Materials Characterization 105 (2015) 118–128

Contents lists available at ScienceDirect

Materials Characterization

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

Characterization of complex carbide–silicide precipitates in a Ni–Cr–Mo–

Fe–Si alloy modified by welding
D. Bhattacharyya ⁎, J. Davis, M. Drew, R.P. Harrison, L. Edwards
Institute of Materials Engineering, Australian Nuclear Science and Technology Organization, Lucas Heights, NSW 2234, Australia

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

Article history: Nickel based alloys of the type Hastelloy-N™ are ideal candidate materials for molten salt reactors, as well as for
Received 9 December 2014 applications such as pressure vessels, due to their excellent resistance to creep, oxidation and corrosion. In this
Received in revised form 21 April 2015 work, the authors have attempted to understand the effects of welding on the morphology, chemistry and crystal
Accepted 1 May 2015
structure of the precipitates in the heat affected zone (HAZ) and the weld zone of a Ni–Cr–Mo–Fe–Si alloy similar
Available online 5 May 2015
to Hastelloy-N™ in composition, by using characterization techniques such as scanning and transmission elec-
Keywords:
tron microscopy. Two plates of a Ni–Cr–Mo–Fe–Si alloy GH-3535 were welded together using a TiG welding
Transmission electron microscopy process without filler material to achieve a joint with a curved molten zone with dendritic structure. It is evident
Energy dispersive spectroscopy that the primary precipitates have melted in the HAZ and re-solidified in a eutectic-like morphology, with a chemistry
Focused ion beam and crystal structure only slightly different from the pre-existing precipitates, while the surrounding matrix
Nickel based alloy grains remained unmelted, except for the zones immediately adjacent to the precipitates. In the molten zone,
Hastelloy-N the primary precipitates were fully melted and dissolved in the matrix, and there was enrichment of Mo and
Welding structures Si in the dendrite boundaries after solidification, and re-precipitation of the complex carbides/silicides at some
grain boundaries and triple points. The nature of the precipitates in the molten zone varied according to the
local chemical composition.
Crown Copyright © 2015 Published by Elsevier Inc. All rights reserved.

1. Introduction found that after welding, some non-equilibrium intermetallic phases

Among the high temperature alloys, nickel based alloys have been
preferred in the power industry, in both the fossil fuel power plants
and nuclear power reactors [1,2] (such as light-water reactors, LWR,
pressurised water reactors, PWR, boiling water reactors, BWR, and
molten-salt reactors, MSR). In general, because of their low void swell-
ing characteristics [3,4], higher creep resistance than stainless steels
[5] and high corrosion resistance[6], Ni based alloys are preferred candi-
date materials for Gen IV reactor in-core components where the tem-
peratures encountered would be greater than 600 °C. Ni based alloys
with Cr, Mo and Fe as major alloying elements, such as Hastelloy N™,
are particularly suitable for molten salt reactors because of their high
resistance to oxidation, corrosion and creep at elevated temperatures
and molten salt environments [2,7,8]. The alloys frequently contain car-
bides of the form M6C (η) and M23C6 [7,9,10]. The Si content in these al-
loys not only improves oxidation resistance, but also stabilizes the
complex silicide/carbide phases of the form Ni2(Mo,Cr)4(Si,C) and
Ni3(Mo,Cr)3(Si,C) [7,11]. These phases appear in the form of equiaxed
particles (with some faceting visible) in stringers parallel to the rolling
or extrusion direction, and can be as large as 2–5 μm. It has been
⁎ Corresponding author at: Bldg. 3, Institute of Materials Engineering, ANSTO, New
Illawarra Road, Lucas Heights, NSW 2234, Australia.
E-mail address: dhb@ansto.gov.au (D. Bhattacharyya).
such as σ, P and μ may be formed due to rapid cooling and segregation
[12–14]. In some cases, annealing at high temperatures, i.e., N 1300 °C,
has also been seen to cause the transformation of these complex sili-
cide–carbides to an intermetallic δ (Ni,Mo) phase [7]. An understanding
of the evolution of the crystal structure, chemistry and morphology of
these various types of precipitates during welding is of primary impor-
tance as they have been known to affect the ductility and fracture be-
haviour of this class of Ni based alloys. Although there have been
some detailed studies of the nature of the precipitates and their stability
after heat treatment [7,15–17], very few in-depth studies appear in the
literature characterizing different types of precipitates in different
weldment regions of this type of alloy. In the study by Ojo et al. [18],
the authors characterize the microstructure of dendrites and precipi-
tates in the weld fusion zone of IN 738LC alloy, while Unfried-Silgado
et al. also focuses their attention on the microstructure of matrix and
precipitates in the welded zone of Ni–Cr–Fe alloy with and without
Mo additions [19] In this paper, the authors have attempted to charac-
terize the precipitates as they appear in various locations in the welded
sample, such as the HAZ or weld metal, using advanced sampling tech-
niques such as focussed ion beam (FIB) milling to extract site-specific
samples for transmission electron microscopy (TEM). The information
thus obtained has been complemented with advanced characterization
techniques such as high-resolution energy dispersive spectroscopy
(EDS) and electron backscatter diffraction (EBSD) to gain a deeper

