You are on page 1of 7

Journal of Alloys and Compounds 460 (2008) 379–385

High temperature diffusion induced liquid phase joining


of a heat resistant alloy
N.P. Wikstrom, A.T. Egbewande, O.A. Ojo ∗
Department of Mechanical and Manufacturing Engineering, University of Manitoba, Winnipeg, Manitoba, Canada R3T 5V6
Received 15 June 2007; accepted 19 June 2007
Available online 22 June 2007

Abstract
Transient liquid phase bonding (TLP) of a nickel base superalloy, Waspaloy, was performed to study the influence of holding time and temperature
on the joint microstructure. Insufficient holding time for complete isothermal solidification of liquated insert caused formation of eutectic-type
microconstituent along the joint centerline region in the alloy. In agreement with prediction by conventional TLP diffusion models, an increase
in bonding temperature for a constant gap size, resulted in decrease in the time, tf, required to form a eutectic-free joint by complete isothermal
solidification. However, a significant deviation from these models was observed in specimens bonded at and above 1175 ◦ C. A reduction in
isothermal solidification rate with increased temperature was observed in these specimens, such that a eutectic-free joint could not be achieved by
holding for a time period that produced complete isothermal solidification at lower temperatures. Boron-rich particles were observed within the
eutectic that formed in the joints prepared at the higher temperatures. An overriding effect of decrease in boron solubility relative to increase in its
diffusivity with increase in temperature, is a plausible important factor responsible for the reduction in isothermal solidification rate at the higher
bonding temperatures.
© 2007 Elsevier B.V. All rights reserved.

Keywords: High temperature alloys; Metals

1. Introduction nique for difficult-to-weld superalloys due to its technological


and economical advantages. The major advantage of this tech-
Heat resistant nickel base superalloys are used for manufac- nique over conventional brazing is that the process can achieve
turing hot section components of aero and land-based gas turbine joints free of eutectic-type microconstituents that are usually
engines, due to their remarkable high temperature mechanical detrimental to mechanical properties. Nevertheless, optimiza-
properties and hot corrosion resistance. Exposure to harsh oper- tion of the process variables that control the bond microstructure
ating conditions often causes these components to be damaged is vital in achieving eutectic-free joint. A number of studies [4–8]
by cracking and surface erosion during service. It is usually have been previously done to model microstructural develop-
more economically attractive to repair damaged parts instead ment during TLP bonding. These studies normally model the
of a complete replacement. Traditional repair techniques, such isothermal solidification stage of the process using a binary
as welding are commonly used in the repair of many superalloy phase relationship between the base and filler alloy. However,
components. However, precipitation hardened nickel base super- the phase relationships that are actually encountered in multi-
alloys, like Waspaloy, containing relatively high amounts of Al component systems that are used in industrial applications may
and Ti are generally considered to be difficult to weld due to their not comply with simple binary system approximation. Recent
high susceptibility to heat-affected zone (HAZ) cracking [1,2]. work [9,10] on numerical modeling of TLP isothermal solidifi-
As a result, a process known as transient liquid phase (TLP) cation process in ternary systems has suggested that the actual
bonding [3], has evolved as an attractive alternate joining tech- phase relationships encountered during TLP bonding of multi-
component systems could result in a significant modification
of isothermal solidification behavior compared to conventional
∗ Corresponding author. Tel.: +1 204 4747972; fax: +1 204 2757507. expectations. Therefore, experimental investigation of the influ-
E-mail address: ojo@cc.umanitoba.ca (O.A. Ojo). ence of process parameters on microstructure of TLP bonded

0925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2007.06.066
380 N.P. Wikstrom et al. / Journal of Alloys and Compounds 460 (2008) 379–385

Fig. 1. (a) SEM secondary electron micrograph of joint prepared at 1100 ◦ C for 1 h. (b) SEM secondary electron micrograph of joint prepared at 1100 ◦ C for 6 h. (c)
SEM secondary electron micrograph of joint prepared at 1100 ◦ C for 8 h.

