2: 9Reduction in the oxygen activity at a metal surface due to theformation of an oxide scale.
Figure IT.3: 14Schematic illustration of a flame spray gun.
Figure IT.4: 15Schematic illustration of an electric arc spray gun.
Figure IT.S: 16Schematic illustration of a detonation spray gun.
Figure IT.6: 18Schematic illustration of a plasma spray gun.
Section III: EXPERIMENTAL PROCEDURE
Figure ITlI: 21Typical sample used in this investigation. (a) top; (b) side.
Figure ITl2: 23Schematic diagram of the apparatus used for sulfidation testing.
Figure ITl3: 23Surface view of a typical mount used in this investigation.
Figure IIl4a-d: 26Coating thickness detection via automated image analysistechniques. (a) Coincident thresholding of mounting media anddark coating constituents; (b) separation of mounting mediawith the use of a size limitation; (c) identification of coatingarea; and (d) correspondence of superimposed vertical lines withlocal coating thickness.
viii
Figure ill.S: 31Detection of coating porosity with the use of fluorescencemicroscopy. (a) Coating structure after 320 SiC grit grind asviewed with the use of brightfield illumination; and (b) thesame area as in (a), but viewed with th<:? use of fluorescencelighting. The porous regions, which appear bright in (b), caneasily be thresholded with image analysis procedures.
Section IV: RESULTS
Figure IV.I: 34As-sprayed microstructure of coating A (top), B (middle), and C(bottom) in unetched condition. The substrate ("A"), coating("B II), and mount material ("C II) are shown in the coating Cmicrograph.
Figure IV.2: 35As-sprayed microstructure of coating A (top), B (middle), and C(bottom) in unetched condition. (A: oxides, B: voids)
Figure IV.3: 36Surface appearance of (a) coating A, (b) coating B, and (c)coating C.
Figure IV.4: 38Elemental distribution in coating A cross-section. (a) SEImicrograph; (b) and (c): EDS Ni and Cr x-ray dot maps. Arrowindicates typical oxide, which is Cr-enriched and Ni-depleted.
Figure IV.S: 39Elemental distribution in coating B cross-section. (a) SEIimage; (b), (c), and (d): Fe, Cr, and Al EDS x-ray dot mapsrespectively. Arrows 1 and 2 denote oxides containing AI, andAl and Cr, respectively.
Figure IV.6: 40Elemental distribution in coating C cross-section. (a) SEIimage; (b), (c), and (d): EDS Ni, Cr, and Al x-ray dot maps,respectively. Arrows 1, 2, and 3 denote oxides containing Cr,AI, and both Cr and AI, respectively.
IX
Figure IV.7: 41Microstructure of coating A after 1000 hours exposure to air at600°C. (arrow indicates the. corrosion scale at thecoating/substrate interface)
Figure IV.S: 41Microstructure of coating B after 1000 hours exposure to air at600°C. (arrow indicates corrosion scale at the coating/substrateinterface)
Figure IV.9: 42Microstructure of coating C after 1000 hours exposure to air at600°C. (arrow indicates corrosion scale at the coating/substrateinterface)
Figure IV.IO: 42Surface oxides on the coating C surface after exposure tooxidizing conditions.
Figure IV.II: 43Coating thickness as a function of oxidation exposure time.
Figure IV.12: 44Average outgrowth length as a function of oxidation exposuretime.
Figure IV.13: 44Average outgrowth width as a function of oxidation exposuretime.
Figure IV.14: 45Schematic illustration of a corrosion outgrowth on a coatingsurface.
Figure IV.15: 45Percent coating surface area covered with corrosion product as afunction of exposure time.
Figure IV.16: 47Percent increase in coating internal oxide content as a functionof oxidation exposure time.
x
Figure IV.17: 47Thickness at the substrate/coating interface as a function ofoxidation exposure time... _
Figure IV.18: 49Surface of coating A after various exposure times to air at600°C. (a) General view; (b) chromium-containing "feathery"morphology; (c) "blocky" morphology; (d) blocky morphologyafter 1000 hours.
Figure IV.19: 50Surface of coating B after various exposure times to air at600°C. (a) General view; (b) chromium-containing "feathery"morphology; (c) duplex feathery and "blocky" morphology; (d)aluminum-containing corrosion product.
Figure IV.20: 51Surface of coating C after various exposure times to air at600°C. (a) General view; (b) chromium-containing "feathery"morphology; (c) aluminum-containing corrosion product; (d)iron-chromium-aluminum product.
Figure IV.21: 53EDS spectrum corresponding to the high-chromium containingcorrosion product found on the coating A and C surfaces afterexposure to oxidizing conditions.
Figure IV.22: 53EDS spectrum corresponding to the high-aluminum containingcorrosion product found on the coating B surface after exposureto oxidizing conditions.
Figure IV.23: 54EDS spectrum corresponding to the corrosion product shown inFigure IV.20d.
Figure IV.24: 55Microstructure of coating A after 1000 hours exposure to S02 at600°C. (A: corrosion outgrowth, B: intercoating corrosion, C:interfacial corrosion scale)
Xl
Figure IV.2S: 55Microstructure of coating B after 1000 hours exposure to S02 at600°C. (A: corrosion outgrowth, B: intercoating corrosion, C:interfacial corrosion scale)
Figure IV.26: 56Microstructure of coating C after 1000 hours exposure to S02 at600°C. (A: corrosion outgrowth, B: intercoating corrosion, C:interfacial corrosion scale)
Figure IV.27: 57Average outgrowth cross-sectional length as a function ofsulfidation exposure time.
Figure IV.28:- Average outgrowth cross-sectional width as a function of
sulfidation exposure time.
57
Figure IV.29: 58Percent coating surface covered with corrosion scale as afunction of sulfidation exposure time.
Figure IV.30: 59Percent increase in coating internal oxide content as a functionof sulfidation exposure time.
Figure IV.31: 59Substrate/coating interfacial corrosion scale thickness as afunction of exposure time.
Figure IV.32: 61Coating A surface after exposure to sulfidizing conditions.
Figure IV.33: 61Coating C surface after exposure to sulfidizing conditions.
Figure IV.34: 62EDS spectrum corresponding to the corrosion scale formed oncoatings A and C after exposure to sulfidizing conditions.
Figure IV.35: 62Needle-like morphology found on the coating A scale.
xu
Figure IV.36: 63Corrosion morphology found on the nickel-containing coatingsafter exposure to sulfidizing conditions. Arrows indicate therecent nucleation of spherical corrosion products on the needle-like corrosion product.
Figure IV.37: 63Corrosion structure on coating C surface showing the coexistanceof spherical and needle-like morphologies.
Figure IV.38: 64Nickel and sulfur containing corrosion product found on thecoating A surface.
Figure IV.39: 65Aluminum- and sulfur-containing corrosion product found on thecoating B surface, and corresponding EDS spectrum.
Figure IV.40: 66Corrosion product found on the coating B surface, whichcontains iron, chromium, and aluminum; and corresponding EDSspectrum.
Figure IV.41: 68SEI image and corresponding EDS/WDS x-ray dot maps for asurface-to-substrate region on the coating A cross-section. A:surface corrosion scale; B: corrosive species in the splatboundaries; C: nickel-depleted regions within the coating.
Figure IV.42: 68SEI image and corresponding EDS/WDS x-ray dot maps for asurface-to-substrate region on the coating C cross-section. A:surface corrosion scale; B: corrosive species in the splatboundaries; C: nickel-depleted regions within the coating.
Figure IV.43: 69Microstructure of coating A after 43 thermal cycles. Arrowsindicate regions where spallation has apparently occurred.
Figure IV.44: 70Microstructure of coating B after 43 thermal cycles.
Figure IV.45: 70Microstructure of coating C after 43 thermal cycles.
Xlll
Figure IV.46:Microstructure of coating A after 3,000 thermal cycles.
Figure IV.4J:Microstructure of coating C after 3,000 thermal cycles.
Section V: DISCUSSION
71
71 .
Figure V.l: 79Uncoated substrate corrosion scale thickness as a function ofoxidation exposure time.
Figure V.2: 79Uncoated substrate corrosion scale thickness as a function ofsulfidation exposure time.
Figure V.3: 82MFP as a function of coating porosity.
Figure V.4: 83MFP as a function of coating thickness.
Figure V.S: 83Interfacial scale thickness as a function of mean free path for allcoatings and both test environments.
Ii)
Figure V.6: 84Percent coating surface covered with corrosion as a function ofexposure time for both isothermal and cyclic heat treatments.
Figure V.7: 88Schematic model of "non-reactive" coating behavior.A=substrate material; B=splat; C=substrate attack.
Figure V.S: 89Schematic model of "reactive" coating behavior. A=corrosionscale on coating surface; B=substrate material; C=progression ofcorrosive specie(s) into coating thickness via splat boundaries,voids, and oxides; D=substrate attack.
XIV
ABSTRACT
The'·high-temperature oxidation and sulfidation behavior of various thermal
spray coatings has been investigated. Three coatings were selected for analysis based
on their commercial availability and promotion as a means of fossil-fired utility boiler
waterwall tube protection. The coating chemical compositions, thicknesses, and
porosity contents and morphologies derived from the selected coatings allowed for the
investigation of these parameters.
Coated test coupons were examined in the as-sprayed and laboratory-tested
conditions with the use of various microscopy and analytical techniques. Laboratory
testing included isothermal exposure to 600°C air and pure SOz in separate tests at
hold times up to 1000 hours. Cyclic oxidation studies were also conducted where
samples were repeatedly cycled to 600°C and held at temperature for approximately 9
hours. In a similar procedure, samples were subjected to 3,000 thermal cycles
between 315°C and 525°C, where heating and cooling times were approximately 10
seconds and 3 minutes respectively.
Although chemical composition was found to dictate coating response during
corrosion testing, coating structural aspects were found to govern the extent of
underlying substrate attack. The "mean free path" (MFP), or the mean distance from
coating surface to substrate via fast diffusion paths such as splat boundaries and
porosity, was determined to be a significant factor for substrate attack. The thickness
of the corrosion scale at the coating - substrate interface resulting from the corrosion
1
process was limited at high MFP's, corresponding to thick and dense coatings. In
addition, the corrosion rate at this interface was significantly reduced when compared
to uncoated substrates when effective coatings were used. Such coatings were able to
form protective oxides at splat boundaries and within voids, thus limiting corrosive
specie migration and the extent of substrate attack. The result of this process was a
decrease in substrate corrosion kinetics from linear-type for uncoated substrates to
logarithmic-type when coatings were used. Conversely, coatings with excessive
porosity were not able to form protective oxides at all intercoating fast diffusion paths,
and substrate oxidation kinetics remained linear with the use of these coatings.
As a result of cyclic oxidation testing, the percent coating surface covered with
corrosion outgrowths was decreased by as much as 'a factor of six when compared to
equivalent oxidation exposure times. This difference was attributed to the stresses that
arise from thermal expansion mismatch and subsequent outgrowth spallation. The role
of porosity morphology became apparent from cyclic thermal testing. Elongated and
tapered intersplat pores were found to crack at their "tips", leading to void linkup and
possible subsequent coating spallation. The stresses resulting from thermal
fluctuations, and the concentration of these stresses at the pore tips, were responsible
for this failure mechanism. Coating thickness losses after 3,000 cycles were as high
as 17.5% due to this phenomenon.
2
I. INTRODUCTION
Fossil fuel fIred power plants in the United States produce an annual combined
power output capacity of approximately 300,000 megawatts.1 Boiler tube failures in
these power plants constitute the largest single cause of forced outages, which can cost
up to an estimated $700,000 per day for large modern units? Due to the desire for
greater plant efficiency and -availability, the control of boiler component degradation
has become a primary industry concern and it has become necessary for utility plants
to seek life extension measures for their present equipment. As boiler tube failure is
the most common form of boiler damage, the preservation of these tubes is of
foremost interest.
Boiler tube failures are the result of many degradation mechanisms, some of
the most damaging being fireside corrosion and corrosion-erosion. In a recent study3,
40% of North American utilities surveyed reported serious fIreside corrosion problems
in their boilers at costs up to $1.7 million per boiler per year. Lost generating
capacity per boiler ranged from 1.1 to more than 11.6 days per year as a result of
forced outages. The corresponding power losses were as high as 162,900 MWh per
boiler per year, a cost in excess of $4 million per boiler annually. Similar conditions
have been reported in Canada4, where approximately 25% of the annual forced outages
have been due to boiler tube failures in recent years, the majority of which have been
caused by corrosion and erosion mechanisms.
