PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

Practical Observations and Interpretation

of Oxide Growth and Exfoliation in Steam
Barry Dooley and Ian Wright

ABSTRACT
Over the last 40 years oxide growth and exfoliation (OGE) in superheater and reheater tubing have been responsible
for a number of power plant problems which seriously have affected reliability. In the same time period, the authors
have collected a data base of scale morphologies that has been used to describe in detail the progression of oxide
scale development to the point where failure can occur. The concomitant evolution of knowledge of the factors
that determine the mode of scale growth and failure in steam has provided the foundation for defining the specific
stages in that progression, understanding differences among ferritic and austenitic alloys, and for categorizing
the influence of plant operating characteristics. In particular, while tube/steam temperature and the maximum
temperature drop at plant shutdown are major variables, the specific cycle chemistry used for the plant feedwater
has very little influence. Key stages in the progression of scale growth to the point of failure are identified as OGE
indices that are specific for ferritic and austenitic steels. These indices are intended to be used proactively to
determine the current condition of a superheater or reheater on the path to exfoliation and possible plant damage.
Also, by analyzing samples of exfoliant or oxide deposits responsible for damage, the origin of the oxide, and thus
a possible superheater or reheater problem, can be identified retroactively.

1 BACKGROUND
Oxide growth and exfoliation of alloys used in ◾ Deposits in steam turbines causing loss of
superheater and reheater steam circuits have performance and capacity.
caused numerous problems in the power indus- ◾ Exfoliated scale in condensate, contributing
try since the late 1960s. Some of these include: to the total iron level.
◾ Short-term overheating (STO) in pendant su- Concern over the first two issues led in the mid-
perheaters (SH) due to accumulation of ex- 1970s to efforts to collect and begin to organize
foliated oxide from austenitic alloys at tight data on the various oxide growth and exfolia-
(hairpin) bends. tion (OGE) morphologies and associated service
◾ Solid particle erosion (SPE) of initial blade conditions, so that the sources of these damage
rows/stages in high-pressure (HP) and inter- mechanisms could be identified and understood
mediate-pressure (IP) steam turbines due to [1–3]. The present authors have been instrumen-
exfoliation from ferritic steels. In extreme tal in recent years in collaborations to combine
cases damage extends through to the initial the accumulating observations with the devel-
low-pressure (LP) stages. opment of insight into the mechanisms of oxide
◾ Longterm overheating (LTO) of ferritic alloy growth and exfoliation [4–6]. These efforts led
tubing due to excessive growth of steam-side to the concept of an Oxide Growth and Exfolia-
oxides due to poor superheater/reheater de- tion Index (OGEI), which is published for the first
sign, resulting in tubes operating above the time in this paper. Understanding of similarities
Oxidation Limit (OL) for the alloy, or to in- and differences among alloys and alloy classes
creasing temperatures as a result of increas- was recognized to be key to defining the steps
ing oxide thickness acting as an insulator. leading to scale failure, as well as for project-
ing the conditions likely to initiate an exfoliation
◾ Deposits throughout steam circuits, some-
event and the form taken by the material lost.
times causing sticking of valves.
Assembly of these observations into two sets
◾ Erosion of steam turbine control valves. of indices has resulted in a methodology for in-
◾ Erosion of bypass valves. dicating whether and when such problems can

374 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

Parameter Range
Ferritic Alloys T11, T12, T22, T5, T9, T23, T91
Austenitic Alloys
Standard austenitics TP304H, TP316H, TP321H, TP347H, TP309, TP310HCbN
Fine-grained TPS304H
Fine-grained & shot-peened TP347HFG
Environmental Baselines
Steam temperature (Ts) saturated and superheated: 1 004–1 202 °C/540–650 °C
Steam pressure (Ps) Sub-critical and super-critical: 290–4 873 psi (20–336 bar)
Tube temperature (Tm) Range up to 1 202 °F/650 °C
Feedwater cycle chemistry AVT and OT with oxygen levels ranging from 1–400+ µg · L–1
Table 1:
Materials and environments of superheater/reheater steam circuits.
AVT(R) all-volatile treatment under reducing conditions
AVT(O) all-volatile treatment under oxidizing conditions
OT oxygenated treatment

be expected by examining superheater/reheater

tube samples removed for metallurgical analy-
sis. In addition, this approach allows identifica-
tion of the exact source of the exfoliated oxide
that has caused the damage delineated above.
To help in understanding the practical impor-
tance of OGE, the following introductory sections
are included to summarize the main features
which relate directly to the suggested OGEI. The
materials and the steam environments of focus
for this paper are included in Table 1. The compo-
sitional specifications for the alloys are summa- Figure 1:
rized in Table 2. Schematic diagram of a cross section of the initial scale typically
formed in steam on ferritic steels containing up to 9 wt% Cr.
1.1 General Introduction to Oxide Growth
Generally, scales growing in steam on the al-
loys listed in Table 2 exhibit morphologies that case for the original L1 on ferritic steels, where
consist typically (at least initially) of two main new layers may form in addition to (or from)
layers. The layer next to the alloy usually is re- growth of the original L1. These changes result
ferred to as an "Fe-Cr spinel" (abbreviated in this largely from the participation of chromium in the
paper as 'L1'), and the main outer layer is "mag- scale-forming process and its effect on slowing
netite" (Fe3O4, abbreviated in this paper as 'L2'), scale growth, and can be used to discriminate
see Figure 1. In simple parlance, the interface in particular among the behaviors of alloys T22,
between L1 and L2 represents the original alloy T23, and T91. The changes that occur in the orig-
surface, with L1 taken to grow inwards, and L2 inal L2 are similar for the ferritic and austenit-
outwards. In fact, both layers are basically mag- ic steels: while the thickness of the original L2
netite (which has a spinel structure), but while increases with time, under some circumstances
L1 contains elements incorporated from the al- the magnetite is further oxidized to hematite
loy as the oxide front progresses inward (chiefly (Fe2O3). The hematite forms along the original
Cr, but also Mn, Si, Mo, etc.), L2 is essentially L2-steam interface at the expense of the magne-
pure magnetite by virtue of its growing by out- tite, and may eventually occupy a large fraction
ward diffusion of iron ions. of the space that otherwise would be described
as 'L2.'
With continued scale development, changes
occur in the morphologies of both original lay-
ers that require more detailed descriptions than
suggested by 'L1' and 'L2.' This is especially the

PPCHEM JOURNAL 21 ▪ 2019/06 375

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

Composition [wt%]
Alloy
Cr Ni Si Mn C S P Mo V Nb Ti W Cu
T12 0.8– 0.5 0.3– 0.05– 0.025 0.025 0.44–
1.25 0.61 0.15 0.65
T11 1.0– 0.5– 0.3– 0.05– 0.025 0.025 0.44–
1.5 1.0 0.6 0.15 0.65
T22 1.9– 0.5 0.3– 0.05– 0.025 0.025 0.87–
2.6 0.6 0.15 1.13
T23 1.9– 0.5 0.1– 0.04– 0.01 0.03 0.05– 0.2– 0.02– 1.45–
2.6 0.6 0.1 0.3 0.3 0.08 1.75
T5 4–6 0.5 0.3– 0.15 0.025 0.025 0.45–
0.6 0.65
T9 8–10 0.25– 0.3– 0.15 0.025 0.025 0.9–
1.0 0.6 1.1
T91 7.9– 0.18– 0.25– 0.06– 0.012 0.025 0.8– 0.16– 0.05–
9.6 0.56 0.66 0.15 1.1 0.27 0.11
TP304H 18–20 8–10.5 0.75 2 0.04– 0.03 0.045
0.1
TP316H 16–18 10–14 0.75 2 0.04– 0.03 0.045 2–3
0.1
TP321H 17–19 9–12 0.75 2 0.04- 0.03 0.045 a
0.14
TP347H 17–19 9–13 0.75 2 0.04– 0.03 0.045 b
0.1
TP347HFG 17–20 9–13 0.75 2 0.06– 0.03 0.04 b
0.1
Super304H 17–19 7.5– 0.3 1 0.07– 0.001 0.004 0.3– 2.5–
10.5 0.13 0.6 3.5
TP309 22–24 12–15 1 2 0.2 0.03 0.045

