Journal of Materials Processing Technology 75 (1998) 137 – 142

Joining of Si3N4 to Si3N4 using rapidly-solidified CuNiTiB brazing

filler foils
Huaping Xiong, Chuangeng Wan, Zhenfeng Zhou *
Welding Di6ision, Department of Materials Science and Engineering, Jilin Uni6ersity of Technology,
Changchun 130025, People’s Republic of China
Received 23 December 1996

Abstract

Joining of Si3N4 to Si3N4 was performed using newly developed rapidly-solidified CuNiTiB brazing foils of different thickness,
three-point bending tests then being employed at room temperature to measure the strength of the joints. The microstructures of
the joints were observed and the element area distribution of Cu, Ni and Ti was examined by scanning electron microscope with
an X-ray wave-dispersion spectrometer. The results of the strength test show that when using rapidly-solidified brazing filler foils
of suitable thickness (for example, 40 mm and 80 mm) the joint strength was increased greatly (up to 402 MPa and 380 MPa
respectively) compared with the use of paste brazing filler of the same alloy. Moreover, the fracture form of the Si3N4/Si3N4 joints
also varied with the change of strength. The thickness of the brazing foils used has an important effect on the corresponding joint
strength, reasons for which are discussed. © 1998 Elsevier Science S.A.

Keywords: Rapid-solidification; Brazing filler foils; Silicon nitride

1. Introduction compositions were designed, then the interfacial reac-

Silicon nitride (Si3N4) ceramic is a kind of structural
ceramic material with combinations of many superior
properties and has been used widely in industry. How-
ever, it is not only difficult but also expensive to
manufacture ceramic workpieces with large dimensions
and complicated shapes. Additionally, ceramic work-
pieces with larger dimensions are more difficult to
produce without defects. Therefore, it becomes espe-
cially important to obtain reliable ceramic/ceramic or
ceramic/metal joints. Generally, AgCuTi brazing filler
metals are commonly used in ceramic joining [1,2], but
this brazing alloy contains a rather high quantity (up to
57 wt.%) of Ag, which is an expensive metal. Therefore,
it is necessary to develop new kinds of brazing filler
metals with similar properties to those of the AgCuTi
alloy, but without silver and other expensive metals.
Recently, the authors [3] investigated the wetting
behavior of CuNiTi alloys on Si3N4. On this basis,
several paste CuNiTiB brazing alloys with different
* Corresponding author. Fax: +86 431 5683397.
tions of the joints brazed with them were studied sys-
tematically, the brazing alloy composition
corresponding to the highest strength value being deter-
mined [4]. The present paper mainly investigates the
joining of Si3N4/Si3N4 using rapidly-solidified foils of
this brazing alloy.
2. Experimental materials and methods
The composition of paste brazing alloy is shown in
Table 1. The purities of the Cu, Ni, Ti and B powders
are 99.7 wt.%, 99.8 wt.%, 99.0 wt.% and 97.1 wt.%,
respectively. The powders were mixed uniformly in a
mortar and the mixed powder was melted twice at 1473
K for 30 min in a crucible of 99.0% Al2O3 with a cover
in a vacuum resistance heated furnace. The surface part
of the ingot obtained, which might have interacted with
the crucible, was removed by an emery wheel, the main
part of the alloy ingot being used as the matrix alloy.
The matrix alloy was broken into pieces and melted
again. Then the liquid alloy was ejected onto the outer

0924-0136/98/$19.00 © 1998 Elsevier Science S.A. All rights reserved.