The use of high-temperature claddings and metallic coatings on boiler tubes has
3
been a somewhat successful method of outage risk reduction. Coatings have proven to
bea reasonably cost-effective means for enhancing the corrosion and erosion
resistance of waterwall tubes. Although much of the utility industry has experimented
with various surface modification techniques, a systematic selection and application
procedure has not been determined. Furthermore, the critical coating properties which
determine the success or failure of a coating in the boiler environment have not been
identified. As a result, utilities have been forced to select coatings on an ineffective
and expensive trial-and-error basis, underscoring the need for a coating selection
process. Such critical coating parameters in the case of thermal spray coatings may
include coating thickness; substrate roughness; alloy content and chemical distribution;
grain size; the size, shape and distribution of voids, oxides and splats; or various
combinations of all these parameters.
The objective of this research effort is to investigate the structure-property
relationships of various thermal spray coatings as related to high-temperature oxidation
and sulfidation behavior. The coatings to be examined will include the following: (1)
electric arc sprayed Metalspray Tafaloy 45CT (51wt% Ni- 45 wt% Cr- 4wt% Ti), (2)
plasma sprayed Metco 465 (65wt% Fe- 27wt% Cr- 2wt% Mo), and (3) plasma
sprayed Metco 468NS (62wt% Ni- 26.5wt% Cr- 7wt% Al- 3.5wt% Co- l.Owt% Y203)'
Through the identification of the critical microstructural characteristics which govern
the severity of coating attack when placed in aggressive environments, quality
assurance guidelines can be established so that coating performance may be
anticipated.
4
II. BACKGROUND
ILA Corrosion
Oxidation, sulfidation, and molten salt accelerated oxidation have been
identified as prominent forms of waterwall fireside corrosion.s-s Oxidation and
sulfidation mechanisms are discussed, as these are the phenomena of interest for this
research effort.
ILA.1 Oxidation
Most metals are readily oxidized at elevated temperatures. As can be seen in
Table IT.I, the oxygen partial pressures required for the oxidation of various metals at
600°C are low enough to promote this form of corrosion in most high-temperature
service environments, and certainly under boiler conditions.
Table II.!: Equilibrium oxygen partial pressures required for the oxidation ofvarious metals at 600°C. (Thermodynamic data acquired from ref. 9)
Reaction Po2 required for
reaction at 600°C
3Fe + 202 =Fe30 41.063 x lO-25atm
2Cr + 3/2 O2 =Cr20 32.327 x 1O-36atm
Ni + 1/2 O2 =NiO 6.135 x 1O-2°atm
2AI + 3/2 O2 = Al20 3 1.228 x 10-56atm
In multicomponent systems, elements with greater affinities for oxygen will
preferentially oxidize over those elements with lesser affinities.10.1l As some metals
5
fonn non-protective scales which are subject to spallation and accelerated attack, more
reactive coatings or alloy additives which develop coherent and protective oxide layers
are used to inhibit metal wastage. Depending on the elements involved, oxidation can
be controlled by the diffusion of oxygen into the metal or alloy, the diffusion of
reactive elements to the metal-gas interface, and/or the diffusion of these species
through a resultant oxide scale.
1I.A.2 Sulfidation
Like oxidation, sulfidation is a corrosion mechanism inherent to fossil-fired
boilers. A direct relationship exists between the sulfur content in a given fuel and the
subsequent extent of sulfidation that occurs in the boiler. The sulfur content of
bituminous coals range from 0.5-3.5%. In addition, ash content is usually 3.0-14.0%,
with ash containing 0.1-30.0% S03. 12 Sulfur is not a desirable oxidant, as it can
penetrate a protective oxide layer on the metal surface. Sulfidation damage can
manifest itself in one of three mechanisms:13
(1) It ties up reactive elements such as chromium and aluminum, making them
unavailable for protective oxide fonnation.
(2) It encourages the formation of an irregular metal/scale interface, and the
eventual development of a two-phase oxide + metal region.
(3) It may result in low-melting point sulfides along
the grain boundaries, giving way to rapid alloy destruction via intergranular
liquid phase attack.
6
When a metal such as iron is heated in a gaseous environment containing
oxygen and sulfur, the following reactions may occur on the metal surface:
Fe + 1/2 O2 --> FeO
Fe + 1/2 S2 --> FeS
with the equilibrium conditions,
(aa)eq = (P02112) =exp(.OoPeaIRT)
(as)eq = (PS2112) =exp(·0°p~T)
(1)
(2)
(3)
(4)
(5)
where aa and as are the activities of oxygen and sulfur respectively, and .00peO and
·OopeS are the standard free energies of formation for FeO and FeS. It is expected
that FeO and FeS will form on the metal surface providing the values of aa and as in
the gas are greater than the equilibrium values (i.e. ~ > (ao)eq and ~ > (~)eq)'
However, the most stable phase is determined by the reaction
FeS + 1/2 O2 = FeO + 1/2 S2
with the equilibrium condition
(aslaa)eq =exp {(.00PeS-·Oopeo)/RT}.5 (6)
Oiven the above equations, the following cases can be predicted:10
(1) (aa)gas > (aa)eq and (as)gas < (ag)eq' FeO is the only stable phase and
forms on the metal surface.
(2) (aa)gas < (aa)eq and (as)gas > (ag)eq' FeS is the only stable phase and forms
on the metal surface.
(3) (aa)gas > (aa)eq and (as)gas > (as)eq' Both FeD and FeS appear to be stable
phases, but the phase that actually forms is determined by equation (6)
7
where one of the two following cases prevail:
(a) (as!aa)gas > (as!aa)eq' In this case, reaction (5) proceeds to the left and
therefore FeS is the stable phase where the metal is in contact with
the gas phase.
(b) (as!aa)gas < (as!aa)eq' In this case, reaction (5) proceeds to the right and
FeO is the stable phase.
The conditions which govern the formation of such phases in the Fe-O-S
system are summarized in the phase stability diagram as shown in Figure ILL If FeD
is the more stable oxide phase, it will grow as a layer on the metal surface. The
formation of this layer will, however, cause a reduction in the oxygen activity at the
metal surface as the layer grows thick enough to permit equilibrium conditions at both
the Fe/FeO and FeO/gas interfaces. Figure n.2 illustrates this principle for a metal
"A". In addition, most oxide scales contain defects in the form of cracks or grain
boundaries that may serve as diffusion paths for sulfur.14 If the sulfur can penetrate
the oxide scale in this way, FeS will form in the areas where the oxygen activity has
been reduced and the as/ao ratio has been increased. lo
As sulfide particles nucleate and grow within the base metal, they begin to line
up along microstructural features such as grain boundaries. In the case of many tube
steels, these sulfides can be chromium-rich. The sulfides then oxidize, forming
8
·20
~ ·30!
l/)N F•..9 .CO
.50 t1 70% AIR
o 90% AIR
o 100% AIR
.60 0 120'1. AIR
·70L---L_--'-_'---'-_-'-_...J..----'-10 ·60 ·50 ·CO ·30 ·20 ·10
lOG P02 (ATM)
Figure D.1. Stability diagram for the Fe-S-O system at 900°F. (from [17]).
ao =(ao)eq
oxygen activity (ao)...Gas /0 = (ao)gas
Figure D.2. Reduction in the oxygen activity at a metal surface due to theformation of an oxide scale. (from [7])
9
chromium oxide intrusions into the metal. The base metal between the intrusions
consequently becomes depleted in chromium. In addition, if the external protective
oxide scale has spalled as a result of thermally induced cracking or erosion, the
normally protective chromium oxide coating cannot redevelop. Instead, the oxide or
sulfi~e of the bulk base metal forms, thus accelerating the oxidation rate. 13•14
An alternative mechanism13 of sulfidation suggests that sulfur establishes paths
in the protective oxide which allow the transport of the base metal outward. The
diffusion rates in sulfides are much greater than those in oxides, so that a very small
volume fraction of connected sulfide network is sufficient to effectively bypass the
protective oxide. The base metal oxide would then grow on the outside of the
protective scale, forming voids in the metal and leading to the mechanical break-up of
the protective coating.
The most obvious means of sulfidation reduction is the use of low-sulfur fuels.
However, since fossil-fired power plants must desulfurize flue gas regardless of coal
quality, l it is economically advantageous to use more readily available, high sulfur
containing coals. Practiced remedial actions in response to sulfidation have included
the readjustment of combustion parameters, the application of sprayed metal and
diffusion coatings, air blanketing of the walls, the use of thicker-walled tubes, and the
installation of co-extruded tubes and claddings.3•5
10
II.A.3 The use of thermal spray coatings for corrosion protection
Thermal spraying involves the deposition of molten material onto a substrate to
form a protective coating. The coating material is melted in and projected from a gun
in the form of fine particles, 5-200 microns in size. A high velocity gas stream carries
the particles to the substrate, where they impact, flow, and rapidly solidify as they
interlock with the substrate surface and themselves to form a coherent coating.IS
Thermal spray coatipgs have recently emerged as a widely employed waterwall.
tube protection measure. One distinct advantage of these coatings over competitive
coatings is their ability to be applied in-situ. An optimization of thermal spray
coatings with respect to the boiler environment may therefore establish them as the
coatings of choice for the utility industry.
There are many general requirements that a coating must meet in order to be
effective. The coating must provide adequate resistance to the corrosion and erosion
mechanisms present in the boiler without compromising the mechanical properties of
the coated component. For boiler tube application, the coating should not significantly
reduce the heat transfer through the waterwall. Finally, the coating material and
application process must be economically acceptable.
Thermal spray coatings have offered protection against sulfidation in the
laboratory and in application, and have proven to be a cost-effective means for
corrosion resistance. For example, an arc sprayed Cr-Ni coating applied to high
temperature sulfurous environments exhibited excellent corrosion resistance after 18
months in service.16 Laboratory tests have shown many plasma spray coatings to be
11
---I
corrosion resistant in environments similar to those present in a boiler. For example,
Fe-Cr-AI-Mo coatings have been found to be effective on a variety of carbon and
stainless steels in a synthetic ash environment.1? MgZr03 has been shown to be an
extremely stable coating on stainless steel in synthetic ash environments.IS Table ILlI
is a listing of plasma spray coatings which have proven to be effective means of
corrosion resistance.
The paper industry has recognized the benefits of plasma spraying, and these
coatings have become the standard method of tube protection in Black Liquor
Recovery Boilers (BLRBs). As in PCF boilers, the fireside attack of BLRB tube walls
occurs by a combination of chemical attack, such as sulfidation and oxidation, and
physical material removal by erosion.22 Paper companies operating BLRBs have been
moderately successful in protecting their boiler tubes with good quality plasma spray
coatings.22,23
Experience with thermal spray coatings is limited among utilities, but many
companies have been successful with their use. In a five-year examination,24 Metco
2218 (multi-layer coating), Metco-444 (9Cr-7AI-5.5Mo-5Fe-Ni balance), WCT-18997,
and WCT-18991 all performed "fair to good" when used for the protection of boiler
tubes. In another study,25 Cr-Fe and Cr-Ni arc spray coatings performed well in the
protection of economizer tubes in an eighteen month in-service experiment. Likewise,
a panel sprayed with Metco-465 (27.5Cr-6AI-2Mo-Fe balance) was shown to have
negligible deterioration on the fireside wall thickness and had not spalled after 2 1/2
years of service.26
12
Table n.ll: Corrosion resistant plasma spray coatings
Coating refs. Uses
Al20 3 15 oxidizing or reducing atms
Ni-20Cr- ~ 19,20 oxygen barrier
Co-Mo-Cr 20 chemical corrosion atrns
Ni-Co-Cr-AI-Y 22,23 superior oxidation resistance
Cr20 3 20 corrosion and cavitation
25Cr-llNi-7.5W 15 oxidation resistance
62Co-28Mo-8Cr-2Si 15 oxidation resistance
95Ni-5AI 15 oxidation resistance
Zn 15 corrosion resistance
Monel 15 corrosion resistance
Ni-Cr-AI-Y 21 oxidation/corrosion resistance
Co-Cr-AI-Y 21 sulfate hot corrosion
Fe-Cr-AI-Y 21 high sulfur environments
ILB Thermal Spray Techniques
There are many types of spraying processes presently being used for coating
purposes. The most successful of these include flame spraying, electric arc spraying,
detonation gun CD-gun) spraying, and plasma spraying??