TP310HCbN/HR3C 24–26 19–22 1 2 0.04– 0.03 0.045 0.2–

0.1 0.6
Table 2
Compositional specifications of alloys. The preferred ranges for key elements are shown, while a single value denotes
the maximum allowable.
a: 4 x (C + N) to 0.7 max
b: 8 x (C + N) to 1 max
Cr chromium Si silicon C carbon P phosphorus V vanadium Ti titanium Cu copper
Ni nickel Mn manganese S sulfur Mo molybdenum Nb niobium W tungsten N nitrogen

Because at the temperature of operation (Table pendent on the partial pressure, p, of oxygen
1) these oxides are semiconductors, they grow from the equilibrium dissociation of steam, via:
by diffusion of iron ions (mostly Fe2+) outwards
and oxidant inwards (the oxidant typically is con- H2O (g) ⇔ H2 (g) + 1/2O2 (g) (1)
sidered to consist of oxide ions (O2–)) [4]. Hence, for which:
growth of the magnetite layer (L2) occurs at the
oxide-steam interface, and Fe-Cr spinel (L1) at k1= (pO2)eq0.5 × pH2/pH2O (2)
the oxide-metal interface. Scale thickening as a and k1 is calculated from the known free energy
result of these counter diffusion processes ini- of the dissociation reaction. As shown in Figure 2
tially follows a parabolic rate law. values of pO2 from the dissociation of steam
It is also important to note that the oxide growth are several orders of magnitude higher than
processes are not dependent on the cycle chem- those for the formation of the various oxides
istry used in the feedwater circuits of the plants considered. "Reducing" conditions associated
(AVT(R), AVT(O) or OT) [7], nor specifically on with AVT(R) feedwater treatments (usually
the dissolved oxygen level. Oxide growth is de- < 10 µg · kg–1 oxygen plus a reducing agent) re-

376 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

enly as being related to the oxygen level in the

steam. The general further development of the
duplex layers and hematite are covered later in
the OGEI sections (Section 2).

1.2 General Introduction to Exfoliation

The initial duplex oxide scales grow in an es-
sentially stress-free state at high temperatures
because of the equal counter flux diffusion pro-
cesses indicated previously. Any stresses expe-
rienced are accommodated by adjustment of the
morphology of the growing scale and possibly
by creep. For the ferritic alloys that subsequently
develop multi-laminated scales, growth stress-
es are generated in the overall oxide layers be-
cause of the confined space. Additional stresses
can arise during scale growth from physical con-
siderations such as constraint from the curved
tube surfaces, and from operational factors such
as periodic variations in steam temperature and
Figure 2:
pressure. However, the largest and most conse-
Comparison of the oxygen partial pressures (pO2) for the
formation of the oxides of interest; those resulting from the quential stresses are from the differential con-
equilibrium dissociation of steam also are included [4]. traction that occurs between the alloy and scale
(and among scale components) during cooling
upon unit shutdown.
sult in an effective pO2 of the order of 10–9 atm
(log10pO2 value of –9.0), which also is several or- Table 3 lists values of the coefficients of thermal
ders of magnitude higher than for the oxides. For expansion (CTE) for ferritic and austenitic steels
OT conditions involving oxygen additions of 20 and oxides of interest, and illustrates the high-
or 200 µg · kg–1, the effective pO2 ranges from 2.2 er values for austenitic steels and the large dif-
x 10–6 to 2.9 x 10–5 atm (log10pO2 values of –5.6 to ferences between the steels and magnetite and
–4.5), higher than for the dissociation of steam. hematite. No separate value for the CTE of Fe-Cr
Overall, from thermodynamic considerations, spinel (i.e., L1) is shown, since practical meas-
all the oxides (including hematite) are expected urements have indicated that it closely tracked
to be stable under all boiler steam conditions, the upper bound of the values for "pure" magne-
though wüstite (FeO) often may be absent due tite (as listed in the "revised" values in Table 3).
to kinetic considerations. This finding was true for the range of chromium
levels found in Fe-Cr spinels formed on T91 and
Often, hematite is observed to be absent from
standard austenitic steels [8].
such scales, and this also is a result of overrid-
ing kinetic factors. Since magnetite is thermo- Examples of calculated oxide-metal strains af-
dynamically more stable than hematite (i.e., can ter cooling from 1 112 °F/600 °C to ambient are
form at lower oxygen partial pressures), and shown in Table 4. The cooling strains are direct-
since hematite grows by further oxidation of ly related to the magnitude of the temperature
magnetite, in situations where the availability of drop, not to the rate of temperature change. This
iron ions at the scale-steam interface is great- is an important practical point as it implies that
er than that of oxidant, magnetite will form in exfoliation on austenitic materials does not oc-
preference to hematite. This typically is the sit- cur until a boiler has cooled almost to ambient,
uation during the early stages of scale growth and that shorter shutdowns ("hot," "warm") may
and/or when outward transport of iron ions out- not be accompanied by exfoliation. From these
ward from the alloy is less impeded by blocking. CTE data it is evident that the cooling strains on
When such blocking does occur, a point will be ferritic steels are tensile, and are smaller than the
reached where the relative availability of iron compressive strains experienced on austenitic
ions at the magnetite-steam interface diminish- steels. Also shown is the large influence of the
es, and the existing magnetite will start to be presence of hematite (assumed in calculations
oxidized to hematite. This is an important point to be a uniform layer along the outer magnetite
because very often the appearance of hematite (L2)-steam interface); a value for 20 % hematite
along this interface has been interpreted mistak- (by thickness) is shown as an example, but this

PPCHEM JOURNAL 21 ▪ 2019/06 377

PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

CTE (K–1 x 10–6) [1] CTE (K–1 x 10–6) Revised 2019 [9]
572 °F/300 °C 1 112 °F/600 °C 572 °F/300 °C 1 112 °F/600 °C
T22 ferritic 14.0 16.2 12.9 13.7
T91 ferritic — — 12.2 12.4
TP300-series 18.6 a
19.5 a
17.5 b
18.7b
austenitics
Magnetite 14.25 16.5 10.4±1.3 13.4±1.4
Hematite 11.9 12.9 10.0 11.8

Table 3:
Examples of differences in coefficients of thermal expansion for alloys and oxides.
a: TP316
b: TP347

T22 Ferritic TP316 Austenitic

Scale with 0 % Fe2O3 1 x 10–4 tension 1.8 x 10–3 compression
Scale with 20 % Fe2O3 0.5 x 10 tension
–4
2.0 x 10–3 compression
Table 4:
Examples of oxide-metal strains on cooling from 1 112 °F/600 °C to ambient for tube materials with scales having different levels
of hematite along the oxide-steam interface [10].

