Materials Characterization 100 (2015) 120–134

Contents lists available at ScienceDirect

Materials Characterization
journal homepage: www.elsevier.com/locate/matchar

Characterization and correlation of mechanical, microstructural and

ultrasonic properties of power plant steel
Magdy M. El Rayes ⁎,1, Ehab A. El-Danaf, Abdulhakim A. Almajid
Mechanical Engineering Department, College of Engineering, King Saud University, P.O. Box 800, 11421 Riyadh, Saudi Arabia

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

Article history: In the present work 912% Cr tempered martensitic steel which is normally used in gas turbine-shrouds in elec-
Received 27 May 2014 tricity generation power plants was characterized through mechanical, microstructural and ultrasonic testing.
Received in revised form 17 November 2014 The steel tested was in the as-received (virgin), artificially aged at 700 °C for three different periods of time name-
Accepted 21 November 2014
ly 336, 840 and 1344 h, and retired from real operating conditions. The mechanical and microstructural test
Available online 17 December 2014
results were assessed and correlated with ultrasonic testing parameters. Artificial aging caused carbides to disso-
Keywords:
lute within the martensite–ferrite matrix and to coarsen and disintegrate at the prior austenite grain boundaries.
Electron backscatter diffraction Aging reduced the percentage of low angle grain boundaries and increased the average misorientation angle,
Martensite both of which indicate the occurrence of static recrystallization. Energy dispersive spectroscopy elemental anal-
Aging ysis identified that the dominant carbide types are Cr23C6 and VC, in which the former was mainly located at the
Dislocations prior austenite grain boundaries whereas, the latter was spread within the entire microstructure. Softening and
Grain boundaries the consequent loss in mechanical properties were detected as a function of prolonged aging times due to the ac-
Recrystallization companying microstructural phases and carbide dissolution which in turn were correlated with ultrasonic veloc-
ity and attenuation. The retired condition had the lowest strength, highest sound velocity and almost similar
attenuation as that of the virgin which possessed highest strength and lowest sound velocity.
© 2014 Elsevier Inc. All rights reserved.

1. Introduction temperature increased, the ultrasonic velocity decreased because of

Numerous components operating in electricity generation power
plants generally suffer from degradation due to their long-term opera-
tion in high temperature. Hence, reliability and safety are crucial issues
in the operation of these plants. This necessitates the assessment of the
risk associated with material degradation through aging and conse-
quently failure which implies to know the potential mechanisms of deg-
radation and their accumulation. Therefore, in order to extend plant life
safely at a minimum cost, it is important to monitor the material charac-
teristics of the components under service, which manifests the need for
non-destructive techniques (NDT) to be implemented. Characterizing
material degradation in the high-temperature components using ultra-
sonic testing aided by microstructural and mechanical tests was the
focus of numerous research works.
Kumar et al. [1] have characterized 9Cr–1Mo modified steel ultra-
sonically by subjecting it to a series of heat treatments consisting of
soaking the steel for constant duration at different temperatures
(1073–1623 K, 800–1350 °C) followed by oil quenching. Ultrasonic ve-
locity and attenuation measurements were used for characterizing the
microstructure obtained by various heat treatments. As the soaking
⁎ Corresponding author.
E-mail address: melrayes@ksu.edu.sa (M.M. El Rayes).
1
P.O. Box 800, 11421, Riyadh, Saudi Arabia.
the increase in the volume fraction of martensite in the structure.
There were sharp changes in the ultrasonic velocities corresponding to
the lower and upper critical temperatures. This was attributed to the
increase in the amount of martensite with the increase in soaking tem-
perature, which is reflected also in the variation in hardness. The atten-
uation was found to decrease due to the increase in the amount of
martensite which has lower attenuation. Furthermore, the ultrasonic at-
tenuation was found to be at a minimum in the sample quenched from
just above the Ac3 (937 °C) temperature where the prior-austenite
grain size was also at a minimum. After 1100 °C, attenuation increased
sharply due to the rapid increase in grain size which is associated with
the carbide dissolution. Heat treated steel 38Cr Mo Al was characterized
using ultrasonic testing [2]. The treatment included three types, all of
which started with austenitizing at 920 °C then either oil quenching,
or oil quenching followed by tempering or air cooling. A good correla-
tion was found between hardness and sound velocity. The quenched
microstructure gave the lowest sound velocity which gradually
increased with increasing the tempering temperature between 200–
600 °C. The attenuation coefficient of quenched followed by tempering
samples was lowest whereas normalized ones were highest. Earlier
work [3] has correlated the mechanical properties with ultrasonic ve-
locity and attenuation of lCr–1Mo–0.25 V steel. In addition to reference
material, it was heat-treated by the artificially accelerated aging method
at a constant temperature of 630 °C and four different durations. It was

http://dx.doi.org/10.1016/j.matchar.2014.11.034
1044-5803/© 2014 Elsevier Inc. All rights reserved.
 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134 121