The various spray technologies are used to produce coatings which are utilized in a
broad range of applications and have been employed in industries such as gas turbine,
aerospace, electronics, paper, automobile, material extraction and manufacturing, and
many others. Each thermal spray process is discussed in further detail below.
13
n.B.1 Flame spraying
In flame spraying, an oxyacetylene gas torch is used to spray material supplied
as wire, rod or powder (Figure 11.3). The stream of burning gases carries the particle~,
atomized and molten, to the substrate material. The main advantages of the process
are low capital cost and ease of operation, as flame spray guns are inexpensive, light,
and compact. Compared to other coating methods, however, particle velocities and
temperatures are low, producing more porous (up to 20% porosity), lower-density
coatings of lower bond strength. IS•V
Gas nozzle
I
Fuel gas
/Aircap
Burning gases
Figure n.3. Schematic illustration of a flame spray gun.
14
ILB.2 Electric arc spraying
As shown in Figure ITA, an electric arc spray system contains two wire
electrodes that strike an arc to melt wire feed. Compressed air, flowing at
approximately 30 fe/min, atomizes the melted wire and propels the resulting droplets
onto the substrate. The rate of spraying may vary from 3 to 200 pounds of wire per
hour, depending on current and the type of metal sprayed. The main advantage of this
process is the formation of a strong bond between the substrate and the coating.IS
+ d.c. supply
Figure ITA. Schematic illustration of an electric arc spray gun.
15
ILB.3 D-gun spraying
The D-gun consists of a water cooled steel tube closed at one end. At the
closed end is a spark plug and a series of valves through which feed powder, gas, and
air flow. This gun is schematically illustrated in Figure II.5. The spark plug ignites
the mixture, setting off detonations at a rate of 4.3 or 8.6 times per second to propel
particles out of the barrel onto the workpiece at velocities up to 2,500 ft/sec and at
temperatures of about 3000oK. The process is mainly used to coat components which
are operated under conditions of severe abrasive wear. The coatings have very high
bond strengths and densities, and very low porosities.15,27
Spark plug
~ /'Powder ~ ,I Barrel~~
+ ~ ~
:/-
~ ~ ~
Nitrogengas
".::~••:J _Oxygen gas
tAcetylene gas
Figure II.S. Schematic illustration of a detonation spray gun.
16
II.B.4 Plasma spraying
The plasma gun consists of a tungsten cathode and a surrounding cylindrical
copper anode which extends beyond the cathode to form a nozzle. Figure II.6 is a
schematic representation of a plasma spray gun. An inert gas - nitrogen, hydrogen,
argon, or helium - flows through the space between the charged electrodes, where it is
ionized to form a plasma. Most spraying is conducted at powers of about 40 kW,
with current and voltage at about 500A and 80V respectively. The gas temperatures
are approximately 6,700°F-13,900°F (9,700°C-13,700°C) with hydrogen and nitrogen,
26,500°F (14,700°C) with argon, and 35,500°F (l9,700°C) in the case of helium. The.r
ability to reach such high temperatures allows high melting point materials, such as
ceramics, to be applied by this process. Gas velocities are controlled by nozzle bore
and power level, and usually range from 650 to 1650 ft/s.21•27
The coating material is supplied in powder form and is directed by a tube into
the jet of plasma that develops in the nozzle. The flame melts and accelerates the
particles towards the target surface. When the droplets of coating material arrive at
the target, they solidify into the surface shape surface which has been roughened prior
to spraying. As the droplets flatten out on the surface, the substrate acts as a heat sink
and solidification takes place in perhaps a millionth of a second.21 The plasma spray
process results in a thick, strongly bonded protective coating.21
17
Insulator
Gassupply
Powder injection
Copperanode
Opening for plasmaflame and molten
particles
Tungstencathode
Electrical and water supply
Figure ll.6. Schematic illustration of a plasma spray gun.
II.C The Effect of Coating Structure on Corrosion Behavior
Coating quality is defined in tenns of various microstructural parameters.
Although quantification of such features for optimum microstructures has not been
determined, these parameters are recognized to influence coating perfonnance as
follows:
1. Volume Percent Porosity. Porosity has been shown to generally degrade
coating mechanical properties such as fracture strength.28 Other properties, such as
electrical and thennal conductivities, are reduced by pores which effectively decrease
the coating cross-sectional area.28 Porosity may develop as a result of one or more
occurrances during the spraying process. If the coating solidification rate is greater
18
than the deposit rate, impinging droplets will impact a solidified coating surface. This
results in poor welding between successive droplets and leads to porosity.29 A high
level of porosity can also arise from presolidified droplets which impinge as solids
because they were either not melted by the spraying process or resolidified in flight.
The resultant porosity is irregularly shaped and generally interconnected.29 In addition
to these factors, porosity can arise from solidification shrinkage, gas entrapment, or a
low spraying velocity.29-3o
2. Coating Thickness and Uniformity. An increase in coating thickness may
enhance coating ability to protect the substrate from corrosion and erosion damage, but
it also may increase internal coating stresses.27,31 The exact role of thermal spray
coating thickness is not well-defined for the boiler environment.
3. Oxide Distribution and Shape. The mehanical properties of thermal spray
coatings are influenced by oxides at intersplat boundaries. Specifically, the presence
of such oxides results in a reduction in cohesive strength and bulk hardness.31 In
addition, oxides which envelop splats during spraying can be detrimental to coating
bond strength.30 The oxides which exist in thermal spray coatings either pre-existed in
the feed material, or come about by spraying in air at high temperatures.30
4. Bond Strength. The bond between a thermal spray coating and its substrate
is essentially a pure mechanical bond. A strong bond is therefore the result of a clean,
rough interface.27,3o.32 Bond strength has also been found to be inversely proportional
to coating thickness, and directly proportional to spray velocity.27 A low bond
strength may result in coating spallation.
19
5. Alloy Composition. An optimum coating alloy composition for boiler tube
applications has yet to be determined. However, elements which can form a protective
scale are desireable for corrosion resistance.
As can be seen from the above discussion, the influence of processing
parameters on coating structure ,is complex; such parameters may include spray
distance, particle velocity, gas flow, powder feed rate, powder and substrate prehe?,t,
gun power, nozzle type, and the raw material powder characteristics.27,33,34
20
r
m. EXPERIMENTAL PROCEDURE
IILA Corrosion testing
IILA.1 Oxidation testing
Five samples of each coating were simultaneously oxidized in a Hayes electric
furnace in laboratory air at 600°C, with one sample of each coating being removed for
analysis at 50, 100, 250, 500, and 1000 hours. Each_specimen consisted of a coated
low-alloy tube steel section which had an approximate size of 1" in length, 1/2" in
width, and 1/2" in thickness. A typical sample is shown in Figure TILL
,---------,----------------_.,
B
Jinch.. . 11•I. I. ! I I. J I I. II
Figure Ill!. Typical sample used in this investigation. (a) top; (b) side.
21
. III.A.2 Sulfidation testing
The sulfidatiort test parameters were the same as that for the oxidation test.
Samples were placed in a Lindberg tube furnace equipped with a 2.5" diameter
alumina tube. The apparatus is schematically shown in Figure III.2. The sulfur
dioxide flow rate was kept constant at approximately 15rnL/min. Nitrogen was used
to purge the system before and after samples were removed.
IILAJ Cyclic oxidation testing
Three samples of each coating were simultaneously subjected to a thermal.,
cycling program, where one thermal cycle represents heating to 600°C in
approximately 30 minutes, an average hold time of approximately 9 hours, and cooling
to room temperature in approximately 30 minutes. Specimens were removed and
examined after 12, 25, and 43 cycles.
IILB Cyclic thermal testing
Coated boiler tubes were placed within the coils of a Lepel induction furnace,
as schematically shown in Figure III.2. The tube surface temperature was
continuously monitored with the use of a Williamson 8200 optical pyrometer. This
test apparatus was configured such that power was supplied to the furance in a 10
second interval whenever the specimen surface was at a temperature below 315°C.
The result of this configuration is a repeating thermal cycle, where a cycle represents a
10 second heating time to 525°C and a subsequent 3 minute cooling period back to
22
III~IIIII.Nitrogen
Sulfur Dioxide
Tube Furnace
/Gasexit
Figure n1.2. Schematic diagram of the apparatus used for sulfidation testing.
/'Tube Sample
--.. ~
Pyrometer
/' StabilizingWeights
Induction Furnace
Figure 111.3. Schematic illustration of the apparatus used for thermal cycling.
23
315°C. Specimens were subjected to 3,000 of these thermal cycles.
III.C Metallographic sample preparation
Both the as-sprayed and laboratory-tested coating specimens were prepared
with the same procedures. Proper metallographic techniques were essential to
minimize specimen distortions and observe accurate coating characteristics. In
addition, since image analysis procedures were employed for quantitative
metallography, artifacts that produce a false sense of light intensity such as edge
rounding were minimized using procedures discussed below.
IILC.1 Sectioning
In order to minimize coating deformation, sectioning was performed with a low
speed saw and diamond blade. Blade speed was held at approximately 120 rpm for a
4" wheel, and the vertical load on the sample was kept at approximately 260g.
Cutting oil was used as lubricant. The direction of cutting, or the tangential direction
of blade rotation, was always into the coating surface. For the laboratory-tested
samples which were covered with a brittle oxide or sulfide scale, the sample surfaces
were coated with epoxy prior to sectioning.
III.C.2 Mounting
Buehler cold-setting epoxide was used as mounting material to avoid any
coating degradation which may occur when using hot-press mounting techniques. A
24
fluorescent powder dye, Struers "Epo-dye", was added to the liquid epoxide for the
purpose of porosity detection, as will be discussed in section I1I.C.
Standard metallographic techniques were employed in order to minimize
"rounding" of coating edges. Two samples were mounted per mount in the orientation
shown in Figure IlIA. In this way, the two coated surfaces provide support for each
other and the planar area of the softer epoxy between the samples is minimized. In
addition, Struers "hard filler" powder was placed between the coating surfaces as a
further edge retention measure.
Hard fillerpowder
Substrate
Figure 111.4. Surface view of a typical mount used in this investigation.
25
III.C.3 Metallography
,Specimens were ground with an automatic polisher with steps including 320,
400, and 600 silicon carbide grit. Grinding parameters were kept at 300 rpm wheel
speed and 25 psi pressure. Polishing was performed by hand with steps including 6
micron diamond paste for approximately 1 minute, 0.05 micron alumina slurry for
approximately 30 seconds, and colloidal s~lica for 30 seconds.
IILD Sample analysis techniques
Both the as-sprayed and laboratory-tested coating samples were analyzed in the
same manner. Microstructural observation and photomicrography was conducted with
the use of a Reichert-Jung MeF3 metallograph. Additional characterization procedures
included the use an ETEC Autoscan scanning electron microscope (SEM) and a mOL
840F SEM, a KEVEX energy-dispersive x-ray spectrometer (EDS), and a mOL 733
Superprobe for wavelength-dispersive x-ray spectrometry (WDS) elemental dot
mapping. Semi- and fully-automated image analysis techniques were employed to
obtain quantitative microstructural coating information. Semi-automated techniques
were used to measure the size of corrosion "outgrowths" in cross-section. For this
purpose, a Donsanto Micro-plan IT digitizing pad was used in conjunction with a
Nikon Optiphot and an IBM personal computer. For fully-automated measurements, a
LECO 2001 image analysis system was employed in conjunction with a Nikon
Photophot light optical microscope. Specifically, coating features were measured via
automated image analysis by the following procedures:
26
i. Coating thickness- The task of thickness detection by fully-automated
image analysis procedures can be difficult because the substrate and mount materials
are often of the same greylevel as many of the coating components and therefore
cannot easily be differentiated. The solution to this problem lies within
microconstituent continuity; that is, the coating elements are noncontinuous, whereas
the substrate and mount materials are continuous in at least one dimension. When
detected and thresholded by greylevel, the coating components of a given greylevel
range will therefore be observed as many particles, each of which is smaller in some
dimension than the mount or substrate material which has also been detected. This is
illustrated in Figure III.5a, where the mount material and coating porosity and oxides'Q
have been thresholded together and placed in the green binary image. As shown in
Figure III.5b, the coating constituents are separated from the mount material by
placing size limitations on the binary image features. The same concept is used to
separate the coating bulk from the substrate material. By manipulating the image such
that the mount and substrate materials are all that remain, the coating can be identified
as the image portion which has not been detected. As shown in Figure III.5c, the
coating retains its characteristic shape, but is now observed as a continuous entity.