joint in multi-component alloy system is considered crucial and 8 h, respectively. The microstructure of the samples, as
and necessary for calibrating theoretical models. Therefore, the obtained by SEM operated in secondary electron imaging mode,
objective of the present work was to study the influence of hold- is shown in Fig. 1(a)–(c). As shown in Fig. 1(a) and (b), the
ing time and temperature, on microstructure of TLP bonded joint microstructure of the joint bonded for 1 and 6 h consisted of a
in Waspaloy superalloy. continuous distribution of centerline eutectic constituent. The
average width of the centerline eutectic constituent decreased
2. Experimental procedure from 36 ␮m after 1 h of bonding to 11 ␮m after 6 h. Holding
for 8 h resulted in formation of a single solid solution phase,
Wrought Waspaloy superalloy was used in this study. The chemical composi-
Fig. 1(c). SEM–EDS compositional analyses of the three main
tion of the alloy was 18.36 Cr, 14.02 Co, 3.73 Mo, 0.02 Nb, 0.02 W, 0.61 Fe, 1.49
Al, 3.12 Ti, 0.032 C, 0.005 B, balance Ni (wt.%). The filler alloy used was Nicro- phases within the eutectic, shown in Table 1, suggested them to
braz 150 foil with a composition of 15 Cr, 3.5 B, balance Ni (wt.%) at 100 ␮m be nickel-rich boride, chromium-rich boride, and nickel base
thickness. TLP bonding of the alloy was performed in a vacuum furnace at a solid solution phase whose composition was similar to the
pressure of 10−4 to 10−5 Torr, and within a temperature range of 1100–1225 ◦ C isothermally solidified ␥ solid solution region immediately adja-
for 1–8 h of holding time. Subsequent to bonding, the samples were sectioned
cent to the centerline eutectic constituent. Boron was detected
transversely by electro-discharge machining (EDM) and prepared by standard
metallographic techniques for microscopic examination. Sections of the joint
were subsequently etched in a solution of 40 mL H2 O + 480 mL HCl + 48 mL
CuCl2 for 5 s. The microstructure of the samples was examined by optical micro- Table 1
scope and by a JEOL 5900 scanning electron microscope (SEM) equipped with EDS compositional analyses of phases present within the centerline eutectic in
an ultra thin window Oxford energy-dispersive spectrometer (EDS). Waspaloy
Phase Composition (at.%)
3. Results and discussion
Al Co Cr Mo Ni Ti W

3.1. Effect of holding time on joint microstructure Cr-rich boride – – 87.87 8.39 3.50 0.24 –
Ni-rich boride 0.37 3.93 10.10 0.43 79.85 5.19 0.14
Gamma (␥) 0.86 4.81 18.93 0.83 73.33 1.23 –
To study the effect of holding time on joint’s microstruc- Ni–Ti-rich 2.02 9.31 4.23 0.59 66.25 16.87 0.19
ture, 100 ␮m gap samples were bonded at 1100 ◦ C for 1, 6,
N.P. Wikstrom et al. / Journal of Alloys and Compounds 460 (2008) 379–385 381

