Professional Documents
Culture Documents
Development of Constitution Diagram For Dissimilar Metal Welds in Nickel Alloys and Carbon and Low-Alloy Steels PDF
Development of Constitution Diagram For Dissimilar Metal Welds in Nickel Alloys and Carbon and Low-Alloy Steels PDF
THESIS
By
2010
2010
ABSTRACT
increased desire to understand the metallurgy that controls the properties of welds
between these materials. The increased use of these combinations has been driven
by their successes in applications where austenitic stainless steel weld metals have
properties, and close match in coefficient of thermal expansion with steels make
weldability issues of their own to mitigate. Among the more formidable concerns
the composition transition zone at the fusion boundary, which often accompanies
during service. The Schaeffler diagram has been demonstrated to have an ability to
been developed based on 300 series stainless steels its overall predictive capability
ii
The goal of this study is to use previously established techniques to develop a
predictive diagram which has greater accuracy when applied to dissimilar welds
between carbon and low-alloy steels and nickel-base alloys. Button melt samples
were created by mixing various proportions of carbon and low-alloy steels and
response data generated by the Magne-Gage was analyzed using statistical methods
and fit to a model in order to allow prediction of magnetic response based on alloy
content. The model was then used for construction of a predictive diagram for
iii
DEDICATION
iv
ACKNOWLEDGMENTS
Boian Alexandrov whose guidance on this project was pivotal to its success, and
Systems Engineering Department at the Ohio State University, and Dr. Jerry Gould of
the Edison Welding Institute for their advice and assistance with statistical analysis
Thanks to Professor Suresh Babu for his advice during parts of this project,
committee.
Thanks to the members of the Welding and Joining Metallurgy Group, both
overstated. Thank you to Adam Hope for his work digitizing the Schaeffler diagram.
Thank you to Jeff Rodelas for his assistance with EBSD work.
Finally, I would like to thank the many people in my life whose support,
advice, and encouragement during this project proved essential to its completion.
v
VITA
Columbus, Ohio
FIELD OF STUDY
vi
TABLE OF CONTENTS
Abstract ....................................................................................................................................... ii
Dedication.................................................................................................................................. iv
Acknowledgments .................................................................................................................... v
Vita ............................................................................................................................................... vi
1. Introduction ...................................................................................................................... 1
2. Background ....................................................................................................................... 3
2.1.1 Composition..................................................................................................................................... 3
2.2.1 Composition................................................................................................................................... 13
vii
2.2.2 Metallurgy and Mechanical Properties............................................................................... 14
3. Objectives ......................................................................................................................... 39
4. Experimental Procedures........................................................................................... 41
4.1 Materials............................................................................................................................... 41
viii
4.9 Diagram Development .................................................................................................... 58
6. Conclusions ...................................................................................................................... 85
References ................................................................................................................................ 87
ix
LIST OF TABLES
Table 5.1: X-ray diffraction and Magne-Gage analysis results for selected
samples ............................................................................................................................. 73
Table 5.2: Point Counting, X-ray diffraction and Magne-Gage analysis results
Table B.1: Button melt sample composition and ferrite number data ............... 98
x
LIST OF FIGURES
Figure 2.1 Fe-C phase diagram (region of interest for steels) .................................. 8
Figure 4.2: Schaeffler diagram enlarged to show study focus area ...................... 45
Figure 5.1: TCP phase within interdendritic regions of 8630-625 85% dilution
Figure 5.7: Ferro-fluid staining of F22-800H 55% button melt sample ............. 66
Gage .................................................................................................................................... 71
Figure 5.13: X-ray diffraction spectra of X80-C276 73% dilution button melt
sample ............................................................................................................................... 74
xii
Figure 5.14: X-ray diffraction spectra of F22-800H 60% dilution button melt
sample ............................................................................................................................... 75
Figure 5.15: Plot of martensite phase fraction vs. FN for this study .................... 79
formulae ........................................................................................................................... 83
xiii
1. INTRODUCTION
CHAPTER 1
INTRODUCTION
fabrication of pressure vessels to name just a few. The nickel-base filler metals offer
several advantages over the use of austenitic stainless steels typically associated
[1, 2].
in stainless steel welds has been the subject of study for over 50 years. The
Schaeffler diagram, created in 1947 [3], and published in its final form in 1949 [4]
was one of the first such diagrams. Due to its large composition range it has been
metal combinations [5-7] including those between nickel-base filler metals and
carbon and low-alloy steels. This diagram was created based on the 300 series
1
stainless steels, and a select number of ferritic and martensitic stainless steels. For
unknown.
for prediction of the presence of martensite in the weld metal of dissimilar metal
welds involving nickel-base alloys and carbon and low-alloy steels. The relative
amount of martensite present in the fusion zone has been closely attributed to
dilution level required for martensite formation based on the Schaeffler diagram.
boundary has been shown to increase the susceptibility of these welds to cracking
welds where nickel-base alloys and carbon and low-alloy steels are involved
highlights the need for a diagram that is specifically intended for prediction of weld
such a diagram using combinations of nickel-base filler materials and carbon and
low-alloy steels. Such a diagram will provide information about the microstructure
of weldments between materials that the Schaeffler diagram is not intended for and
2
2. BACKGROUND
CHAPTER 2
BACKGROUND
Carbon and low-alloy steels account for 95% of the world’s construction and
fabrication metals [8]. Low cost and favorable mechanical properties make these
common due the large quantity being utilized. Based on composition, strength, heat-
1. Carbon Steels
5. Chromium-molybdenum steels
2.1.1 Composition
Carbon steels form the basis from which all the other steel alloy groups are
created. Due to its ability to impart significant strength when added to iron in small
3
amounts, carbon is the most important alloying element in steel. As a result the Fe-C
system is the basis for all carbon and low-alloy steel grades. Carbon steels contain
carbon contents of less than 1.0 wt.%. Carbon steels are designated in order of
high carbon steel [8]. In addition to carbon, manganese, silicon, and copper may also
be present in carbon steels as a result of the steel making practice [9]. Manganese
contents in carbon steels are kept below 1.65 wt.%, and copper and silicon contents
product thickness. In general the contents of these elements individually are kept
below 1.0 wt.%. Carbon is present in these alloys in the range of 0.05-0.25 wt.%.
Manganese is present at levels below 2.0 wt.% for increased formability and
weldability [9].
HSLA steels but can also include increased amounts of varying elements to promote
specific properties. Addition of 8-9% nickel to A553 steel results in increased low-
the HSLA and quenched and tempered steels. However, these steels contained
increased amounts of various elements relative to these other steels. The increase in
4
alloying additions results in the ability to be heat-treated to high-strength following
welding, as the name would suggest. Carbon contents in these alloys range from
0.25 wt.% to 0.45 wt.%. Because of cracking concerns associated with heat-
treatment of these alloys sulfur and phosphorus levels are kept below 0.040 wt% or
lower dependent on steel melting technique and intended final strength level [9].
niobium, and nitrogen are also used to increase high-temperature properties. These
boundary sliding thus increasing creep strength. Alloying additions are also tailored
such that the desired matrix of ferrite, bainite, or martensite can be achieved with
proper heat treatment. Chromium contents in these steels range from 0.5–9 wt.%,
and molybdenum contents range from 0.5-1.0 wt.%. Carbon contents in these steels
2.1.2 Metallurgy
phase diagram and the non-equilibrium transformations that occur under certain
conditions. The compositions previously discussed are chosen on the basis of the
effects different alloy additions have on these reactions. Additionally, because of the
5
importance of welding as a means for joining these materials for many applications
The Fe-C phase diagram shown in Figure 2.1 [9] provides a starting point for
that pure iron will exist as a variety of different phases, each with different
equilibrium conditions, or very slow cooling, austenite will transform into ferrite (α)
or pearlite, a mixture of ferrite and cementite (Fe3C) for the range of carbon
other phases may form such as bainite and martensite. Each of the aforementioned
Llewellyn et. al [10]has reported that there are five practical options for
3. Precipitation strengthening
6
4. Transformation strengthening
5. Dislocation strengthening
Carbon and low-alloy steels derive their mechanical properties based on any
strengthening methods is beyond the scope of this review, for detailed information
the reader is referred to the references cited throughout this section. The following
7
Figure 2.1 Fe-C phase diagram (region of interest for steels)
steels. Generally speaking ferrite is the softest and therefore weakest of the
to iron increases the strength of ferrite by solid solution strengthening, with carbon
being potentially potent in this regard [9]. Reducing the ferrite grain size also has a
beneficial effect on grain size, with the particular advantage that toughness is not
elements that form a fine dispersion of particles can be added such that at heat
grain boundaries by these particles. Fine austenite grain size will result in fine
the austenite to ferrite transition occurs can also be utilized to control grain size.
steels. Final microstructure in carbon steels will consist of a mixture of ferrite and
rolling is the preferred processing route as it eliminates the need for heat treatment.