http://dx.doi.org/10.1016/j.matchar.2015.05.001
1044-5803/Crown Copyright © 2015 Published by Elsevier Inc. All rights reserved.
 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128
insight into the evolution of these precipitates as a function of the max-
imum local temperature reached during welding. This method of
extracting samples from different parts of the cross-section of a welded
specimen provides a unique opportunity for studying the effect of
heating to a range of temperatures in a single sample, and has been
found very useful by the authors here in getting a wide view of the var-
ious types of phase transformations occurring during the welding
process.
2. Experimental methods
The material used for the welding experiment was a piece cut from
an extruded and solution annealed bar of Ni based alloy with the
nominal composition close to that of Hastelloy® N, obtained from the
Shanghai Institute of Applied Physics (SINAP). The chemical composi-
tion was determined by the LECO combustion method and inductively
coupled plasma atomic emission spectroscopy (ICP-AES), and is given
(in wt.%) in Table 1 below.
Two pieces were cut from the rod of diameter approximately 17 mm
(Fig. 1(a)), and welded together in the manner shown in Fig. 1(b), using
a tungsten inert gas (TIG) weld, without any filler. The top piece and the
bottom piece were rotated 90° with respect to each other before the
welding, as shown in Fig. 1(c). The welding was performed at a current
of 56 A at 9.9 V. The dimensions of the welded pieces are shown in the
figure. The process of welding caused the alloy in the top part to melt
and flow downwards, and subsequently solidify in the curved shape,
as shown schematically in Fig. 1(d) and (e).
A cross-section of the weld was cut and mounted in a conductive Ba-
kelite mould, and polished to SEM quality finish (better than 0.05 μm).
The sample was characterized using a Zeiss® Ultra Plus™ scanning
electron microscope (SEM) to examine the microstructure, and
orientation maps were obtained using electron backscatter diffraction
method (EBSD). The chemical composition of the various phases was
obtained in the SEM using energy dispersive X-ray spectrometry.
EBSD and EDS analysis were done using the HKL™ and AZTEC™ soft-
ware and detectors manufactured by Oxford Instruments™. When
some pixels were un-indexed in the EBSD maps, these were “cleaned
up” based on the orientation of the surrounding pixels using the “Clean-
up” procedure in HKL® Tango™. The un-indexed pixels were replaced
only when they had a minimum of 7 indexed neighbours (out of a
possible 9).
Samples for cross-sectional transmission electron microscopy (TEM)
from specific areas in the weld, HAZ and base metal were obtained
using a Zeiss® Auriga™ focused ion beam (FIB) instrument, with
a Cross-beam™ configuration. These samples were first cut out
using relatively high accelerating voltage for the Ga ions (30 kV), and
high currents (N 600 pA). They were then thinned to electron transpar-
ency using lower voltages (2–5 kV), and lower currents (100–240 pA),
to try to minimize the damage caused by Ga ion implantation and colli-
sion cascades. The samples were then examined using a Jeol® 2200 FS™
and a Jeol® 2010 F™ 200 keV TEM. The structure of the precipitates and
the matrix was determined using selected area diffraction techniques,
and the chemical composition by energy dispersive X-ray spectrometry
(EDXS) using an Oxford Instruments® EDS detector.
Differential thermal analysis (DTA) [20] was performed in order to
determine the melting point of the alloy and of any of the minor phases,
using a Seiko® DTA/TGA instrument. The sample was heated in an Ar at-
mosphere at ~20 °C/min to ~ 1468 °C and held at this temperature for
1 min, and cooled at ~ 100 °C/min to about 1200 °C. The furnace was
Table 1
Chemical composition of the Ni–Mo–Cr–Fe alloy in wt.%.
Mo Cr Fe Mn Si Al C Ni
16.0 6.40 3.62 0.58 0.43 0.06 0.05
119
subsequently switched off and the sample allowed cooling to room tem-
perature within the furnace.
3. Results
3.1. Preliminary characterization of the weld microstructure
An SEM backscatter image of the cross-section of the weld is shown
in Fig. 2(a). It is clear from this image that the weld metal spread out in a
curved manner due to gravitational force and resolidified in long den-
dritic structures. The dashed line in the image gives the approximate fu-
sion line boundary between the weld metal and the heat affected zone
(HAZ). The EBSD generated grain boundary map in Fig. 2(b) shows
the random high angle grain boundaries in black and the Σ3 twin
boundaries in green. It is clear that in the melt zone, the material
did not form any twins in the dendritic grains formed after re-
solidification. In fact, there is a clear distinction between the dendritic
grains in the weld zone on the one hand and the grains in the HAZ
and the parent metal on the other, in that almost all twin boundaries ap-
pear in the base metal/HAZ grains, and almost none appear in the weld
zone. This fact, along with the shape of the grains, can be used as a very
good guideline to demarcate the weld fusion boundary.
The SEM image in Fig. 3(a) shows a magnified view of the region
near the boundary of the weld metal and base metal, with the weld
joint between the two bars going horizontally, and the stringers of pri-
mary precipitates on the upper left side of the weld line, which were
present in the as received material. The precipitates do not appear in
stringers in the lower bar below the weld joint, because the lower
piece was rotated 90° with respect to the upper piece prior to welding,
and therefore, the section shown here is cut transverse to the direction
of the stringers for the lower part of the sample. This situation is shown
schematically in Fig. 1(c). The precipitates constituting the stringers are
more or less equiaxed in general, with the size in the range of 1–5 μm,
and are slightly faceted, as shown in Fig. 3(b). As is evident from
Fig. 3(a), the pre-existing precipitates melted and dissolved in the
weld metal, and the stringers are no longer visible in the re-solidified
dendritic grains. However, newly formed fine precipitates are visible
in the weld metal, which are shown more clearly in the high magnifica-
tion image in Fig. 3(c), where the distribution of these precipitates along
the dendrite boundaries is apparent. As can be seen from a careful ob-
servation of this image, the boundaries between two dendritic columns
or lamellae in the same dendrite colony have segregation of solute
(which was confirmed to be mostly Mo and Si by EDS, not shown
here), whereas most of the precipitation occurred in the dendrite
boundary triple points. The image in Fig. 3(d) shows the structure of
one of these newly formed precipitates in detail, and it is clear that
these precipitates have a eutectic-like morphology. There are also traces
of diffusion of elements between the matrix and the eutectic like precip-
itates, visible in the upper part of the image.
In the heat affected zone (HAZ), i.e., within a band of about 150 μm of
the fusion line, the pre-existing precipitates – which were almost
equiaxed – are transformed to a new kind of precipitate, which also
has a eutectic-like morphology. The image in Fig. 4(a) shows a row of
precipitates forming a stringer, running from left to right in the image.
The left part of the image is comprised mainly of the base metal,
which may have experienced some increase in temperature, but not
large enough to cause any significant amount of phase transformation.
Thus we see almost unchanged precipitates in this area. As we move
to the right, the maximum temperature reached is higher. One of the
precipitates which was part of an original stringer is marked as 1, and
is shown in a higher magnification image in Fig. 4(b). This precipitate
has started to show some melting and re-solidification at the grain
boundary, and possibly even some internal redistribution of elements,
as is indicated by the network like pattern of brighter “strings”. On mov-
ing further to the right, the temperature increases even more, and the
Bal
pre-existing globular precipitates are replaced by a new kind of
120 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128