mal solidification under equilibrium condition, the only solid


phase which forms is the solid solution phase, and formation of
other phases is basically prevented [3]. In situations where suf-
ficient time is not allowed, any residual liquid could transform
during cooling from the bonding temperature into eutectic-type
solidification products.
The solidification transformation behavior of residual liq-
uid during TLP bonding of nickel using a Ni–B–Cr ternary
filler alloy was studied and modeled numerically using Scheil
simulation by Ohsasa et al. [11]. A ternary centerline eutectic,
consisting of Ni base solid solution phase (␥), Ni boride (Ni3 B)
and Cr boride (CrB), was predicted in the bonded specimens.
Their simulation results showed that during solidification of the
residual liquid in a sample held at 1100 ◦ C, Ni-rich ␥ phase
formed as the primary phase, followed by the eutectic reac-
tion L → ␥ + Ni3 B at 1042 ◦ C. Solidification was reported to be
completed via a ternary eutectic reaction L → ␥ + Ni3 B + CrB
at 997 ◦ C. In addition, Gale et al. [12] in their work on TLP
bonding of nickel using Ni–Si–B ternary filler alloy observed,
with the use of high temperature X-ray diffraction, a deposit of
Ni plus Ni3 B eutectic mixture in a sample joined at 1150 ◦ C for
5 min. The ternary centerline eutectic observed in the present
work, which was similar to those reported by Ohsasa et al.
[11], is believed to be formed by solidification reaction(s) dur-
ing cooling due to an insufficient diffusion time for complete
isothermal solidification to occur during holding at the bond-
ing temperature. Similar eutectic reaction product consisting of
nickel-rich boride, chromium-rich boride, and nickel base solid
solution phase were also observed in alloy IC6 (Ni–15.9 Al–7.78
Mo–0.125 B–0.05C (at.%)) specimens that were bonded with
the same Ni–Cr–B filler alloy (Fig. 3(a), (b) and (c) and
Table 2), despite the fact that the IC6 base alloy does not contain
chromium. Therefore, the result of Scheil simulation reported by
Ohsasa et al. [11] reasonably describes the solidification behav-
ior of residual liquid insert during TLP bonding of these alloys
with Ni–Cr–B filler alloy.

3.2. Effect of bonding temperature on joint microstructure

To study the effect of bonding temperature on joint’s


Fig. 2. (a) SEM–EDS spectra showing boron peak in chromium-rich boride of microstructure, 100 ␮m gap sized samples were bonded at
the centerline eutectic. (b) SEM–EDS spectra showing boron peak in nickel-rich
1100, 1145, 1175, and 1225 ◦ C, respectively, for 6 h. The SEM
boride of the centerline eutectic.
microstructure of the sample bonded at 1100 ◦ C is presented in
Fig. 1(b), while Fig. 4(a)–(c) shows that of samples bonded at
1145, 1175, and 1225 ◦ C, respectively. An incomplete isother-
in the boron-rich particles but could not be quantified due to the mal solidification of liquid interlayer occurred in the sample that
inability of the software to quantify light elements accurately
(Fig. 2(a) and (b)).
During TLP bonding, an interlayer, which is sandwiched Table 2
between the base alloys, melts and rapidly attains equilibrium at EDS compositional analyses of phases present within the centerline eutectic in
liquid–solid base alloy interfaces through a dissolution process. alloy IC6
Following this, a solid-state diffusion of melting point depres- Phase Composition (at.%)
sant element occurs from the liquid interlayer to the solid base Cr Al Ni Mo
alloy. This results in a decrease in the volume of liquid that can be
maintained at equilibrium, causing isothermal solidification to Cr-rich boride 62.88 – 24.77 12.35
Ni-rich boride 11.61 1.28 86.22 0.89
proceed inward from the solid mating surface. Due to an absence Gamma 14.55 3.12 81.54 0.79
of solute rejection at the solid–liquid interface during isother-
382 N.P. Wikstrom et al. / Journal of Alloys and Compounds 460 (2008) 379–385

Fig. 3. (a) SEM secondary electron micrograph of the centerline eutectic in alloy IC6. (b) EDS spectrum showing boron peak in chromium-rich boride in centerline
eutectic in alloy IC6. (c) EDS spectrum showing boron peak in nickel-rich boride in centerline eutectic in alloy IC6.