9
HSLA steels are strengthened on the same principals as carbon steels.
larger role than in carbon steels where grain refinement is more significant. HSLA
bainite and martensite that are stronger than ferrite and cementite. The increased
hardenability, the ease of martensite formation. Alloy additions are also made to
alloys will consist of tempered martensite, with or without the presence of second
phases such as precipitates and retained austenite. These steels often require
quench and tempered grades. For this reason their metallurgy is closely related to
the quenched and tempered steels. The most important difference is that the
10
increase in alloying additions increases the hardenability making the formation of
martensite easier, which has significant impact on the welding metallurgy of these
steels as will be discussed. Additionally, the higher carbon levels exhibited in these
steels make the martensite formed harder and more brittle. Like quenched and
mechanical properties.
these steels is similar to the other more highly alloyed classes already discussed.
The addition of chromium to these steels account for their increased oxidation
temperatures at which these steels are often used by forming precipitates that
increase resistance to grain boundary sliding. Other elements, such as V, Nb, and Cr,
will also participate in precipitation reactions that have similar effects. Complex
heat treatments are generally required for reasons already discussed as well as the
The welding metallurgy of all the above steels is quite complicated as well.
They are welded with a variety of filler materials to suit particular needs, which can
result in a variety of issues related to the welding metallurgy of a given system. The
main concern when welding these steels as will be discussed in more detail below is
the formation of hard brittle martensite within the fusion zone and heat affected
zone (HAZ). This is particularly problematic in the HTLA grades that are highly
11
alloyed resulting in high hardenability. When welded, these grades are likely to form
hard martensite in the HAZ. Loss of mechanical properties can also occur for a
number of other reasons during welding. Grain growth in the HAZ of alloys where
ferrite grain size plays an important strengthening role can reduce base metal
back into solution during welding and don’t have a chance to re-precipitate on
cooling. The welding metallurgy can be quite complicated and will vary based on the
2.1.3 Weldability
joined [8]. Different alloy groups are susceptible to solidification cracking, lamellar
tearing, and reheat cracking. The number of different weldability concerns varies in
when joining carbon and low-alloy steels is cold cracking; also known as hydrogen
steel alloys. This type of cracking is associated with hard, brittle martensite often
encountered in the heat-affected zone (HAZ) of carbon and low-alloy steels. When
this hard brittle microstructure is combined with stress, and hydrogen from many
common welding processes, cracking occurs. Although the details of the phenomena
surrounding this type of cracking are disputed and not well understood, general
methods for avoiding it do exist. These focus mainly on eliminating one of the
contributing factors mentioned above. Post weld heat treatment is often applied to
12
welds where brittle microstructures would be expected to form in the HAZ upon
cooling. Furthermore, steel grades such as the HSLA and quenched and tempered
steels, where hardenability has been reduced by lowering the amount of alloying
additions, such that martensite formed on cooling is of a softer and less brittle
nature if formed at all, have been developed with joining in mind. The prevalence of
this form of cracking has led to many ways of mitigating it in the industry.
2.1.4 Applications
Carbon and low-alloy steels find use in the greatest variety of applications of
any engineering alloy. The carbon varieties are popular in construction of a number
of structures such as buildings and bridges [11]. The HSLA grades are used in
applications ranging from ship hulls to automobiles. The quench and tempered
equipment utilized by the power industry [8]. Carbon and low-alloy steels remain
popular due to the wide range of mechanical properties they offer that can be
2.2.1 Composition
they derive the majority of their strength from alloying additions that act as solid
solution strengtheners. Several classes of these alloys exist based upon the primary
13
alloying additions. Nickel-copper alloys were among the first nickel-based alloys to
be produced and form the basis of the 400 series alloys in use today [12]. Chromium
Additional systems are based on Ni-Mo, and Ni-Fe with both of these alloying
additions having large maximum solubility in nickel and acting as solid solution
strengtheners. Additions of Co, W, Ta may also be added for their effects as solid-
solution strengtheners. Niobium, along with many of the elements already listed
as austenite. The Ni-Cu series will remain fully austenitic to room temperature due
to the complete solubility that exists between these two elements. Carbon
concentrations over 0.02 wt. % will lead to the formation of Cr rich M23C6 carbide at
temperatures below 950ºC during the cooling cycle. MC type carbides can form
conjunction with carbon. In Ni-Mo and Ni-Cr-Mo alloys the topologically close
packed (TCP) phases σ, μ, and Ρ may form if sufficient Mo and Cr is present. These
phases appear at interdendritic regions and are therefore are associated with a
[12]. The practical aspects of the presence of these phases will be discussed in
subsequent sections.
14
The mechanical properties of solid-solution strengthened nickel-base filler
metals in the as-welded condition are similar to those of their matching base
materials. They often may have superior properties to matching base metals due to
the inclusion of Al, Ti, Mn, Nb, and Mo in the filler metals imparting additional
strength. Many of these filler metals also exhibit reasonable elevated temperature
second phases. Postweld heat treatment has been shown to result in precipitation
the undesirable phase δ (Ni3(Nb,Mo,Cr,Fe,Ti)) in alloy 625 when carried out in the
phase, coupled with dissolution of body centered tetragonal γ’’ phase (Ni3Nb), which
precipitates out at the low end of the temperature range and imparts strength in
alloy 625 after PWHT, results in a reduction of all mechanical properties of the
weldments.
2.2.3 Weldability
type of cracking is associated with the presence of liquid films along solidification
grain boundaries that lead to cracking at stresses below the yield stress. Increased
15
solidification temperature range is closely associated with an increase in
during solidification can result in the formation of low melting point phases that
tendency to expand the solidification temperature range when added to nickel. The
strong tendency of these elements to segregate into the liquid during solidification,
wetting of liquid films along grain boundaries, add to their propensity to increase
metals. Additions of Mn, Si, Al, and Ti have been shown to combat some of these
effects and reduce the solidification cracking susceptibility [14]. In alloys with
greater concentrations of elements, such as Nb, Cr, and Mo, which participate in
formation of these phases at the end of solidification. These phases tend to form at
lower temperatures than the alloy solidus thus increasing the solidification
and C-276 have shown an increased risk of solidification cracking susceptibility due
to the presence of the TCP phases discussed above [15]. Cieslak has reported on the
alloy 625 caused by the formation of laves phase and MC type carbides [16]. It is
16
noteworthy that although alloy 625 often shows poor resistance to solidification
aided by the fact that it often will generate sufficient liquid to backfill and heal
metals are generally quite weldable as a result of either tight control of alloying
additions, small solidification temperature ranges, and the ability to backfill and
heal cracks.
state phenomenon and occurs at temperatures between the solidus (Ts) and 0.5Ts.
ductility over this temperature range, and the resultant stresses from welding will
cause cracking. The mechanism for DDC is not well understood, and many theories
exist to try and explain the phenomenon. DDC is found to be present along migrated
base alloys. Recent studies [17-19] have indicated that DDC is related to grain
boundary sliding that occurs at high temperatures. It has been shown [12, 17-20]
also be a concern when such alloys are used in situations where corrosion
17
resistance is important. This can be attributed to the tendency for elemental
composition gradients which can lead to localized attack [12]. PWHT can have both
and time it is carried out at. When stress corrosion cracking is a concern the use of
PWHT to relieve residual stress can be beneficial. Cortial et al. [13] has reported an
discussed earlier that results in formation of γ’’ and δ phases. PWHT of these alloys
must therefore be carried out carefully to insure the best mix of properties.
2.2.4 Applications
and their matching filler metals would therefore often be expected to be present in
many of the same applications where joining of these materials would be necessary.
These include use in the power generation, chemical processing and petrochemical
industries [12]. Use in these industries is often closely related to their superior
corrosion resistance at high temperature [21]. They are also used as cladding
material for variety of carbon, low-alloy, and stainless steels [8, 12, 21].
18
2.3 Weld Metal Constitution Diagrams
for almost a century. The majority of the research into constitution diagram
development has been focused on stainless steels that exhibit a final microstructure
martensite. The carbon and low-alloy steels and nickel-base alloys will be shown to
welded. For this reason, the focus of this section will be to present background on
those constitution diagrams that have been developed which are capable of
austenitic and duplex austenitic plus ferritic systems. Many of these diagrams still
diagrams were the building blocks for modern constitution diagrams used to predict
subject have been undertaken. Olson [22] has compiled a review of constitution
diagram history up to 1985. While reviewing many of the same studies as Olson,
19
et al. [5], have also presented a thorough review of the subject. Those authors have
Strauss and Maurer [24] produced the first of these diagrams in 1920. Their
wrought steels. Nickel and chromium percentages were the y-axis and x-axis
Riedrich and Hoch [25]and is presented in Figure 2.2. Their diagram included the
addition of austenite-ferrite stability lines. This diagram was applicable for alloys
normal contents of carbon, silicon, and manganese this diagram was useful in
20
Figure 2.2: Modified Strauss-Maurer Diagram [25]
the given element, as they will in all equations that follow. The equation published
by Newell and Fleischman would become the standard for constitution diagram
that promoted austenite and ferrite onto opposing sides of their equations. These
groupings were the predecessors to the chromium and nickel equivalent formulas
used in modern day constitution diagrams. Much of the research into constitution
21
diagrams has been in altering of the nickel and chromium equivalents to ensure
equation to improve its accuracy when applied to welding. The constant 8 was
ferrite formation. Presented with austenite forming elements on the right and
ferrite forming elements on the left, as it would eventually be written, their equation
Creq Cr 1.5Mo 2 Nb
Thomas [29] reported the following equation for the stability of austenite
relative to δ-ferrite:
22
This equation was significant for two reasons. It included terms for silicon and
boundary featured on the previously discussed diagrams. This trend would become
combine the concepts of equivalency formulae with the Maurer diagram. The
Maurer diagram was used in his initial study; however, he modified it to include
nickel and chromium equivalents as the y-axis and x-axis respectively. His
Multiplying factors for chromium, nickel, manganese, and carbon were chosen based
factors for silicon, molybdenum, and niobium based on his own experience.