a b
1 2 Weld

2 1

c Stringers of precipitates d
TIG weld with no filler
material
1 1

2 2

e
Weld metal

Fig. 1. (a) Schematic of cut sections for welding, (b) plan view of weld, (c) schematic showing sections of the two pieces being joined, (d) formation of new surface after welding without
filler material, and (e) schematic showing extent of molten weld zone.

precipitate, which has a eutectic like morphology. This is visible in the
magnified image of the precipitate marked as 2, shown in Fig. 4(c). It
is interesting to note that the matrix grains surrounding precipitate 2
did not melt during the welding process, as is evident from the grain
structure, with twins in it. New precipitation in a thin film-like form at
the grain boundaries surrounding precipitate 2 is also visible in this
image. Such precipitation is mostly confined to the random high angle
grain boundaries, and the twin boundaries are seen to be devoid of
these fine film-like precipitates.
3.2. Estimation of temperature variation in the weld sample
In order to obtain an estimate of the maximum temperatures
achieved at different regions in the welded sample, two complementary
approaches were taken.
First, three different pieces of the same material as the welded sam-
ple were taken from the same bar, and heated to 1150 °C, 1200 °C and
1250 °C, respectively, and held for 1/2 hour each in an Ar atmosphere.
These were then fast cooled in Ar in a crucible to room temperature

a b

Weld metal

Base metal

Fig. 2. (a) SEM backscatter image showing base metal, weld metal and joint, and (b) EBSD map with grain boundaries in black and twin boundaries in green.
 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128 121

a
b
Stringers

Joint line

c d

Fig. 3. (a) Stringers of precipitates in the base metal replaced by fine grain boundary precipitates in the dendritic boundaries in the weld metal, (b) magnified image of a precipitate in the
base metal, far away from the weld zone or the HAZ, inside one of the stringers, (c) magnified image of the structure in the weld metal, showing segregation of heavier elements at the
dendrite boundaries and precipitation at the triple points, and (d) a high magnification image showing the structure of one of the dendrite boundary precipitates in detail.