concluded that the ultimate and yield strengths as well as the Vickers
hardness decreased whereas the elongation of the material increased
as the aging time increased. Attenuation coefficient based on scattering
increased as the degradation occurs due to the increase in aging which is
accompanied by an increase in grain size and precipitates at grain
boundaries. Another work [4] has studied the degradation mechanism
of SUS 316-L where it was found that the change in microstructure
between the virgin and the degraded material can be detected by an
ultrasonic velocity which was correlated with the tensile strength.
Vasudeven et al. [5] have characterized ultrasonically an isothermally
annealed 20% cold worked-Ti-modified austenitic stainless steel (D9)
using ultrasonic velocity measurement. This work showed that the var-
iation in ultrasonic velocity is sensitive to the variation of aging time and
also hardness. The aim of the present work is to correlate the micro-
structural and mechanical properties with ultrasonic measurements
conducted on gas turbine-third stage shroud supplied from an interna-
tional manufacturer. For the purpose of comparison, these shrouds were
divided into five conditions, starting with the as-received (virgin),
artificially-aged for three different aging times and finally a shroud
that retired from real service conditions.
2. Materials and methods
The as-received (virgin) shrouds having the chemical composition
in wt.%: C: 0.328; Cr: 11.55; Ni: 0.196; Mn: 0.256; Si: 0.505; V: 0.0831;
and Ti: 0.012, belonging to the 9–12% Cr-martensitic–ferritic steel,
normally used in various power plant applications, are used in this
study. These shrouds were subjected to accelerated aging at 700 °C in
an electric resistance furnace for different durations namely 336, 840
and 1344 h. In addition to these materials and for the purpose of

Fig. 1. Martensitic ferritic microstructure showing the prior austenite grain boundaries for different conditions: a) 0 h (virgin); b) 336 h; c) 840 h; d) 1344 h and e) 21,000 h (retired) Note:
Solid arrow points: prior austenite grain boundaries; dotted arrow points: block boundaries.
122 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134

comparison, retired shrouds which were taken from a power plant real
operation for 21,000 h at about 550 °C were added to the materials in-
vestigated in this work. The virgin, heat treated and retired shrouds
were wire cut in the form of (W × L × T) 50 × 50 × 30 mm blocks for
ultrasonic testing as well as 11 mm diameter rods for extracting micro-
structural, hardness and tensile test samples. The microstructure was
examined by Field Emission Scanning Electron Microscopy (FESEM)
7600 JEOL. For SEM test the samples were prepared according to stan-
dard metallographic sample preparation which includes grinding
using SiC sand paper then polished using diamond paste of 1.0 and
0.05 μm and finally etched with Vilella reagent to reveal the sample's
microstructure. For electron back scattered diffraction (EBSD) the sam-
ples were polished with colloidal silica as a final step prior to scanning.
XRD was used for the determination of the precipitated phases present.
A Brucker model D8 Discover X-ray diffractometer having CuKα radia-
tion (λ = 1.5406 Å) was used to analyze the specimens. The diffraction
patterns of various samples investigated were recorded at 40 kV and
40 mA in the range of 5–125° (2θ) at a step size of 2.0°. The diffraction
pattern was collected automatically by the data acquisition system
stored in the computer. The background data was subtracted from the
diffraction pattern to obtain the final spectrum. The peaks obtained
were compared with those available in PDF 00-001-1261 database.
The ultrasonic measurement samples were machined using end
milling then surface ground on both sides in order to ensure complete
parallelism between faces. In order to eliminate roughness, visible ir-
regularities and any oxides which might affect the ultrasonic measure-
ments the samples were further processed similar to the metallographic
preparation steps. For ultrasonic measurements, the pulse-echo tech-
nique and direct contact method were applied to obtain ultrasonic ve-
locities and attenuation coefficients using ultrasonic pulse–receiver
equipment, Karl Deutsch (Model-Echograph-1085). Ultrasonic mea-
surements for all samples were obtained by using commercial NDT

Fig. 2. Carbide and precipitate morphology with different aging times at different magnifications. Left column: locations taken at ×2000 with a white circle magnified at ×6000 shown in
the right column. a) 0 h (virgin); b) 336 h; c) 840 h; d) 1344 h and e) 21,000 h (retired).
 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134 123

Fig. 2 (continued).

ultrasonic-longitudinal wave transducers of 4 MHz. A constant force
was applied to the probe against the specimen surface so as to have a
constant thickness of couplant layer at the interface between the spec-
imen surface and the probe. Ultrasonic velocity was determined by di-
viding twice the specimen thickness by time of flight (TOF) obtained
between zero crossing of the first and second back-wall echoes using
Eq. (1) as referred in [3]. In order to check result repeatability, ten ultra-
sonic readings of each specimen were averaged to represent the data
obtained, which gave an error of around ±0.5%.
error in all readings of about ±2%. In order to evaluate the mechanical
behavior of all shrouds, tensile specimens were extracted from the cen-
ter of the rod along its axis and were cut according to ASTM E-08. Tensile
tests were conducted at room temperature and at a cross-head speed
velocity of 2 mm/min using Instron machine model 3385 H. The aim
of this work is an attempt to correlate the ultrasonic parameters; name-
ly sound velocity and attenuation with aging time in terms of the varia-
tions in microstructural phases as well as mechanical properties of
martensitic–ferritic steel used in gas turbine third-stage shrouds.