Balmforth [23] notes that the absence of a nitrogen term in Schaeffler’s nickel
equivalency formula, despite its strong effect as an austenite stabilizer, was likely
23
Schaeffler’s original diagram was created by plotting data points from 34
filler material on 4340 steel and mild steel plate, as well as undiluted weld metal on
the modified Maurer diagram. Using the chromium and nickel equivalents he
determined, data points were determined based off of tested chemical compositions
redrawn to coincide with the data points. This diagram is presented Figure 2.3.
Included in Figure 2.3 are the data points from Schaeffler’s original work, as well as
lines corresponding to the modified Maurer diagram illustrating how it was adapted
by Schaeffler.
24
Figure 2.3: The Schaeffler Diagram (1947) [3]
(Creq 16) 2
Nieq 12
12
where Nieq and Creq are his nickel and chromium equivalents as reported above. He
notes that his equation only differs from those proposed by Newell and Fleischman,
Schaeffler published modified versions of his diagram in 1948 [30], and again
in 1949 [4]. The 1948 version of his diagram featured straight lines for the
boundaries and included iso-ferrite lines allowing for better prediction of weld-
version. However, the adjustment of the chromium equivalent in his final version is
resulted in revision of the multiply factors for silicon, niobium, and molybdenum, as
well as a slight relocation of the phase boundaries. His final chromium equivalent
was:
Schaeffler’s final diagram is presented in Figure 2.4. This diagram was reported to
be accurate to within 4% ferrite when used for prediction of microstructure in 300
series weld materials. The left side of the diagram was determined to be
The final version of the Schaeffler diagram still finds use today as will be discussed
in subsequent sections.
26
Figure 2.4: The Schaeffler Diagram (1949)
to focus on specific regions of interest within earlier diagrams [23]. The reader is
referred to the previously mentioned reviews [5, 22, 23] for details in development
microstructure as well as those concerned with ferritic plus martensitic weld metal
focused on those that predict weld metal microstructures consisting austenite plus
martensite.
27
Balmforth [23] reports that Eichelman and Hull [31] were the first to study
When the Ms is assumed set to be 20ºC this equation is shown to reduce to:
This equation demonstrates a manganese to nickel ratio close to the 0.5 reported in
Schaeffler’s nickel equivalent. Andrews [32] study of Fe-Mn-Ni weld metal with
The manganese to nickel ratio illustrated here is almost the opposite of that
predicted by Schaeffler.
well as the work of Andrews, and of Eichelman and Hull, conducted a study to
the position of the austenite – austenite plus martensite on the Schaeffler diagram.
Several welds were made that resulted in nominal 9 wt% Cr and 1.2 wt% Cr
concentrations with varying levels of Mn and Ni. It was shown that the equation of
28
Eichelman and Hull correlated well to the 1.2% Cr data generated, while the
equation of Andrews produced good correlation with the 9% Cr data. On this basis
an equation was generated which predicts austenite formation in weld metal with 0-
16 wt% Cr:
Two significant features of this equation were noted. First, the interaction of
austenite former as is seen in stainless steels. Second, the nature of chromium to self
interact as demonstrated by the Cr2 term. The author also proposed a modification
29
Béres [34] has also proposed a modified version of the Schaeffler diagram.
He notes that the Schaeffler diagram was designed using stainless steels with carbon
contents of less than 0.1 wt%. However, the effect of carbon as an austenite
stabilizer is related to its ability to depress the Ms, which is drastically reduced at
levels above 0.1 wt% as are often typical in carbon and low-alloy steels, and could
published formulae for Ms as a function of alloying additions Ni, Mn, Cr, Mo, and Si
Creq Cr Mo 1.5Si
0.2
Nieq Ni 0.5Mn 10C
C
These equations combined with the Ms equation:
MS MsC 21(Ni 0.5Mn) 16.8(Cr Mo 1.5Si)
4.2
M sC 454 210C
C
Below chromium equivalents of 18, the left side of the diagram is used with the
30
concentration. The use of his modified nickel equivalent, Nieqβ is required in this
region of the diagram. The right side of the diagram is to be used as the original
2.4.1 Applications
carbon and low-alloy steels in a variety of applications. When carbon and low-alloy
steels are welded to stainless steels using a stainless filler metal the large difference
9.5-10 µin/in/°F) of these two groups can lead to cracking in elevated temperature
service (greater than 425ºC). Nickel-alloys possess a CTE of 8 µin/in/°F, that lies
31
between carbon and low-alloy steels and stainless steels. For this reason they are
frequently used to successfully join these alloy groups for use in the power-
alloys are often used to clad carbon and low-alloy steels in application where
2.4.2 Weldability
welds made on carbon and low-alloy steels and those made with solid-solution
strengthened nickel-base filler metals, weldability concerns exist that are specific to
Although the carbon and low-alloy steels have a variety of different final
microstructures based upon composition and processing, they are all similar in that
they will solidify as body-centered cubic (BCC) delta ferrite. The different crystal
structures of the solidification phases of the two alloy groups give rise to an
boundaries, which have been linked with cracking in some instances. The
epitaxial growth of the weld metal off of the base metal as is typically observed in
austenitic (FCC) filler material is unable to grow epitaxially from the ferritic (BCC)
heat-affected zone of the base metal. The result is heterogeneous nucleation within
the weld metal. As the HAZ of the base metal cools through the δ-ferrite region into
32
the austenite region it becomes FCC and a high-angle boundary is left between the
HAZ and fusion zone. This resultant boundary has high energy and is mobile, and
migrates into the FCC weld metal as a result. As the material cools to room
temperature the boundary is locked into place and the final microstructure is left
with this so called Type II boundary which runs essentially parallel to the fusion
Figure 2.7: Schematic representation of Type II boundaries and their formation [35]
The microstructural zone in which Type II boundaries are found within the
fusion zone is commonly referred to as the transition region. Along with Type II
martensite that forms on cooling as well as after PWHT [36]. The formation of
diffusion of carbon from the higher-carbon base material into the weld metal. The
33
Schaeffler diagram discussed previously can be used to predict the presence of
martensite in nickel-base weld metals diluted by carbon and low-alloy steel base
metals as shown in Figure 2.8 [6, 12]. Although the dilution levels necessary to cause
region (~10μm) will always exist that includes the high dilution levels necessary for
Figure 2.8: Example usage of Schaeffler diagram as it relates to dissimilar metal welding [12]
Numerous studies have been conducted that examine factors that contribute
to the formation of this layer of martensite, and the deleterious effects it has on joint
properties [6, 7, 36]. It has been previously noted that martensite is frequently
associated with hydrogen induced cracking. Rowe et al. [6] investigated the
34
susceptibility to hydrogen induced cracking of welds made with two austenitic
stainless steel filler metals as well as one nickel-base filler metal. They found that
welds made with all three filler metals showed increased cracking susceptibility
when hydrogen was introduced through the shielding gas. Metallographic analysis
showed that the cracking was associated with a layer of martensite that formed
within the transition region that exhibited high hardness relative to the weld metal.
induced cracking of the three filler metals tested. This was attributed to the
compared to the two austenitic stainless steel filler metals examined, which resulted
Additionally, cracking could not be induced in multi-pass welds made using the
same materials. This was attributed to the increased ability of hydrogen to diffuse
away from the martensite given the higher temperatures associated with multi-pass
present in the transition region to a minimum is apparent from this study; as well as
induced cracking.