was bonded for 6 h at 1100 ◦ C. This resulted in the formation phase relationships between liquid interlayer and solid base
of continuously distributed eutectic constituents of nickel-rich alloy, and depend on classic solutions to Fick’s laws of diffusion.
boride, chromium-rich boride, and nickel base solid solution in According to these models, an increase in bonding temperature
the joint’s centerline. In contrast to the situation at 1100 ◦ C, a is expected to increase isothermal solidification rate, and pro-
holding time of 6 h at 1145 ◦ C resulted in a complete isothermal duce a eutectic-free joint in a reduced time, tf , compared to
solidification of liquid interlayer, as the joint was entirely free the time needed at lower temperatures. However, it has been
of centerline eutectic constituents (Fig. 4(a)). A further increase suggested by Sekerka [14] that an increase in bonding tempera-
in temperature resulted in formation of centerline eutectic con- ture may not necessarily result in a faster solidification process,
stituents in the joints bonded at 1175 and 1225 ◦ C, respectively, as the actual rate of solidification will depend upon details of
Fig. 4(b) and (c). This eutectic was present after 6 h of bonding the phase relationships in complex alloy and on diffusivities
at the respective temperatures, even though its formation was of solutes across the solid–liquid interface. In agreement with
prevented at a lower bonding temperature of 1145 ◦ C. It was this, Sinclair et al. [9,10] suggested that deviation from isother-
also observed that the eutectic product in the 1175 and 1225 ◦ C mal solidification rate predicted by conventional TLP diffusion
joints contained a phase in addition to those observed at the lower models, which are generally based on binary phase relationships,
bonding temperatures (1100 and 1145 ◦ C). SEM–EDS compo- might be encountered during TLP bonding of multi-component
sitional analysis of the extra phase, Table 1, suggested it to be a systems. It was proposed that if the solubilities and/or diffu-
Ni–Ti-rich intermetallic phase. sivities of the two diffusing solutes (normally melting point
Analytical and numerical models have been developed for depressants) capable of controlling the isothermal solidification
isothermal solidification process during TLP bonding [4–8,13]. process are very different, the isothermal solidification stage
These models were mostly based on the use of pseudo-binary could be divided into two parabolic regimes. The first would be
N.P. Wikstrom et al. / Journal of Alloys and Compounds 460 (2008) 379–385 383

Fig. 4. (a) SEM secondary electron micrograph of joint prepared for 6 h at 1145 ◦ C. (b) SEM secondary electron micrograph of joint prepared for 6 h at 1175 ◦ C. (c)
SEM secondary electron micrograph of joint prepared for 6 h at 1225 ◦ C.

dominated by the “faster” solute, and the second by the “slower” tition coefficient of Ti in nickel alloy is less than one; therefore
of the two. an increase in its concentration in the liquid phase would result
In the present work, diffusion of boron, which is essentially in a depression of solidification temperature [16]. A consider-
an interstitial atom in nickel and thus with a higher diffusivity able enrichment of this element in the liquated insert during
compared to substitutional solutes, is believed to be isother- TLP bonding at 1175 and 1225 ◦ C could, as suggested by Sin-
mal solidification rate controlling factor during TLP bonding clair et al. [9,10], be responsible for the commencement of a
at 1100 and 1145 ◦ C. Thus, an increased diffusivity of boron solidification regime characterized by a slower isothermal solid-
with an increase in bonding temperature from 1100 to 1145 ◦ C ification rate that was observed in the present work. Although,
resulted in a reduction in isothermal solidification completion manufacturers of filler alloys tend to avoid the use of solute
time. However, during the bonding process, continual interdif- elements that can retard isothermal solidification rate, a consid-
fusion of elements between the liquid insert and the base metal erable enrichment of liquated insert with base alloying elements
would cause the composition of the liquid to be continually during TLP bonding of superalloys could result in an undesir-
modified as solidification progressed. Notably, a bonding tem- able sluggish isothermal solidification rate, particularly at higher
perature of 1175 ◦ C is close to the solvus temperature of Ti-rich temperatures.
␥ precipitate particles, which are present in the matrix of solid Besides the possible role of the second solute, Ti, in reduc-
base alloy. The dissolution of the particles would allow more Ti ing the isothermal solidification rates at higher temperatures,
atoms to be available within the base alloy ␥ matrix [15], result- the decrease in solubility of boron with increase in temperature
ing in a steep concentration gradient between the base alloy ␥ above its eutectic temperature in nickel could also be another
matrix and liquid interlayer. Subsequently, this can result in a important contributing factor. According to Fick’s 2nd law of
significant enrichment of the liquid interlayer with Ti, which is diffusion, and assuming a constant diffusion coefficient:
expected to increase with an increase in bonding temperature
from 1175 to 1225 ◦ C. As shown in Fig. 2(b) and (c), the for- ∂C ∂2 C
=D 2
mation of Ni–Ti-rich intermetallic eutectic constituent within ∂t ∂x
the samples bonded at 1175 and 1225 ◦ C, respectively, sug- where ∂C/∂t is the change in solute concentration with time at
gests an enrichment of the ternary liquated interlayer with Ti a given position in the base metal, which could provide an indi-
during isothermal solidification process. The solidification par- cation of isothermal solidification rate, and D is the diffusion
384 N.P. Wikstrom et al. / Journal of Alloys and Compounds 460 (2008) 379–385