Measurement of coating thickness can now take place by superimposing a template of
parallel lines which are normal to the coating thickness, retaining that portion of the
lines which intersect the coating area, and measuring the length of these lines. Figure
III.5d shows the resulting correspondence between the coating microstructure and the
generated lines which are used to measure coating thickness.
27
,-
For all coatings analyzed, 51 thickness measurements were taken per field, and
no less than 10 fields were evaluated per coating. The corresponding minimum cross
sectional coating area used for thickness detection was 9.36E+06 square microns. The
macro written to perform thickness measurements is listed in Appendix 1.
11. Porosity- Pores, oxides, and mounting material can possess similar greylevels
when observed with the use of a light optical micros~ope and brightfield illumination.
For porosity detection, thermal spray coatings should be vacuum mounted in
fluorescent epoxy. Samples are ground with 320-grit SiC paper until the surface
epoxy layer has been removed. By using fluorescent lighting microscopy and
a color camera in conjunction with an image analysis system, porosity is easily
detected (Figure llI.6a&b) and differentiated from all other coating constituents. This
detection technique avoids the problems associated with trying to distinguish between
voids and separate particles which may be of similar greylevels, sizes, and shapes. It
should be noted that the mount material will also be detected, but can be eliminated
with a size limitation. The size, shape, and distribution of pores can now be easily
measured. It is necessary to separate the coating from the substrate and mount
materials in order to determine porosity area percent, and this can be performed under
brightfield illumination in the same manner as described for coating thickness
measurement.
28
Figure III.5a-d. Coating thickness detection via automated image analysistechniques. (a) Coincident thresholding of mounting media and dark coatingconstituents; (b) separation of mounting media with the use of a size limitation;(c) identification of coating area; and (d) correspondence of superimposedvertical lines with local coating thickness.
29
The minimum cross-sectional area considered for the porosity detection of any
of the coatings analyzed was 4.8E+06 square microns. The macro written to perform
porosity measurements is listed in Appendix II.
iii. Oxide detection- . Oxides can be readily thresholded with brightfield
illumination, and that contribution from porosity can be subtracted using the porosity
detection techniques described above. The minimum cross-sectional area evaluated for
the oxide detection of any coatings considered was 8.2E+06 square microns.
IV. Interfacial scale thickness - The corrosion scale at the substrate/coating
interface resulting from laboratory testing is thresholded along with coating oxides.
As in thickness testing, an image of vertical lines is superimposed over the binary
image, and these two images are combined such that all that remains are lines which
mark the intersection of these two images. In order to separate the lines which
correspond to the scale from those which correspond to the oxides, they must be
interactively selected by the operator. The length of these lines is then automatically
measured, and the result is the thickness of the interfacial scale.
30
Figure llI.6. Detection of coating porosity with the use of fluorescencemicroscopy. (a) Coating structure after 320 SiC grit grind as viewed with theuse of brightfield illumination; and (b) the same area as in (a), but viewed withthe use of fluorescence lighting. The porous regions, which appear bright in (b),can easily be thresholded with image analysis procedures.
31
IV. RESULTS
IV.A As-Sprayed Microstructural Characterization
IV.A.! LOM cross-sectional observations
The as-sprayed microstructures of Metalspray 45ct, Metco 465, and Metco
468NS (hereafter coatings A, B, and C) can be seeh in Figure IV.1. As discussed
below, the typical coating morphologies, the substrate-coating interfaces, as well as the
relative coating thicknesses are observed in these micrographs. Measurements of
coating thickness are summarized in Table IV.I.
At higher magnifications, coating structural details become more pronounced.
The size, shape, and distribution of oxides and voids within each coating can be seen
by examinatioB of Figures IV.2a-c. The oxides ("A" arrows), which appear grey when
observed with the use of light optical microscopy,35 are a result of splat oxidation
during the spraying process and/or pre-existed in the coating feed material. As such,
the placement of these oxides is largely limited to the outer splat edges and splat
boundaries. The amount of oxides in a given coating is dependent on the chemical
composition of the feed material and various spraying parameters such as spray
temperature, velocity, distance, and environment,30,31
Two different void morphologies can be observed from Figures IV.2a-c ("B"
arrows). Whereas coating A (Figure IV.2a) possesses an elongated, tapered porosity
morphology, coating B (Figure'IV.2b) is characterized by large, equiaxed pores. As
shown in Figure IV.2c, coating C is extremely dense; porosity is not readily apparent
32
from the micrograph. Like oxide content, the porosity content and morphology is
largely dependent on various spraying parameters. The oxide and porosity contents for
each coating are listed in Table IV.I. !
Table IV.I. As-sprayed microstructural characterization results for the coatingsexamined
COATING THICKNESS POROSITY OXIDE
(microns) (area %) (area %)
A 543 ± 60 0.7 ± 1.0 26.0 ± 0.4
B 605 ± 67 5.5 ± 0.8 11.0 ± 0.7
C 993 ± 16 0.1±0.1 30.5 ± 2.2
IV.f,..2 SEM/EDS Observations
The surfaces of coatings A, B, and C as viewed with the use of scanning
electron microscopy can be seen in Figures IV.3a-c, respectively. These micrographs
show the surface roughness for each coating and provide a basis for comparison with
the laboratory-tested coating surface appearances. The interconnected porosity which
exists through the thickness of coating B, and the resulting surface roughness can be
seen in Figure IV.3b. In contrast, the relatively smooth surface of coating C is shown
in Figure IV.3c. A more quantitative assessment of surface roughness can be gained
by examination of the standard deviation for coating thickness data (Table IV.l),
which shows coating B to be the roughest and coating C to be the smoothest.
33
A --200/-lm
Figure IV.l. As-sprayed microstructure of coating A (top), B (middle), and C(bottom) in unetched condition. The substrate (" A"), coating ("B") and mountmaterial (" C") are shown in the coating C micrograph.
34
Figure IV.2. As-sprayed microstructure of coating A (top), B (middle), and C(bottom) in unetched condition. (A: oxides, B: voids)
35
Figure IV.3. Surface appearance of (a) coating A, (b) coating B, and (c) coatingC.
The distribution of chemical components within cross-section of each
coating can be observed with the use of EDS x-ray dot mapping. Figures IVA-6 show
all coatings to be fairly uniform in chemical distribution, with the exception of oxides
which are found to be rich in the more reactive coating constituents (note arrows in
36
Table IV.II. Chemical composition and oxide constitution for all coatings.
Coating Composition Elements present incoating oxides
51 NiA 45 Cr chromium
-4 Ti
65 FeB 27 Cr aluminum,
6 Al aluminum-chromium2Mo
62 Ni26.5 Cr aluminum,
C 7 Al chromium3.5 Co1 YZ0 3
micrographs). Table IV.IT summarizes the elemental compositions and indicates the
elements which constitute the oxides for all coatings.
IV.B Oxidation Test Results
IV.B.1 LOM Cross-Sectional Observations
The typical structure of each coating after 1000 hours exposure can be seen in
Figures IV.7-9. Several microstructural changes result from exposure time. These
include the formation of oxides in the form of outgrowths at coating surfaces, an
increase in intercoating oxide contents, and the formation of an oxide scale at
coating/substrate interfaces (arrows in figures). Surface outgrowths are more easily
observed at higher magnifications, and the typical structure of these outgrowths is
shown in Figure IV.lO (arrows). The change in coating thicknesses with exposure
time is plotted in Figure IV.II. Although coating thickness was found to generally
37
decrease with exposure, the large fluctuations in data can be attributed to variations
resulting from the spraying process rather than from oxidation testing.
Figure IVA. Elemental distribution in coating A cross-section. (a) SEImicrograph; (b) and (c): EDS Ni and Cr x-ray dot maps. Arrow indicatestypical oxide, which is Cr-enriched and Ni-depleted.
38
Figure IV.S. Elemental distribution in coating B cross-section. (a) SEI image;(b), (c), and (d): Fe, Cr, and AI EDS x-ray dot maps respectively. Arrows 1 and2 denote oxides containing AI, and AI and Cr, respectively.
39 )
Figure IV.6. Elemental distribution in coating C cross-section. (a) SEI image;(b), (c), and (d): EDS Ni, Cr, and Al x-ray dot maps, respectively. Arrows 1, 2,and 3 denote oxides containing Cr, AI, and both Cr and AI, respectively.
40
Figure IV.7. Microstructure of coating A after 1000 hours exposure to air at600°e. (A arrows: intercoating corrosion; B arrows: corrosion scale at thecoating/substrate" interface)
Figure IV.S. Microstucture of coating B after 1000 hours exposure to air at600°e. (A arrows: intercoating corrosion; B arrows: corrosion scale at thecoating/substrate interface)
41
Figure IV.9. Microstructure of coating C after 1000 hours exposure to air at600°C. (A arrows: intercoating corrosion; B arrows: corrosion scale at thecoating/substrate interface)
"'-"-, ./" .- . . -=::ai. \;. .,'~
'1~~~.-_.I '2'Cfl1m~'r-c='\ . />- y'-
Figure IV.IO. Surface oxides on the coating C surface after exposure to oxidizingconditions.
42
-Cfj 1000 Ir::
~¢0J-lV.~ 800
fE ¢-
iCfj
'tCfj 600 ~
Q)
f §CoolineAr:: Cooline B~ CoalingCV 400.~
f..=E-cOf.) 200
fr::.~...,~
0 0U 0 200 400 600 800 1000 1200
Exposure Time (hours)
Figure IV.ll. Coating thickness as a function of oxidation exposure time.
The oxidation outgrowth size can be seen in Figures IV.12-13, which show the
average outgrowth length and width respectively for all time increments. These
measurements were made in cross-section, such that an outgrowth "length" denotes its
distance along the coating surface, while its "width" represents the distance it has
grown out from or into the coating surface (Figure IV.l4). Due to the excessive free
surface area of coating B resulting from its irregular surface topography and porous
nature, outgrowth data was not collected. Error bars are not included in these graphs
since, for many data points, one standard deviation was as large as its corresponding
average data point. A measure of the corresponding coating surface area percent
covered with these outgrowths is shown in Figure IV.IS, which shows an approximate
linear relationship. The data contained in this figure was obtained by summing the
43
uoo1000800600400200
Exposure Time (hours)
oo--.----r-..--.,.--.----,-......--,----...-,-.............,o
30
0
0
20 00
00 o CoatineA
0 o CoatlngC
0 00
10
Figure IV.12. Average outgrowth length as a function of oxidation exposure time.
-er:,c::oJ-cU.,.6-
200
0
150
00
0 0o Coatlne A
10 8o CoatlngC
0
Exposure Time (hours)
uoo1000800600400200oD--.----r-..--.,.--.----,-......--,----...-,--.--l
o
Figure IV.B. Average outgrowth width as a function of oxidation exposure time.
44
Coating
Figure IV.14. Schematic illustration of a corrosion outgrowth on a coatingsurface.
10.,..-------------------.,
o Coatlne: Ao CoalingC
uoo
o
1000800600
o
o
400
o
o
200
DO
o
4
o
Exposure Time (hours)
2
oD-.......-__r~-...--.......-__r--.-.,..___.......-__r--.---4
o
6
8
Figure IV.15. Percent coating surface area covered with corrosion product as afunction of exposure time.
45
outgrowth lengths and comparing the summations with the total sample cross-sectional
lengths.
In addition to the formation of surface oxides, the increase in the coating
internal oxide content with exposure time can be used as a measure of coating
reactivity. Figure IV.16 shows the performance of each coating in this respect.
Coating A is seen to be the most reactive coating, with a 159% peak oxide content
increase from its initial value. In contrast, coatings B and Chad 34% and 33% peak
oxide increases, respectively.
Because of corrosive specie migration down splat boundaries and oxides, a
corrosion scale eventually develops at the coating/substrate interface. The thickness of
this interfacial scale as a function of exposure time is shown in Figure IV.17. Error
bars are not shown in this figure because many data points have a standard deviation
as large as the data values themselves. The wide scatter in data is attributed to the
measurement of both affected and unaffected substrate areas. However, the mean
values as shown in the figure do reflect trends which can be used to identify the role
of various microstructural parameters on coating performance.