Velocityðm=sÞ ¼ 2  thickness=timeðsÞ: ð1Þ

This was further verified by a function in the ultrasonic equipment

which automatically calculates the ultrasonic velocity when locating
two successive echoes [1st and 2nd] at the extreme sides of the screen
with an error of maximum ± 1.0%. The ultrasonic attenuation values
were calculated [6–8] based on the reduction of the amplitude of an ul-
trasound pulse, measured in decibels per millimeter (dB/mm) and
given as:

20 A
α¼ log 0 ð2Þ
2x A1

where α is the attenuation coefficient [dB/mm], x is the thickness of the

sample measured in the test [mm], A0 is the amplitude of the first echo
in dB and A1 is the amplitude of the second echo. The constant 2 (two) is
because the pulse-echo technique is used. The measurements were
repeated ten times giving a maximum error of about ±4.0%, and were
obtained considering longitudinal waves and frequency of 4 MHz.
Vickers macro-hardness tests were performed using 10 kgf and were Fig. 3. Carbide morphology in the retired condition. Note: carbide clusters are enclosed
performed 5 times for each heat treated specimen with an average within ellipse.
124 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134

3. Results and discussion
3.1. Microstructure
The microstructure was characterized in different conditions name-
ly; virgin (as-received), artificially aged for different periods of time
(336, 840 and 1344 h at 700 °C) as well as the retired condition which
have served in real operation conditions for 21,000 h at 540–550 °C.
Generally, the microstructure of all conditions was found to be typically
tempered martensite. Fig. 1 presents the microstructure constituents,
for all conditions, showing large prior austenite grains (annotated by
white solid arrows) containing martensite laths (dark) having different
orientations at which block boundaries exist (white dotted arrows). In
addition, it was noted from the same figure that the prior austenite
grain size ranged between 50–60 μm as measured by the linear intercept
method, for all samples. This was attributed to the presence of thermally
stable and un-dissolved carbides along the austenite grain boundaries
which inhibited the growth of austenite grains as reported by Thomas
Paul et al. [9] who conducted his work on the same steel type like that
used in this work.
Fig. 2(a–e) shows a closer look at the carbide distribution and mor-
phology corresponding to different conditions. In general, these carbides
(white) precipitated within the martensite laths, at lath boundaries and
along prior austenite grain boundaries (PAGBs) forming a continuous net-
work, which is identical to that found in Bursak et al. [10] and Golanski
and Slania [11] who applied in their work the same Cr Mo steel. Based
on ten images taken from each condition [at ×6000] and with the aid of
image analysis software, statistical calculations on area fractions occupied
by carbides resulted in around 19.8, 8.3, 6.5, 5.4 and 15.6% corresponding
to 0, 336, 840, 1344 and 21,000 h, respectively.

