Interfacial reaction of infrared brazed NiAl/Al/NiAl and
Ni
3
Al/Al/Ni
3
Al joints
T.Y. Yang
a, S.K. Wu
b,*, R.K. Shiue
caDepartment of Mechanical Engineering, Kuang Wu Institute of Technology, Taipei 112, Taiwan, Republic of China bInstitute of Materials Science and Engineering, National Taiwan University, Taipei 106, Taiwan, Republic of China cDepartment of Materials Science and Engineering, National Dong Hwa University, Hualien 974, Taiwan, Republic of China
Received 1 December 2000; accepted 24 January 2001
Abstract
The early-stage microstructural evolution of NiAl/Al/NiAl and Ni3Al/Al/Ni3Al in temperature ranges between 800 and 1200C
for 2–290 s was studied by infrared brazing. Al3Ni phase is firstly formed in both Al-rich melt and the interface between nickel
aluminide base material and Al-richmelt for specimens brazing at 800C. The solid-state interdiffusion between Al
3Ni and base
material results in the formation of the interfacial Al3Ni2phase, and further growth of Al3Ni2phase at 800C is impeded by the
Al3Ni interlayer due to its stoichiometry. For specimens brazing at 1000C, the reaction changes from hypereutectic into peritectic
reaction. The Al-rich melt dissolves more Ni atoms with the increment of brazing temperature to 1000C. The Al
3Ni2phase is now
initially formed in both the Al-rich melt and joint interface. The growth of Al3Ni2interlayer at 1000C is much faster than that at
800C. Transport of Al atoms in forming Al
3Ni2phase at 1000C is greatly increased due to the contact between Al3Ni2and liquid
Al-richmelt. The Al3Ni shown in the joint is formed upon cooling cycle of the infrared brazing. The microstructural evolution in
Ni3Al/Al/Ni3Al joint is similar to that in NiAl/Al/NiAl except for the formation of NiAl phase between Al3Ni2and Ni3Al. It can
also be attributed to the solid-state interdiffusion between Ni3Al and Al3Ni2interlayer. However, the intermediate phase Ni5Al3,
which is stable below 700C, is not found in the experiment. # 2001 Elsevier Science Ltd. All rights reserved.
Keywords:A. Intermetallics, miscellaneous; D. Microstructure
1. Introduction
Withever increasing demand of materials withbetter performance in bothcreep strengthand oxidation resis-tance, great attention has been focused in developing nickel aluminides, including NiAl and Ni3Al. Meanwhile,
many interfacial reactions withvarious Al–Ni inter-mediate phases, such as Al/Ni [1–8], Al/Ni3Al [9–11], Al/
NiAl [12] and Ni3Al/NiAl [13], have been extensively
stu-died. Most researches were focused on temperatures below the melting point of pure aluminum, 660C.
Therefore, solid-state diffusion dominated the reaction kinetics in the joint. Some experiments were performed at temperatures between 660 and 800C, so the liquid
diffusion of aluminum played an important role in the process [8,9]. In addition to the change of microstructure
in the joint, a much faster reaction rate was expected if liquid diffusion was involved in the experiment.
Infrared brazing makes use of infrared energy gener-ated by heating a tungsten filament in a quartz tube as the heating source, providing rapid heating and cooling up to 3000C/min; therefore, infrared brazing is highly
suitable in studying the mechanism of early-stage reac-tion kinetics in the joint [14–16]. With the aid of accu-rate thermal cycle control, the transient microstructural evolution of the brazed joint can easily be unveiled by this technique. According to the Al–Ni binary alloy phase diagram, two peritectic reactions can be found at 854 and 1133C [17]. The brazing of NiAl/Al/NiAl and
Ni3Al/Al/Ni3Al at 800, 1000 and 1200C may result in
different reaction kinetics. The purpose of this study is infrared brazing two nickel aluminides, NiAl/Al/NiAl and Ni3Al/Al/Ni3Al, in the temperature range between
800 and 1200C. With the aid of a fast infrared heating
rate, the early-stage of reaction paths in brazing nickel aluminides will be studied.
0966-9795/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. P I I : S 0 9 6 6 - 9 7 9 5 ( 0 1 ) 0 0 0 1 1 - 5
www.elsevier.com/locate/intermet
* Corresponding author. Tel.: 7846; fax: +886-22363-4562.