46
...C 200Q,I...c --c -'UQ,I c'tl.... ISO><0 cOIlC.......Cll C C CoaUnSAC 100 A CoaUnSBU o CoaUnsC
C C....Q,IfIl CCllQ,I SO...VC A 0 0.... 0
A A A...C 0 0Q,IV 0...Q,I 0 200 (OIl 'llO BOO 1000 nOll
~
Exposure Time (hours)
Figure IV.16. Percent increase in coating internal oxide content as a function ofoxidation exposure time.
Exposure Time (hours)
uoo1000800600400200o+--'--r--~'-'-~~---'-~---.--r-~---I
o
so
0
40
030
o CoalineA~ CoalineB
0 CoalingC
200 0 0
00~D
t.10 t. t. t.
Figure IV.17. Thickness at the substrate/coating interface as a function ofoxidation exposure time.
47
IV.B.2 SEM/EDS Surface Observations
Figures IV.18-20 provide a look at the coating surfaces and the primary
corrosion products formed during the oxidation process for coatings A, B, and C,
respectively. In all cases, the corrosion products are in the form of the previously
dicussed small, localized outgrowths. Figures N.18a, IV.19a, and IV.2Oa show the
random placement of these outgrowths on coatings A, B, and C, respectively.
Several different outgrowth morphologies are observed during the 1000 hour
exposure time for coating A. The feathery morphology shown in Figure N.18b was
found on test coupons after all exposure times. A similar corrosion product was found
by other investigators36 during a study which involved the oxidation of an Fe-13 wt%
Cr arc spray coating at 750°C. This morphology is believed to develop into the
morphology shown in Figure N.18c, which has a more block-like structure emerging
from its previous feathery appearance. This new structure was noted at 250 hours and
all subsequent exposure times. After 1000 hours, many outgrowths were characterized
by the block-like surface appearance, which has become more pronounced at this time
interval. Figure IV.18d shows this structure, which contains some feather-like features
which are reminiscent of the original morphology shown in Figure IV.18b. The
typical EDS spectrum obtained for these structures in shown in Figure IV.21, which
indicates a high chromium concentration.
48
Figure IV.IS. Surface of coating A after various exposure times to air at 600°C.(a) General view; (b) chromium-containing "feathery" morphology; (c) "blocky"morphology; (d) blocky morphology after 1000 hours.
49
Figure IV.19. Surface of coating B after various exposure times to air at 600°C.(a) General view; (b) chromium-containing "feathery" morphology; (c)aluminum-containing corrosion product; (d) iron-chromium-aluminum product.
50
Figure IV.20. Surface of coating C after various exposure times to air at 600°C.(a) General view; (b) chromium-containing "feathery" morphology; (c) duplexfeathery and "blocky" morphology; (d) aluminum-containing corrosion product.
51
Because of its similar composition to coating A, many of the corrosion
products formed on coating C were analagous in composition and morphology to those
shown in Figure IV.I8. The familiar chromium-containing "feather-like" structure,
which formed after only 50 hours of exposure time, is shown in Figure IV.20b. After
SOD hours, this structure can be seen to contain a duplex morphology (Figure IV.20c)
where both the "feather-like" and "block-like" morphologies coexist in the same
outgrowth. The typical EDS spectrum obtained for these outgrowths is similar to that
shown in Figure IV.21. In addition to these corrosion products, aluminum oxides were
also present at all exposures, the morphology of which is shown in Figure IV.20d.
The corresponding EDS spectrum obatined for the outgrowth shown in Figure IV.20d
is shown in Figure IV.22.
Corrosion products formed on the coating B surface can be seen in Figures
IV.19a-e. As for coatings A and C, the feathery outgrowth morphology shown in
Figure IV.19b was found to contain a high chromium concentration (EDS spectrum
similar to that shown in Figure IV.21). Other products which seemed to develop from
this morphology on the other coating surfaces, however, were not found for this
sample. As was expected, aluminum also quickly reacted with the test environment,
and the corresponding corrosion product can be seen in Figure IV.19c (EDS spectrum
similar to that shown in Figure IV.22). This structure was found at all exposure
intervals. One addition corrosion product, shown in Figure IV.19d, was found to
contain all three coating reactive elements, namely chromium, aluminum, and iron.
The corresponding EDS spectrum in shown in Figure IV.23.
52
-_-.-'..-._-.-'-'...-...'_.......'---.'-.-.......'.\..•_,-'.._.
Cr'
r
A~,·\~~--"-,-,w..,-.,.,...._,_w_,.,_,
l~BB HR OXID 45CT VFS: 786~ 1~. 24
Figure IV.21. EDS spectrum corresponding to the high.chromium containingcorrosion products found on the coating A and C surfaces after exposure tooxidizing conditions.
Cr'
A1
re
"." .•_._,.!\'I"_J
MI)
Ill\"'flf/\"~'/'\')""'·I/"'\~·".·"'\"~i"~\.-.'Ir'ty.~/..~,~
Cr'
/\5~~ HR 465 OXID vrs: 1128 lB.24
Figure IV.22. EDS spectrum corresponding to the high-aluminum containingcorrosion product found on the coating B surface after exposure to oxidizingconditions. .
53
A1
..........J..,..,.
Ck"
Mo•
J\"",."""'w""'.~-''''~-\"Fe
UFS:' 13~9 1~,24
Figure IV.23. EDS spectrum corresponding to the corrosion product shown inFigure IV.20d.
IV.C Sulfidation Test Results
IV.C.1 LOM Cross-Sectional Observations
The microstructures of coatings A, B, and C after exposure to the sulfidation
environment can be seen in Figures IV.24-26, respectively. As in oxidation testing,
the important occurances here include the formation of corrosion outgrowths at the
coating surfaces ("A" arrows), an increase in the coating oxide content ("B" arrows),
and the formation of a corrosion scale at the substrate/coating interface ("C" arrows).
It should be noted that, here, "oxide" denotes oxides and/or sulfides which may have
formed from the corrosion process.
54
Figure IV.24. Microstructure of coating A after 1000 hours exposure to 802 at600°C. (A: corrosion outgrowth, B: intercoating corrosion, C: interfacialcorrosion scale)
100 fLm
Figure IV.25. Microstructure of coating B after 1000 hours exposure to 802 at600°C. (A: corrosion outgrowth, B: intercoating corrosion, C: interfacialcorrosion scale)
55
Figure IV.26. Microstructure of coating C after 1000 hours exposure to S02 at600°C. (A: corrosion outgrowth, B: intercoating corrosion, C: interfacialcorrosion scale)
For coatings A and C, outgrowth size, as measured in terms of cross-sectional
length and width, is plotted as a function of exposure time in Figures IV.27 and IV.28
respectively. As noted from these figures, both coatings were extremely reactive in
the test environment. In fact, a corrosion scale, rather than outgrowths, was found on
coating C samples at all exposure times greater than 50 hours. Scale formation was
found on coating A after 1000 hours of exposure. The corresponding coating surface
area covered with corrosion is plotted in Figure IV.29, which shows that 100% of the
coating C surface was quickly covered with scale.
56
-C1:J
=0~V 15000
.pooj
e0- 0
..t::....00
= 10000Q)
IoJ o CoatineA
..t:: o CoatingC....~0 5000~
00....::s0Q)
0',On 0
000fI:S 0 200 400 600 800 1000 1200~(IJ
:> Exposure Time (hours)<Figure IV.27. Average outgrowth cross-sectional length as a function ofsulfidation exposure time.
300,------------------,
o200
0
0 0100 0
DO0
o00
CoatineACoating C
o
12001000800600400200
Exposure Time (hours)
ou---.---r-...---.--.....--r-~_._-....-r__-.--_lo
Figure IV.28. Average outgrowth cross-sectional width as a function ofsulfidation exposure time.
57
The fluctuations in coating oxide content with exposure time can b~ seen in
Figure IV.30. As was the case for oxidation testing, coating A is the most reactive
coating under sulfidizing conditions. The peak oxide content for coating A was found
to be 175% greater than its as-sprayed condition, as compared to 68% and 57% for
coatings Band C respectively. Although this data suggests coating A to be much
more reactive than coating C, it is important to note that coating C is approximately
twice as thick as coating A, and the calculation of area percent oxide is subsequently
skewed. The thickness of the scale at the substrate/coating interface as a result of the
corrosion process is plotted as a function of exposure time in Figure IV.31. As was
noted for oxidation testing, this scale was the thickest for coating B and the least thick
for coating C. Again, these trends can be attributed to coating structural aspects.
Q)-~uCJ)
uo...t=.....• ,...c
~100 0 0 0 0
~Q) 80~QJ 0;> o CoalineA0 60 0 o CoalingC
UQ)
40U 0~
'-I-c~
= 20CJ) 0..... ,nt:= 0Q) 0 200 400 600 800 1000 UOOU~Q)
Exposure Time (hours)~
Figure IV.29. Percent coating surface covered with corrosion scale as a functionof sulfidation exposure time.
58
..c: 200Q,I..c:0 cUQ,I"t'.... 150><0coc:...... C CoaIl:>SAnl0 100 C b. CoaU.S B
U o CoallnSC
c: C.... C
C b.Q,IlI.Inl
50 0III
'" 0 b. 0U 0 b.c: b....... 0c:Q,IU D
'" D 200 too 600 eoo 1000Q,I
~
Exposure Time (hours)
Figure IV.30. Percent increase in coating internal oxide content as a function ofsulfidation exposure time.
Exposure Time (hours)
8 CoalineACoatine B
6 CoatingC
UOO
o
o
1000800600400200
40
0
30
0 020 0 00
610 6
( 6
0+-.........-r-.........---.,-~-r----r----.--~-.---1- 0
Figure IV.3!. Substrate/coating interfacial corrosion scale thickness as a functionof exposure time.
59
IV.C.2 SEM/EDS Surface Observations
The corrosion products found on coatings A and C
surfaces after exposure to sulfidizing conditions were composed only of nickel and
sulfur. As will be shown in section IV.C.3, WDS results indicate the presence of
oxygen as well. Two distinct nickel-sulfur-(oxygen) morphologies were observed.
The most prominent of these is shown in Figures IV.32-33, which display a spherical-
shaped corrosion product morphology found on coatings A and C respectively. The
EDS spectrum obtained for this morphology is shown in Figure IV.34. Several
investigators have shown this morphology to consist of a duplex layer of NiO +
N· S 38-4113+x 2'
As was shown in the previous section, both coatings became completely
covered with this scale. The needle-like morphology shown in Figure IV.35 was
found to cover large portions of both coatings A and C. These needles acted as
nucleations sites for additional spherical particle growth, as can be seen in Figure
IV,36 (arrows). Figure IV,36 was taken off an edge of the coating A surface,
suggesting a sulfidation mechanism of outward nickel migration. It follows from the
above noted investigations38,39 that these needle-like outgrowths are comprised of
Ni3S2• Figure IV.37 shows the coexistance of these two morphologies on coating C
after 1000 hours of exposure. One additional, less prevalent nickel-sulfur-(oxygen)
corrosion product was observed on the coating A surface, as shown in Figure IV.38.
60
Figure IV.32. Coating A surface after exposure to sulfidizing conditions.
Figure IV.33. Coating C surface after exposure to sulfidizing conditions.
61
s N1
N1
urs: 864 10.24
Figure IV.34. EDS spectrum corresponding to the corrosion scale formed oncoatings A and C after exposure to sulfidizing conditions.
Figure IV.35. Needle-like morphology found on the coating A surface.
62
Figure IV.36. Corrosion morphology found on the nickel-containing coatingsafter exposure to sulfidizing conditions. Arrows indicate the recent nucleation ofspherical corrosion products on the needle-like corrosion product.
Figure IV.37. Corrosion structure on coating C surface showing the coexistanceof spherical and needle-like morphologies.
63
Figure IV.38. Nickel and sulfur containing corrosion product found on thecoating A surface.
The discontinuous aluminum-containing corrosion particles found on the
coating B surface after exposure to oxidizing condition (Figure IV.20c) were also the
primary corrosion product/morphology observed after exposure to sulfidizing
conditions. Aluminum was also noticed to react with sulfur, and the outgrowths
shown in Figure IV.39 are the result (arrows). The only other typical corrosion
product found, an iron-chromium oxide, is shown in Figure IV 040. The EDS spectra
for each of these morphologies is provided with the photomicrographs.
64
A1
S
1. -
VFS: 8881 le,24
Figure IV.39. Aluminum- and sulfur-containing corrosion product found on thecoating B surface, and corresponding EDS spectrum.