effect continues to increase with increasing lev- While maps such as Figure 3 can indicate when
els of hematite. scale failure is expected, and generally by which
mode, an important practical consideration is
Calculations of the critical strains at which in-
to be able to define the conditions under which
dividual oxides will fail as a function of scale
actual loss of scale will occur, and the charac-
thickness and mode of failure have been used to
teristics of the material exfoliated. For scales
create "oxide failure maps" such as that shown
formed on ferritic alloys that are subjected to
in Figure 3 [after reference 1]. Note that in this
tensile stresses on cool down the map indicates
map, negative values of strain are tensile in the
that the likely first step in exfoliation is through-
oxide, and positive values are compressive in
scale cracking. Actual loss of scale requires the
the oxide. Such maps allow visualization of the
further step of decohesion, which would occur
critical strain or oxide thicknesses (dcrit) at which
preferentially along planes of weakness in the
failure will occur according to the specific mode
scale, such as voids along interfaces.
expected for given alloy/scale combinations. For
scale failure in tension, there exist models de-
For scales formed on austenitic alloys that are
scribing several different modes, including by
subjected to compressive stresses on cool down
through-scale cracking, and by the formation of
the tendency will be for the scale to buckle. Sce-
multi-laminations [11], clearly indicating the in-
narios for expected modes of exfoliation range
fluence of scale morphology.
from local decohesion along a plane of weak-
Superimposed on the map in Figure 3 are results ness and lifting of the affected scale, followed
from measurements on scales formed on ferrit- by through-cracking of the scale at the ends of
ic and austenitic steel superheater tubes. These the buckle and detachment of the lifted scale, to
clearly indicate that scales on ferritic steels fail (in (absent an obvious plane of weakness) initiation
tension) at greater thicknesses than on austen- by the formation of inclined shear cracks in the
itic steels (in compression). The vertical lengths scale, followed by local decohesion between
of the superimposed data points in Figure 3 fur- these cracks, lifting, and loss of the affected
ther illustrate the influence of the presence of flake of scale [11]. In the exfoliation routes where
hematite in such scales. As expected from the decohesion is the final step, separation of oxide
results in Table 4, the presence of hematite in segments may not occur immediately following
the scale on a ferritic steel decreases the tensile the cracking event; additional stresses from con-
stress/strain and effectively delays exfoliation to tinued oxidation due to ingress of steam along
greater scale thicknesses, whereas for an austen- the cracks may be needed [12]. Overall, in any
itic steel hematite causes earlier exfoliation. of these scenarios it is obvious that the detailed

378 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

morphology of the scale plays an

important role in determining the
route taken by the exfoliation pro-
cess, and the form of the material
that detaches.

2 INTRODUCTION
TO THE OXIDE GROWTH AND
EXFOLIATION DATA BASE
This section of the paper compris-
es two main parts. The first de-
scribes the morphological devel-
opments with increasing time for
the scales formed on ferritic steels,
and shows the progression of OGE
as described by OGEI Numbers
1 to 5. The second part shows the
progression of OGE for austenitic
steels by OGEI Letters A to E. The
description of each index includes
the morphological features of each
stage. For some alloys in a given
class there can be major differ-
ences in the morphologies of the
scales formed in steam, for example, between ogy. At higher temperatures this initial morphol-
ferritic steels T22 and T23 or T91, and between ogy breaks down into the OGEI 2 and 3 morphol-
austenitic alloys with and without shot peening. ogies at relatively early operating times.
Such differences are highlighted because of the
importance of being able to recognize the source
OGEI 2: This index stage involves the
of exfoliant found to be associated with damage first breakdown of the initial duplex pair.
or failure. Recent comprehensive compilations Initial breakdown involves the formation of a
of representative scale morphologies are avail- second duplex pair of oxides at the original Fe-
able [5,13]. Cr spinel/tube interface (i.e., the alloy-L1 inter-
face), followed by the initiation of further duplex
2.1 Oxide Growth and Exfoliation pairs at the same interface. Figure 5 shows an
Indices for Ferritic Steels – Alloys example: the original duplex pair still is identifi-
Considered: T11, T12, T22, T23, T91
OGEI 1: In this stage the initial, charac-
teristic duplex scale structure is formed.
On all the ferritic steels of interest for service in
steam (typically with chromium contents below
about 11 wt%) an initial duplex oxide is formed
that consists of an inner layer of Fe-Cr spinel
(L1) and an outer layer of magnetite (L2), as
suggested in Figure 1. This morphology is seen
most often on alloys T11, T12, and T22 operating
at the lower temperatures of superheaters and
reheaters, and an example is shown in Figure 4.
There is no evidence of any laminations or for-
mation of alternate layers beneath this initial
duplex pair, and no hematite decoration on the
outside of the magnetite layer (L2), even after
very long times of operation. At "low tempera- Figure 4:
tures" this morphology remains for long times, Initial duplex oxide on alloy T22 after 158 000 h at a steam
and may never change into an OGEI 2 morphol- temperature of 900 °F/480 °C.

PPCHEM JOURNAL 21 ▪ 2019/06 379

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

able at this stage, even though a number of lam-

inations (repetitive duplex pairs) are present. In
addition, voids can be seen along the interfaces
between duplex pairs. In some cases, especially
where these voids are more continuous, there
will be indications of hematite along the outer
magnetite surface (Figure 6 shows an example).
These repeating pairs of Fe-Cr spinel and mag-
netite layers at the alloy-original L1 interface
are much thinner than the original duplex pair
(L1/L2), and are always oriented with the Fe-Cr
Figure 5:
Example of OGEI 2 on alloy T22 showing initial formation of laminations
spinel nearest the alloy surface. Also, the Fe-Cr
at the base of the original Fe-Cr spinel (L1) of the initial duplex pair spinel in these pairs has a very small grain size,
after 149 000 h (3 120 starts) at an estimated temperature of 980 °F/526 °C while in the corresponding magnetite layer the
(note: the two black lines were inscribed to measure oxide thickness). grains are significantly larger, apparently colum-
nar, and include intermixed porosity [6].
OGEI 3: In this stage there starts to be a
difference in the morphologies of scales
formed on alloys T11/T12/T22, T23,
and T91.
For alloys T11/T21/T22, OGEI 3 is characterized
by an increasing number of distinct laminations
consisting of repetitive duplex pairs which grad-
ually consume the original Fe-Cr spinel (L1).
Lines of voids are evident within the laminated
structure. Examples of these features in scales
on alloys T12 and T22 are shown in Figure 7.

Figure 6:
Further example of OGEI 2 on alloy T22 showing initial formation of lam-
inations at the base of the original Fe-Cr spinel (L1) of the initial duplex
pair, and formation of hematite on the outer surface of the original mag-
netite layer (L2).

Figure 7:
Example of OGEI 3 for (a) 155 µm-thick scale on alloy T12,
and (b) 475 µm-thick scale on alloy T22 from secondary and
tertiary reheater outlets, respectively, of the same boiler
(Ts = 1 056 °F/569 °C). b

380 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

Figure 8:
Example of OGEI 3 for alloy T23 after 31 000 h at 1 050 °F/565 °C, showing (right bottom): the varying thickness of the alternate
layers, and (right top): hematite formation at the original L2-steam interface.

For alloy T23, OGEI 3 is characterized by an in- gest alternating layers with different structures
creasing number of alternate layers of Fe-Cr spi- or morphologies like T22 [6] as illustrated in Fig-
nel and magnetite (duplex pairs) which always ure 7b. The bands have a higher chromium lev-
vary considerably in thickness, Figure 8, unlike el than the laminations, and consist of discrete
the laminations in the scales on T11 and T22, grains or subgrains of Fe-Cr spinel (FeCr2O4) sur-
which typically are of similar thickness. Further, rounded by magnetite (Fe3O4).
unlike for T22, on T23 these alternate layers ini-
This change in the morphology of the overall in-
tiate immediately below the original magnetite
ner layer compared to the lower-Cr ferritics is
layer (at the original L1-L2 interface). In addition,
directly related to the distribution of chromium
there is usually some obvious hematite along
in the ferritic-martesitic lattice of the 9-Cr alloys
the outer magnetite-steam interface (as indicted
[5]. The specific morpho-logy of the martensit-
in Figure 8).
ic laths in the alloy depends on its composition
As with alloy T22, the alternate layers of mag- and heat treatment, which in turn influence the
netite and Fe-Cr spinel on T23 are readily identi- structure and chromium content of the Cr-rich
fied from X-ray maps for iron and chromium, as bands in the scale.
shown in Figure 9.
During initial oxide growth, chromium enters
For T91, the detailed morphology of the overall the inner layer (L1) with a similar distribution to
inner layer differs from that on the lower-Cr al- that in the alloy, as the chromium carbide lat-
loys in that instead of alternating Cr-rich/Fe-rich tice of the alloy is oxidized in-situ. Further, these
laminations, the distribution of chromium in this bands appear to be present in the overall inner
layer takes the form of Cr-rich bands. In fact, the layer almost from the beginning of scale forma-
overall inner layer on T91 has no features to sug- tion. Hence, blocking of outward iron ion trans-

Figure 9:
Elemental X-ray maps for the alternate duplex layers formed in the original inner layer on alloy T23, indicating that these pairs consist of
alternate magnetite (Fe-rich) and Fe-Cr spinel (Cr-rich) layers, and showing the co-location of Mn, Si, and W with the Cr.