Fig. 4. Carbide morphologies and their dimensions corresponding to various material conditions; a) 0 hrs. (virgin); b) 336 hrs.; c) 840 hrs.; d) 1344 hrs.; and e) 21,000 hrs. (retired). Left
column: the microstructure at which carbide measurements were taken; right column: duplicate microstructure with carbide dimensions.
 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134
Fig. 4 (continued).
In specific, Fig. 2(a and e) represents the virgin and retired condi-
tions respectively, which shows that carbides/ precipitates (white)
were heavily dispersed and eventually interconnected/coagulated
forming clusters within the entire microstructure. Fig. 2(b, c, and
d) which corresponds to the artificially-aged samples for 336, 840 and
1344 h respectively, shows that accelerated aging at 700 °C caused car-
bides to coalesce, dissolute and consequently diminish especially within
the martensite laths as well as to spherodize mainly within ferrite
islands as will be shown later. Fig. 3 shows a typical example of carbide
clusters found in the retired sample.
Higher magnification was applied in order to better visualize carbide
morphologies and measure their size as well as to assist in differentiat-
ing between two different phases namely; martensite laths and ferrite
islands as shown in Fig. 4(a–e). Golanski and Slania [11] and Krauss
[12] found the same phases of martensite and ferrite in the same type
of steel used in the present work.
In the virgin condition, Fig. 4(a), most carbides were spread within the
whole microstructure and had either rod or spherical/globular shape,
having continuous beads or separate rods between 421 ± 46 nm long
and average diameters of around 290 ± 23 nm, respectively. This result
was in line with that obtained in earlier work [11]. With the retired con-
dition; shown in Fig. 4(e), slightly less carbides were noted compared to
the virgin one, however, significant amounts of carbides and precipitates
still exist forming a continuous network. These carbides also had rod and
spherical shapes in the average of 983 ± 65 nm long and diameters of
388 ± 39 nm, respectively. The close amounts of carbides in the retired
[15.6%] compared to the virgin [19.8%] sample, is due to the thermal sta-
bility of these carbides during operation at 550 °C which can maintain this
stability up to 650 °C as reported by [13,14].
On the other hand, Fig. 4(b, c, and d) shows that carbides along
the PAGB have a rod-like shape in the average of 1.447 μm long
125
and started to coarsen and disintegrate due to aging for 336 h
as in Fig. 4(b) (sometimes called bead carbide-stringed [15]).
In addition, from the same figure, spherical carbides which were
mainly found in ferrite islands had an average diameter of 337 nm.
Prolonged aging caused more coarsening/spherodization of carbides
at the PAGBs leading to further disintegration and also more sphero-
dization of spherical carbides, which have survived during treat-
ment, leading to average diameters of 282 and 257 nm for 840 and
1334 h, respectively.
Apart from the influence of aging treatment on carbides, ferrite
islands were also affected by aging in which it evolved and coars-
ened on the expense of martensite laths. This result was also report-
ed in a number of publications [12,15–18]. The reason is that the
boundaries of the martensite crystals are associated with interfacial
energy which provides a driving force for the coarsening of the fer-
rite crystals that evolve from the as-quenched martensitic crystals
[12]. Fig. 5(a) shows an example of the method followed for mea-
suring the area of ferrite islands in the 1344 h sample, where the
measurements were conducted ten times at various locations at
the same magnification (× 4000) in each sample using a software
available with SEM. Fig. 5(b) shows the variation of ferrite area frac-
tion with increasing aging time. It is worth noting that the percent-
age increase in ferrite content was around 9% between 0 and 1344 h
which corresponds to the artificial aging at 700 °C, whereas it was
about only 5% between 1344 and 21,000 h at about 550 °C in real op-
erating conditions.
3.2. EDS carbide elemental analysis
Energy dispersive spectroscopy (EDS) was conducted fifteen
times on each sample. This was done in order to further understand
126 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134
the carbide types which dominate/influence the material properties.
Carbide shape, size and location were the criteria behind which the
selection of locations for EDS elemental analysis was done.
As mentioned above in the microstructure section, the carbides
were generally found at PAGBs, within martensite laths and ferrite
islands and at their boundaries. Carbides located at the PAGBs
were relatively larger than all other carbide types found within the
microstructure [17]. They were the dominant carbide type and con-
sistently identified with all types of material conditions being M23C6
type as shown in Fig. 6, which shows typical examples of it with dif-
ferent treatment conditions. This carbide type is normally Cr-rich
carbide but may also be W, Mo, V, Fe and B as agreed upon in various
publications [11,17,19]. M23C6 was found since the early stages of
tempering because they nucleate easily at the PAGB (shown in
Fig. 6), martensite laths and block boundaries.
Aging (subsequent tempering) promoted the precipitation of
fine MX carbides/particles causing them to be finely dispersed in the
matrix as reported by [19]. The formation of MX precipitates occurs
when strong carbide formers are added to the alloy (e.g. V, Nb, Ta, Ti),
which is V in the present work. Fig. 7 represents some examples of
this type of carbide at different treatment conditions. It was found
[20] that MX particles usually form on dislocations within the matrix
or on sub-grain boundaries where they inhibit/pin the dislocation
movement and consequently provide high creep resistance as sug-
gested by Golanski and Slania [11]. They also reported that with aging,
no significant change was found in size or chemical composition of
MX, which is in line with that reported by [19].
Fig. 5. Methodology of ferrite islands area calculation in 1344 hrs. sample; a) Left: location of measurement, Right: duplicate showing the borders of ferrite islands used for measurement.
b) Variation of ferrite content with different material conditions.
3.3. X-ray diffraction (XRD)
XRD patterns of all samples, shown in Fig. 8, have in common five
diffraction peaks (peak # 1–5) with different intensities indicating
that the corresponding carbides/phases or intermetallic compounds
were consistently present irrespective to the material condition.
The possible sources of these peaks are identified on each. The
peaks corresponded to different M23C6 carbides and different inter-
metallic compounds were similar to those reported in [19]. This re-
sult is confirmed by that obtained in EDS section presented earlier
and agrees well with that reported in several publications such as
[16,21,22]. It was noted that artificial aging and real operation rep-
resented by 8 weeks and retired conditions respectively resulted
in the detection of peaks that corresponded to elements such as Si,
Fe and V which indicate the possibility of carbide dissolution of
what used to be strong carbide formers.
3.4. Electron backscattered diffraction (EBSD)
Fig. 9(a–e) displays in the left column the orientation imaging grain
boundary maps representing the true grain boundaries (boundaries
with misorientation angle greater than 15°), whereas, the right column
displays their corresponding histogram for the misorientation angle dis-
tribution for 0 (virgin), 336, 840, 1344 and 21,000 h (retired) conditions
respectively, with the percentage of low angle grain boundaries (LAGBs)
and average misorientation angle noted on each histogram.
 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134 127

It is evident from Fig. 10(a) that the low angle grain boundaries
(LAGBs) representing the extent of sub-grain boundaries were continu-
ously reduced in the order of virgin, prolonged aging times and ending
with retired conditions due to their recovery under the effect of long
elevated temperature exposure. This result is confirmed in earlier
publications [12,17] where similar findings are reported. The explana-
tion to recovery may be attributed to high dislocation density within
martensitic crystals in the virgin condition produced by lattice deforma-
tion/distortion. This dislocation substructure contributes high strain
energy to martensitic microstructures and provides a driving force
for recovery and recrystallization during further tempering (aging).
Hence, low-angle, low-energy boundaries are eliminated by recovery
mechanisms (annihilating dislocations) whereas, high-angle, high-
energy boundaries are rearranged by grain growth mechanisms [12].