coefficient, and ∂2 C/∂x2 is the change in concentration gradient another important factor contributing to the observed reduction
with distance. At any given instant, the concentration gradient in isothermal solidification rate of the materials. It is plausible
(∂C/∂x) of a diffusing solute in the base metal is influenced by that the slower rate of isothermal solidification at the higher
the solute solubility limit and thus ∂2 C/∂x2 is also dependent temperatures (above 1175 ◦ C) was a consequence of overriding
on maximum solute solubility. Therefore, it can be seen from effect of lower solubilities of boron relative to its higher dif-
this equation that the diffusion-controlled isothermal solidifi- fusivity at these temperatures. Therefore, the decrease in boron
cation rate is not only dependent on diffusion coefficient, but solubility with increase in temperature is an important factor that
also on solubility, both of which can be influenced differently needs appropriate consideration in the selection of processing
by bonding temperature. An increase in temperature will cause parameters for optimizing TLP bonding of Waspaloy to obtain
an exponential increase in solute diffusivity according to the a high quality and reliable eutectic-free joint.
Arrhenius equation. However, according to the Ni–B phase dia-
gram the solubility of boron in nickel decreases with increase 4. Summary and conclusions
in temperature above the eutectic temperature. A reduction in
the solubility of boron at higher temperatures (1175 ◦ C and 1. Isothermal solidification of liquated Ni–Cr–B filler occurred
above) in the Waspaloy superalloy used in the present work during TLP joining of Waspaloy superalloy and the extent
could cause a decrease in the rate of change of concentration of the solidification increased with increasing holding time.
gradient (∂2 C/∂x2 ) in the base metal during bonding. This could, Incomplete isothermal solidification of the liquated insert
according to Fick’s 2nd law, reduce boron diffusion and, as such, resulted in formation of centerline eutectic-type product
decrease the rate of isothermal solidification at these higher tem- along the joint, the thickness of which increased with increase
peratures. Boron-containing particles were observed within the in gap size.
centerline eutectic product in joints prepared at the high tem- 2. Complete isothermal solidification in 100 ␮m joint, which
peratures where reduction in isothermal solidification rate with precluded the formation of the centerline eutectic, occurred
increase in temperatures was observed (Fig. 4(b) and (c)). The within 8 h and 6 h of holding time at 1100 and 1145 ◦ C,
presence of boron-rich particles within the centerline eutectic, respectively.
suggests that boron diffusion out of liquated insert was reduced 3. A deviation from conventional expectation was observed in
at these higher brazing temperatures. A decrease in solubility of joints bonded at 1175 and 1225 ◦ C. Contrary to the con-
boron in nickel from 320 ppm at 1125 ◦ C to 210 ppm at 1225 ◦ C ventional expectation, a decrease in isothermal solidification
has been previously related to reduction in isothermal solidifica- rate occurred with increase in temperature to these tempera-
tion kinetics during diffusion brazing of pure Ni, in the literature tures, which resulted in the formation of centerline eutectic
[5]. along the joints even after 6 h of holding time that produced
Likewise, based on analytical study of TLP bonding of binary a eutectic-free joint at 1145 ◦ C.
systems, Tuah-Poku et al. [7] developed the following equation 4. An increased interdiffusion induced enrichment of base
to express the dependence of isothermal solidification comple- alloying elements, particularly Ti, in the liquated insert at
tion time, t, on bonding temperature: 1175 and 1225 ◦ C and decrease in boron solubility were
suggested as plausible factors contributing to the observed
 