PPCHEM JOURNAL 21 ▪ 2019/06 381

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

Figure 10:
Example of OGEI 3 for T91 after 63 600 h at 1 051 °F/566 °C,
showing a distinct banded structure in the overall inner layer,
voids at the base of the original magnetite, and hematite in
the original magnetite (in this scanning electron microscope
back-scattered image, hematite appears darker than magne-
tite).

port will occur from the start and increase as the with chromium in the bands would significantly
number of bands increases with time. Typically, increase their blocking power. Additionally, in
hematite is considered to form earlier on T91 some situations the voidage in the scales on T91
compared with alloys T11/12/22 and T23. spreads along the base of the original magne-
The OGEI 3 stage for alloy T91 usually is associat- tite, and may increase sufficiently that delami-
ed with the formation of a semi-continuous layer nation of the original magnetite (and associated
of voids along the base of the original magnetite hematite) eventually occurs.
(original L1-L2 interface), and hematite decora- Overall, OGEI 3 for this range of ferritic steels
tion on the outer surface of the original magne- is a key stage that allows unique distinctions to
tite (original L2-steam interface). These features be made by analysis of oxide flakes collected in
result initially from increased restriction of the an area of plant failure or damage. Some of the
outward diffusion of iron ions by the blocking laminated or banded structure "consumes" the
action of the chromium-rich bands. Fe-Cr spinel/L1 of the original duplex pair until
Figure 10 shows an example of the typical ap- it disappears or is not visibly distinct. For alloys
pearance in scale formed on T91 in steam of T11, T12, and T22 this inner structure becomes
these Cr-rich bands in the overall inner layer, multi-laminated. For T23 there are very distinct
voids along the base of the original magnetite, wider layers of unequal thickness. On T91 this in-
and hematite along the outer surface of the orig- ner structure takes the form of non-continuous
inal magnetite, all of which are a direct conse- bands of more highly chromium-rich regions
quence of reduced transport of iron ions across within the overall inner layer. In all cases, voids
the overall inner layer. The corresponding ele- that start to appear at the base of the original
mental X-ray maps shown in Figure 11 clearly magnetite are associated with the laminations,
identify the high chromium levels in the bands layers, and bands. Formation of hematite at the
in the overall inner layer, and suggest that while outer surface of the original magnetite usually
Si also is present, it only co-locates with chromi- appears coincidentally with these developments.
um in the bands later in the oxidation process
(nearer the alloy surface). The presence of silicon

Figure 11:
Elemental X-ray maps for the scale shown in Figure 10, indicating the Cr-rich bands in the overall inner layer, as well as the co-location
with the Cr of Mn in the top portion, and of Si in the bottom portion of this layer.

382 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

OGEI 4: This stage involves initiation of

exfoliation (essentially, the origination of
cracks).
For alloys T11, T21, and T22, as the thickness of
the multi-laminated structure increases, cracks
start to appear perpendicular to the alloy/scale
interface. Most often such cracks do not propa-
gate through the full thickness of the multilayers,
and may be contained "within" the scale, as sug-
gested in Figure 12 for alloy T22; the original Fe-
Cr spinel is indicated in location No. 3, while the
multi-laminated structure consisting of multiple
duplex pairs is found in location No. 2. In such Figure 12:
cases, no scale exfoliation may have occurred. Example of OGEI 4 for alloy T22 showing initiation of exfoliation as
cracking perpendicular to the alloy-scale interface (arrows).
Cracks similar to those seen in T22 develop per-
pendicular to the alloy-oxide interface in scales
on T91, but may not be obviously damaging to
the integrity of the scale; some apparently "heal"
(Figure 13).

OGEI 5: This is the stage where actual

scale loss by exfoliation occurs.
For T11, T12, T22, and T91 the mode of exfoliation
involves variants of crack formation perpendicu-
lar to the alloy-oxide interface, followed by sep-
aration along planes of weakness in the scale.
Such planes usually involve voids associated
with the multi-laminations (T11, T12, T22), or
voids accumulated along the base of the original
magnetite (T91). Sometimes exfoliation takes
place perpendicular to the alloy-oxide interface.
Exfoliant from T23 is similar to T11/T12/T22, but
the alternate layers of spinel and magnetite in Figure 13:
the inner layer involved are much more distinct Example of OGEI 4 for a 475 µm-thick scale on T91 showing
(wider, and of unequal thickness). propagation of cracks perpendicular to the alloy surface in
the overall inner layer and the original magnetite, voidage
Exfoliation from T11/12 and T22 can produce along the original duplex interface, a banded structure in the
flakes having a range of sizes, and loss may oc- original L1, and hematite on the outer magnetite surface (af-
cur multiple times from the same location, espe- ter 15 500 h at 1 026 °F/552 °C).
cially on T22, Figure 14. Such
exfoliant typically comprises
thick flakes of oxide contain-
ing the magnetite (L2) from
the original duplex pair and
part of the original Fe-Cr spinel
layer (L1) containing (usually
several) multi-laminations, as
indicated in Figure 15. These
flakes of oxide can become
entrained in the steam flow
and cause solid particle ero-
sion (SPE) in steam turbine
components. The cracks lead-
ing to exfoliation most often
do not traverse the full scale
thickness. Hematite typically

PPCHEM JOURNAL 21 ▪ 2019/06 383

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

(a) (b)
Figure 15:
Examples of OGEI 5 for T22 showing thick multi-laminations and examples of locations from which exfoliation has occurred: (a) showing the
morphology of the scale after a region has exfoliated, and (b) a nominally 360 µm-thick scale from which two prior exfoliation events have
occurred (locations indicated by arrows).

is thought to relate to the varying thicknesses of

the laminations in the main inner layer formed
on T23.
For T91 the most common mode of exfoliation
involves delamination of relatively long lengths
of the outer magnetite layer with any associ-
ated hematite, as illustrated in Figure 16. The
voids between and associated with the layers/
bands in the original Fe-Cr spinel layer increase
Figure 16:
and coalesce to separate the Fe-Cr spinel from
An example of the typical exfoliation mode of T91, with separation of the original magnetite. These scale features are
the full thickness of the original magnetite layer. referred to as "delaminations," and start most
frequently at the base of the original magnetite
(Figure 17a). Occasionally, the delamination pro-
is present on these thick, laminated exfoliants
cesses on T91 result in an exfoliant consisting of
(located on the steam interface of the original
the original magnetite layer plus some fraction
L1/magnetite of the initial duplex), but often is
of the main inner layer. Sometimes there is ox-
difficult to see because the flakes are very thick.
ide bridging associated with this delamination
Exfoliation from T23 consists of much larger process to form closed "blisters," as shown in
flakes than T11/22 (similar to those from aus- Figure 17b; such features have resulted in tube
tenitics), as shown in Figure 14b. This behavior overheating and boiler tube failures (BTF).

(a) (b)
Figure 17:
Example of OGEI 4 for T91, showing (a) distinct chromium rich bands in L1 and separation along the original L1-L2 interface on T91 after
61 000 h at temperatures up to 1 250 °F/670 °C, and (b) leading to the development of closed blisters (after 63 600 h at 1 051 °F/566 °C).