65
sre
A1
Cl'"
re
J\,\_\.,I,."''r''.~~'\''.'-Jl'".''.-.''JI_\.'''''''vrs: 2386 H3,24
Figure IVAO. Corrosion product found on the coating B surface, which containsiron, chromium, and aluminum; and corresponding EDS spectrum.
66
IV.C.3 SEMjEPMA Cross-Sectional Observations
The EPMA x-ray dot mapping results obtained for coatings A and C are
displayed in Figures IV.41-42, respectively. Each figure shows one surface-to-
substrate region on each coating after 1000 hours of exposure time. The surface
corrosion scale, the presence of corrosive species in the splat boundaries, and nickel-
depleted coating regions are indicated in these figures.
IV.D Cyclic oxidation test results
The microstructures of coatings A, B, and C after 43 thermal cycles can be
seen in Figures IV.43-45, respectively. Outgrowth size was measured for coatings A
and C after 25 and 43 cycles each in the same manner as for the isothermal test
samples. The test results are summarized in Table N.IV. There was no significant
change in coating thickness due to the cyclic oxidation process.
IV.E Cyclic thermal test results
The microstructures of coatings A and C after 3,000 thermal cycles is shown in
Figures IV.46 and IV.47 respectively. As a result of the test procedure, the thickness
of coating A was decreased by 17.5%, while there was no significant difference in the
thickness for coating C.
67
Figure IV.4I. SEI image and corresponding EDS/WDS x-ray dot maps for asurface-to-substrate region on the coating A cross-section. A: surface corrosionscale; B: corrosive species in the splat boundaries; C: nickel-depleted regionswithin the coating.
Figure IV.42. SEI image and corresponding EDS/WDS x-ray dot maps for asurface-to-substrate region on the coating C cross-section. A: surface corrosionscale; B: corrosive species in the splat boundaries; C: nickel-depleted regionswithin the coating.
68
Table IV.Ill. Outgrowth size data for cyclic oxidation test results.
Coating / Outgrowth length Outgrowth width % coating surfacenumber of (microns) (microns) area covered with
cycles outgrowths
A/25 11.4 ± 2.6 10.8 ± 4.1 0.6
N43 16.6 ± 12.1 9.1 ± 5.7 1.8
C/25 7.4 ± 2.5 10.2 ± 2.5 0.2
C/43 13.7 ± 4.3 13.7 ± 4.3 0.6
Figure IV,43. Microstructure of coating A after 43 thermal cycles. Arrowsindicate regions where spallation has apparently occurred.
69
:.;:--. '
-~ 100/Lm
Figure IV.44 Microstructure of coating B after 43 thermal cycles.
Figure IV.45. Microstructure of coating C after 43 thermal cycles.. ,
70
Figure IV.46. Microstructure of coating A after 3,000 thermal cycles.
-100f-Lm
Figure IV.47. Microstructure of coating C after 3,000 thermal cycles.
71
v. DISCUSSION
V.A Oxidation and Sulfidation
V.A.l Thennodynamic considerations
In the oxidation of coatings A and C, both of which contain large amounts of
nickel and chromium, the EDS results indicate the fonnation of chromium oxides on
the coating surfaces. The preferential oxidation of chromium over nickel is
thermodynamically supported. Using thermodynamic data from Gaske1l9, the standard
free energies of fonnation for one mole of NiO and CrZ0 3 can be calculated to be
160.5 KJ and -893.3 KJ respectively at 600°C. One would therefore expect the
formation of CrZ0 3 and a reduction of any NiO.1o,l1 Although the formation of Alz0 3,
with a standard free energy of fonnation equalling -1,401.6 KJ/moI9, is expected in
both coatings Band C, its presence is limited due to the low aluminum content in
these coatings. For coating B, oxides of the more abundant reactive coating elements,
namely Fe304 and CrZ03, are almost equal in stability with standard free energies of
formation -818,029.3 and -893,320.0 respectively. Disregarding any kinetic influence,
both of these phases can therefore be expected to form on the coating surface. The
thermodynamic expectations for all coatings are in agreement with experimental
results.
The EDS and EPMA observations obtained for coatings A and C show the
sulfidation process to be controlled by the reaction(s) between Ni and SOz' A
thermodynamic analysis of the nickel sulfidation process has been outlined
72
\ .-
elsewhere?7-40 Two possible reaction scenarios are suggested:
NiO + S03 = NiS049Ni + 2NiS04 =8NiO + Ni3Sz
Worrell and Ra041 have pointed out that although NiS04 is thermodynamically the
most stable of the above phases at 600°C, its formation is extremely slow. The fIrst
of the two reaction scenarios outlined above is therefore preferable. The formation of
these corrosion products explains the 'corresponding reaction kinetics. Researchers41
have found that the Ni3Sz phase provides a rapid transport path for nickel through the
.outer duplex scale to the gas-scale interface, resulting in reaction rates 104 to 106 times
faster than nickel oxidation rates.
Anderson and Kofstad40 suggest that for chromium-containing nickel alloys, it
is reasonable to assume that CrZ0 3 will form as well, as this would be the most
thermodynamically stable compound. However, because SOz can penetrate this scale
and nickel simultaneously oxidizes, the result is a non-protective scale consisting of a
V.A.2 Surface corrosion
Because of large data scatter, it is difficult to obtain any kinetic information
from outgrowth size. However, a measure of surface reactivity can be acquired from
Figures IV.I5 and IV.29, which illustrate the percent coating surface covered with
corrosion after exposure intervals in oxidation and sulfidation environments,
respectively. Because of the relatively slow reaction kinetics for the oxidation of
73
coatings A and C, and the subsequent large percentage of surface area available for
corrosion-product nucleation, Figure IV.15 shows a linear relationship for all exposure
intervals. A linear relationship also exists for sulfidizing conditions (Figure IV.29),
but here, 100% of the coating surface is covered with scale after 100 hours of
exposure for coating C and after 1000 hours for coating A. The rapid sulfidation
kinetics is explained by the reactions between nickel and sulfur dioxide, as discussed
in section V.A.l.
V.A.3 Intercoating corrosion
Figures IV.16 and IV.30 show the percent increase in coating oxide contents
after exposure to oxidation and sulfidation environments, respectively. These plots
suggest the reaction kinetics to be largely parabolic in nature, and thus, diffusion
controlled. The tendency for oxide content to rapidly increase with short exposure
times and then approach a steady-state value at longer times can be explained by the
following corrosion mechanism. Corrosive specie(s) progress from the environment
into the coating cross-section via splat boundaries, voids, and oxides. If the coating
ma.t~rial is reactive, splats will be attacked from their boundaries inwards. In the case
of oxidation, the formation of chromium oxide at the splat edges is probable (refer to
Table II.I) since each coating contains chromium. This oxide exhibits so-called
"protective" behavior, and the rate of further oxidation is subsequently diminished. 11,43
Whereas Figure IV.16 shows the increase in coating oxide content to become
essentially zero after a certain time for oxidation conditions, Figure IV.30 suggests the
74
\,
amount pf coating oxide content to continue to rise even after 1000 hours of exposure
for sulfidizing conditions. This could be attributed to the formation of non-protective
sulfides and/or oxides which can enhance the corrosion process.
The differences in coating responses shown in Figures IV.l6 and IV.30 can be
attributed to differences in coating characteristics, such as their reactivities, ease of
corrodent penetration, amount of oxide already present which may be protective,~
and/or free surface area available within each coating. Since coatings A and C have
similar compositions, they are expected to behave similarly. The differences between
these coatings as shown in Figures IV.16 and IV.30 must therefore be due to
microstructural differences. For example, since coating C is extremely dense and
possesses a high concentration of chromium and aluminum oxides, it lacks internal
free surfaces which are easily attacked and contains many internal protective oxides
which limit the corrosion process. In addition, since coating C has a thickness
approximately twice that of coating A, it has twice the cross-sectional area available
for corrosion. Thus, for a given amount of corrosion which can occur, the
corresponding calculated increase in intercoating oxide area percent for coating C will
be roughly half that for coating A. Even with this correction, however, the increase in
oxide content for coating A is significantly greater than that found for coating C.
For the sulfidation of the nickel-containing coatings, the mechanism of
corrosion is illustrated in the EPMA x-ray dot maps shown in Figures IVAI and ry.42
for coatings A and C, respectively. These figures illustrates three points of interest:
75
(i) The presence of corrosive species, Le. oxygen and sulfur, is limited to splat
boundaries, voids, and oxides. In the sulfidation of solid nickel specimens,
researchers37 have found that the inward transport of sulfur as atoms or molecules ~s
unlikely because the sulfur activity in the S02 environment is several orders of
magnitude less than that for the sulfides within the corrosion scale. For sulfur to
migrate into the bulk metal, it would have to diffuse against the chemical potential
gradient in the scale. It can therefore be reasoned that since the presence of splat
boundaries, voids, and oxides in thermal spray coatings provide a means for S02
diffusion, the coating sulfidation process is unique and distinct from that of bulk
alloys.
(ii) The migration of nickel from the coating bulk to the surface, where it
reacts with the environment, is evident. This occurance leaves behind a porous
corrosion "band" in the bulk coating. This behavior is in agreement with the
-sulfidation theory discussed in section n.A.2, which suggests an outward migration of
reactive elements to form a non-protective scale. A study involving the high
temperature corrosion of nickel in S02 environments also suggests this outward
diffusion of nickel37• It is further noted in the same study that the outward migration
of nickel may result in a detachment of the scale.
(iii) The porous regions left behind by outward nickel migration are thought to
have assisted in the corrosion process once the corrosive specie(s) have reached these
regions. These intercoating areas contain high levels of chromium and sulfur, and
provide fast diffusion paths for sulfur and oxygen migration.
76
L -
V.A.4 Corrosion Scale Formation at the Coating/Substrate Interface
V.A.4.i Kinetics of Scale formation
The thickness of the coating/substrate interfacial scale as a function of exposure
time has been plotted in Figures IV)7 and IV.31 for oxidation and sulfidation
environments, respectively. The corresponding corrosion rates have been determined
from these plots. Table V.I summarizes the type of oxidation which has occurred for
each coating and environment as well as the rate .constant for each. As shown in this
table, the corrosion rate of the substrate beneath the coating was in most cases
logarithmic in nature. This corresponds to a limiting oxide layer thickness where
further oxidation becomes negligible. In the case at hand, this behavior is attributed to
the formation of protective chromium oxide layers at coating splat boundaries, which
clog fast diffusion paths and limit subsequent substrate attack. The only exception to
this type of behavior was found for the coating B system in oxidizing conditions,
which best resembles linear rate oxidation kinetics. In this case, the rate of alloy
corrosion is unaffected by oxide formation and proceeds at a constant rate. Although
coating B does contain chromium and is expected to form internal chromiumoxide._
layers, the porous nature of this coating prohibits the oxide from becoming protective.
As can be seen from Figures IV.17· and IV.31, a linear oxidation rate is a catastrophic
form of attack, while logarithmic is more desireable.
For comparison purposes, the thickness of the corrosion scale on the uncoated
substrate regions were measured (Figures.V.1 and V.2). As can be seen from these
figures, the coating C substrate formed the thickest corrosion scale for both exposure
77
-atmospheres, while the coating A substrate formed the thinnest. These plots also show
that all three alloys formed a thicker stale during oxidation as opposed to sulfidation.
The corresponding oxidation rate types and constants are shown in Table V.II.
"~'~;\~
Table V.1. Coated substrate corrosion'rate type and corresponding rate constantsfor all coatings and both-test environments.
Coating/Environment Oxidation Type Rate Constant
A/oxidation logarithmic ke = 3.75
B/oxidation linear kL =3.33E-2
C/oxidation logarithmic ke = 1.34,
A/sulfidation logarithmic ke = 3.03
B/sulfidation logarithmic ke = 6.8}
C/sulfidation logarithmic ke = 1.57
Table V.II. Uncoated substrate corrosion rate'type and corresponding rateconstants for all substrates and both test environments.
Substrate/Environment Oxidation Type Rate Constant
A/oxidation'" linear kL = 8.06E-2
B/oxidation linear kL = 0.246
C/oxidation linear kL = 0.691
A/sulfidation logarithmic ke = 48.3
B/sulfidation logarithmic ke = 57.2
C/sulfidation parabolic Is, = 4.21
78
QJ,...c 800-r------------------,fUv~
rJ'JfJ'J=QJ 0 600~J-(
fU vJ-(.~
~8,.Q --- 400:s fJ'J
rJ'Jgj
"'a =QJ~~vfU·~
O..t=vE-c~ 1200
Exposure Time (hours)
o Coating A, oxo Coating B, oxA Coating C, ox
Figure V.l. Uncoated substrate corrosion scale thickness as a function ofoxidation exposure time.