(RT), and cross-sectional samples were polished to be examined under
the SEM. It was found that no clearly detectable microstructural changes
occurred in the precipitates. So it became clear that the temperatures
attained in the band that showed eutectic-like precipitate in the HAZ
were greater than 1250 °C. Following this, small pieces of the as-
received material were subjected to differential thermal analysis
(DTA) to determine the melting point of the alloy. This was done by
heating the sample to a maximum of ~1468 °C(±1 °C). This tempera-
ture was chosen since it was about the maximum that the instrument
could reach. The DTA curve in Supplementary Fig. S1 shows a very
strong dip above ~1435 °C indicating the melting of the alloy. Although
the alloy probably starts to melt at ~1435 °C (±5 °C), the melting pro-
cess is not complete even at the maximum temperature reached, i.e.,
1468 °C. The melting process is probably completed slightly above
1470 °C, estimated from the slope of the curve before the temperature
was held constant. Further careful examination of the DTA curve re-
vealed a smaller dip in the heat absorption curve between 1355 °C
and 1395 °C, with a minimum at 1385 °C, which is very likely indicative
of the melting of the pre-existing precipitates.
When the micrograph in Fig. 3(a) is examined carefully, the width
of the HAZ (i.e., where the pre-existing precipitates have melted,
but the prior γ grain structure is intact), varies from 150 to 220 μm.
This means that in this region, there is a temperature drop of
about (1435 − 1355=) 80 °C to (1470 − 1355=) 115 °C, thus giving
a temperature gradient of about 0.37 to 0.76 °C/μm. The temperature
gradient in the weld metal is, therefore, expected to be lower than this
range.
3.3. Characterization of the precipitates in different zones
The goal of this study is to understand the chemical and structural
nature of the precipitates in various regions of the material, ranging
from the base metal, through the HAZ to the weld metal. In order to
do this, the focused ion beam instrument was used to lift out TEM sam-
ples of the precipitates and the matrix from the various regions of
interest.
3.3.1. Pre-existing precipitates
The image in Fig. 5(a) shows a bright field (BF) TEM image of such a
sample take from the base metal about 40 μm away from the HAZ/base-
metal boundary. It shows the cross section of a pre-existing precipitate
embedded in the matrix. The selected area diffraction pattern (SADP) of
the precipitate shown in Fig. 5(b) was taken from the precipitate. It
clearly matches the pattern of an FCC/Fd3m structure taken on a
[− 112] zone axis. The lattice parameter was determined to be
~10.9 Å (with an error of ±2–2.5%). There seems to be no specific orien-
tation relationship between the precipitate and the matrix.
Energy dispersive X-ray spectroscopy (EDS) was performed on the
sample (in the TEM) to determine the chemical composition of the pre-
cipitate and the matrix. The C content of the precipitate was estimated
assuming the C content of the matrix to be 0. The measured C content
in the matrix was taken as “background level” and subtracted from
the C measurements to get the “true” value. Thus the difference of the
measured C content in the precipitate and that in the matrix was
taken to be the true C content of the precipitate. The concentrations of
122 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128

1 2

b c

Fig. 4. Progressive phase transformation with increase in temperature. (a) SEM image showing precipitates in stringers near the HAZ, with the base metal to the left, HAZ in the middle, and
weld metal towards the right of the image, within 150 μm of the fusion line, (b) magnified view of precipitate marked as 1 in (a), and (c) magnified view of precipitate marked as 2 in (a).

all the other elements were re-adjusted to achieve a normalized sum of
100% for all the elements. The results of the EDS analysis after the cor-
rection for the C content are shown in Table 2 below.
It should be noted that the C content in this estimate and all the
other measurements in this paper is not very accurate, since it is subject
to the usual errors in EDS measurements for concentrations of light ele-
ments. Therefore these quantities are only given as estimates for
guidance.
This precipitate composition is close to Ni2(Mo,Cr)4(Si,C), a variation
of the Ni2Mo4C phase, which has a lattice parameter of 11.25 Å when
synthesized separately [21].
3.3.2. Precipitate from heat affected zone (HAZ)
The SEM image in Fig. 4(c) shows a typical precipitate from the HAZ,
which has a eutectic-like structure, while the surrounding grain struc-
ture remains intact. It is very likely that the precipitate has formed by
melting and resolidification of the pre-existing precipitate in the matrix,
since the eutectic-like precipitate is formed in the same row as the pre-
cipitates in the stringer to its left, and the transformation temperature
(1355 °C–1395 °C) is higher than the solid–solid phase transformations
generally reported in the literature, such as the work by Klarstrom et al.
[9]. In order to understand the chemical composition and crystallo-
graphic structure of these precipitates, a FIB cross-sectional sample