384 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

A further variation of exfo-

liation behavior observed
with T91 is loss of the full
scale thickness. The rel-
atively few reported oc-
currences of this variant
have involved service in
heat recovery steam gen-
erators (HRSGs) that have
experienced large thermal
transients related to at-
temperation, condensate
management, and inade-
quate drains [14]. The cause
of this mode of scale loss
has been attributed to the
development of voidage
along the metal-oxide interface, usually early in
life.
Figure 18a illustrates the mass of exfoliant col-
lected from a P91 header in an HRSG, and the
size and shape of individual flakes. These flakes
clearly comprised the full thickness of the scale,
Figure 18b.
2.2 Oxide Growth and Exfoliation
Indices for Austenitic Steels – Alloys Figure 19:
Schematic diagram of a cross section of a typical initial scale formed
Considered: TP304H, 316H, 321H,
in steam on "standard" austenitic steels.
347H and HFG, 309, 310HCbN/
Alloy HR3C, and Internally-Treated
Tubes (e.g., Shot-Peened) of TP304H
A further outcome of this absence of significant
Unlike the ferritic steels considered, in which in-
morphological changes is that the scales formed
creasing alloy chromium content results in dis-
in steam on austenitics maintain an essentially
tinct changes in the morphology of the inner Fe-
single duplex structure throughout their lives,
Cr spinel layer of the original duplex scale, for
which allows the layers to be unambiguously
austenitic steels the major effect is an increase
described as 'L1' and 'L2'.
in the protectiveness of the Fe-Cr spinel without
major morphological changes. While the level of OGEI A: In this stage the characteristic
chromium in the inner layer on austenitic steels duplex scale structure develops.
initially is inherently higher than on these fer- The morphology of the initial scales formed on
ritic steels, with time the chromium level along "regular" austenitic steels can be very variable,
the interface with the austenitic alloy can in- depending on their actual levels of Cr, Ni, Mn,
crease further, so that the inner layer becomes Si, Nb, and S, for instance, as well as differences
increasingly protective. This is because the Fe-Cr in the surface condition of the tubes and the ex-
spinel has a variable crystal structure (typical- posure conditions experienced during initial op-
ly described as FexCr2–xO4) that can range from eration. Nevertheless, the scales rapidly assume
magnetite (Fe3O4) to chromite (FeCr2O4, with 46 the mostly duplex structure which typifies OGEI
wt% Cr). As a result, the overall inner layer on A, with a few areas where a very thin, protective
austenitic steels can have a chromium gradient (Cr-rich), single layer persists for some time, as
while maintaining the same crystal structure/ suggested in Figure 19. The inner layer of the du-
morphology. plex structure is an Fe-Cr spinel (L1, magnetite
containing Cr) and the outer layer is magnetite
(L2), while the single-layered, thin scale typical-
ly is FeCr2O4, the highest-Cr version of the Fe-Cr
spinel. At this point there are no voids at the L1-
L2 interface, and no hematite in the outer part of

PPCHEM JOURNAL 21 ▪ 2019/06 385

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

Although the spinel structure of FeCr2O4 is en-

ergetically favored in austenitic steels, such that
it forms more readily than the higher-Cr oxide
chromia (Cr2O3, 68 wt%) [15], with time the ox-
ide at the base of L1 on the higher-Cr and the
fine-grained alloys can exceed the maximum of
46 wt% Cr corresponding to FeCr2O4, indicating
that chromia also may be present. In such areas
the basal scale would be expected to be more
protective for longer times.
OGEI B: This stage is characterized
by the development of voids between
the original duplex layers and the
appearance of hematite in L2.
As the total oxide thickness increases and the
ionic diffusion slows, voids begin to form at the
Figure 20: L1-L2 interface, and gradually become more vis-
Optical photomicrograph of metallographically-prepared ible, Figures 22a and b. In addition, the amount
cross section of scale formed on a "regular" austenitic of hematite in L2 increases until it occupies
(TP304H) showing an example of OGEI A. a large percentage of the cross section of that
layer, though this feature can vary widely. These
changes result from increasing blockage of
movement of iron ions and oxidant in L1 due to
L2; Figure 20 is an example of the OGEI A duplex
the formation of increasingly Cr-rich oxide along
scale structure on TP304H.
the base of L1. Where such Cr-rich oxide com-
For alloys capable of forming a more protective pletely occupies surface alloy grain boundaries,
initial scale, such as the higher-Cr austenitics growth of L1 may be markedly slowed (or effec-
TP309 and TP310HCbN/HR3C, as well as fine- tively stopped), with the result that the thickness
grained austenitics (TP347HFG, Super304H), of L1 becomes very uneven (as illustrated in Fig-
in stage OGEI A the bulk of the initial scale is ure 22). Note that there is never (or almost nev-
much thinner, and first appears mostly as ar- er) any initiation and growth of laminations in L1
eas of thin, single-layered, Cr-rich oxide with in the oxides formed on these austenitic steels.
isolated duplex islands (L1+L2); in some areas
the L1-alloy interface is decorated with Cr-rich OGEI C: In this stage there develops an
oxide which may encircle associated alloy grain increasing density of voids along the L1-
boundaries (Figure 21). With time the areas of L2 interface and an increasing fraction
thin, single-layered oxide typically become con- of L2 is converted to hematite.
verted to the double-layered structure, so that With increasing time and thickening of the total
the overall scale morphology eventually resem- oxide the number of voids formed along the L1-
bles that on 'regular' austenitics, but remains L2 interface increases as blocking of transport
notably thinner. through L1 continues to increase, and eventual-
ly a semi-continuous line of voids is developed,
Figure 23. No breakdown of the structure of L1
is observed; instead this layer may become in-
creasingly irregular in appearance due to con-
tinuing inward growth in some alloy grains, ad-
jacent to regions where such growth has been
effectively stopped by the formation of Cr-rich
oxide along encircling alloy grain boundaries.
OGEI D: In this stage separation
occurs along the L1-L2 interface, and
Figure 21:
exfoliation of L2 begins.
Example of OGEI A on higher-Cr austenitic steels; the nodules have a The stresses between the oxides and austenitic
duplex structure (L1+L2), and a thin layer of Cr-rich scale connects the steels that develop upon cooling are compres-
nodules and partially encircles alloy grains near the surface. sive and, further, since hematite has a signifi-

386 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

(a) (b)
Figure 22:
Examples of initiation of voidage along the L1-L2 interface on austenitic steels, and growth of hematite at the L2-steam interface:
(a) TP304H, and (b) TP347H (4 100 h at Ts = 1 085 °F/585 °C).

(a) (b)
Figure 23:
Examples of (a) development of voidage along the L1-L2 interface on "normal" austenitic steel, TP347H, and (b) growth of hematite
at the L2-steam interface.

Figure 24:
Example of OGEI D for a "regular" austenitic, TP347H: separation has occurred
along the L1-L2 interface and the whole thickness of L2 has been lost.

Figure 25:
cantly lower CTE than the alloy and magnetite, its presence in Detail of exfoliation of the outer magnetite layer (L2)
L2 effectively increases the stress differential between L2 and from TP347H, showing the clean separation between
the alloy/L1. With a sufficient temperature drop these compres- the Fe-Cr spinel (L1) and magnetite layers.
sive stresses are able to lead to scale exfoliation. It is important
to realize that in normal boiler operation the maximum pos-
sible temperature drop, typically from service temperature to ed by the areas of semi-continuous
near ambient, is required to generate compressive stresses of voidage along the L1-L2 interface
the required magnitude. This implies that exfoliation events will becoming preferred paths along
not occur during shorter periods of shutdown (such as over- which stress relief can occur.
night, or at a weekend) when the tubes do not cool down fully.
Typically, exfoliation initiates once
Exfoliation from these austenitic steels results in detachment of the density of voids/cracks along
the full thickness of L2 (including any hematite), as illustrated the L1-L2 interface and hematite
in Figures 24 and 25, and in some cases L2 is lost over long in L2 reaches a critical percentage.
lengths of surface. However, loss of any part of L1 rarely occurs The authors suggest that this could
since L1 effectively is keyed to the alloy by the very irregular involve typically more than 70 % of
alloy-L1 interface. This results in the planes of weakness creat- the interface between L1 and L2.