QJ,...cfU 200V
rJ'J-rn
QJ =~O AfU bJ-(.~
l!J.~8 0 Coating A, sulf
,.Q--- 0 Coating B, sulf100 l!J. 0 0 l!J. Coating C, sulf:s fJ'J AA 0rJ'JfJ'J
QJ 0
"'a = 0QJ~ 0~v 0 0fU·~
O..t= 00vE-c=
0
~0 200 400 600 800 1000 1200
Exposure Time (hours)
Figure V.2. Uncoated substrate corrosion scale thickness as a function ofsulfidation exposure time.
79
As can be seen from Table V.II, all substrates oxidized at a linear, or
catastrophic, rate in oxidizing environments with the fastest attack occurring for
substrate C and the slowest for substrate A. Conversely, substrate attack was not as
severe in sulfidizing conditions, where substrates exhibited either logarithmic or
parabolic corrosion rates. The corrosion behavior of these alloys are dictated by alloy
composition, as shown in Table V.III. For example, the high chromium concentration
of substrate A accounts for its relatively good oxidation resistance when compared to
the other alloys.
Table V.ITI. Substrate compositions as determined by wet chemical analysis.
Substrate -~ Carbon (wt%) Chromium (wt%) Molybdenum (wt%)
A~.
0.11 1.21 0.50
B 0.22 0.10 0.02
C 0.23 0.45 0.01
A comparison of oxidation rates for coated and uncoated substrates can be
gained by inspection of Tables V.I and V.II. In this respect, coatings A and C
effectively protected their substrates, as linear substrate oxidation rates were reduced
to logarithmic. On the other hand, the oxidation rate for substrate B remained linear
even when coated. Although the application of coating B resulted in an order of
magnitude difference in linear oxidation kinetics, the failure of the coating to limit
substrate corrosion such that no further oxidation occurred after a certain exposure
time makes it inferior to the oxidation protection offerred by coatings A and C. As
80
mentioned earlier, the differences in coating performance are attributed to coating
structure. For example, inspection of Tables V.I and V.II indicates that coating C
provided the best substrate protection of all coatings in both environments; in
oxidizing conditions, the coating C substrate was reduced from the highest linear rate
to th~:.lowest logarithmic rate, while for sulfidizing conditions, it was reduced from the
only parabolic rate to the lowest logarithmic rate. Conversely, coating B provided the
least substrate protection resulting in the highest substrate corrosion kinetics for both
test environments. Since coating C is the most dense and thick coating of those
studied, and coating B is the most porous by an order of magnitude, these results are
expected.
V.AA.ii Interfacial Scale Thickness as a Function ofCoating Microstructure
The thickness of the coating/substrate interfacial scale (IS) resulting from test
procedures was used to determine the relationship between coating efficiency and
coating microstructure. A parameter, "mean free path (MFP) to substrate", was
developed and used as a measure of coating microstructure. This parameter denotes
the mean distance from coating surface to substrate via splat boundaries, voids, and
oxides. The dependence of MFP on coating microstructural features, such as porosity
and thickness, is shown in Figures V.3 and VA, respectively. Figure V.3 indicates
MFP to be inversely proportional to coating porosity content, as there is a shorter, less
tortuous path with increasing porosity. As expected, MFP and coating thickness are
~
directly proportional. Figure VA shows this relationship, although the data point
81
corresponding to coating B is low because of the high porosity content in this coating.
The relationship between the thickness of the scale a~ the coating/substrate
interface resulting from the corrosion process and the minimum MFP for all coatings
and both test environments is shown in Figure V.5. In the function shown on the y-
scale of this plot, [(1ST - ISTo) (relative UST], the terms are defined as follows:
1ST = Interfacial scale thickness [microns]ISTo = 1ST in as-sprayed condition [microns]. This term is necessary so that
the contribution to 1ST from spraying, and not the corrosion process,is substracted.
UST = thickness of corrosion scale found on uncoated substrate regions."Relative UST" refers to UST in comparison to the thickest USTamong substrates for the particular time interval of interest. This termis necessary to normalize the 1ST measurement with respect to thevarious substrate reactivities.
Figure V.5 shows that IS thickness due to corrosion is limited at high MFP's,
indicating that substrate attack is therefore minimal for coatings which are dense and
thick.3000
~
.... -~ enV°allngCQ.cs::
(1) 0(1) t 2000~ • .-c
.~ eCoaling As::- ok'
~ (1)(1) ....
~e1000....
(1) enCoating B00..0 \,!?' ~ ::s
~rJ'J(1)
> 0< .... 00 1 2 3 4 S 6
Coating Percent Porosity
Figure V.3. MFP as a function of coating porosity.
82
3000 ,(/
...c:.... -~ rJJ CoatlngC
~= ~C(1) 0
-(1) tJ-4.,..c 2000
~e Coaling A=- O~~ (1)(1) ....
~S1000....
(1) rJJCoating BOO~
~ = cFJ-4rJ'J(1)
:> 0< .... 0500 600 700 800 900 1000
Coating Thickness (microns)
Figure V.4. MFP as a function of coating thickness.
100~-----------------......,
E-:'(/)-I
f-4(/)--
80
60
0
040
0
20 o. II
• 0013
0
00 1000
o
2000 3000
o Coating A,oxo Coating B, oxb. Coating C. ox• Coating A, lull• Coating B,lUllA Coating C. lull
Minimum Mean Free Path (microns)
Figure V.5. Interfacial scale thickness as a function of mean free path for allcoatings and both test environments.
83
V.B The Effect of Cyclic Oxidation Testing
The difference between oxidation under isothermal and thermal cycling
conditions can be seen in Figure V.6. As evident from this plot, these different
oxidizing conditions promote a significant difference in the percent of coatings A and
C which are covered with outgrowths. Although there is not enough data to establish
oxidation trends, the comparatively small surface area covered with outgrowths
resulting from thermal cycling is attributed to the stresses that arise from thermal
expansion mismatch and subsequent outgrowth spallation.
6-r------------------,
0
0
o A.isothermalo C, isothermal
[J• A, cyclicIII C, cyclic
0
[J
100
•l'iI
200 300 400 500 600
Approximate Exposure Time (hours)
Figure V.6. Percent coating surface covered with corrosion as a function ofexposure time for both isothermal and cyclic heat treatments.
84
V.C The Effect of Cyclic Thermal Testing
The results of cyclic thermal testing illustrate the effects of porosity shape.
The as-sprayed coating A structure, as shown in Figure IV.la and IV.2a, contains
intersplat porosity which is elongated and tapered. With the introduction of thermal
cycles and corresponding stress fluctuations resulting from thermal expansion, such
pores may extend as cracks due to the stress concentrations at the crack tips.
Referring to previous micrographs, a typical void in the as-sprayed coating A
microstructure may have a length of 150 microns and a width of 10 microns. The
corresponding radius of curvature at the void "tip" can be calculated as44
1: = b2ja = 0.333,
where a = half major axisb = half minor axis.
Since the stress at the void tip can be given as44
the stress at this typical void tip will be approximately 30 times the stress found in the
coating bulk. The coating is therefore likely to yield or crack at the void ,tips, thus
promoting void "extension". Furthermore, given the equation44
where Kc is the coating material fracture toughness, O"ys is the coating material yield
strength, and ~ is a critical crack length, it can be seen that once a void grows to
some critical size as dictated by the coating strength, the coating fracture toughness
will be reached and catastrophic failure (uncontrolled crack growth and subsequent
localized spallation) will result. After 3,000 thermal cycles, the stucture of coating A
85
\
appears as found in Figure IVA6. It appears from this micrograph that the voids have
grown in size due to the thermal cycing process, and regions of the coating near the
surface have spalled due to this phenomenon (arrows). This failure mechanism
resulted in a 17.5% reduction in average coating thickness after 3,000 cycles.
In contrast to coating A, coating C possesses no visible porosity (Figures IV.1c
and IV.2c) and the stresses which arise from thermal cycling must therefore act upon
the coating as a continuous entity rather than becoming localized at intersplat void
tips. As a result, cracking within the coating does not occur, but the coating may
become separated from its substrate as shown in Figure IVA7. This type of spallation
is not uncommon to thermal spray coatings since the bond between the coating and
substrate is mechanical in nature. Bond strength is therefore relatively weak when
compared to other coating types (such as diffusion coatings) and is sensitive to
substrate conditions such as roughness and cleanliness.
V.D The Effect of Coating Composition,
Depending on chemical composition, a coating will behave in either a
"reactive" or "non-reactive" manner for a given environment. This behavior is thought
to be independent of coating structure. In this regards, the performance of each
coating for each of the test environments is summarized in Table V.IV. In considering
the data in this table, it should be noted that the coating C oxide content values are
d~ceptively low since this coating is roughly twice as thick as the other coatings, and
the calculation of area percent intercoating oxide content is subsequently skewed.
86
Table V.IV. Coating reactivity based on surface scale formation and intercoatingoxide content. (O=oxida.tion test results; S=sulfidation test results)
• •. ,~nt-s.mPle Maximum percentCoating surface cove d with scale increase in intercoating
oxide content
A 9.97(0), 100(S) 159(0), 175(S)
B N/A; N/A 34(0), 68(S)
C 6.78(0); 100(S) 33(0), 57(S)
"Non-reactive" behavior is schematically modelled in Figure V.7. As can be
seen from this figure, the coating structure remains unaffected by exposure to a
corrosive environment. At some point in time, however, the corrosive specie(s) have
diffused through the substrate material via splat boundaries. Subsequent attack is
noted in the form of small, localized regions in the substrate. With some further time
increment, these localized corrosion regions grow along the substrate/coating interface
to form an interfacial corrosion scale. An example of non-reactive coating behavior is
illustrated in Figure IV.9, which. presents the structure of coating C after 1000 hours in
oxidizing conditions. It is noted from this figure that the coating structure remains
virtually unchanged from its as-sprayed condition, but substrate attack is considerable.
From microstructural observations of coatings which behaved in a non-reactive
fashion during laboratory test procedures, the following failure mechanisms are noted:
(i) the formation of a corrosion scale at the coating - substrate interface due to the
migration of corrosive specie(s) along splat boundaries, oxides, and voids, and
(ii) the possible formation of small, localized corrosion products at free surfaces.
87
!-.
--:::::::::::::::::::::::,,]{::;:;:::::::::::8:::::;::;::::::::;:::::::;::::::::::::::::::::::::T::::::::::::::::::::::::::::::::::::::::~::::::::::;::::::::::::;::I:::::::;::::::::;:::::::::::::::::::::::::::::::;:~;:::::::::::
;::::::::::::::::;:::::::;:::::::::::::::::;::::::::::::::::::::::::1:::::::::::::::::::::::::::;::;:::::::::::::::::::::::::;:::::::::::'1::::::::::1::::::::::::::::::::::::::::::::::::::::~::::::::::::::::::::::::::'[::::::::::::::::-
:::::1:'::::::::::::::::':::::::':::':':':"'::":::':::::':,%:,:::,::I-:::,:l::,:::,:':':::::::'::i:::::::':::':,:::::::::,:::,,:;:'::,:::,,:::1-:::::1::::,::::,::,::,:,::,:,,::,::::::::::::,:::::::::::::::::::,::::::::::::::::::-:::;::::::::::::::::::::::::::::::::::::::::::::::::::::::::,::::::]'::::::,:::::::::::::::::::::::::::::::::::::::::,:,::::::::::::::::Y:::::::::- :::::,::::::;,:::::::::::::::,:::::::::::::::::::::j::::::::::::::::1'::::::::::::::::--
-.;:::::::::: :::::::::::::~:f:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::T:::::::::::::::::::::i:::::::::::::::::::::::::::::::::::::::::::::I::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
;:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::I::::~:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::E:::::- :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::1:-:::::::::::::::-::::]:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::I::;::l-:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::1:":1":::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::;:::;:::-::::::::::::;::::::::::::::::::::::::::::::;::::~:;::::;::::::;:::I::::::::::;;;::::;;:::::::::::;;:;;:::;:::::::;:;:::::;;;:::::::::::T:::;:::::- ::;:;:::;:::::::::::::::::;:;;;:::::::::::::::;:::::;:::;:::::;:;:;:'[::::;:::::}:--
-
Figure V.7. Schematic model of "non-reactive" coating behavior. A=substratematerial; B=splat; C=oxide within coating; D=coating void; E=substrate attack.