Fig. 6. EDS analysis of Cr- rich carbides (Cr23C6) at PAGBs for different treatments. a) 0 hr (virgin); b) 336 hrs.; c) 840 hrs.; d) 1344 hrs.; e) 21,000 hrs. (retired); Left column: Carbide
morphology; Middle column: element histogram; Right column: Quantitative EDS results.
128 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134

Fig. 7. Examples of MX (VC) carbides found with different treatments: a) 336 hrs; b) 840 hrs. and c) 1344 hrs. Left column: Carbide morphology; Middle column: element histogram; Right
column: Quantitative EDS results.

In addition, the average misorientation angle has gradually increased angle type with low misorientation angles, whereas the prior austenite
(Fig. 10b), in the same order because of the high temperature exposure grains and block boundaries are of higher misorientation angle type.
leading to static recrystallization. It was reported in most literature such This can explain the existence of the abnormal misorientation angle dis-
as [19,23], that martensite laths and the sub-grains within it are of low tribution shown in Fig. 9 right column, which shows two separate/

Fig. 8. X-ray diffraction patterns of the studied steel at different conditions.

 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134 129

isolated groups of misorientation angle distribution. The first group is
located at the low misorientation angle which corresponds to martens-
ite laths and sub-grain boundary, whereas the second was located at the
high misorientation angle range corresponding to PAGBs and block
boundaries.
The average grain size was measured using the linear intercept
method by applying a 40 line-raster for increased sensitivity in the hor-
izontal and vertical directions, applying a critical misorientation angle of
10°. It can be noted from Fig. 11 that with all aging times; until 1344 h,
the grain size was almost stable having a slight tendency to increase
with increasing aging time, where the grain size was limited to less
than 1 μm indicating a condition of restricted grain growth, which is
due to the presence of thermally stable carbides such as Cr23C6 and VC
present at the prior austenite grain boundaries as well as within and
at the martensite laths. This was confirmed by the conclusion reached
by [24] who indicated that in spite of aging temperature and the large
driving force for recovery due to the high dislocation density of
martensite, the pinning effect of carbide particles on the grains and
even sub-grains is responsible and necessary to maintain fine grains.
In the retired condition, however, an abrupt increase in grain size to
about 3.5 μm was noted due to the increase in HAGBs (59%) as depicted
from Fig. 9(e).
3.5. Room temperature mechanical properties
The hardness and mechanical property results were compatible with
the microstructural ones where it was found that increasing the aging
time softens the material and also reduces the ultimate and yield
strengths as shown in Fig. 12.

Fig. 9. Orientation imaging grain boundary maps and their corresponding histogram for the misorientation angle distribution for a) 0 hrs. (Virgin); b) 336 hrs. c) 840 hrs. d) 1344 hrs. and
e) 21,000 hrs. (Retired).
130 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134

Fig. 9 (continued).

The highest hardness and consequently strength in the virgin condi-

tion are referred to the presence of martensite with high dislocation
density and also, the high amount of sub-grain boundaries expressed
as percentage of LAGBs. As mentioned above that martensite laths and
sub-grains within the laths, which are mainly responsible for micro-
structural strengthening, are of low angle type with low misorientation,
as agreed upon in all literature. In addition, sub-grains act as obstacles to
dislocation motion during deformation/tensile testing.
Elevated temperature exposure during aging caused a reduction in
the dislocation density, reduction in the fraction of sub-boundaries
and less dispersion and dissolution of carbides, all of which were
reflected on the reduction of hardness and strength. Fig. 13 shows the
influence of increasing the LAGBs on the hardness.
Here, it should be noted that in spite of the large amount of carbides
found in the retired condition which were somewhat similar to that in

Fig. 10. The influence of different aging times on a) Low angle grain boundaries and
b) Average misorientation angle. Fig. 11. The influence of different aging times on grain size.
 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134 131

Fig. 12. Influence of aging time on a) hardness and b) mechanical properties.