PPCHEM JOURNAL 21 ▪ 2019/06 387

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

(a) (b)
Figure 26:
Oxide flakes exfoliated from alloy TP347H after reheater service: (a) showing representative lengths of exfoliant, and (b) a cross section of
a typical flake, showing it to consist of magnetite with a significant fraction of hematite.

Figure 26 illustrates metallographically-pre- TP347H (Figure 26) of 48 µm for flakes with rela-
pared cross sections of such exfoliated oxide tively thick layers of hematite, to 62 µm for flakes
flakes, which clearly consist of magnetite con- with very little hematite. The observation further
taining various fractions of hematite. In all cases supports the discussion that the presence of he-
there are indications of hematite along the prior matite increases the cooling stress differential,
L2-L1 interfaces. leading to exfoliation occurring from thinner
scales.

Figure 28:
Optical cross section of scale formed on shot-peened TP304H
Figure 27: after RH service for 23 000 h at 1 112–1 157 °F/600–625 °C.
Scale formed on fine-grained alloy TP347HFG after very high
temperature service, showing exfoliation of a thin (22 µm) out- The adherent OGEI A-stage scale shown in cross
er layer of scale consisting of 74 % hematite/26 % magnetite.
section in Figure 28 was formed on shot-peened
TP304H after service for 23 000 h at 600–625 °C
in a final reheater. No exfoliation had occurred,
On fine-grained alloys where a high chromium
even at a total scale thickness of 65 µm, where-
content rapidly develops at the base of L1 and
as the scale formed on "normal" TP304H in the
quickly shuts down outward transport of iron
same unit had completely exfoliated in the same
ions, L2 is more rapidly converted to hematite
time. The elemental scanning electron micro-
than on "regular" austenitics. Figure 27 shows
scope (SEM) X-ray images shown in Figure 29
a cross section of the scale formed on TP347H-
suggest that the outer layer (L2) was quite ad-
FG after service at very high temperatures
herent without obvious voids along the L1-L2
(Tm > 1 200 °F/650 °C) on which hematite was at
interface, so that the apparent interfacial crack
least 74 % of the thickness of L2. The maximum
evident in the optical image (Figure 28) was like-
meas-ured chromium content at the base of L1 was
ly a result of metallographic preparation. There
59 wt% (FeCr2O4 is 46 wt%; Cr2O3 is 68 wt%), and
appears to be an essentially continuous Cr-rich
exfoliation had occurred that involved the full
oxide along the base of L1, which accounts for
thickness of L2 (hematite and remaining magne-
the protective nature of this scale.
tite). The total scale thickness from which exfoli-
ation had occurred was relatively small (approx-
imately 35 µm), producing flakes of estimated
thicknesses of 18–23 µm. These compare with
an average thickness of flakes exfoliated from

388 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

Figure 29:
SEM X-ray images for the scale shown in Figure 28.

OGEI E: Following exfoliation of the

outer magnetite layer (L2) there is some
regrowth of oxide.
Figure 30 shows a cross section of the scale on a
"regular" austenitic steel immediately following
an exfoliation event in which the full thickness
of the outer magnetite layer (L2) was lost. With
subsequent exposure, growth of new oxide will
occur on the newly-bare surface of L1, the rate
of growth depending on the extent of comple-
tion of continuity of the layer of Cr-rich oxide at
the base of L1. The new oxide (new L2) could be Figure 30:
magnetite (as in OGEI A) or hematite, depending Scale remaining on the surface of "regular" austenitic steel TP304H
following an exfoliation event after 22 000 h at Ts = 1 005 °F/538 °C, in
of the prevailing rate of transport of iron ions. In
which all of the outer magnetite layer (L2) has exfoliated.
some cases, especially on fine-grained or shot-
peened alloys, hematite forms almost 100 % of
the new growth. Figure 31 indicates the growth
of hematite directly on the remaining original L1
on TP347HFG.

PPCHEM JOURNAL 21 ▪ 2019/06 389

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

(a) (b)
Figure 31:
Illustrations of the regrowth of scale directly on the remaining inner layers (L1) on austenitic steels: (a) initial regrowth of hematite on L1,
and (b) development of a continuous new outer layer consisting solely of hematite on fine-grained alloy TP347HFG.

3 APPLICATION OF OGEI TO GENERATING

PLANT PROBLEMS
Section 2 has provided progressive morpholo-
gies of the oxides that grow on ferritic and aus-
tenitic steels in steam circuits of power gener-
ating plants. This progression goes from first
operation with scales consisting of duplex pairs
(Fe-Cr spinel and magnetite) to exfoliation on
cool down due to differences in the coefficients
of thermal expansion of the oxides and alloys.
Both the growth and exfoliation processes can
cause a wide range of reliability problems in Figure 32:
Damage to first stage of an IP steam turbine due to SPE.
these plants, sometimes in the superheater or re-
heater (RH), where the oxides actually grow, but
also in other plant components in which exfoli- 3.1 Solid Particle Erosion of Steam
ated oxide flakes cause tube blockages. Because Turbine Components
of differences in the morphologies and modes
This mechanism was researched extensively
of exfoliation of the oxides grown on some of
in the 1970s and 1980s and it is generally un-
the alloys employed as SH and RH tubes, rec-
derstood that the erosive particles are thick,
ognition of key features of recovered/collected
multi-laminated oxide layers exfoliated from
exfoliant allows the source of the damaging par-
the low-Cr ferritic alloys (T11, T12, T22), such as
ticles to be identified. This approach can be used
illustrated in Figure 15. If they reach the turbine
in root-cause investigations of the problems af-
itself, damage can be severe, as illustrated in
ter the events but, probably more importantly,
Figure 32.
also to proactively identify future problems if
samples of SH/RH tubing are removed for met- 3.2 Short-Term Overheating Boiler Tube
allurgical analysis and identification of the OGEI. Failures
Some of the well-understood problem areas can This has been a worldwide BTF problem since
be identified as follows: the late 1960s (first cases involved austenitic
steel TP316 in Scottish boilers). It has occurred
predominantly in platen superheaters, but has
also more recently occurred in a number of plat-
en reheaters. The usual operating scenario is
that the unit has operated continuously at high
base load for long periods (sometimes up to a
year or more) without shutting down to ambient.
This allows the oxide morphology on austenitic
alloys to progress to OGEI C as illustrated in
Figure 23. Most often the unit then has a forced

390 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

(a) (b)
Figure 33:
Massive deformation of tubes at the bottom of a platen superheater due to a STO failure (a) (courtesy of John Begg, Eskom, South
Africa), from blocking by exfoliated oxide at a tube bend (b).