2. Experimental procedures
Two-nickel aluminides, Ni–50.0 at.% Al and Ni–24.0 at.% Al–500 ppm B, were used as base material. Master alloys were prepared by melting pure metal pellets in an induction furnace withAr protected atmosphere, fol-lowed by 1200C50 h homogenization treatments. The
homogenized NiAl and Ni3Al ingots were subsequently
cut into approximate 1083 mm specimens. All joined surfaces were polished by SiC paper up to grit 600 before infrared brazing. Pure aluminum foils (99.997 wt.%) with100 mm in thickness and approximately the same size as the base metal were used as filler metal. The aluminum foil was sandwiched between the above two base metals. Boththe base metal and aluminum foil were alternatively cleaned by an ultrasonic machine in acetone and ethyl alcohol solution prior to brazing.
There were two types of brazed joints, NiAl/Al/NiAl and Ni3Al/Al/Ni3Al, performed in the study. To
enhance the absorptivity of brazing specimens to the infrared rays, all specimens were clamped between two graphite plates, and a thermocouple was in contact with the brazing specimen. An ULVAC SINKO-RIKO RHL-P610C infrared furnace with the heating rate of 3000C/
min and Ar atmosphere was used throughout the experiment. Table 1 summaries all process variables used in the study. There is a time delay between the actual specimen temperature and programmer temperature. Based on the data of temperature recorder, there are two time periods including heating and holding time shown in the table. The brazing time specified in the following paragraphs is the actual specimen holding time in the experiment.
The brazed specimens were cut by a low speed dia-mond saw. Their cross-section were first ground by SiC papers, and subsequently polished by 0.3 mm alumina powder. The polished cross section of the brazed speci-mens was examined using a Philips XL30 scanning electron microscopy (SEM) equipped withan energy dispersive spectroscopy (EDS). Quantitative chemical analysis was performed using a JEOL JXA-8600SX
electron probe microanalyzer (EPMA) equipped witha wavelengthdispersive spectrometer (WDS).
3. Results and discussion
3.1. Microstructural evolution of the infrared brazed NiAl/Al/NiAl
Fig. 1(a)–(c) shows backscattered electron images (BEIs) of NiAl/Al/NiAl specimens brazed at 800C2 s,
1000C16 s, and 1200C2 s, respectively. Two EPMA
line scan profiles indicating Ni (upper line scan) and Al (lower line scan) are also included in the figure. Based on the EPMA quantitative chemical analysis, four phases can be identified in the joint. They are Al-rich, Al3Ni,
Al3Ni2and NiAl as displayed in Fig. 1. It can be noted
that the microstructure of the brazed joint changes pro-minently as the brazing temperature increasing from 800 to 1000C. A similar microstructure is observed as
the brazing temperature increases from 1000 to 1200C.
To explain the microstructural evolution of the joint for different brazing temperatures, an Al–Ni binary alloy phase diagram is included in Fig. 2.
A few Al3Ni islands are observed in Al-richmatrix as
illustrated in Fig. 1(a), and it can be explained by the Al– Al3Ni hypereutectic reaction at 800C. Al3Ni phase is
initially formed in both the Al-rich melt and the interface between NiAl and Al-richmelt. According to the Al–Ni binary alloy phase diagram, the liquid melt is not in contact withAl3Ni2phase, so there is no Al3Ni2phase
at the beginning of the reaction [17,18]. However, the solid state interdiffusion between Al3Ni and NiAl
inter-face results in the formation of Al3Ni2phase, and there
is no Al3Ni2phase around Al3Ni islands in the Al-rich
melt. Consequently, different formation mechanism can be expected in forming Al3Ni2 and Al3Ni phase. The
formation of Al3Ni2phase at the interface must be
fol-lowed by the formation of Al3Ni phase. There is no
con-tact between the Al3Ni2 phase and Al-rich melt as
displayed in Fig. 1(a). Withincreasing brazing time from 2 to 20 s at 800C, Al
3Ni islands in the Al-rich melt,
Al3Ni and Al3Ni2phases at the interface grow steadily,
as shown in Figs. 1(a) and 3(a). However, both Al-rich melt and Al3Ni islands are consumed if further
extend-ing the brazextend-ing time to 290 s at 800C [Fig. 3(b)]. The
continuous growthof the Al3Ni2phase at the interface
between Al3Ni and NiAl is observed in the experiment.
It indicates that the Al3Ni2 phase is more stable than
Al3Ni phase in the NiAl/Al/NiAl brazing.