πWo2 2h exp(Q/RT ) reduction in isothermal solidification rate at the higher bond-
t= ing temperatures.
16Do (C␣s )2

where Wo is the initial liquid width, Do the diffusion coef- Acknowledgments


ficient, Q the activation energy, T the temperature of bonding
and C␣s is the equilibrium solidus composition of the melting Financial support for this work from the University of Man-
point depressant. They qualitatively showed that, as the tem- itoba is acknowledged and one of the authors A.T. Egbewande
perature increases, isothermals solidification completion time, is grateful to the Manitoba Government for the award of a Man-
t, would decrease via the exponential term. Nevertheless, an itoba graduate scholarship.
increase in bonding temperature simultaneously reduces the
solidus composition (solid solubility), C␣s , and, thus, will tend to References
increase the time required for complete isothermal solidification.
This indicates that time, t, may not continue to monotonically [1] S.D. Duvall, W.A. Owczarski, Weld. J. 46 (9) (1967) 423s–432s.
decrease with increase in temperature but instead will tend to [2] O.A. Ojo, N.L. Richards, M.C. Chaturvedi, Scripta Mater. 50 (2004)
641–646.
increase as the temperature reaches and exceeds a critical value [3] S.D. Duvall, W.A. Owczarski, D.F. Paulonis, Weld. J. 53 (4) (1974)
where the influence of reduced solubility overrides higher dif- 203–214.
fusivity. Enrichment of the liquated inert with Ti, capable of [4] Y. Nakao, K. Nishimoto, K. Shinozaki, C.Y. Kang, Trans. Jpn. Weld. Soc.
depressing the melting temperature, may, thus, not be solely 20 (1) (1989) 60–65.
responsible for the reduction in isothermal solidification rate [5] J.E. Ramirez, S. Liu, Weld. J. 71 (1992) 365s–375s.
[6] W.D. MacDonald, T.W. Eagar, Metall. Trans. A 29A (1998) 315–326.
during the TLP bonding in the present work. Interplay between [7] I. Tuah-Poku, M. Dollar, T.B. Massalski, Metall. Trans. A 19A (1988)
increase in diffusivity and decrease in solubility of melting 675–686.
point depressant boron with increase in temperature could be [8] Y. Zhou, W.F. Gale, T.H. North, Int. Mater. Rev. 40 (5) (1995) 181–196.
N.P. Wikstrom et al. / Journal of Alloys and Compounds 460 (2008) 379–385 385

[9] C.W. Sinclair, J. Phase Equilib. 20 (4) (1999) 361–369. [13] O.A. Ojo, N.L. Richards, M.C. Chaturvedi, Sci. Technol. Weld. Join. 9
[10] C.W. Sinclair, G.R. Purdy, J.E. Morral, Metall. Trans. A 31A (2000) (2004) 532–540.
1187–1192. [14] R.F. Sekerta, Physical Metallurgy of Metals Joining, TMS-AIME, St Louis,
[11] K. Ohsasa, T. Shinmura, T. Narita, J. Phase Equlib. 20 (3) (1999) 199– MO, 1980, pp. 1–3.
206. [15] R. Rosenthal, D.R.F. West, Mater. Sci. Technol. 15 (12) (1999) 1387–1394.
[12] W.F. Gale, E.R. Wallach, Metall. Trans. A 22A (1991) 2451–2457. [16] M.A. Taha, W. Kurz, Z. Metallkd. 72 (8) (1981) 546–549.

You might also like