outage in equipment other than in a steam cir- 3.3 Longterm Overheating

cuit, which requires the unit to be cooled to am- A very common BTF in SH/RH circuits is
bient. On return to service, just as the boiler is longterm overheating/creep (LTOC) due to ex-
approaching full/high load, a BTF will occur in a cessive growth of steam-side oxide and accom-
SH/RH platen element. Sometimes the primary panying loss of tube wall thickness. The tubing
damage is immense, with grossly twisted tubes, most often has a black and cracked appearance
as indicated in Figure 33a. The maintenance staff on the outside surface of the tubing (Figure 34).
will repair the damage, and typically on return to
service another BTF in another SH/RH platen will
occur. This cycle has often been repeated 3 or 4
times, until it is recognized that these are short-
term overheating failures due to exfoliated oxide
blocking the circuit at a bend at the bottom of the
SH/RH platen; an example is shown in Figure 33b.
The plant operator will then use various tech-
niques to identify the blocked area, clear the
blockage, and return the boiler to service. It is
not unknown for further BTFs to occur.
The exfoliated oxide causing the blockage even-
tually will be identified as coming from the aus-
tenitic alloy tubing (as illustrated in Figures 24
and 25). There also are some observations that Figure 34:
suggest that the oxide could originate from alloy Typical appearance of a SH tube after failure due to LTO and creep.
T91 (such as illustrated in Figures 16–17). Here,
using the described OGEI approach, the exact
source of the exfoliated oxide can be determined For over 40 years boiler manufacturers have ad-
by relating the metallography to OGEI 4 or OGEI dressed the issue of excessive oxide growth by
D. It is also quite common to find in the debris limiting tube temperatures to an Oxidation Lim-
causing the blockage examples of exfoliant from it (OL), which was previously referred to as the
other SH/RH alloys located upstream of the fail- maximum temperature limit for tubing. These
ure location; again, using the OGEI, it is possible temperatures were derived from laboratory ox-
to identify the exact source. As expected from idation experiments and were well published
the discussion in Section 2.2, it is not surprising for the low-chromium ferritic steels [2,16,17]. A
to note that the authors do not know of any STO most important criterion in boiler design is not
BTF directly related to any exfoliation from shot- to include tubing in a superheater or reheater
peened austenitic alloys. that will exceed the OL for that tube material.

PPCHEM JOURNAL 21 ▪ 2019/06 391

 PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

Oxidation Limit experience of the authors instances of

Tube Material overheating (LTOC) of austenitics are
°F °C
rare, the relevance of these OLs for
T12 1 022 550 austenitic steels to boiler operation is
T11 1 049 565 not clear. They are listed here for com-
T22 1 076 580 pleteness.
T23 1 103 595 3.4 Erosion of Valves and De-
T91 1 120 605 posits in Steam Turbines
TP300-series austenitics (< 25 wt% Cr) 1 292 700 Deposits in steam turbines often in-
TP300-series austenitics (> 25 wt% Cr) 1 382 750 fluence generation and performance.
Here collection and metallurgical ex-
amination of oxide samples from the
areas of damage can lead to identifi-
cation of an OGEI stage, which in turn
Unfortunately, the same laboratory experiments can identify sections in the SH/RH where there
have not been undertaken for newer ferritic might be an unknown overheating problem
steels (such as T23 and T91), and thus there has causing rapid steam oxidation and/or early ex-
been a number of recent examples of the inclu- foliation.
sion of these alloys in new boilers with design
temperatures based only on maximum stress
limits, which typically are higher than the OL.
As a consequence, such tubes have experienced
LTOC very early in life. Based on the authors'
practical experience and from the OGE indices
in Section 2.1, an updated list of OLs including
these newer materials is suggested in Table 5. A
recent publication has indicated a similar limit
for T91 of 1 112 °F/600 °C [18].
It is well understood that as the oxide grows in
thickness the resistance to heat transfer of the
overall tube wall increases (rule of thumb is for
every mil (25 µm) of oxide the temperature of the (a)
tube wall will increase by about 4 °F/2 °C). Proac-
tive use of the OGEI can help to avoid this mech-
anism if tube samples are removed for metallur-
gical analysis, and the findings compared with
the morphologies for the OGEI progressions.
OLs also have been established for austenitic
steels used in steam boilers for almost 40 years
[2,17,18], and a distinction was made between
alloys containing less or more than 25 wt% Cr;
the prevailing values are shown in Table 5. These
values were derived from considerations of re-
duction of creep strength due to loss of tube wall
section from the effects of environmental deg- (b)
radation, and typically are significantly lower Figure 35:
than those associated with service in air [19,20]. Oxide deposits in a steam turbine: (a) typical appearance,
Metal loss from corrosion on both the steam and (b) deposit identified as stage OGEI 3 oxide from a T23
side and fireside was considered. In addition to reheater.
any influence of reduced wall section on creep
behavior, there is the further consideration that Figure 35 shows an example of deposit in a
continuing increase in the thickness of corrosion steam turbine which was identified by metallur-
scales will increase resistance to heat transfer gical examination as originating from a stage
through the tube wall, increasing tube tempera- OGEI 3 oxide formed on alloy T23 (as illustrated
ture, and reducing lifetime. However, since in the in Figure 8). Elemental mapping showed that the

392 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

inner layer contained W and V (as in Figure 9). ious situations where deposition of exfoliant
This further identified an emerging problem in causes either damage or loss of plant perfor-
the T23 section of the reheater, so that the local mance.
overheating could be addressed. In other cas-
es, the depositing oxide may be from austenitic 5 ACKNOWLEDGMENTS
and/or ferritic steels, but in each case the OGEI This has been a working lifetime activity of the
can be identified and thus the location and ma- authors, and we are grateful to the many peo-
terial of concern. ple, organizations, and plants worldwide that
Deposits in Tubing and Other Places; have supported aspects of this work and pro-
Sticking Valves vided superheater and reheater samples. Exten-
sive thought and discussion on steam-grown
The same process can be undertaken to identify
oxides have been provided by: John Stringer,
exactly the source of the exfoliant. Also, in many
Steve Paterson, Peter Tortorelli, Michael Schütze,
cases the authors have used the OGEI process
Adrian Sabau, Bruce Pint, and Tony Fry. Discus-
to show that sticking valves were not related to
sions with Jeff Henry on Oxidation Limits have
SH/RH exfoliation, but instead due to excessive
also been useful. One of us (IGW) is grateful for
levels of feedwater corrosion products (particu-
continued support from the US Department of
late and soluble).
Energy, Office of Fossil Energy, and the Electric
3.5 Possible Increase of Total Iron Moni- Power Research Institute while at the Oak Ridge
tored in the Condensate National Laboratory (ORNL). Morphological
characterization was conducted in association
This is currently a very important item, such that
with Wendy Weiss (Structural Integrity Associ-
the International Association for the Properties
ates), and Hu Longmire, Larry Walker, and Dor-
of Water and Steam (IAPWS) is developing new
othy Coffey of ORNL. Diane Dooley helped with
guidance for monitoring total iron on fast- and
the word processing and initial development of
frequent-start units. The same process as de-
some of the tables.
scribed in the other examples can be used to
identify the source of the total iron in the con-
densate. 6 REFERENCES
[1] Armitt, J., Holmes, D. R., Manning, M. I.,
4 SUMMARY, CONCLUSIONS, AND TO- Meadowcroft, D. B., Metcalfe, E., The Spalling
of Steam-Grown Oxide from Superheater and
WARDS THE FUTURE Reheater Tube Steels, 1978. Electric Power Re-
The paper describes two OGE indices: a) OGEI 1 search Institute, Palo Alto, CA, USA, EPRI Report
to 5 for ferritic steels, and b) OGEI A to E for aus- No. FP-686.
tenitic alloys. In each stage the progression of
[2] Dooley, R. B., McNaughton, W. P., Boiler Tube
the oxide morphologies can be identified from
Failures: Theory and Practice, Vol. 1: Boiler Tube
the initial magnetite/Fe-Cr spinel duplex pair to
Fundamentals, 1996. Electric Power Research
those associated with exfoliation. In the case of
Institute, Palo Alto, CA, USA, EPRI Technical Re-
ferritic steels exfoliation usually involves multi-
port No. 105261.
ple layers of oxide, whereas for the austenitics
exfoliation involves only the outer magnetite [3] Dooley, R. B., "Opening Remarks", presented
layer. Following the formation of the initial du- at the EPRI-NPL Workshop on Scale Growth and
plex growth on ferritic steels, the morphologi- Exfoliation in Steam Plant, 2003 (National Phys-
cal development is clear and distinctly different ical Laboratory, Teddington, England). Electric
on alloys T11/T12/T22 compared to T23 and T91, Power Research Institute, Palo Alto, CA, USA.
and the associated OGEIs can be used to define
[4] Wright, I. G., Dooley, R. B., "A Review of
the exact stage in the progress to exfoliation.
the Oxidation Behavior of Structural Alloys in
For austenitic alloys the morphological path to
Steam", International Materials Reviews 2010,
exfoliation is dependent on the formation of a
55(3), 129.
chromium-rich oxide at the alloy/Fe-Cr spinel
interface, and the OGEIs here allow distinction [5] Wright, I. G., Dooley, R. B., "Morphologies of
between austenitics that will exfoliate early or Oxide Growth and Exfoliation in Superheater
not at all. Examples have been included on the and Reheater Tubing of Steam Boilers", Materi-
application of OGEIs to plant problems such as als at High Temperatures 2011, 28(1), 40.
short- and longterm overheating tube failures,
solid particle erosion of turbine blades, and var-