Dupont and Kusko [7] also investigated the formation of martensite along the
fusion boundary of a carbon steel clad with nickel-base and austenitic stainless steel
filler metals. Their study matched the findings of Rowe et al., noting that nickel-base
35
filler materials resulted in the formation of less martensite within the transition
region. They used concentration gradients measured from the base material into an
undiluted region of the filler metal to determine the martensite start temperature
suppressing the Ms temperature the distance into the fusion zone the Ms
temperature was calculated to be above room temperature was ~10μm for the
nickel-base filler metal compared to ~40μm for the austenitic stainless steel filler
metal. As a result, martensite is able to form a greater distance into the cladding
material when the austenitic stainless steel filler metal was used. The agreement
with the work of Rowe et al. can be attributed to the fact that as the necessary
dilution to form martensite was increased by the use of nickel-base filler metals
over stainless steel filler metals, the Ms temperature was being maintained at or
below room temperature up to this dilution as is shown in the work of Dupont and
Kusko.
Gittos and Gooch [36] also confirmed the presence of martensite near the
steel cladding. They also discussed the effect of PWHT on the transition region. The
composition gradient that exists following welding results in suppression of the Ac1
temperature in the transition region and the PWHT temperatures in the range of
650-720 ºC overshoot the Ac1 resulting in the formation of fresh austenite. PWHT
causes carbon diffusion from the base material into the transition region and weld
36
material to occur due to the difference in carbon concentration between the base
and filler materials. The results of this are formation of Cr-rich carbides within the
transition region and weld metal. The carbides themselves contribute to hardening
in this region and also tie up Cr resulting in a lower level in the matrix that increases
result is that the PWHT has a limited effect on increasing ductility in the transition
region. However, the larger problem associated with PWHT of dissimilar metal
Carbon migration from carbon and low-alloy steel base materials into the
weld metal in dissimilar metal welds is associated with the formation of a soft
ferritic microstructure in the HAZ [12, 36, 37]. Gittos and Gooch [36] noted this
nickel-base and austenitic stainless steel filler metals, and that it had a lowered
hardness compared to the as-welded HAZ and was the location of many failures
between materials in the joint. Therefore, these types of failures are more
associated with dissimilar metal welds where an austenitic stainless steel filler
material is used because of the large difference of CTE. This mismatch in CTE results
in high local strain along the interface between the carbon/low-alloy steel and the
austenitic stainless steel filler metal. Over time at high temperature, grain boundary
37
sliding occurs in the soft ferritic region resulting in creep failures. Nickel-base filler
metals are often the solution for this type of problem. Nickel-base alloys possess a
CTE that closely matches carbon/low-alloy steels, and are also stronger than the
carbon denuded ferritic HAZ. As a result they will transfer the strain caused by
mismatches in CTE between different materials onto the austenitic stainless steel
increased when they are used to weld carbon and low-alloy steels. In these
the filler metal by the base material. As previously noted, the solidification
susceptibility of a given alloy. Increased levels of Fe, Cr, Cu, Si in nickel-base weld
and low-alloy steels high levels of impurities such as S and P will result in locally
38
3. OBJECTIVES
CHAPTER 3
OBJECTIVES
The objective of this study was to evaluate the ability of the Schaeffler
diagram to predict weld metal microstructure for when carbon and low-alloy steels
are joined with solid-solution nickel-base filler materials. The data compiled
throughout this process will be used to accomplish the main goal of development of
a new weld metal constitution diagram aimed specifically at welds between these
two materials from which samples in this study were created. The specific
materials.
40
4. EXPERIMENTAL PROCEDURES
CHAPTER 4
EXPERIMENTAL PROCEDURES
In order to meet the objectives of the study a range of data was required that
covered the majority of the martensite plus austenite region of the Schaeffler
diagram, as well as that area immediately surrounding it. Much of the procedure
that follows has been adapted from the work of Balmforth [23], whose work
provided a good starting point for the following study. This section contains the
4.1 Materials
Carbon and low-alloy steels and nickel-base alloys for this study were
selected on the basis of their position when plotted on the Schaeffler diagram, their
use in the industry and their availability. The materials used in this study are
presented along with their composition in Table 4.1. The selection of these materials
covered a wide range of the Schaeffler diagram as shown in Figure 4.1. Material
combinations for sample production were selected that allowed the easiest drawing
whose connecting ties lines were roughly parallel to each other. Initial material
41
combinations are illustrated in Figure 4.2, which demonstrates how tie-lines were
within each material combination for sample production. One dilution was selected
that would fall in the middle of the martensite plus austenite region of the Schaeffler
diagram. The two other dilutions were selected such that they fell just within the
austenite region on the other side of the austenite – austenite plus martensite
boundary, and just within the martensite region on the other side of the martensite
– martensite plus austenite boundary. The initial dilutions selected are illustrated
42
Table 4.1: Chemical composition of materials used in this study
Material
Mn
Mo
N*
Nb
Cu
Co
Cr
Ni
W
Al
Si
Ti
V
P
C
S
8630 0.299 0.856 0.009 0.008 0.257 0.770 0.782 0.363 0.002 0.004 - 0.035 0.009 0.198 - 0.008
A553 0.044 0.582 0.004 0.001 0.189 8.970 0.140 0.010 0.001 0.004 0.004 0.018 0.001 0.034 0.050 0.012
Steels
A508 0.182 1.449 0.005 0.002 0.181 0.920 0.125 0.514 0.002 0.002 - 0.016 0.004 0.040 - 0.025
X80 0.023 1.370 0.014 0.004 0.113 0.013 0.040 0.010 0.059 0.042 - 0.031 0.005 0.033 - 0.003
F22 0.141 0.430 0.011 0.022 0.292 0.128 2.420 1.030 0.004 0.002 - 0.025 0.017 0.190 - 0.010
P92 0.120 0.430 0.010 0.001 0.280 0.120 8.270 0.380 0.010 0.010 0.045 0.006 0.174 0.190 1.615 -
625 0.017 0.037 0.003 0.001 0.084 64.40 22.29 8.680 3.660 0.212 - 0.102 - 0.014 - -
FM82
Ni-Base
0.040 2.810 0.004 0.001 0.030 70.40 21.00 0.005 2.210 0.050 - - - 0.090 - 0.050
Alloys
FM52M 0.020 0.760 0.003 0.001 0.110 61.37 29.06 0.040 0.820 0.210 - 0.050 - 0.020 - -
C276 0.003 0.330 0.007 0.006 0.300 57.75 16.06 16.01 0.179 0.010 0.014 0.030 0.020 0.060 3.350 0.200
800H 0.068 0.710 0.006 0.001 0.380 30.90 20.90 0.250 0.040 0.330 - 0.280 - - - -
*Note: Omission of value does not reflect lack of presence in given alloy
43
Figure 4.1: Schaeffler diagram with materials of interest plotted
44
Figure 4.2: Schaeffler diagram enlarged to show study focus area
45
4.2 Sample Production
creating weld metal microstructures. For this reason the button melting technique
was utilized as the means of sample production for this study. A schematic of the
button melting setup is presented in Figure 4.3. The intended final sample size was
four grams. Masses of base materials used to create the samples were determined
by calculation using the intended dilution and intended final sample mass. The
common convention for dilution is weld metal dilution by the base metal was used.
Masses of base materials for use in creation of button melt samples were
Because base materials were available in varying forms (pipe, sheet, welding wire,
etc.) they often needed to be sectioned to the necessary mass. This was usually
accomplished by sectioning with an abrasive saw and wet grinding on a belt sander.
Base materials were cleaned in ethyl alcohol and dried following this procedure.
They were then massed on a high precision digital scale to within 0.01 gram of the
target mass.
After being massed out, base materials were placed within the water-cooled
copper crucible of the button melting apparatus. The chamber was then closed and
purged with argon. Melting of samples was accomplished with the GTAW torch
using a DCEN polarity and a 2% thoriated tungsten electrode with a nominal current
setting of 150 A. The current was controlled with a foot pedal and was gradually
increased to its maximum as applying excessive current before base materials had
46
been allowed to melt could result in ejection of material. The GTAW torch is swirled
at maximum current, as shown in Figure 4.3 to try to create motion within liquid
button that would eliminate concentration gradients. Current was similarly ramped
down gradually following melting. Samples were flipped upside down and re-melted
47
Figure 4.3: Schematic representation of button melting setup
4.3 Characterization
Following button melting, samples were sectioned in half using a LECO VC-
mounted in Bakelite. Samples were polished to final polish using two methods. The
48
base unit equipped with semi-automatic head and the use of LECO magnetic
CAMEO disks, and LECO FAS disks. The polishing steps for this procedure were a
grinding until planar using the blue CAMEO disk, followed by use of the silver
CAMEO disk with 6μm diamond suspension, followed by use of 3μm diamond
paste on a adhesive backed polishing cloth attached to FAS disk, before finishing
with colloidal silica followed by water to flush the pad and remove residue from
samples. Wheel speed of 250 RPM was used for all steps but colloidal silica when the
speed was reduced to 150 RPM. Cylinder head pressure varied based on application.