As the brazing temperature increases from 800 to 1000C, the reaction changes from hypereutectic into
peritectic reaction as demonstrated in Fig. 2. In sucha case, the Al-rich melt dissolves more Ni atoms as increas-ing brazincreas-ing temperature to 1000C. Contrary to the
pre-vious case, the Al3Ni2phase is firstly formed in both the
Table 1
Process variables used in infrared brazing two-nickel aluminides NiAl/Al/NiAl (heating+holding) Ni3Al/Al/Ni3Al (heating+holding) 800C 25+2 s 25+2 s 25+20 s 25+20 s 25+290 s 25+110 s 1000C 30+6 s 30+2 s 30+16 s 30+16 s 30+46 s 30+96 s 30+96 s 1200C 49+2 s –
Al-richmelt and joint interface. Moreover, the growth of Al3Ni2 at the bonding interface is much more
pro-minent than that of Al3Ni2in Al-richmelt as shown in
Fig. 1(b). The transport of Ni atoms are impeded by the thick Al3Ni2 phase at interface, so the growth rate of
Al3Ni2phase at the interface is much faster than that at
Al-melt. It is also observed that the growth of interfacial Al3Ni2at 1000C is much faster than that at 800C, as
compared between Fig. 1(b) and Fig. 3(a). As discussed earlier, the formation of Al3Ni2 phase at 800C is
caused by solid-state interdiffusion of reacting elements, Al and Ni. It is expected that the transport of Al atoms in forming Al3Ni2phase at 1000C is expedited by the
contact between Al3Ni2and liquid Al-richmelt.
The Al3Ni phase shown in Fig. 1(b) is formed upon
the cooling cycle of infrared brazing. When the melt cools down to peritectic reaction at 854C, the peritectic
reaction will be proceeded as below. L þ Al3Ni2 ! Al3Ni
This can explain that some Al3Ni2 islands are
enclosed by Al3Ni phase as shown in Fig. 1(b). Since the
formation of Al3Ni solely depends upon cooling cycle of
brazing, the size of Al3Ni remains nearly the same as the
brazing time increases from 6 s to 46 s, as illustrated in Fig. 4 (a) and (b). Consequently, the formation of Al3Ni
phase during the cooling cycle can be supported by both the microstructural observation and the Al–Ni binary phase diagram. Based on Fig. 4(c), it is noted that the formation of Al3Ni2phase consumes almost all Al-rich
melt in the joint at 1000C for 96 s.
As the brazing temperature further increases to 1200C, the microstructure of the joint is continuously
changed. Based upon Al–Ni binary phase diagram, NiAl should be the first phase formed in the melt at 1200C. However, the Al-rich melt contains insufficient
Ni atoms to form NiAl. Because the Ni concentration in the melt is lower than 50 at.%, there is no possibility to form NiAl in the melt. This is consistent with the microstructural observation in the brazed joint as dis-played in Fig. 1(c). There is no NiAl phase observed in the joint. However, the Ni atoms dissolved in Al-rich melt at 1200C are much more prominent than those at
1000C. Upon the subsequent cooling cycle, more Al
3Ni2
and Al3Ni phases in the joint are expected. In such a case,
less Al-rich phase is obtained in the final microstructure as displayed in Fig. 1(c). The Al-rich melt experiences two peritectic reactions during the cooling cycle, and some Al3Ni2 are enclosed by Al3Ni phase. Similarly,
bothAl3Ni2 and Al3Ni phases forms upon cooling
cycle, so it is expected that the volume fraction of these phases is independent of brazing time. This is able to explain why the interfacial thickness of Al3Ni2brazed at
1200C for 2 s is not thicker than that brazed at 1000C
for 16 s as compared between Fig. 1(b) and (c). The interfacial Al3Ni2 phase will grow at 1000C, but it is
not formed yet at 1200C.
Fig. 1. Backscattered electron images (BEIs) and EPMA line-scanning profiles of the NiAl/Al/NiAl specimens: (a) 800oC2 s, (b) 1000C
16 s, (c) 1200C2 s (a, b, c and d represent Al-rich, Al
3Ni, Al3Ni2,
3.2. Microstructural evolution of infrared brazed Ni3Al/
Al/Ni3Al
Fig. 5 shows the cross-section of an Ni3Al/Al/Ni3Al
joint after brazing at 800C withvarious time periods. It
is noted that part of the Al3Ni phase is fractured, and it
could be caused by grinding process before metallo-graphic observation. The microstructure of Ni3Al/Al/
Ni3Al brazed at 800C is similar to that of NiAl/Al/NiAl
brazed joint except coarser Al3Ni and Al3Ni2phases in
the joint. Since Ni3Al base material contains more Ni
than NiAl base material, the melt can easily dissolve more Ni atoms from Ni3Al during infrared brazing. The
Al3Ni phase is firstly formed in both melt and interface
during brazing, and the Al3Ni2 phase can also be
observed at the interface between Ni3Al and Al3Ni
phase. Similarly, the Al3Ni2 phase is formed by
solid-state interdiffusion between Al3Ni and Ni3Al substrate.