PII S 0 9 2 4 - 0 1 3 6 ( 9 7 ) 0 0 2 9 4 - X
138 H. Xiong et al. / Journal of Materials Processing Technology 75 (1998) 137–142
Table 1
Chemical composition of paste brazing alloy (wt.%)
Ni Ti B Cu
5.0–25.0 16.4 – 28.2 0–3.0 Balance
surface of a copper wheel rotating at : 1800 rpm. The
rapidly-solidified brazing foils prepared by such means
were originally 40– 50 mm thick and 5 – 7 mm wide.
Both sides of these foils were polished mechanically
until the thickness decreased to 20 mm. All of the foils
were cut by scissors to the dimension 3× 4 mm. A
number of foils of 20 mm thickness were laminated
together to provide brazing foils of different thickness.
The ceramic bars used are hot-pressed sintered Si3N4
with the dimension 3×4 ×20 mm. Before joining, the
brazing foils and the Si3N4 bars to be joined were
cleaned ultrasonically in acetone and dried by air blow-
ing. The brazing foils were placed between two Si3N4
bars and the assembly then held to form a butt joint by
a kind of organic glue in a specially designed iron jig as
shown in Fig. 1. During heating the organic glue de-
composed and volatilized, so that no impurities re-
mained. The vacuum was kept at between 3.0× 10 − 5
and 5.0 ×10 − 5 torr during the alloy melting and braz-
ing process. Brazing experiments were carried out with
a heating rate of 15 K min − 1, a brazing temperature of
1353 K and a brazing time of 10 min. The cooling rate
was 15 K min − 1 until 673 K, then about 1.5 K min − 1
to room temperature. The joint strength of the butt
joint was measured by the three-point bend test (shown
in Fig. 2) with a cross-head speed of 0.5 mm min − 1 at
room temperature. The microstructures of the brazing
alloys and the joints were observed, and the element
area distribution of Cu, Ni and Ti were examined by a
scanning electron microscope (SEM) with an X-ray
wave-dispersion spectrometer (XRWDS). The composi-
tions of the different brazing alloy zones of joints with
different thickness were determined using an energy-dis-
persion spectrometer (EDS). The thickness of the inter-
facial reaction layers and of the brazing alloy zones of
Fig. 1. Iron jig used in the joining experiments.
Fig. 2. Schematic diagram of the three-point bend test for the
measuring of the Si3N4/Si3N4 joint strength.
the joints were measured, as was the micro-hardness
(Hv) of the brazing alloy zones of the joints.
3. Experimental results and discussion
3.1. Microstructures and element area distribution of
the Si3N4 /Si3N4 joints
Joints of Si3N4/Si3N4 were produced using rapidly-
solidified brazing filler foils of 20, 40, 80 and 120 mm
thickness, Fig. 3(a,b,c) showing the morphology of the
joints obtained with such foils. The microstructures and
element area distribution images of Cu, Ni and Ti of
the joint brazed with rapidly-solidified foil of nominal
40 mm thickness (the actual thickness of the joint is
:36 mm) are shown in Fig. 4, the figure indicating that
Cu was richer at the central part of the joint than at the
interface, whilst the elements Ni and Ti segregated to
the Si3N4. According to Fig. 4(c,d), the segregation
zone of the active element Ti was wider than that of Ni,
and Ti segregated more closely to Si3N4. This signifies
that only Ti in the brazing alloy participated originally
in the reaction at interface ‘a’, close to the Si3N4 (refer
to Fig. 4(a)); then at interface ‘e’, a little further away
from the Si3N4 (refer to Fig. 4(a)), both Ni and Ti being
involved in the reactions. That the element area distri-
bution of Cu, Ni and Ti of the joints obtained with the
rapidly-solidified brazing alloy was similar to that ob-
tained with the paste brazing alloy shows that similar
interfacial reactions occurred, thus forming similar in-
terfacial reaction products (refer to Fig. 5).
The Si3N4/Si3N4 joints brazed with the paste brazing
alloy showed a maximum strength value of 338.8 MPa
[4], whilst the strength values were evidently increased,
up to 402 MPa and 380 MPa respectively, when
rapidly-solidified brazing foils of 40 mm and 80 mm
thickness were used. Fig. 6 shows the microstructures of
the CuNiTiB brazing alloy before and after the rapid-
solidification treatment. It is clear that the microstruc-
ture of the latter is more homogeneous than that of the
former, inducing a more compact interfacial mi-
crostructure (refer to Fig. 3(a,b,d)), thus it is favorable
for the improvement of the joint strength to use the
rapidly-solidified brazing filler foils. Moreover, the frac-
ture form of the joints used in the strength test also
varied with the increase of the joint strength. Fig. 7(a,b)
 H. Xiong et al. / Journal of Materials Processing Technology 75 (1998) 137–142 139

Fig. 3. Microstructures of Si3N4/Si3N4 joints made using the rapidly-solidified CuNiTiB brazing foils of thicknesses (a) 40 mm, (b) 80 mm, (c) 120
mm and (d) using the paste CuNiTiB brazing alloy (etched first by HF + HNO3 solution, and then by 5% FeCl3 solution).