In general pressure was about 8-9 psi/sample when 1.5in samples were used, and
about 5-6 psi/sample when 1.25in samples were used. These levels were generally
reduced 1-3 psi/sample during colloidal silica step. Time at each step was also
varied; the author suggests 3 min on each step following rough grinding except for
final polishing with colloidal silica, where 1.5 min polish and 1.5 min water flush
should be used initially. Samples were rinsed in soapy water followed by ethyl
alcohol between each step listed. Manual polishing was performed on Leco
were wet ground to 800 grit, followed by polishing with 3μm, then 1μm diamond
paste. Samples were rinsed in cold water, followed by ethyl alcohol, and then
ultrasonically cleaned between each step. Samples intended for ferrite number (FN)
was unnecessary. They were simply sectioned following procedure outlined above,
49
ground to 800 grit and rinsed in water, followed by ethyl alcohol and dried in hot
air.
microstructure of melted buttons. These etchants and procedure for their use are
listed in Table 4.2. Additionally, the use of ferro-fluid, a colloidal suspension of Fe3O4
particles was used to differentiate between austenite and martensite. The Fe3O4
phase. A single drop of ferro-fluid is placed on the surface of a mounted sample; the
sample is then rinsed in a bath of petroleum ether and agitated to remove the ferro-
fluid. The sample is then placed in another bath of petroleum ether and agitated
50
Table 4.2: List of etchants used in this study
without the aid of a polarizer or differential image contrast filter. Samples were
contents of button melted samples was determined using two optical analysis
51
methods. The first involved use of image analysis software ImageJ [38]to selectively
austenite and martensite present. The image analysis software was then used to
convert micrographs to black and white images, fine tune brightness and contrast to
distinguish phases, selectively threshold a given phase based on black or white level,
and determine amount of threshold area based on pixel count relative to total pixels.
Point counting was also used as a means of determining austenite and martensite
phase fraction. Point counting was performed in accordance with ASTM E562. Point
ImageJ was used superimpose a grid onto micrographs, as well as tune contrast as
necessary to give greater phase differentiation. The grid was chosen such that on
each micrograph three regions of 100 points existed for analysis. For each sample
sample. Phase fraction was determined for each sample based on the standard.
Micrographs used for phase fraction analyses were taken in the region of button
melted sample where dendritic solidification was observed as this was the most
the martensite content of the button melt samples. Despite the fact that FN is
investigators [39] to quantify the amount of martensite present in weld metals. The
magnet from a sample to a dial reading that can be converted into a FN. The FN is
This calibration results in a plot of FN versus dial readings. The calibration involves
taking dial readings from several calibration test specimens that have an established
FN. Plotting of these readings versus the FN established for each calibration
specimens resulted in a several data points which a line can be fit through. This
calibration was carried out prior to use of the Magne-Gage for this study and is
Because the Magne-Gage was designed with the intention of identifying small
amounts of ferrite in austenitic weld metal the spring can only provide enough force
by itself to free the magnet from samples with a small overall percentage of ferro-
Figure 4.4 for all counterweights used as part of this study. As can be seen, use of
because calibration blocks only existed up to 76 FN, actual samples were used with
calibration weights of 20.5g, 27g, and 35g. FN number measurements were made for
53
all button melt samples. Measurements were made until three successive readings
were obtained that deviated by no more than three dial marks, and the average of
these three readings was taken as the dial reading from which a FN was calculated.
determined by XRD. X-ray diffraction (XRD) studies were performed using a Scintag
XDS2000 diffractometer. The system has a Cu anode and 2kW sealed tube X-ray
source. The configuration was the theta-theta type goniometer. The X-ray detector
54
was an i-Ge solid-state type. Accelerating voltage and operating current were 45 kV
and 20 mA respectively.
Measurements were made over 2θ values from 37º to 87º at a rate of one
degree per minute. These values were chosen because they allowed for
measurement of four distinct peaks, the austenite (200) and (220) peaks, and the
calculated based on the direct comparison method [41]. This method relates
martensite phase fraction (cα) to austenite phase fraction (cγ) by the equation:
I R c
I R c
Where Iγ and Iα are the integrated intensities of a chosen peaks of austenite and
values are calculated by first taking the total number
martensite respectively. These
of counts in a given peak from a 2θ value safely on one side of that peak to a 2θ
value safely on the other side. Following that, the counts per second value
associated with the starting and finishing 2θ values are each multiplied by half the
number of second associated with scanning from one value of 2θ to the other. These
numbers are then added together and represent an estimate of the total number of
subtracted from the sum of the counts across a given peak as calculated initially to
give a measure of integrated intensity for a given peak. Rγ and Rα depend on the
Bragg angle, hkl, and the phase in question and can be calculated from data
published in the reference above. Once integrated intensities and R-values have
55
been calculated for crystallographic planes of interest, a value for the ratio of cγ to cα
can be calculated from the above expression. This value can then be used to
c c 1
Comparisons were made between each of the peaks measured resulting in a total of
four values was taken as the martensite phase
four values. The average of these
fraction.
martensite and austenite. All hardness indents were made with 25g load in order to
Material compositions from Table 4.1 above were collected from various
sources. Materials for this study were largely obtained from past research
conducted members of the Welding and Joining Metallurgy Group at the Ohio State
from the vendor. Other base materials were sent out for chemical analysis.
Worthington Steel and MSI performed chemical analysis of base materials. Chemical
56
Balmforth [23] demonstrated that the composition of buttons could be
calculated accurately based on dilution. This method assumes that the amount of
value of X in the steel alloy and nickel-base alloy used to create it by:
regression model fit. Normality of response data was achieved by applying the
FN 70
FN Norm sinh
20
The equation above was developed by assessment of the normality of FN data based
on mean and standard deviation. Observation of the data scatter led to the choice of
were determined by trial and error until the data with highest normality was
achieved. Both original and normalized response data was analyzed in Minitab 15
statistical software package using regression techniques in order to fit a model that
development.
welds was developed based on regression models determined previously. The terms
in the regression equation were separated into equivalency formulae based on their
tendency to increase or reduce FN. Those terms that had a positive effect on FN
they tended to promote austenite. Martensite equivalent was placed on the x-axis
and austenite equivalent on the y-axis. Boundary lines were determined based on
FN expected to give approximately 0, 50, and >95 volume percent martensite, and
plotted on the diagram. Diagrams were tested for relative conformity with FN data
58
5. RESULTS AND DISCUSSIONS
CHAPTER 5
Initial button melt samples as plotted in Figure 4.2 were prepared using
the fully martensitic, fully austenitic, and duplex austenite plus martensite regions
melted samples. Application of Etchant No. 2 from Table 4.2 proved the most
successful at distinguishing martensite from austenite. The results of this and all
other etchants utilized as part of this study varied greatly from sample to sample
based on the different compositions of each button melt sample. Additionally, with
the exception of Etch No. 4, etching response was very non-uniform across the
containing nickel-base filler metals, such as C-276 and 625, the TCP phases σ and P
may be present in small amounts. This is illustrated in Figure 5.1 where small
Figure 5.1: TCP phase within interdendritic regions of 8630-625 85% dilution button melt sample (Etch 2)
60
Figure 5.2: : Predominantly austenitic microstructure of F22-800H 55% dilution button melt sample (Etch 4)
100% dilution 8630 base metal that would be expected to be fully martensitic. Small
islands of retained austenite are present, as is typical of all highly diluted samples.
Figure 5.4 depicts the typical microstructure of a duplex austenite plus martensite
sample. The martensite is present at dendrite cores appears dark, while the
austenite present within the interdendritic areas appears light. During solidification,
the liquid will become enriched in solute, thereby lowering the Ms temperature
relative to the first to solidify dendrite cores. As the sample cools, those regions that
61
are solute lean, the dendrite cores, whose Ms and Mf temperatures remain above
room temperature transform to martensite. The solute rich regions, liquid which
will transform to austenite upon cooling, whose Ms and Mf temperatures are below
Appendix A.
Figure 5.3: Predominantly martensitic structure with pockets of retained austenite in 100% 8630 button melt
sample (Etch 2)
62
Figure 5.4: Duplex martensite + austenite microstructure of P92-800H 74% button melt sample (Etch 5)
4.6. Measurements are presented in Figure 5.5 and Figure 5.6.The results of the
measurements are consistent with the expectation that these samples illustrate two
distinct phases within their microstructure. The relatively high value of 393 HV is
also within the hardness range that would be expected for martensite. Although the
value of 261 may be a little less than expected, it should be noted that the size of the
increased amount of martensite within the sample, and the increased carbon within
Figure 5.5: Vickers hardness indents in selected phases of F22-800H 63% dilution button melt sample
64
Figure 5.6: : Vickers hardness indents in selected phases of F22-800H 60% dilution button melt sample
had sufficient difference in phase contrast that phase fraction quantification could
be achieved by means of the first method discussed in 4.3.3. Achieving the level of
etchants were tried but they all gave similar results. The lath structure of martensite
ferro-magnetic nature of nickel. Based on this, and phase fraction results which
were inconsistent with results obtained using the Magne-Gage (as discussed below),
the use of image analysis as the primary method of phase fraction quantification
This project was concerned with creating samples that by prediction of the
66
Schaeffler diagram would result in microstructures that were free of ferrite. This
was a key aspect of the choice of the Magne-Gage for martensite detection. As
magnetic response from austenite within button melt samples would be expected to
the Magne-Gage would prove difficult [23]. The additional ferromagnetic influence
for the magnetic response of the sample and to what magnitude each was
responsible. No ferrite was detected during microscopy of button melt samples used
as part of this study so this was not believed to be an issue that would preclude the
used to demonstrate the ferrite free final microstructure of button melt samples.