With the increment of brazing time, the interfacial Al3Ni2phase grows constantly; the volume fraction of
Al-richmelt is decreasing after brazing as illustrated in Fig. 5. The consumption of Al atoms in the melt results in growthof Al3Ni2phase at the interface. As discussed
earlier, the Al3Ni2 phase is more stable than Al3Ni in
Ni3Al/Al/Ni3Al joint according to the experiment.
Fig. 6 shows the microstructure of the infrared brazed joint at 1000C for 2, 16 and 96 s, respectively. As the
brazing temperature increases to 1000oC, Al
3Ni2 is the
first phase precipitated from the Al-rich melts. Mean-while, the Al3Ni2phase is also formed at the bonding
interface, and grows continuously withthe brazing time
increasing. The Al3Ni phase is formed upon cooling cycle
of brazing. Therefore, the volume fraction of Al3Ni phase
in the joint is not depended upon brazing time only. The consumption of Al in the melt is caused by formation of Al3Ni2 phase. The Al-rich melt is almost completely
depleted for the specimen brazed at 96 s, and no Al3Ni
phase in the joint after brazing as shown in Fig. 6(c). The intermediate phase NiAl, with a composition between Al3Ni2and Ni3Al, is not observed until the brazing time
increases to 96 s at 1000C as shown in Fig. 6(c). As
discussed earlier, the formation of NiAl can be attrib-uted to the solid-state interdiffusion between Al3Ni2and
Ni3Al interlayer. It is expected that more incubation
time is necessary due to the slow solid-state interdiffu-sion of the reacting elements. This is consistent with the result of other researches performed at lower tempera-tures [3,9,11,12]. However, the intermediate phase Ni5Al3, which is stable below 700C, is not found in the
experiment.
3.3. The growth of Al3Ni2interlayer during the infrared
brazing
Based upon previous discussions, the reaction mechanism in both Al3Ni and Al3Ni2phases is different
as the brazing temperature changes from 800 to 1000C.
The Al3Ni phase is developed upon cooling cycle at
1000C infrared brazing, and it is formed during 800C
brazing. According to the experiment, the Al3Ni phase
is less stable than Al3Ni2 phase in both NiAl/Al/NiAl
and Ni3Al/Al/Ni3Al joints. The microstructure of the
brazed joint is primarily dominated by Al3Ni2phase for
longer infrared brazing time, e.g. 96 or 290 s. Moreover, the formation mechanism of the Al3Ni2interlayer may
be different between 800 and 1000C infrared brazing.
The solid-state interdiffusion of Al and Ni atoms between Al3Ni and substrate (NiAl or Ni3Al) dominates
the formation of Al3Ni2phase at 800C. Since the
dia-meter of Al atom is smaller than that of Ni, it is rea-sonable that the diffusion of Al atoms is faster than that of Ni atoms. The formation of Al3Ni2interlayer is then
rate controlled by the diffusion Al atoms. According to the binary phase diagram shown in Fig. 2, Al3Ni is a
stoichiometric compound, and it can only dissolve very limited range of Al concentration. The flux of Al atoms from the Al-rich melt into the Al3Ni2phase is, therefore,
greatly impeded by the formation of Al3Ni interlayer.
Since the formation of a Al3Ni interlayer can greatly
deteriorate the transport of Al atoms, the growth of
interfacial Al3Ni2phase at 800C is muchslower than
that at 1000C as compared between Fig. 1(b) and 2(a).
On the other hand, the mass transport of Al atoms in Al-richmelt into the Al3Ni2 phase at 1000C is not
Fig. 3. Interfaces of NiAl/Al/NiAl brazed at 800C for (a) 20 s, (b)
290 s (a, b, c and d represent Al-rich, Al3Ni, Al3Ni2, and NiAl phases,
respectively).