the virgin one, however, the retired condition was notably softer and structure, hence possessing low sound wave propagation velocity com-
possessed lower strength than the virgin one. The reduction of sub- pared to ferrite which has low resistance to sound thus allowing higher
grain boundaries in the matrix which represent obstacles to dislocation speeds. These results agree with those obtained by [25–27].
motion as well as the recovery of high dislocation density, and also the The evolution of sub-grains and sub-grain boundaries within martens-
relatively larger grain size and high percentage of HAGBs [59%] may be ite laths contributes in the reduction of sound velocity. Gür and Keleş [28]
the reasons for this softening. have reported that the structures consisting of sub-grains and their
boundaries have an effect on lattice straining and interrupt the matrix
3.6. Ultrasonic velocity and attenuation continuity thus lowers the elastic modulus and consequently lowers the
propagation rate of sound velocity. This is confirmed in Fig. 14(b) which
Fig. 14(a) shows the influence of varying aging time on the ultrason- shows the inverse relation between LAGBs % (representing sub-grains
ic velocity. Increasing the aging time is accompanied by a gradual and their boundaries) and sound velocity.
increase in sound velocity. This is due to the type of microstructure as
well as the density of carbides/precipitates corresponding to their
respective treatment condition. As mentioned earlier, the virgin micro-
structure is predominantly composed of martensite laths with high
dislocation density as well as occasional ferrite islands and also carbides
that are dispersed within the entire microstructure. Gür and Tuncer [25]
reported that the ultrasonic velocity in martensite is essentially affected
by changes in the modulus of elasticity of individual grains, the crystal
lattice distortion level and the orientation of grains. The increase in lat-
tice distortion and the subsequent increase in dislocation density reduce
sound velocity. This was confirmed earlier by Papadakis [26] who report-
ed that the elastic moduli decrease in the order of ferrite–pearlite–bai-
nite–martensite and based on the proportional relation between the
elastic modulus and sound velocity it can be stated that martensite has
lower sound velocity when compared to ferrite, which was found in
higher amounts in the retired condition. In addition, martensite presents
high resistance to ultrasound waves, because of its compact and fine Fig. 13. Influence of increasing the LAGBs on hardness of the microstructure.
132 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134

Fig. 14. The relationship between sound velocity and: a) aging time and b) LAGBs.

Fig. 15 shows that increasing aging time gradually reduces the atten- coefficient that is higher than its aged counterparts and more or less
uation coefficient till 1344 h. However, the retired sample that was sub- similar to the virgin condition. The presence of large amount of carbides
jected to real operating conditions of 21,000 h exhibited an attenuation precipitated within and around the grain and grain boundaries in the

Fig. 15. Influence of aging time on attenuation coefficient.