The composition of sample P92 – 800H 88% dilution has a calculated chromium
content of 9.8 wt% and a calculated nickel content of 3.9 wt%. This represents the
largest chromium to nickel ratio of any button created during this project. The large
chromium to nickel ratio suggests that this sample is the most likely to form delta
JMatPro was used to calculate phase fraction versus temperature data for this
67
during equilibrium conditions this sample will solidify as delta ferrite, but will
within the delta ferrite during solidification may enrich it to the point that regions
shown in Figure 5.9, delta ferrite will become enriched in both chromium and nickel
as solidification proceeds. The last to solidify areas will contain approximately 9.6
wt% chromium and 3.8 wt% nickel. The graphical output from JMatPro of phase
100% austenite will still occur. This demonstrates that it is unlikely delta ferrite will
Figure 5.8: JMatPro graphical output of solidification nature of compositional makeup of P92-800H 88% dilution
68
Figure 5.9: JMatPro graphical output of nickel and chromium segregation within δ ferrite during solidification
Figure 5.10: JMatPro graphical output of solidification nature of enriched δ ferrite in P92-800H 88% dilution
69
The results of the Magne-Gage are presented graphically in Figure 5.11.
Qualitatively the results agree with the prediction provided by the Schaeffler
increase in FN.
three regions within each curve. Moving from high to low dilution each curve seems
followed by a level of moderate slope where data exists. This can be explained by
microstructural makeup. Kotecki [40] has shown that increasing the amount of iron
in fully ferritic samples will result in an increase in FN, with FN approaching 200
being observed for commercially pure iron. The effect of composition on magnetic
results in little magnetic response. The Magne-Gage will have difficulty resolving
any change in magnetism due to alloy content when the microstructure is primarily
austenite, so the slope of the curve would be expected to be essentially zero until
martensite is present within the sample. Once martensite is encountered within the
sample the FN will raise steadily with increasing dilution as the amount of
70
materials will retain some austenite to room temperature even at 100% base
material. This results in the FN at high level being affected somewhat by small
changes in the amount of retained austenite from sample to sample; but more
consistently by the increasing level of iron present in the sample as the dilution
increases. Once the level of retained austenite reaches these levels (believed to be
around 2-5 vol%) the FN increases to a lesser degree with increasing dilution. More
71
5.5 X-ray Diffraction
martensite phase fraction for selected buttons, such that a correlation between
martensite phase fraction and FN could be determined. The results of XRD analysis
are presented in Table 5.1 below, and are superimposed on the previously
presented FN data in Figure 5.12 XRD results from highly diluted samples are in line
peak to peak were seen in two samples as evidenced by the large standard deviation
observed in analysis of F22-800H 66% dilution and 63% dilution. This could be the
result of due to texture effects present in button melt samples due to preferred
solidification that occurs from button melting apparatus. The result of XRD analysis
between 54.3-76.1 volume percent in samples with FN of around 2. This result has
spectra from these samples, X80-C276 73% dilution and F22-800H 60%, are
presented in Figure 5.13 & Figure 5.14 respectively. Examination of the spectra from
these samples reveals that distinct peaks are present for both martensite and
austenite. This confirms the presence of martensite in these samples, despite the
possible inaccuracy of the quantitative result presented in Table 5.1. This result is
preceding sections.
72
Table 5.1: X-ray diffraction and Magne-Gage analysis results for selected samples
Figure 5.12: FN versus dilution of button melt samples determined by Magne-Gage with superimposed XRD
analysis values
73
Figure 5.13: X-ray diffraction spectra of X80-C276 73% dilution button melt sample
74
Figure 5.14: X-ray diffraction spectra of F22-800H 60% dilution button melt sample
XRD analysis in the hope that the results would correlate better to FN data. Point
75
Table 5.2: Point Counting, X-ray diffraction and Magne-Gage analysis results for selected samples
predicted from XRD analysis as shown in Table 5.2. Martensite phase fractions are
slightly greater than those predicted by XRD in most cases. This is a result of heavy
segregation of alloying elements to the last to solidify region at the top of the button
melt samples. Due to this segregation, coupled with reduced cooling rates
experienced in this area, a fully austenitic will be present to some varying extent.
The contribution of this area to the overall phase fraction would be accounted for in
XRD analysis, but not point count analysis, which was confined to the region of
intended for detection of ferrite respond less when used on samples consisting of
also play a role in the reduced magnetic response of button melt samples. The
addition of alloying elements can be shown to reduce the Curie temperature of pure
iron, which could account for the weakened magnetic response of samples tested as
The sample size used for FN number testing may also play a role in the
reduced magnetic response exhibited by certain button melt samples. The Magne-
Gage was designed for samples whose total mass is approximately 4 grams, about
twice the size of the sample size used for testing. This may result in some lowering
of the FN. The presence of retained austenite between martensite laths may also
contribute to FN response. This would cause the martensite phase fraction reported
by point counting to be higher than the true martensite phase fraction within the
77
sample. The discrepancies reported between the characterization techniques used
XRD and point counting data to define an upper bound and the assumption that a
ferrite number of zero corresponds to a sample free of martensite. The upper bound,
where greater than 95% volume fraction martensite is present was predicted to
occur at greater than 135 FN. For samples examined as part of this study, martensite
Martensite % 0.7037FN
The equation is represented graphically in Figure 5.15. Ferrite number data from
button melt samples that were 100% steel base material are plotted in the top right
hand corner of this diagram. These buttons would be expected to represent as close
to a fully martensitic microstructure. Data from XRD and point counting analysis are
also plotted on this diagram. As can be seen a line drawn through these points
would intersect the y-axis at a value much greater than zero volume percent
martensite. This is due to the reasons discussed in section 5.6. As a result the
78
Figure 5.15: Plot of martensite phase fraction vs. FN for this study
4.8. The regression model chosen for diagram development yielded the following
This model was chosen based on the highest overall R2 value of those tested
combined with the low average p-values of its terms. Note that this equation reports
79
FN in the normalized form, so use of the equation reported in 4.8 used to normalize
values. This has been done to calculate predicted values of FN based on composition
for button melt samples such that a plot of measured FN versus predicted FN can be
160.0
140.0
120.0
100.0
Measured FN
80.0
60.0
40.0
20.0
0.0
-5.0 15.0 35.0 55.0 75.0 95.0 115.0 135.0 155.0
Predicted FN
Although this equation features mostly two-factor interactions that drive its
the same as models developed using only single terms. Niobium was the only
80
composition data. In the above formula it is seen to have the same effect, albeit in
concentration, with manganese, chromium and nickel also playing key roles in
many ways, just another form of an equation for Ms. Austenite forms when alloying
and fully austenitic structures are formed. The above equation also demonstrates
this behavior. Alloy additions for the most part tend to decrease the FN, just as they
do the Ms temperature, and eventually the FN drops below zero, indicating a fully
austenitic sample.
The above equation was separated into those terms that increased and
equivalency formulae:
formula above is plotted on the x-axis and the austenite equivalency from above on
81
the y-axis. Borders of the duplex martensite plus austenite region have been created
approximately 0, 50, and <95 volume percent martensite. These numbers were
martensite phase fraction, as well as the position these values are found on curves in
Figure 5.12. The diagram is presented again in Figure 5.18 with the addition of FN
data superimposed on it. The vast majority of the data falls into the expected region
Figure 5.17: Preliminary constitution diagram for microstructure prediction in dissimilar metal welds
82
Figure 5.18: Preliminary constitution diagram for microstructure prediction in dissimilar metal welds with FN data
joints. It has been created from limited data that therefore may not accurately
predict microstructure in all steel and nickel base alloy systems. Furthermore,
83
Specific attention should be brought to the gray band that borders the
Gage. This has led to the addition of a gray band to the diagram that represents the
In its present form this diagram should be used as an analytical tool only.
This diagram is not qualified for use in weldments intended for actual service, and
should therefore not be used for this purpose. The use of the diagram as a means of
welds between carbon and low-alloy steels and solid-solution nickel base filler
materials is an example of where this diagram may provide benefits over its
84
6. CONCLUSIONS
CHAPTER 6
CONCLUSIONS
dissimilar metal welds between carbon and low-alloy steels and nickel-base
carbon and low-alloy steels and nickel-base filler materials that have
compositions within the ranges of the data used to create the diagram.