Fig. 4. Interfaces of NiAl/Al/NiAl brazed at 1000C for (a) 6 s, (b) 46
s, (c) 96 s (a, b, c and d represent Al-rich, Al3Ni, Al3Ni2, and NiAl
limited by the Al3Ni interlayer. The Al3Ni2 phase is a
non-stoichiometric compound, and it can dissolve some range of Al concentration. Therefore, it is expected that the growth of Al3Ni2 phase at 1000C is muchfaster
than that at 800C brazing. Fig. 7 shows the average of
interfacial thickness of Al3Ni2phase for various brazing
conditions. It demonstrates that the growth rate of the Al3Ni2at 1000C is about one order of magnitude
lar-ger than that at 800C, and this is consistent with the
previous discussion. Fig. 5. Interfaces of Ni3Al/Al/Ni3Al brazed at 800C for (a) 2 s, (b) 20
s, (c) 110 s (a, b, c, d and e represent Al-rich, Al3Ni, Al3Ni2, NiAl and
Ni3Al phases, respectively).
Fig. 6. Interfaces of Ni3Al/Al/Ni3Al brazed at 1000C for (a) 2 s, (b)
16 s, (c) 96 s (a, b, c, d and e represent Al-rich, Al3Ni, Al3Ni2, NiAl
4. Conclusion
For specimens brazing at 800C, the Al
3Ni phase is
firstly formed in both the Al-rich melt and the interface between nickel aluminide base material and Al-rich melt. The solid state interdiffusion between Al3Ni and
base material results in the formation of the interfacial Al3Ni2phase, and the growth of Al3Ni2phase at 800C
is greatly impeded by the Al3Ni interlayer due to its
stoichiometry.
For specimens brazing at 1000C, the reaction
chan-ges from hypereutectic into peritectic reaction. The Al-richmelt dissolves more Ni atoms withthe increment of brazing temperature to 1000C. The Al
3Ni2 phase is
now initially formed in boththe aluminum melt and joint interface. The growth of Al3Ni2 interlayer at
1000C is much faster than that at 800C. Transport of
Al atoms in forming Al3Ni2phase at 1000C is greatly
increased due to the contact between Al3Ni2and liquid
Al-richmelt. There is no Al3Ni interlayer to limit the
growthof Al3Ni2phase. The Al3Ni phase shown in the
joint is formed upon cooling cycle of the infrared braz-ing. The thickness of Al3Ni2 at 1000C is about one
order of magnitude larger than that at 800C.
The microstructural evolution in Ni3Al/Al/Ni3Al
joint is similar to that in NiAl/Al/NiAl except the for-mation of a NiAl interlayer between Al3Ni2and Ni3Al.
It can be attributed to the solid-state interdiffusion between Ni3Al and Al3Ni2 interlayer. However, the
intermediate phase Ni5Al3, which is stable below 700C,
is not found in the experiment.
Acknowledgements
The authors gratefully acknowledge the financial support of this research by the National Science Council (NSC), Republic of China under NSC Grants 88-2216-E149-002 and 88-2216-E002-026.
References
[1] Castleman LS, Seigle LL. Trans TMS-AIME 1957;209:1173. [2] Castleman LS, Seigle LL. Trans TMS-AIME 1958;212:589. [3] Janssen MMP, Rieck GD. Trans TMS-AIME 1967;239:1372. [4] Nastasi M, Hung LS, Mayer JW. Appl Phys Lett 1983;43:831. [5] Tarento RJ, Blaise G. Acta Metall Mater 1989;37:2305. [6] Bertoti I, Mohai M, Csanady A, Barna PB, Berek H. Surf
Inter-fact Anal 1992;19:457.
[7] Cardellini F, Mazzone G, Montone A, Vittori Antisari M. Acta Metall Mater 1993;42(7):2445.
[8] Tsao C-L, Chen S-W. J Mater Sci 1995;30:5215. [9] Metelnick MP, Varin RA. Z Metallkde 1991;82:346. [10] Fan W, Varin RA, Wronski Z. Z Metallkde 1994;85:522. [11] LieblichM, Gonzalez-Carrasco JL, Caruana G. Intermetallics
1997;5:515.
[12] Jung SB, Minamino Y, Yamane T, Saji S. J Mater Sci Lett 1993;12:1684.
[13] Otoshi Y, Fujiwara K, Horita Z, Nemoto M. Mater Trans JIM 1998;39(1):225.
[14] Lee S-J, Wu S-K, Lin R-Y. Acta Metall Mater 1998;46:1283. [15] Lee S-J, Wu S-K, Lin R-Y. Acta Metall Mater 1998;46:1297. [16] Shiue R-K, Wu S-K, O J-M, Wang J-Y. Metall Mater Trans
2000;31A:2527.
[17] Massalski TB. Binary alloy phase diagrams. Materials Park: (OH) ASM International, 1990 (p. 183).
[18] Baker H. ASM Handbook (Vol. 3). Materials Park (OH): ASM International, 1992.
Fig. 7. The average thickness of Al3Ni2interlayer for various infrared