a b

Precipitate

Matrix

Fig. 5. Precipitate in base metal, about 40 μm from the visual HAZ area edge. (a) BF image showing primary precipitate in matrix and (b) SADP of precipitate taken at [−112] zone axis,
showing FCC structure.
 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128
Table 2
Concentration of different elements (in atomic %) for both matrix and precipitate in the
base metal, after correcting for the surface carbon from contamination of the sample.
Mo Cr Fe Mn Si C Ni
Matrix 14.56 7.89 4.49 0.77 1.57 0.00 70.37
Precipitate 50.22 5.17 1.27 0.19 9.53 4.49 28.75
was taken from one such precipitate. A BF TEM image of the precipitate
from the HAZ is shown in Fig. 6(a), where a few of the ligaments of the
precipitate are on a zone axis. It is clear that at least some, if not all, lig-
aments have the same crystallographic orientation. An SADP from the
precipitate “arm” or “ligament” marked with an arrow is shown in
Fig. 6(b). A measurement of the d-spacing values of the diffraction
spots and the angles between them indicates an FCC/Fd3m type crystal
structure taken from a b01-1N zone axis with a lattice parameter of
~10.9 Å (with an error of ±2–2.5%). This is very similar to the lattice pa-
rameter of the pre-existing primary precipitate in the base metal, as
mentioned in the previous section.
The scanning transmission electron microscopy (STEM) BF image in
Fig. 7(a) shows a dotted line across two of the lamellae of the eutectic-
like precipitate in the HAZ, which are marked as 1 and 2 in Figs. 6(a) and
7(a). The concentration profiles obtained from a quantitative scan along
this line, in at.% for each element, are shown in Fig. 7(b) and (c), before
correcting for C content. These plots show, as expected, the precipitate
lamellae rich in Mo, Si and C, and lean in Ni, Cr and Fe. Careful examina-
tion of the Mo curve reveals a dip in the Mo concentration near the
precipitate–matrix interface, indicating that the cooling is not an equi-
librium process, and that the local composition is influenced by the dif-
fusion rate ahead of the interface. For a better estimate of the relative
chemical abundance of different elements in the precipitate, the compo-
sition of the precipitate lamellae in atomic % was calculated by the aver-
age composition in terms of all the elements, and correcting them for
the C concentration apparent in the matrix, as described above. From
such an estimate, the corrected composition of the precipitate lamellae
is found to be close to Ni2(Mo,Cr)4(Si,C), as evident from Table 3 below.
An EBSD map and a corresponding EDS map were obtained from
each of the two such eutectic-like precipitates in the HAZ, marked as 1
and 2 in Fig. 8(a). A magnified band-contrast image of precipitate 1
showing the map area in detail is shown in Fig. 8(b), and an EDS map
showing Mo abundance in the precipitate phase is shown in Fig. 8(c).
An EBSD map showing the orientation in inverse pole figure (IPF) col-
ours with respect to the sample Z axis is shown in Fig. 8(d). In some
cases, it becomes difficult to detect the precipitate phase when the liga-
ments are less than ~ 100 nm thick. The matrix and the precipitate
a b
Fig. 6. (a) BF TEM image of precipitate in the HAZ – molten and resolidified in Eutectic-like structure – with one lamella at a zone axis and (b) [01–1] ZAP for precipitate indicated by arrow
in (a).
123
ligaments can be easily distinguished in most cases by comparison of
this EBSD map with the Mo distribution map. It is clear from the EBSD
map that the ligaments of the matrix phase located in between those
of the precipitate phase have the same orientation as the surrounding
matrix grain. In another instance, precipitate 2 was subjected to an
EBSD and EDS scan, the BC map from which is shown in Fig. 8(e), the
Mo distribution map in Fig. 8(f), and the IPF (Sample Z) map in
Fig. 8(g). In this case, we see that the matrix ligaments in between the
precipitate ligaments always take the orientation of the grain that it is
contiguous with. Thus we see matrix orientations Am, Bm and Cm, in be-
tween the precipitate ligaments when they are contiguous with grains
having the corresponding orientations. At least four more such
eutectic-like precipitates in the HAZ area were examined, and in all
these cases, the orientation of the matrix-phase inside the precipitate la-
mella cluster was the same as the matrix grain surrounding the precip-
itate, except when one of the grains contiguous with the precipitate
lamellae was below the surface. In those cases, further polishing to re-
move a small thickness of material showed the continuity of the matrix
phase inside the precipitate cluster with the matrix grain of the same
orientation.
EBSD maps from different clusters of precipitates in the HAZ did not
show any fixed orientation relationship with the matrix. TEM studies
also showed no set of misfit dislocations around the precipitates.
3.3.3. Precipitate from weld zone
A cross sectional TEM sample was made from a eutectic-like precip-
itate from the molten weld zone, located close to the fusion line and
about 670 μm above the two samples from the base metal and the
HAZ, using the FIB. A BF TEM image from this sample is shown in
Fig. 9(a). A cluster of precipitates is visible, where the orientation is
aligned to the [01 −1] zone axis for the precipitate. All the ligaments
in the cluster near the surface have almost the same orientation.
An SADP from the precipitate lamella marked as 1 in Fig. 9(a) is
shown in Fig. 9(b), from which it can be deduced that the precipitate
has an FCC structure with lattice parameter of ~ 11.2 Å (with an error
of ±2–2.5%). The “matrix phase” inside the precipitate also has a single
orientation, as observed in the SADP shown in Fig. 9(c), which was
taken from an area which included both the matrix and the precipitate.
In this diffraction pattern, the {111} matrix planes are in strongly
diffracting condition, and are almost coincident with the {333} diffrac-
tion spot from the precipitate. The precipitates did not show any consis-
tent orientation relationship with the matrix, although in this particular
case, there does seem to be a good matching between low index planes
of the matrix and the precipitates. The angle between the precipitate
[110] zone axis and the matrix [− 112] zone axis is about 3.6°. This
124 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128

a b

Fig. 7. (a) Trace of quantitative line scan across lamellae of the eutectic-like precipitate in the HAZ, marked as 1 and 2 in Fig. 6(a), (b) concentration profile of Mo and Ni along the line in
wt.%, and (c) concentration profile of Cr, Si, Fe and C in at.%, corrected for C content.