3. The diagram is limited by the fact that it was created from ferrite number
and therefore variation can exist in ferrite number response of two samples
4. The accuracy of the diagram near the austenite – austenite plus martensite
border is not precise. The ferrite number data generated from Magne-Gage
has been drawn on the new diagram to compensate for this effect.
86
REFERENCES
87
alloy 625. Metallurgical and Materials Transactions A, 1995. 26(5): p.
1273-1286.
14. Savage, W.F., E.F. Nippes, and G.M. Goodwin, Effect of Minor Elements on
Hot Cracking Tendencies of Inconel 600. Welding Journal.Vol.56, 1977.
56(8): p. 245s-253s.
15. Cieslak, M., T. Headley, and A. Romig, The welding metallurgy of
HASTELLOY alloys C-4, C-22, and C-276. Metallurgical and Materials
Transactions A, 1986. 17(11): p. 2035-2047.
16. Cieslak, M.J., The Welding and Solidification Metallurgy of Alloy 625. NI,
SP, Nickel base alloys, Superalloys, 1991. 70(2): p. 49s-56s.
17. Noecker Ii, F.F. and J.N. Dupont, Metallurgical Investigation into Ductility
Dip Cracking in Ni-Based Alloys: Part II. Welding Journal, 2009. 88(3): p.
62s-77s.
18. Noecker Ii, F.F. and J.N. DuPont, Metallurgical Investigation into Ductility
Dip Cracking in Ni-Based Alloys: Part I. Welding Journal, 2009. 88(1): p.
7S-20S.
19. Ramirez, A.J. and J.C. Lippold, High temperature behavior of Ni-base weld
metal. Materials Science and Engineering: A, 2004. 380(1-2): p. 245-
258.
20. Lippold, J.C. and N.E. Nissley, Ductility-Dip Cracking in High Chromium,
Ni-Base Filler Metals. 2008. p. 409-425.
21. Oates, W.R., ed. Welding Handbook: Volume 3 - Materials and
Applications - Part 1. 8th ed. Vol. 4. 1998, American Welding Society:
Miami.
22. Olson, D.L., PREDICTION OF AUSTENITIC WELD METAL MICROSTRUCTURE
AND PROPERTIES. Welding Journal (Miami, Fla), 1985. 64(Compendex):
p. 281. s-295. s.
23. Balmforth, M.C., -, Development of a ferritic-martensitic stainless steel
constitution diagram. 1998. p. xv, 140 leaves.
24. Strauss, B. and E. Maurer, Die hochlegierten chromnickelstahle als
michtrostende stahle. Kruppsche Monatshefte, 1920. 1(8): p. 129-146.
25. Scherer, R., G. Riedrich, and G. Hoch, Einfluss eines gahaltes an ferrit in
austenitischen chrom-nickel-stahlen auf den kornzerfall. Archiv fur das
Eisenhuttenwesen, 1939. 13: p. 52-57.
26. Newell, H.D. and M. Fleischmann, Hot rolled metal article and method of
making same., U.S. Patent, Editor. 1938.
27. Field, A.L., F.K. Bloom, and G.E. Linnert, Development of armor welding
electrodes: relation to the composition of austenitic (20Cr-10Ni)
electrodes to the physical and ballistic properties of armor weldments.
OSRD. 1943.
28. Campbell, H.C. and J.R.D. Thomas, Effect of alloying elements on tensile
properties of 25-20 weld metal. Welding Journal, 1946. 25(11): p. 760-
768.
88
29. Thomas, J.R.D., A constitution diagram application to stainless weld
metal. Schweizer Archiv fur Angewandte Wissenschaft und Technik,
1949. 1: p. 3-24.
30. Schaeffler, A.L., Welding dissimilar metals with stainless electrodes. Iron
Age, 1948. 162(1): p. 72-79.
31. Eichelman, J.G.H. and F.C. Hull. Effect of composition on temperature of
spontaneous transformation of austenite to martensite in 18-8-type
stainless steel. in American Society for Metals -- Meeting, Oct 20-24 1952.
1952. Cleveland, OH, United States: American Society for Metals.
32. Andrews, K.W., Empirical formulae for calculation of some
transformation temperatures. Iron and Steel Institute -- Journal, 1965.
203(Part 7): p. 721-727.
33. Self, J.A., D.K. Matlock, and D.L. Olson, EVALUATION OF AUSTENITIC Fe-
Mn-Ni WELD METAL FOR DISSIMILAR METAL WELDING. Welding Journal
(Miami, Fla), 1984. 63(Compendex): p. 282. s-288. s.
34. Beres, L., Proposed modification to Schaeffler diagram for chrome
equivalents and carbon for more accurate prediction of martensite
content. Welding Journal (Miami, Fla), 1998. 77(Compendex): p. 273-s-
276-s.
35. Nelson, T.W., J.C. Lippold, and M.J. Mills, Nature and evolution of the
fusion boundary in ferritic-austenitic dissimilar metal welds - Part 2: on-
cooling transformations. Welding Journal (Miami, Fla), 2000.
79(Compendex): p. 267s-277s.
36. GITTOS, M.F. and T.G. GOOCH, The interface below stainless steel and
nickel-alloy claddings. Vol. 71. 1992, Miami, FL, ETATS-UNIS: American
Welding Society.
37. Lundin, C.D., DISSIMILAR METAL WELDS - TRANSITION JOINTS
LITERATURE REVIEW. Welding Journal (Miami, Fla), 1982.
61(Compendex): p. 58. s-63. s.
38. ImageJ: <http://rsbweb.nih.gov/ij/>.
39. Kotecki, D.J., Martensite boundary on the WRC-1992 diagram. Welding
Journal (Miami, Fla), 1999. 78(Compendex): p. 180s-192s.
40. Kotecki, D.J., EXTENSION OF THE WRC FERRITE NUMBER SYSTEM. 1982. V
61(Compendex): p. 352. s-361. s.
41. Cullity, B.D., Elements of x-ray diffraction / B. D. Cullity. 2d ed ed. 1978,
Reading, Mass. :: Addison-Wesley Pub. Co. xii, 555 p. :.
42. Jerry Gould, W.P., James Cruz, An Examination of Electric Servo-Guns for
the Resistance Spot Welding of Complex Stack-Ups, in Sheet Metal
Welding Conference XIV. 2010.
43. Ramasamy, S., J. Gould, and D. Workman, Design-of-experiments study to
examine the effect of polarity on stud welding. Welding Journal (Miami,
Fla), 2002. 81(Compendex): p. 19/S-26/S.
44. Hecker, S.S., et al., EFFECTS OF STRAIN STATE AND STRAIN RATE ON
DEFORMATION-INDUCED TRANSFORMATION IN 304 STAINLESS STEEL - 1.
89
MAGNETIC MEASUREMENTS AND MECHANICAL BEHAVIOR. Metallurgical
transactions. A, Physical metallurgy and materials science, 1982. 13
A(Compendex): p. 619-626.
45. Talonen, J., P. Aspegren, and H. Hanninen, Comparison of different
methods for measuring strain induced -martensite content in austenitic
steels. Materials Science and Technology, 2004. 20(Compendex): p.
1506-1512.