PPCHEM JOURNAL 21 ▪ 2019/06 393

PPCHEM OXIDE GROWTH AND EXFOLIATION IN STEAM

[6] Wright, I. G., Dooley, R. B., "Steam-Side Scale [18] Grade 91 Use-Temperature Limits, 2017.
Morphologies Associated with Scale Exfoliation Electric Power Research Institute, Palo Alto, CA,
from Ferritic Steel T22", Materials at High Tem- USA, EPRI Report No. 3002011137.
peratures 2013, 30(3), 168.
[19] Morris, L. A., "Generally-Accepted Maxi-
[7] Technical Guidance Document: Volatile Treat- mum Service Temperature in Air for AISI Stain-
ments for the Steam-Water Circuits of Fossil and less Steels", Handbook of Stainless Steels (Eds.:
Combined-Cycle/HRSG Power Plants, 2015. In- D. Peckner, I. M. Bernstein), 1977. McGraw-Hill,
ternational Association for the Properties of Wa- New York, NY, USA, Chapter 17-1.
ter and Steam, IAPWS TGD3-10(2015). Available
[20] High-Temperature Characteristics of Stain-
from http://www.iapws.org.
less Steels, 1979. Produced by the American
[8] Eberle, F., Kitterman, J. H., "Scale Formation Iron and Steel Institute, Washington, D.C., USA
on Superheater Alloys Exposed to High-Tem- and distributed by the Nickel Development Insti-
perature Steam", Behaviour of Superheater Al- tute, A Designer's Handbook Series No. 9004.
loys in High-Temperature-High-Pressure Steam
(Ed.: G. E. Lien), 1968. American Society of Me-
chanical Engineers, New York, NY, USA, 67. AUTHORS
[9] Holcomb, G. R., "A Review of the Thermal Barry Dooley (B.Eng. with first class honours in
Expansion of Magnetite", Materials at High Tem- Metallurgy and Ph.D. in Oxidation of Metals,
peratures 2019, 36(3), 232. both from University of Liverpool, UK; D.Sc. in
[10] Dooley, R. B., Paterson, S. J., Wright, I. G., Power Plant Chemistry from the Moscow Pow-
"Oxide Growth and Exfoliation in Steam – Plant er Institute (University), Moscow, Russia) is a
Experience", presented at the EPRI-NPL Work- Senior Associate with Structural Integrity Asso-
shop on Scale Growth and Exfoliation in Steam ciates (SI) in the UK. Before joining SI he was
Plant, 2003 (National Physical Laboratory, Ted- a Technical Executive, Materials and Chemistry,
dington, England). Electric Power Research Insti- at the Electric Power Research Institute (EPRI) in
tute, Palo Alto, CA, USA. Palo Alto, California, and Charlotte, North Car-
olina, specializing in Cycle Chemistry, Boilers,
[11] Schütze, M., Protective Oxide Scales and
HRSGs, Steam Turbines, and Materials, as well
Their Breakdown, 1997. John Wiley & Sons Ltd,
as managing the Boiler and HRSG Tube Failure
Chichester, United Kingdom.
Reduction/Cycle Chemistry Improvement and
[12] Evans, H. E., Lobb, R. C., "Conditions for FAC Programs. Before joining EPRI, he provid-
the Initiation of Oxide Scale Cracking and Spall- ed Cycle Chemistry and Materials Services for
ation", Corrosion Science 1984, 24(3), 202. the fossil fleet at Ontario Hydro in Toronto, Can-
ada, and conducted oxidation research in the
[13] Atlas of Oxide Scales Formed on Super-
Materials Division of the Central Electricity Re-
heater and Reheater Tubes, 2015. Electric Power
search Laboratories (CERL) of the former Central
Research Institute, Palo Alto, CA, USA, EPRI Re-
Electricity Generating Board (CEGB) at Leather-
port No. 3002006598.
head, Surrey, England. Dr. Dooley is the author
[14] Dooley, R. B., Anderson, R. A., "Trends in or co-author of over 290 papers in the areas of
HRSG Reliability – A 10-Year Review", Power- metallurgy, power generation, boiler and HRSG
Plant Chemistry 2019, 21(3), 158. tube failures, FAC, cycle chemistry, steam tur-
bine failures, and life extension and assessment
[15] Hansson, A. N., Pantleon, K., Grumsen, F.
of fossil plants. He is the Executive Secretary
B., Somers, M. A. J., "Microstructure Evolu-
for the International Association for the Proper-
tion during Steam Oxidation of a Nb Stabilized
ties of Water and Steam (IAPWS), where he is
Austenitic Stainless Steel", Oxidation of Metals
an Honorary Fellow and chairs the development
2010, 73(1–2), 289.
of the IAPWS Technical Guidance Documents for
[16] Combustion, Fossil Power Systems: A Ref- the cycle chemistry of power plants. Dr. Dooley
erence Book on Fuel Burning and Steam Gen- organizes and chairs boiler/HRSG and chemistry
eration (Ed.: J. G. Singer), 1981. Combustion forums and users group meetings worldwide.
Engineering, Inc., Windsor, CT, USA, 3rd Edition.
[17] Kitto, J. B., Stultz, S. C. (Eds.), Steam: Its
Generation and Use, 2005. The Babcock & Wil-
cox Company, Barberton, OH, USA.

394 PPCHEM JOURNAL 21 ▪ 2019/06

 OXIDE GROWTH AND EXFOLIATION IN STEAM PPCHEM

Ian G. Wright (B.Sc. Tech., Chemical Engineer- ious universities. He also has co-organized and
ing, M.Sc. and Ph.D., Corrosion Science from the co-chaired several major conferences. At Bat-
University of Manchester) worked at Battelle-Co- telle-Columbus he was involved in alloy devel-
lumbus (1971–1994) and the Oak Ridge Nation- opment and qualification projects, while among
al Laboratory (1995–2010) before establishing his activities at ORNL was management of ma-
WrightHT, Inc. In 1993–94 he was a Visiting Sci- terials aspects of DOE's Advanced Turbines Sys-
entist at EPRI, Palo Alto. His career has focused tems program and the Ultra-Supercritical Steam
on the application of mechanistic understanding Materials program. In addition, he was Director
of materials behavior to address the causes of of the Laboratory's Fossil Energy Program, and
shortcomings, as well as the issues confronted co-founded the UK-US Collaboration on Energy
when applying materials near the limits of their Research. He has authored/co-authored over 200
capabilities. Promotion of these activities has peer-reviewed papers, as well as over 100 oth-
involved serving on materials panels for major ers, and has made over 150 presentations at ma-
United States Department of Energy (DOE) and jor meetings. He is a Fellow of NACE and ASM
EPRI-funded projects, as well as peer review International.
panels for corrosion research at NASA and var-

PPCHEM® journal
appreciates any information on planned conferences, workshops, and
meetings in the field of power plant chemistry. The information received
from event organizers will be edited and printed on a space available
basis at no cost to event organizers.
Visit us at journal.ppchem.com

The
Media Information
2020
and other important information available on the web
Subscription Rates 2020
To Order Individual Articles
To Order Back Issues

To Request a Free Copy of the PPCHEM® journal

journal.ppchem.com

PPCHEM JOURNAL 21 ▪ 2019/06 395