gives rise to a semi-coherent interface between the matrix and the pre-
cipitate in the weld zone, as evident from some TEM images (Supple-
mentary Figure S2) showing interface dislocations in the precipitates
in Fig. 9(a). The satellite peaks in the SADP from both matrix and precip-
itate may be formed due to double diffraction. Convergent beam elec-
tron diffraction (CBED) patterns showed that the 200 spots, which are
kinematically forbidden in a diamond cubic structure, but are present
due to double diffraction, have G-M lines, which are indicative of a dia-
mond glide. This indicated that the structure is likely to be Fd3m, al-
though the Bravais lattice is FCC.
The chemical composition of the precipitate was examined using
EDS in the TEM from various locations. One such quantitative EDS line
scan is shown in Fig. 10(a) as a series of spots in a line from which the
measurements were taken, and which runs from the matrix phase to
the precipitate and across to the matrix phase on the other side. The dis-
tance between the spots was 20 nm. As can be seen from the data plot of
the concentrations of the various elements (in at.%) shown in Fig. 10(b),
which are corrected for carbon content in the matrix, the precipitate is
rich in Mo, Si and C, and lean in Ni, Cr and Fe, as the previous two
types of precipitate. On calculating the average compositions in the ma-
trix and the precipitate, and correcting for the C content as explained
above, the composition of the precipitate and the matrix (in at.%) was
estimated as follows (see Table 4 below).
Table 3
Corrected composition for matrix and eutectic-like precipitate in HAZ in at.%.
Corrected Mo Cr Fe Mn Si
Matrix avg. 14.4 8.2 4.6 0.5 1.5
Ppt. avg. 46.6 6.7 1.8 0.8 6.9
This composition is similar to another known phase, Ni3(Mo,Cr)3(Si,C),
although the C content is higher than the expected ~ 12% after
correction. It has to be remembered, however, that the quantification
of C content in EDS is not very accurate. The main difference between
this composition and that of the precipitates in the base metal and
the eutectic-type precipitate in the HAZ is that the Ni to (Mo + Cr)
ratio is close to 1:1 in the precipitate in the weld metal, while it is
~1:2 in the latter. On closer examination, it is evident that the Ni con-
tent is higher and the Mo content lower in both the matrix and the pre-
cipitate in this sample, as compared to those in the base metal and HAZ
samples.
A second cross-sectional sample from the weld zone was taken from
its middle, i.e., about 700 μm from the fusion line. The crystal structure
was again found to be cubic with lattice parameter of ~ 11.15 Å. The
chemical composition of the matrix and precipitate, after correction
for C, is shown in Table 5 below.
As is evident from this table, the Ni/(Mo + Cr) ratio is close to 1:2 in
this case. It is also clear that the Ni content is lower and the Mo content
higher in both the matrix and the precipitate in this sample, compared
to the weld metal sample described above.
Careful examination of Fig. 3(a) shows segregation of elements at
the grain or lath boundaries in the dendrite formed in the melted region,
as observed in the area marked with a dashed oval. High magnification
backscatter (Z contrast) images of such regions showed that there is en-
richment of some elements at the grain boundaries (Fig. 3(c)), with pre-
cipitation preferred at triple points or quadruple points. Energy
dispersive spectroscopy (EDS) showed these grain boundaries to be
C Ni
richer in Mo and Si, similar to the precipitates themselves. In the interest
0.0 70.8 of space, these results are not shown here, but are provided in Supple-
7.3 29.8
mentary figure S3.
 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128 125

a
b c

1 2

e f

g Am

Cm

Bm

Fig. 8. (a) SEM image showing region from which an EBSD map was obtained, with two eutectic-like precipitates from the HAZ, (b) magnified image (band contrast) of the precipitate
marked as 1 in (a) from which the EBSD map was taken, (c) EDS molybdenum map showing Mo enrichment in the precipitate and (d) EBSD map showing Euler IPF Z direction in colours,
according to colour code shown in the inset, (e) band contrast image showing an eutectic-like precipitate in the HAZ, marked as 2 in (a), (f) EDS molybdenum map showing Mo enrich-
ment in the precipitate and (g) EBSD map showing IPF Z direction in colours, according to colour code shown in the inset.

4. Discussion
From the above results, it is clear that as the temperature increases
from RT to the melting point of the alloy (~ 1435 to ~ 1470 °C), the
pre-existing precipitates first undergo melting at around 1355–
1385 °C in the HAZ. Here, they probably form a eutectic mixture with
the surrounding matrix, and melt to form a liquid pool, while the larger
area in the grain surrounding the precipitate (beyond a few microns
from the original precipitate/matrix phase boundary) does not melt.
After the welding is complete, the temperatures in the sample decrease,
and the pool of liquid surrounded by solid matrix in the HAZ solidifies in
place to the form of eutectic-like lamellae. This is evident from the lines
of stringers being maintained even by the re-solidified precipitates in
the HAZ. During this process, if the lamellae of the matrix phase solidify
in contact with the surrounding matrix grain, it is highly likely that
these lamellae will assume the orientation of the same grain, as ob-
served in the examples given in the paper, and also in more instances
recorded (not shown here for considerations of space). The re-
precipitated phase has a very similar composition and crystal structure
(including lattice parameters) to the pre-existing equiaxed precipitates.
126 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128

a b

Fig. 9. (a) BF TEM image showing a eutectic-like precipitate from the weld zone on zone axis, and (b) SADP from the precipitate showing the [01−1] ZAP, (c) SADP from the precipitate and
matrix together, showing the b011N zone axis pattern (ZAP) for the precipitate, the {111} spots for the matrix, and satellite spots (most likely from double diffraction).