90
A. ADDITIONAL MICROGRAPHS
APPENDIX A
ADDITIONAL MICROGRAPHS
Figure A.1: Predominantly martensitic microstructure of 100% dilution 8630 steel (Etch 5)
91
Figure A.2: 8630-625 85% dilution microstructure (Etch 5)
Figure A.11: F22-800H 55% dilution microstructure showing possible TCP phases (Etch 4)
96
B. BUTTON MELT SAMPLE DATA
APPENDIX B
Button melt sample composition and ferrite number data are reported in
Table B.1 below. The dilution of a given button is listed in the first column between
97
Table B.1: Button melt sample composition and ferrite number data
Ferrite Number
Material
Mn
Mo
Nb
Cu
Co
Cr
Ni
Al
W
Ti
Si
V
P
C
S
8630 0.299 0.856 0.009 0.008 0.257 0.770 0.782 0.363 0.002 0.004 0.000 0.035 0.009 0.198 0.000 0.008
100.00% 0.299 0.856 0.009 0.008 0.257 0.770 0.782 0.363 0.002 0.004 0.000 0.035 0.009 0.198 0.000 0.008 151.4
89.99% 0.271 0.774 0.008 0.007 0.240 7.137 2.934 1.195 0.368 0.025 0.000 0.042 0.008 0.180 0.000 0.007 139.9
85.41% 0.258 0.736 0.008 0.007 0.232 10.056 3.921 1.577 0.536 0.034 0.000 0.045 0.008 0.171 0.000 0.007 117.6
80.35% 0.244 0.695 0.008 0.007 0.223 13.271 5.007 1.997 0.721 0.045 0.000 0.048 0.007 0.162 0.000 0.006 33.0
77.08% 0.234 0.668 0.008 0.006 0.217 15.351 5.711 2.269 0.840 0.052 0.000 0.050 0.007 0.156 0.000 0.006 0.5
51.14% 0.161 0.456 0.006 0.005 0.172 31.863 11.292 4.427 1.789 0.106 0.000 0.068 0.005 0.108 0.000 0.004 0.5
625 0.017 0.037 0.003 0.001 0.084 64.40 22.29 8.680 3.660 0.212 0.000 0.102 0.000 0.014 0.000 0.000
A553 0.044 0.582 0.004 0.001 0.189 8.970 0.140 0.010 0.001 0.004 0.004 0.018 0.001 0.034 0.050 0.012
100.00% 0.044 0.582 0.004 0.001 0.189 8.970 0.140 0.010 0.001 0.004 0.004 0.018 0.001 0.034 0.050 0.012 154.8
94.86% 0.044 0.697 0.004 0.001 0.181 12.127 1.212 0.010 0.115 0.006 0.004 0.017 0.001 0.037 0.047 0.014 152.7
89.05% 0.044 0.826 0.004 0.001 0.172 15.697 2.424 0.009 0.243 0.009 0.004 0.016 0.001 0.040 0.045 0.016 144.7
84.95% 0.043 0.917 0.004 0.001 0.165 18.218 3.280 0.009 0.334 0.011 0.003 0.015 0.001 0.042 0.042 0.018 119.6
83.18% 0.043 0.957 0.004 0.001 0.162 19.303 3.649 0.009 0.373 0.012 0.003 0.015 0.001 0.043 0.042 0.018 84.2
82.06% 0.043 0.982 0.004 0.001 0.160 19.989 3.882 0.009 0.397 0.012 0.003 0.015 0.001 0.044 0.041 0.019 36.1
80.10% 0.043 1.025 0.004 0.001 0.157 21.192 4.290 0.009 0.440 0.013 0.003 0.014 0.001 0.045 0.040 0.020 0.5
Alloy 82 0.040 2.810 0.004 0.001 0.030 70.40 21.00 0.005 2.210 0.050 0.000 0.000 0.000 0.090 0.000 0.050
98
Table B.1: Button melt sample composition and ferrite number data (Continued)
Ferrite Number
Material
Mn
Mo
Nb
Cu
Co
Cr
Ni
Al
W
Ti
Si
V
P
C
S
A508 0.182 1.449 0.005 0.002 0.181 0.920 0.125 0.514 0.002 0.002 0.000 0.016 0.004 0.040 0.000 0.025
100.00% 0.182 1.449 0.005 0.002 0.181 0.920 0.125 0.514 0.002 0.002 0.000 0.016 0.004 0.040 0.000 0.025 153.1
95.06% 0.174 1.415 0.005 0.002 0.177 3.908 1.555 0.491 0.042 0.012 0.000 0.018 0.004 0.039 0.000 0.024 149.2
87.02% 0.161 1.360 0.005 0.002 0.172 8.766 3.881 0.452 0.108 0.029 0.000 0.020 0.003 0.037 0.000 0.022 136.1
82.43% 0.154 1.328 0.005 0.002 0.169 11.539 5.208 0.431 0.146 0.039 0.000 0.022 0.003 0.036 0.000 0.021 116.5
79.88% 0.149 1.310 0.005 0.002 0.167 13.085 5.948 0.419 0.167 0.044 0.000 0.023 0.003 0.036 0.000 0.020 61.8
77.99% 0.146 1.297 0.005 0.002 0.165 14.225 6.494 0.410 0.182 0.048 0.000 0.023 0.003 0.036 0.000 0.019 2.0
FM52M 0.020 0.760 0.003 0.001 0.110 61.37 29.06 0.040 0.820 0.210 0.000 0.050 0.000 0.020 0.000 0.000
X80 0.023 1.370 0.014 0.004 0.113 0.013 0.040 0.010 0.059 0.042 0.000 0.031 0.005 0.033 0.000 0.003
100.00% 0.023 1.370 0.014 0.004 0.113 0.013 0.040 0.010 0.059 0.042 0.000 0.031 0.005 0.033 0.000 0.003 156.0
84.85% 0.020 1.212 0.013 0.004 0.141 8.761 2.466 2.433 0.077 0.037 0.002 0.031 0.007 0.037 0.508 0.033 147.0
77.98% 0.019 1.141 0.012 0.004 0.154 12.726 3.566 3.532 0.085 0.035 0.003 0.031 0.008 0.039 0.738 0.046 133.5
75.82% 0.018 1.118 0.012 0.004 0.158 13.976 3.913 3.878 0.088 0.034 0.003 0.031 0.009 0.040 0.810 0.051 110.7
73.90% 0.018 1.099 0.012 0.005 0.162 15.080 4.219 4.184 0.090 0.034 0.004 0.031 0.009 0.040 0.874 0.054 65.9
73.01% 0.018 1.089 0.012 0.005 0.163 15.599 4.363 4.328 0.091 0.033 0.004 0.031 0.009 0.040 0.904 0.056 2.3
70.13% 0.017 1.059 0.012 0.005 0.169 17.261 4.824 4.788 0.095 0.032 0.004 0.031 0.009 0.041 1.001 0.062 0.3
C276 0.003 0.330 0.007 0.006 0.300 57.75 16.05 16.00 0.179 0.010 0.014 0.030 0.020 0.060 3.350 0.200
99
Table B.1: Button melt sample composition and ferrite number data (Continued)
Ferrite Number
Material
Mn
Mo
Nb
Cu
Co
Cr
Ni
Al
W
Ti
Si
V
P
C
S
F22 0.141 0.430 0.011 0.022 0.292 0.128 2.420 1.030 0.004 0.002 0.000 0.025 0.017 0.190 0.000 0.010
100.00% 0.141 0.430 0.011 0.022 0.292 0.128 2.420 1.030 0.004 0.002 0.000 0.025 0.017 0.190 0.000 0.010 143.1
82.14% 0.128 0.480 0.010 0.018 0.308 5.625 5.721 0.891 0.010 0.061 0.000 0.071 0.014 0.156 0.000 0.008 146.8
78.82% 0.126 0.489 0.010 0.018 0.311 6.644 6.333 0.865 0.012 0.071 0.000 0.079 0.013 0.150 0.000 0.008 145.4
76.15% 0.124 0.497 0.010 0.017 0.313 7.469 6.828 0.844 0.013 0.080 0.000 0.086 0.013 0.145 0.000 0.008 137.6
66.05% 0.116 0.525 0.009 0.015 0.322 10.574 8.693 0.765 0.016 0.113 0.000 0.112 0.011 0.126 0.000 0.007 90.3
63.20% 0.114 0.533 0.009 0.014 0.324 11.453 9.221 0.743 0.017 0.123 0.000 0.119 0.011 0.120 0.000 0.006 54.2
60.17% 0.112 0.542 0.009 0.014 0.327 12.385 9.781 0.719 0.018 0.133 0.000 0.127 0.010 0.114 0.000 0.006 2.4
54.87% 0.108 0.556 0.009 0.013 0.332 14.015 10.760 0.678 0.020 0.150 0.000 0.140 0.009 0.104 0.000 0.005 0.3
800H 0.068 0.710 0.006 0.001 0.380 30.90 20.90 0.250 0.040 0.330 0.000 0.280 0.000 0.000 0.000 0.000
P92 0.120 0.430 0.010 0.001 0.280 0.120 8.270 0.380 0.010 0.010 0.045 0.006 0.174 0.190 1.615 0.000
100.00% 0.120 0.430 0.010 0.001 0.280 0.120 8.270 0.380 0.010 0.010 0.045 0.006 0.174 0.190 1.615 0.000 139.2
87.82% 0.114 0.464 0.010 0.001 0.292 3.868 9.808 0.364 0.014 0.049 0.039 0.039 0.153 0.167 1.418 0.000 141.0
84.93% 0.112 0.472 0.009 0.001 0.295 4.757 10.173 0.360 0.015 0.058 0.038 0.047 0.148 0.161 1.372 0.000 133.2
74.20% 0.107 0.502 0.009 0.001 0.306 8.062 11.529 0.346 0.018 0.093 0.033 0.077 0.129 0.141 1.198 0.000 90.2
68.71% 0.104 0.518 0.009 0.001 0.311 9.750 12.222 0.339 0.019 0.110 0.031 0.092 0.120 0.131 1.110 0.000 33.4
62.91% 0.101 0.534 0.009 0.001 0.317 11.538 12.955 0.332 0.021 0.129 0.028 0.108 0.109 0.120 1.016 0.000 0.4
800H 0.068 0.710 0.006 0.001 0.380 30.90 20.90 0.250 0.040 0.330 0.000 0.280 0.000 0.000 0.000 0.000
Min 0.017 0.456 0.004 0.001 0.141 0.770 0.782 0.009 0.002 0.004 0.000 0.014 0.001 0.036 0.000 0.000
Max 0.299 1.415 0.013 0.018 0.332 31.863 12.955 4.788 1.789 0.150 0.039 0.140 0.153 0.198 1.418 0.062
100