88
o
-...:.:.::-..::::;::.....
':':'::'.':'.:.
Figure V.8. Schematic model of "reactive" coating behavior. A=progression ofcorrosive specie(s) into coating thickness via splat boundaries, voids, and oxides;B=corrosion scale on coating surface. (See Figure V.4 for further description).
89
'.'
"Reactive" coating behavior is schematically modelled in Figure V.8. The
initial form of attack noted from this schematic illustration is the formation of surface
corrosion and the progression of corrosion into the splat boundaries closest to the
coating surface. With an increase in time, a "corrosion path" which extends from the
coating surface to some distance within the coating thickness is notable due to the
migration of corrosive specie(s) along splat boundaries and the subsequent attack of
splats. The splats most closest to the coating surface eventually become completely
engulfed with corrosion product and soon after become part of the surface scale. The
mechanism of substrate attack is analogous to that for non-reactive coatings. An
example of reactive coating behavior is shown in Figure IV.24, which presents the
coating A structure after 1000 hours under sulfidizing conditions.
From microstructural observations of coatings which behaved in a reactive
fashion during laboratory test procedures, the following failure mechanisms are noted:
(i) the formation of a corrosion scale at the coating - substrate interface due to the
migration of corrosive specie(s) along splat boundaries, oxides, and voids; (ii) splat
attack from boundaries inwards; (iii) a substantial increa,se in intercoating oxide
and/or sulfide content; (iv) the possible formation of an "oxide network" and
subsequent cracking; (v) the possiple migration of coating elements to the surface
scale, leaving behind an intercoating porous layer; and (vi) possible spallation
resulting from the linkup of corrosion and/or porous layers.
90
VI. CONCLUSIONS
1. During the high-temperature corrosion testing of thermal spray coatings A, B, and
C, the following phenomena were found to occur:
i) Corrosion "outgrowths" formed on the coating surfaces. The average cross-
sectional length of these outgrowths ranged from t~n{of microns for oxidation/"~-
testing, to tens of millimeters in the case of sulfidation testing. The percent of the
coating surfaces covered with these outgrowths increased linearly with exposure
time until one continuous scale was formed. After oxidation testing, outgrowths
were found to be rich in chromium; aluminum, chromium, and iron; and chromium
and aluminum for coatings A, B, and C, respectively. In the case of sulfidation,
outgrowths were found tosontain nickel; aluminum, chromium, and iron; and
nickel for coatings A, B, and C, respectively.
ii) The amount of intercoating corrosion increased with exposure time. This
increase was found to be either parabolic or logarithmic with time, indicating the
process to be diffusion controlled. The amount of intercoating corrosion which
occurred was attributed to the reactivity between coating elements and the
corrosive specie(s), the ease of corrodent penetration, the presence of protective
oxides within the coating structure, and/or free surface area available within each
coating.
91
iii) A corrosion scale formed at the coatIng/substrate interfaces, the thickness of
which was found to be dependent on coating microstructure. A measure of coating
porosity and thickness was defined in terms of the "mean free path" (MFP) from
coating surface to substrate via splat boundaries, voids, and oxides. Interfacial
scale thickness was found to be limited at high MFP's, corresponding to thick and
dense coatings.
2. Substrate corrosion rates were found to be substantially less for coated substrates
as opposed to uncoated ones. Furthermore, substrate corrosion protection was related
7 to coating structure. Since coating C was the thickest and most dense of the coatings
(highest MFP), it provided the best substrate protection by most significantly reducing
the substrate corrosion rate in both test environments.. Since both coatings C and A
were relatively dense, these coatings were able to reduce substrate oxidation kinetics
from linear to logarithmic by the formation of protective chromium oxides at splat
boundaries, voids, and other fast diffusion paths. Conversely, since coating B was the
most porous coating by an order of magnitude, it offered the least substrate protection.
Because of its porous nature, this coating was unable to form protective oxides within
all fast diffusion paths. Subsequently, substrate oxidation kinetics remained linear
even with the addition of coating B.
3. For the nickel-containing coatings (coatings A and C), surface corrosion kinetics
for sulfidation were approximately two orders of magnitude faster than that for
92
oxidation. Whereas only 9% and 7% of the coatings A and C surfaces were covered
with c?rrosion after 1000 hours of oxidation exposure, a continupus scale existed on
these coatings after 1000 and 100 hours of respective exposure to sulfur dioxide. The
rapid sulfidation kinetics is attributed to the surface formation of Ni3S2, which has
been found by other investigators to provide a rapid transport path for nickel to the
gas-scale interface.
4. For the nickel-containing coatings (coatings A and C), the mechanism of
sulfidation was identified. Sulfur and oxygen was found to progress from the coating
surface towards the substrate by way of splat boundaries, voids, and pre-existing
oxides. Nickel was found to concurrently migrate from the coating bulk to the
surface, where it reacts with the S02 environment to form a non-protective scale. The
porous regions left behind by outward nickel migration are thought to have assisted in
the corrosion process by providing fast diffusion paths for sulfur and oxygen
migration.
5. As a result of cyclic oxidation testing, the percent coating surface covered with
corrosion outgrowths was decreased by as much as a factor of six when compared to
equivalent oxidation exposure times. This difference was attributed to the stresses that
arise from thermal expansion mismatch and subsequent outgrowth spallation. As a
result of cyclic thermal testing, the role of porosity morphology became apparent.
Elongated and tapered pores were found to crack at their "tips", leading to void linkup
93
and possible subsequent coating spallation. This process was responsible for a 17.5%
decrease in the coating A thickness after 3,000 thermal cycles. The stresses resulting
from thermal fluctuations, and the concentration of these stresses at the pore tips, were
responsible for this failure mechanism.
94
VII. REFERENCES
1. Balzhiser R.E. and Yeager, KE., Scientific American, 9(1987)100-107.
2. Boiler Tube Failure: Correction, Prevention, and Control, EPRI report GS-6467,Research Project 1890-7, Palo Alto, California, '1989.
3. Wright, 1.0., Williams, D.N., and Mehta, A.K, in Boiler Tube Failures in FossilPlants (Proc. conf.), EPRI, Palo Alto, California, 1987.
4. McNabb, D., Sidney, D., and Patterson, R.W, in Boiler Tube Failures in FossilPlants (Proc. conf.), EPRI, Palo Alto, California, 1987.
5. Meadowcroft, D.B., Materials Science and Engineering, 88(1987)313-320.
6. Cocubinsky, 1., in Failures and Inspections of Fossil-Fired Boiler Tubes (Proc.conf.), EPRI CS-3272, Palo Alto, California, 1983.
7. Stringer, 1., in High Temperature Corrosion in Energy Systems (Proc. Conf.) ,Michael F. Rothman ed., AlME: New York, 1985, pp. 3-27.
8. Pohl, 1.H., Heap, M.P., and Mehta, A.K, in Failures and Inspections of FossilFired Boiler Tubes (Proc. conf.), EPRI CS-3272, Palo Alto, California, 1983.
9. Gaskell, D.R. Introduction to Metallurgical Thermodynamics, 2nd ed., New York:Hemisphere Publishing Co., 1981.
10. Pettit, F.S., Goebel, lA., and Goward, O.W., Corrosion Science, 9(1969)903-913.
11. Birks, N., Meier, O.H., and Pettit, F.S., Jornal of Metals, 12(1987)28-31.
12. Singer, J.O., ed. Combustion: Fossil Power Systems, 3rd ed., CombustionEngineering, Inc., 1981.
13. Stringer, 1., High Temperature Technology, 3(1985)119-141.
14. Meier, O.H., Pettit,F.S., in High Temperature Corrosion in Energy Systems (Pmc.Conf.), Michael F. Rothman ed., AIME: New York, 1985, pp. 161-176.
15. Mayer, c.A., Welding Design and Fabrication, 55(1982)66-75.
16. Zanchuk, W., Surface and Coatings Technology, 39/40(1989)65-69.
95
17. Chou, F.S., Daniel, P.L., Blazewicz, AJ., and Dudek, R.F., in High Temperature·Corrosion in Energy Systems (Proe. Conf.), Michael F. Rothman, ed., AIME:New York, 1985, pp.327-:343.
18. EPRI report CS-1653. Corrosion Problems in Coal-Fired Boiler Superheater andReheater Tubes - Fireside Corrosion. Palo Alto, California, 1980.
19. Corchia, M., Delogu, P., and Nenci, F., Wear, 119(1987)137-152.
20. Sangam, M., and Nikitich, J., Journal of Metals, 9(1985)55-60.
21. Herman, H., Scientific American, 9(1988)112-117.
22. Guzi, C.E., Thun, D.P, and Zellmer, G.F., in 1986 Fifth International Symposiumon Corrosion in the Pulp and Paper Industry (Proc. conf.), B.c., Canada, 1986.
23. Personal conversation with Carl Lohstroh, Proctor and Gamble Co.
24. Pennsylvania Power & Light. Summary of the Unit #1 1983 Metal Spray Test. Areport prepared by PP&I, PP&L: Allentown, Pa., 1989.
25. GPU. Evaluation of Coated Economizer Tubes from Homer City Unit #2. A reportprepared by GPU Nuclear, GPU: Reading, Pa., 1989.
26. NYSEG. Plasma Spray Metallizing, Homer City Unit #2 Boiler Research andDevelopment Project. A report prepared by NYSEG, NYSEG: Homer City, Pa.,1987.
27. Department of Trade and Industry. Wear Resistant Surfaces in Engineering.London: Crown Publishers, 1986.
28. McPherson, R., Surface and Coatings Technology, 39/40(1989)173-181.
29. Mathur, P., Apelian, D., and Lawley, A., Acta metallurgica, 37/2(1989)429-443.'.,.
30. Apelian, D., Paliwal, M., Smith, RW., and Schilling, W.F., International MetalsReview, 28/5(1983)271-294.
31. Rickerby, D.S., Eckold, G., Scott, KT., and Buckley-Golder, I.M, Thin SolidFilms, 154(1987)125-147.
32. Iwamoto, N., Makino, Y., Umesaki, N., Endo, S., Osaka, and Kobayashi, H., in10th International Thermal Spray Conference (Proc. conf.), Germany, 1983.
96
33. Kvernes, 1., Espeland, M., and Norholm, 0., Scandanavian Journal of Metallurgy,17(1988)8-16.
34. Novak, R.C., Jour. of Engineering for Gas Turbines and Power, 110(1988)617620.
35. Fowler, D.B., Riggs, W., and Russ, J.C., Advanced Materials and Processes,138(1990)41-52.
36. Harris, SJ., and Overs, M.P., Thin Solid Films, 118(1984)495-505.':-
37. Seiersten, M., and Kofstad, P., Corrosion Science, 22/5(1982)487-506.
38. Kofstad, P., and Akesson, G., Oxidation of Metals, 12/6(1978)503-526.
39. Gesmundo, F., de Asmundis, C., and Nanni, P., Oxidation of Metals, 20, Nos. 5/6,1983.
40. A. Andersen and P. Kofstad, Corrosion Science, 24/8(1984)731-743.
41. Worrell, W.L., and Rao, B.K, in High Temperature Corrosion (Proc. conf.), R.A.Rapp, ed., National Association of Corrosion Engineers, Houston, Texas, 1981.
42. Natesan, K, and Baxter, DJ., in Corrosion-Erosion-Wear of Materials at ElevatedTemperatures (Proc. conf.), A.V. Levy, ed., National Association of CorrosionEngineers, Berkeley, California, 1986.
43. Stott, F.H., Materials Science and Technology, 5(1989)734-740.
44. Hertzberg, R.W. Deformation and Fracture Mechanics of Engineering Materials.3rd ed., New York: John Wiley & Sons, 1989.
97
VITA
Scott Thomas Bluni was born to parents Thomas and Carol in Rockville Centre, New
York, on December 16, 1966. He was raised in Massapequa, and later Coram, Long
Island. After graduating from Saint Anthony's High School in Huntington, New York,
he attended Lafayette College in Easton, Pennsylvania. He graduated from Lafayette
with honors in 1989 with a B.S. Metallurgical Engineering Degree. Since then, he has
been pursuing graduate degrees at the Department of Materials Science and
Engineering at Lehigh University, Bethlehem, Pennsylvania. Since receiving an M.S.
-------degree, Scott has been studying the solidification of eutectic zinc-aluminum hot-dip
coatings at Lehigh.
98