This phase is likely to be of the form of Ni2(Mo,Cr)4(Si,C), which is one of
the phases reported in the literature [7].
As the highest temperature attained increases beyond the melting
point of the matrix, the whole alloy melts in those regions. On cooling
after the welding process is completed, these areas which had
completely melted resolidify in the dendritic morphology, forming
large elongated grains with few twins. There is segregation of Mo and
Si in the grain boundaries and triple or quadruple points, with the latter
two types of sites showing precipitation of complex silicide/carbide
phase with high Mo and Si. The two samples taken from the molten

a b
Ni

Mo
C

Fig. 10. (a) STEM BF image of a eutectic-like precipitate from the weld zone, with a trace of a quantitative line scan shown in a dotted line across it and (b) the line profile of composition in
at.%, for Ni, Cr, Mo, Si, C, Fe and Mn, corrected for carbon content in the matrix.
 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128 127

Table 4 5. Conclusions
Chemical composition of matrix and precipitate in weld zone, estimated after correction
for carbon concentration in matrix.
(i) Microstructural studies of the welded cross section of the Ni–Cr–
Mo Cr Fe Mn Si C Ni
Mo–Fe alloy studied show that the weld metal is comprised al-
Matrix avg. 8.47 6.54 5.40 0.95 0.95 0.00 77.69 most entirely of dendritic-type grains, which show long lamellae,
Ppt. avg. 32.50 5.08 3.19 0.19 6.17 16.83 36.04
and are mostly free of the annealing twins seen in the matrix.
(ii) The pre-existing equiaxed precipitates present in stringers in the
base metal are mostly of the type Ni2(Mo,Cr)4(Si,C), with a face-
weld zone, one from high up in the weld and close to the fusion line, and
centred cubic/Fd3m structure and a lattice parameter of ~10.9 Å.
the other lower down and in the middle of the weld zone, showed two
(iii) The pre-existing equiaxed precipitates first melt at a temperature
different compositions. The first sample showed higher Ni and lower
of ~1355–1385 °C, possibly forming a eutectic mixture with the
Mo both in the precipitate and the matrix than the rest of the samples,
surrounding Ni rich matrix material, and re-solidifies in a lamellar
and the precipitate had a Ni:(Mo + Cr) ratio close to 1:1. The second
eutectic-like structure after cooling. The matrix phase in this eu-
sample from the weld metal, lower in the weld and ~700 μm from the
tectic mixture takes on mostly the orientation of the surrounding
fusion line, had Ni and Mo composition closer to that in the base
matrix phase if there is continuity with the surrounding grain.
metal and HAZ samples, and the precipitate had a Ni:(Mo + Cr) ratio
Again, the eutectic-like precipitates show a chemical composition
of 1:2, also similar to them. Thus the re-solidified phase in the weld
close to Ni2(Mo,Cr)4(Si,C), and an FCC crystal structure with a lat-
zone may be of the form Ni3(Mo,Cr)3(Si,C) or Ni2(Mo,Cr)4(Si,C), de-
tice parameter of ~10.9 Å. This indicates that although at this stage
pending on the overall chemical composition in the region, with the re-
the pre-existing precipitates have melted, mixed with the sur-
gions with higher Mo giving the latter type of precipitate. This
rounding matrix and re-solidified, they do not form a new phase.
difference in the composition of the eutectic-type phases formed in
(iv) The pre-existing precipitates melt entirely in the weld zone, and
the two samples taken from different areas in the weld zone may arise
on re-solidification of the metallic alloy, there is segregation of
due to the redistribution of elements by local chemical segregation in
Mo and Si at the dendrite grain boundaries. At triple and quadru-
the molten weld zone. The precipitates in this study differ in one funda-
ple points of intersection between dendritic boundaries, eutectic-
mental aspect from those in the work by Gehlbach et al. [7], in that none
like precipitates are formed. FIB samples taken from two locations
of the eutectic-like precipitates studied here were found to be interme-
in the weld region showed different overall compositions of Mo
tallic such as the δ Ni–Mo reported in that study. As evident from the
and Ni. TEM examination of these precipitates indicates that at
EDS data, all the precipitates studied showed a distinct rise in the C con-
least some of them are of the type Ni3(Mo,Cr)3(Si,C), and some
tent compared with the surrounding matrix, and there were substantial
others have a composition closer to Ni2(Mo,Cr)4(Si,C), judging
fractions of Cr and Si apart from Mo and Ni, thus implying that these are
from their Ni:(Mo + Si) ratios. The overall Ni:Mo ratio in both
carbides of the varieties mentioned in the text above. This may be due to
the precipitate and the matrix was higher in the region of the
a slightly different composition of the present alloy from that studied by
first kind of precipitate than that of the second kind. This indicates
the previous authors. The alloy in the present study has 16% Mo, 6% Cr
that there may be variation in the overall composition of the melt
and 0.43% Si, compared with 16.5% Mo, 7.26% Cr and 0.6% Si in the pre-
from place to place, and such variation could affect the nature of
vious study (all concentrations in wt.%). This makes the alloy slightly
the precipitates formed in the particular region of the weld
richer in Ni, and puts it farther away from the so-called “critical MD”
zone. The composition of the precipitates is not exactly stoichio-
line, thus making it more difficult for composition fluctuations to
metric, but is close to the indicated ideal compositions of the
cause precipitation of the intermetallic δ phase [12]. The lower concen-
M6C type complex silicide/carbide. The lattice parameters of
tration of Si could be another factor contributing to the absence of the δ
both kinds of precipitates were found to be close to 11.2 Å.
phase and other intermetallic phases. It must be taken into account that
(v) No evidence of the formation of δ or other types of intermetallic
these phases are not necessarily in their equilibrium states, as evident
phases was found in either the HAZ or the weld metal, at least
from the dips in the Mo concentration near the precipitate–matrix in-
in the samples examined. All the precipitates examined were
terface, which indicates incomplete mixing in the matrix.
found to be one of the two kinds of complex silicide/carbide men-
The precipitate with the Ni:(Mo + Cr) ratio of 1:1 showed good
tioned in Section 3.
matching of the {111} spot with the {333} spot of the matrix, and
showed signs of semi-coherency at the interface. The precipitates in
the HAZ or the base metal did not show any fixed orientation relation-
ship with the matrix, however, neither any prominent set of misfit dis- Acknowledgements
locations. This may be due to a better lattice match between the
precipitate and the matrix in the first weld zone precipitate than in The authors would like to acknowledge the extremely patient and
the others, as a result of the difference in chemical composition. skilful assistance of Tim Palmer, Kim Lu and Clint Jennison in the prep-
These, and other matters relating to coherency of precipitates with aration of SEM and metallography samples of the highest quality. The
matrix, need to be investigated further, because they may play an im- authors also acknowledge gratefully the assistance of Dr. Gordon
portant role in radiation damage absorption at the interfaces, and Thorogood in carrying out X-ray diffraction experiments and analysis
strengthening of the alloy, if the precipitates are dispersed more finely when needed. Thanks are also due to Dr. Ondrej Muransky for carrying
and uniformly. out helpful FE simulations and for his comments. The research reported
in this paper is partly funded by the grant from the Department of In-
dustry, Innovation, Science, Research and Tertiary Education of the Aus-
tralian Government under the aegis of the Joint Research Centre
established between ANSTO and the Shanghai Institute of Applied
Table 5
Chemical composition of matrix and precipitate in the middle of the weld zone, corrected Physics (SINAP), China (Funding Agreement ACSRF00400).
for C content.
Appendix A. Supplementary data
Mo Cr Fe Mn Si C Ni