 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134
Fig. 16. The relation between hardness and a) ultrasonic velocity and b) attenuation
coefficient.
virgin condition compared to the artificially aged ones may be the rea-
son for promoting scattering and energy loss of the propagating sound
waves. The longer the aging time, more carbide spherodization and
dissolution took place and consequently fewer chances for scattering,
hence lower attenuation coefficient. On the other hand, and in spite
that the retired sample was operating for much longer times, its atten-
uation coefficient value was somewhat similar to that of virgin con-
dition. This could be again due to the large amount of carbides
dispersed within the entire microstructure either individually, or
clustering or in the form of continuous network as shown earlier
in Figs. 2(e), 3 and 4(e).
Here it should be noted that the martensite phase makes the struc-
ture more isotropic, being a single homogeneous phase [14], which
leads to a decrease in its scattering power thus possessing the lowest at-
tenuation compared to ferrite as reported in various publications such
as [1,28,29]. However, the presence of large amounts of carbides in var-
ious sizes in both of the virgin and retired conditions could be the reason
for the similar as well as the higher attenuation coefficient compared to
the artificially aged conditions. Quantitatively, the virgin condition con-
stituted about 76% martensite and 19.8% carbides, whereas, in the re-
tired condition martensite was about 67% and carbides were about
15.6%, all of which led to similar attenuation. In order to assess the con-
tribution of martensite and carbide contents in attenuation, the percent-
age change in their values in the virgin and retired condition was
calculated. From the virgin to retired conditions martensite content
was reduced by 12% (more attenuating power, due to the increase of
ferrite on the expense of martensite) and carbide content was also re-
duced by 12.3% (less scattering events leading to lower attenuation).
In specific, attenuation coefficient was only increased by 3.0%, from
the virgin to the retired condition. This can be interpreted though the
existence of two competing effects that seems to balance out. The
small increase in attenuation coefficient by 3% can point towards that
133
Fig. 17. Correlation between mechanical properties with a) sound velocity and
b) attenuation.
carbides are slightly more influential on sound wave attenuation than
microstructural phases.
To elaborate on the higher attenuation in the virgin and retired
samples, compared to the artificially aged ones, the scattering events
existing in these two samples ought to be discussed. Carbides exist
within the entire matrix material with a relatively higher area fraction
which can be considered as a composite of three different components
having their own features namely; matrix within which martensite
and ferrite phases coexist, matrix/carbide interface in terms of different
matrix phases as well as different carbide types and the carbides them-
selves having different types and shapes and all of which contribute in
increasing the extent of anisotropy as well as the acoustic impedance
of the structure hence increasing attenuation. Several publications [18,
24–26] have confirmed that the higher the extent of anisotropy the
higher the scattering power and consequently higher attenuation.
3.7. Correlation between hardness and mechanical properties with
ultrasonic measurements
Fig. 16(a and b) shows the relation between hardness and sound
velocity and attenuation coefficient respectively of the shrouds under
investigation in its virgin, artificially-aged and retired conditions. It
was found that an inverse relationship exists between ultrasonic veloc-
ity and hardness where higher hardness is accompanied with a reduc-
tion in sound velocity.
The lattice distortion, high dislocation density, low elastic modulus
and large percentage of sub-grains (LAGBs) accompanying the martens-
itic [hardest] structure were the reasons for the low sound velocity.
With longer aging time the material becomes softer due to the recovery
134 M.M. El Rayes et al. / Materials Characterization 100 (2015) 120–134
of dislocations and increased ferrite content thus allowing an increase in
sound velocity as shown in Fig. 16(a).
The attenuation coefficient was also correlated with hardness which
in turn is related to the relative amounts of carbides as well as the
microstructure. Longer aging times [from 0 to 1344 h], as mentioned
earlier, reduces the amounts of carbides thus reducing chances of scat-
tering and loss of sound wave energy leading to lower attenuation. In
addition, the close attenuation values of virgin and retired conditions
were attributed to the anisotropy that these carbides give to the struc-
ture hence higher attenuation. Fig. 17(a and b) shows the relation be-
tween strength and ultrasonic velocity and attenuation coefficient
respectively, where it can be noted that these results were in line with
that obtained with hardness tests.
4. Conclusions
From the above work the following conclusions can be derived and
summarized:
• Artificial aging at 700 °C for different periods of time caused:
– Ferrite coarsening on the expense of martensite.
– Formation of M23C6 which was dominant among the entire micro-
structure, whereas, to a less extent the formation of MX precipitates.
– Reduction of low angle grain boundaries (sub-grain boundaries) as
well as an increase in the average misorientation angle due to static
recrystallization.
– Reduction in dislocation density, fraction of sub-grain boundaries
and less dispersion of carbides.
• Not only martensite laths are responsible for microstructural strength-
ening but also sub-grains within the laths play also a role.
• The ultrasonic velocity is affected by the type of microstructure within
which it passes through and also the amount of carbides and its disper-
sion.
• The evolution of sub-grains and their boundaries due to aging reduces
sound velocity.
• Increasing aging time leads to the dissolution of carbides thus reducing
the attenuation coefficient.
• There is an inverse relation between sound velocity and hardness
whereas softening is accompanied by lower attenuation.
Acknowledgment
This work was supported by the National Science, Technology and
Innovation Plan (NSTIP) strategic technologies program, within the pro-
ject number (08-ADV-209-02) in the Kingdom of Saudi Arabia. Sincere
thanks to Saudi Electrical Company- Riyadh for supplying the retired
power plant steel.
References
[1] Anish Kumar, K. Laha, T. Jayakumar, K. Bhanu Sankara Rao, Baldev Raj, Comprehensive
microstructural characterization in modified 9Cr–1Mo ferritic steel by ultrasonic mea-
surements, Metall. Mater. Trans. A 33 (Jun 2002) 1617–1626 (A).
[2] L. Lin, X.M. Li, J.L. Tan, Comparison among ultrasonic velocity, attenuation and
power spectra of different heat treatment transformation product in 38 Cr Mo Al
steel, Key Eng. Mater. 270-273 (2004) 346–352.
[3] Chang-Sung Seok, Jeong-Pyo Kim, Studies on the correlation between mechanical
properties and ultrasonic parameters of aging 1Cr–lMo–0.25 V steel, J. Mech. Sci.
Technol. 19 (2) (2005) 487–495.
[4] K. Kawashima, T. Isomura, S. Ohta, Characterization of thermal degradation of stain-
less steel with ultrasonic velocities and backscattering noise, Mater. Sci. Forum
210–213 (1996) 283–288.
[5] M. Vasudeven, P. Palanichamy, S. Venkadesen, A Novel technique for characterizing
annealing behaviour, Scr. Metall. Mater. 30 (11) (1994) 1479–1483.
[6] K. Vijayalakshmi, V. Muthupandi, R. Jayachitra, Influence of heat treatment on the
microstructure, ultrasonic attenuation and hardness of SAF 2205 duplex stainless
steel, Mater. Sci. Eng. A 529 (2011) 447–451.
[7] J. Stella, J. Cerezo, E. Rodrı´guez, Characterization of the sensitization degree in the
AISI 304 stainless steel using spectral analysis and conventional ultrasonic tech-
niques, NDT E Int. 42 (2009) 267–274.
[8] Vera Lúcia de Araújo Freitas, Paulo G. Normando, Victor Hugo C. de Albuquerque,
Edgard de Macedo Silva, Antonio Almeida Silva, João Manuel R.S. Tav, Nondestruc-
tive characterization and evaluation of embrittlement kinetics and elastic constants
of duplex stainless steel SAF 2205 for different aging times at 425 °C and 475 °C,
Internet.
[9] V. Thomas Paul, S. Saroja, M. Vijayalakshmi, Microstructural stability of modified
9Cr–1Mo steel during long term exposures at elevated temperatures, J. Nucl.
Mater. 378 (2008) 273–281.
[10] Ján Michel, Marián Buršák, Marek Vojtko, Microstrucutre and mechanical properties
degradation of CrMo creep resistant steel operating under creep conditions, Mat.
Eng. Materiálové inžinierstvo 18 (2011) 57–62.
[11] G. Golanski, J. Slania, Effect of different heat treatments on microstructure and me-
chanical properties of the martensitic GX12CrMoVNbN9-1 cast steel, Arch. Metall.
Mater. 58 (1) (2013). http://dx.doi.org/10.2478/v10172-012-0145-x.
[12] G. Krauss, Steels: Heat Treatment and Processing Principles, ASM International,
Materials Park, OH, 1990.
[13] Wei Sha, From Materials Science to Structural Engineering, Springer- Verlag,
London, 2013, ISBN 978-1-4471-4872-2. ((eBook), Library of Congress Control
Number: 2012955041).
[14] Wei Yan, Wei Wang, Yi-Yin Shan, Ke Yang, Microstructural stability of 9–12%Cr ferrite/
martensite heat-resistant steels, Front. Mater. Sci. 7 (1) (2013) 1–27. http://dx.doi.org/
10.1007/s11706-013-0189-5 (Higher Education Press and Springer-Verlag Berlin
Heidelberg 2013).
[15] Cheng-Hsun Hsu, Hwei-Yuan Teng, Sheng-Chien Chiu, Ultrasonic evaluation of
temper-embrittlement for martensitic stainless steel, Mater. Trans. 44 (11) (2003)
2363–2368.
[16] Hilmar Kjartansson Danielsen, Z-phase in 9–12% Cr steels(Ph.D. thesis) Department
of Manufacturing Engineering and Management, Technical University of Denmark,
DK-2800 Kgs. Lyngby, Denmark, Feb 2007.
[17] G. Golañski, Effect of the heat treatment on the structure and properties of
GX12CrMoVNbN9-1 cast steel, Arch. Mater. Sci. Eng. 46 (2) (Dec 2010) 88–97.
[18] Cheng-Hsun Hsu, Hwei-Yuan Teng, Yeong-Jern Chen, Relationship between ultra-
sonic characteristics and mechanical properties of tempered martensitic stainless
steel, JMEPEG, 13, ASM International, 2004. 593–599. http://dx.doi.org/10.1361/
15477020420828.
[19] David Rojas Jara, 9–12 % Cr heat resistant steels: alloy design, TEM characterization
of microstructure evolution and creep response at 650 °C(Ph. D. Dissertation Thesis)
Fakultaet fuer Maschinenbau der Ruhr Universitaet Bochum, Bochum, 2011.
[20] A. Di Gianfrancesco, S. Tiberi Vipraio, D. Venditti, Long Term Microstructural Evolu-
tion of 9–12% Cr Steel Grades for Steam Power Generation PlantsMetal 2012. (23. -
25. 5. 2012, Brno, Czech Republic, EU).
[21] Arūnas Baltusnikas, Rimantas Levinskas, XRD analysis of carbide phase in heat resis-
tant steels, Mater. Sci. (ISSN: 1392-1320) 12 (3) (2006) (MEDŽIAGOTYRA).
[22] B. Ravi Kumar, Sailaja Sharma, Parikshit Munda, R.K. Minz, Structure and micro-
structure evolution of a ternary Fe–Cr–Ni alloy akin to super martensitic stainless
steel, Mater. Des. 50 (2013) 392–398.
[23] Megumi Kimura, Koji Yamaguchia, Masao Hayakawa, Kazuo Kobayashia, Kenji
Kanazawa, Microstructures of creep-fatigued 9–12% Cr ferritic heat-resisting steels,
Int. J. Fatigue 28 (2006) 300–308.
[24] Kouichi Maruyama, Kota Sawada, Jun-ichi Koike, Strengthening mechanisms of
creep resistant tempered martensitic steel, ISIJ Int. 41 (6) (2001) 641–653.
[25] C. Hakan Gür, B. Orkun Tuncer, Characterization of microstructural phases of steels
by sound velocity measurement, Mater. Charact. 55 (2005) 160–166.
[26] Emmanuel P. Papadakis, Ultrasonic attenuation and velocity in three transformation
products in steel, J. Appl. Phys. 35 (5) (May 1964).
[27] C. Hakan Gür, İbrahim Çam, Comparison of magnetic Barkhausen noise and ultra-
sonic velocity measurements for microstructure evaluation of SAE 1040 and SAE
4140 steels, Mater. Charact. 58 (5) (May 2007) 447–454.
[28] C.H. Gür, Y. Keles, Ultrasonic characterization of hot rolled heat treated steels, Mater.
Charact. 45 (9) (Sep 2003).
[29] Anish Kumar, B.K. Choudhary, K. Laha, T. Jayakumar, K. Bhanu Sankara Rao, Raj
Baldev, Characterization of microstructure in 9% chromium ferritic steels using ul-
trasonic measurements, Trans. Indian Inst. Metals 56 (5) (Oct 2003) 483–497.