Matrix avg. 15.07 8.05 4.44 0.95 1.66 0.00 67.63 Supplementary data to this article can be found online at http://dx.
Precipitate avg. 38.73 7.00 1.49 0.19 5.94 20.36 24.99
doi.org/10.1016/j.matchar.2015.05.001.
128 D. Bhattacharyya et al. / Materials Characterization 105 (2015) 118–128

References
[1] J. Brown, B. Montford, 25th Annual Conference of Metallurgists, Canadian Institute
of Mining and Metallurgy, Toronto, Canada 1986, p. 6.
[2] T. Yonezawa, in: R.J.M. Konings (Ed.), Comprehensive Nuclear Materials, Elsevier
2012, pp. 233–265.
[3] W.G. Johnston, J.H. Rosolowski, A.M. Turkalo, T. Lauritzen, J. Nucl. Mater. 54 (1974)
24–40.
[4] R.M. Boothby, in: R.J.M. Konings (Ed.), Comprehensive Nuclear Materials, Elsevier
2012, pp. 123–150.
[5] A.F. Rowcliffe, L.K. Mansur, D.T. Hoelzer, R.K. Nanstad, J. Nucl. Mater. 392 (2009)
341–352.
[6] S. Fyftich, in: R.J.M. Konings (Ed.), Comprehensive Nuclear Materials, Elsevier 2012,
pp. 71–91.
[7] R.E. Gehlbach, J.H.E. McCoy, International Symposium on Structural Stability in
Superalloys, TMS (1968) 346–366.
[8] V. Ignatiev, A. Surenkov, in: R.J.M. Konings (Ed.), Comprehensive Nuclear Materials,
Elsevier 2012, pp. 221–250.
[9] D.L. Klarstrom, H.M. Tawancy, R. M. F., in: M. Gell (Ed.), Superalloys 1984 (Fifth In-
ternational Symposium), Metallurgical Society of AIME, Champion, Pennsylvania,
USA 1984, pp. 553–562.
[10] J. Zhang, R. Hu, J. Wang, J. Li, Int. J. Innov. Technol. Exploring Eng. 3 (2013) 48–51.
[11] Z. Xu, L. Jiang, J. Dong, Z. Li, X. Zhou, J. Alloys Compd. 620 (2015) 197–203.
[12] J.S. Ogborn, D.L. Olson, M.J. Cieslak, Mater. Sci. Eng. A 203 (1995) 134–139.
[13] M.J. Perricone, J.N. Dupont, Metall. Mater. Trans. A 37A (2005) 1267–1280.
[14] M.J. Cieslak, T.J. Headley, A.D. Romig, Metall. Trans. A 17 (1986) 2035–2047.
[15] H.E. Collins, in: W.J. Boesch (Ed.), International Symposium on Structural Stability in
Superalloys TMS, Seven Springs, PA, USA, 1968.
[16] X. Xie, Y. Zeng, L. Kou, J. Dong, L.M. Pike, D. Klarstrom, in: R.C. Reed (Ed.), Superalloys
2008, TMS, Champion, PA, USA, 2008.
[17] M. Raghavan, R. Mueller, G.A. Vaughn, S. Floreen, MTA 15 (1984) 783–792.
[18] O.A. Ojo, N.L. Richards, M.C. Chaturvedi, Scr. Mater. 51 (2004) 683–688.
[19] J. Unfried-Silgado, A. Ramirez, Met. Mater. Int. 20 (2014) 307–315.
[20] M.J. Vold, Anal. Chem. 21 (1949) 683–688.
[21] K. Kuo, Acta Metall. 1 (1953) 301–304.