show photographs of the joints obtained (upper) and
of the corresponding fracture samples (lower) brazed
with paste brazing alloy and rapidly-solidified brazing
foil of 40 mm thickness. Fracture occurred at the inter-
face of the joints with lower strength value, and the
adhering interfacial products, fine ceramic or metallic
pieces, can be observed on the fracture surface. For
the joints with higher strength values, when fracture
took place, cracks originated from the interface, then
extended into the Si3N4 (refer to Fig. 8).
3.2. Effect of the thickness of the brazing filler foils on
the joint strength
It was found that if the brazing alloy is too thin
(e.g. 20 mm), the corresponding strength value of the
joint is very low (140 MPa), mainly because too-thin
brazing foils could not fill the gap between the two
Si3N4 bars to be joined, which resulted in not only a
decrease of the actual joining area, but also a lack of
foil at the edge of the joint, so that a notch effect was
produced on the Si3N4/Si3N4 joint when loaded. With
the increase of the thickness of the brazing foils (noted
as d0), both the thickness of the central part (namely
the brazing alloy zone) of the joint (noted as d2) and
the thickness of the whole joint (noted as d3) in-
creased. The thickness of the interfacial reaction layers
(noted as d1) increased first (from 6 to 10 mm) and
then tended to be constant. Moreover, the strength
(sb) of the joints increased first, up to the maximum
value of 402 MPa, then tended to decrease (refer to
Fig. 9). When the thickness of brazing foils was in-
creased from 80 to 120 mm, the joint strength de-
creased rapidly, although the thickness of the
interfacial reaction layer increased only slightly (from
10 to 11 mm). To analyze the cause of the above
changes, the compositions of the brazing alloy zone
(refer to Fig. 3(b,c) and Fig. 4(a)) of the joints with
different thicknesses were examined by EDS, as was
the corresponding micro-hardness.
140 H. Xiong et al. / Journal of Materials Processing Technology 75 (1998) 137–142
Fig. 4. Showing (a) the microstructure, and the element area distribution images of (b) Cu, (c) Ni and (d) Ti, of a Si3N4/Si3N4 joint made using
rapidly-solidified CuNiTiB brazing foil of 40 mm thickness.
The results show that with an increase in the thick-
ness of the brazing foils, the Ti content of the brazing
alloy zone of the joints increased (refer to Fig. 9), and
the corresponding micro-hardness also increased, so
that the plasticity of Cu-rich layers at the central part
of the joints was reduced. You Chul Kim et al. pro-
duced the joints of Si3N4/steel using Cu – X binary
alloys (Cu–5 wt.% Cr, Cu – 1 wt.% Nb, Cu – 3 wt.%
V, Cu–5 wt.% Ti and Cu – 10 wt.% Zr) as brazing
filler metals, finding that the Cu – 5 wt.% Cr brazing
alloy with the lowest hardness was the most effective
in reducing the residual stresses resulting from ther-
mal mismatch of the joints [5]. In the joining of ce-
ramic to ceramic, interfacial thermal mismatch equally
exists between the brazing alloy and the ceramic. The
so-called Si3N4/Si3N4 joint is actually composed of
Si3N4/interfacial reaction layer/brazing alloy zone/in-
terfacial reaction layer/Si3N4, in which the interfacial
reaction layer acts as a transition layer from the Si3N4
ceramic to the brazing alloy zone of the joint. How-
ever, this transition layer can not effectively relax the
interfacial thermal stresses between the Si3N4 and the
brazing alloy because of the existence of brittle reac-
tion products. Only a soft brazing alloy zone can
secure this relaxation, as does the interlayer used in a
ceramic/metal joint. When the thickness of the braz-
ing filler foils is 40 mm, that of the interfacial reaction
layer (d1) is 6 mm, the Ti content of the brazing alloy
zone of the joint is only 4.4 wt.% and the correspond-
ing micro-hardness value is very low (Hv 196, refer to
Fig. 9): this soft central layer is favorable for the
relaxation of the interfacial thermal stresses. When the
thickness d0 is increased from 80 to 120 mm, the
thickness d1 increased from 32 to 64 mm (refer to Fig.
9), which will raise the interfacial stresses. On the
other hand, the Ti content of the brazing alloy zone
of the joint increased from 10.0 to 13.8 wt.%, which
increased the brittleness of the brazing alloy zone: due
to the above two effects, the joint strength decreased
rapidly.
142 H. Xiong et al. / Journal of Materials Processing Technology 75 (1998) 137–142
Fig. 8. Schematic diagram of the fractures of the Si3N4/Si3N4 joints
with: (a) lower and (b) higher strength values.
The microstructure of the rapidly-solidified Cu-
NiTIB brazing alloy is more homogeneous than that
without the rapid-solidification treatment. Using
Fig. 9. For Si3N4/Si3N4 joints brazed with rapidly-solidified CuNiTiB
brazing foils, showing the relationships between the micro-hardness;
the Ti content of the central part of the joint; the thickness of the
central part (d2), the thickness of the interfacial reaction layer of the
joint (d1); the thickness of the whole joint (d3); the three-point bend
strength of the joints and the thickness of the brazing foils (d0).
.
rapidly-solidified brazing foils with suitable thickness
(e.g. 40 and 80 mm), joints of Si3N4/Si3N4 with higher
strength values (402 and 380 MPa respectively) were
obtained. The ideal transition structure of the Si3N4/
Si3N4 joints is: Si3N4/interfacial reaction layer with a
suitable thickness/soft brazing alloy zone with a suit-
able thickness/interfacial reaction layer with a suitable
thickness/Si3N4.
With increasing the thickness of the brazing filler
foils from 80 to 120 mm, the thickness of the brazing
alloy zone increased, which resulted in higher interfa-
cial stresses, although the thickness of the interfacial
reaction layer changed only slightly. On the other
hand, the Ti content of the brazing alloy zone of the
joint increased also, which increased the hardness and
brittleness of the brazing alloy zone, so that the inter-
facial thermal stresses could not relax effectively. Due
to the above two effects, the joint strength decreased
rapidly.
Acknowledgements
This research was supported financially by the Na-
tional Education Commission of China. The authors
appreciate the help of Mr Min-Xiu Quan, Director,
and his colleagues, National Laboratory of Rapidly
Solidified Non-Equilibrium Alloys, Shenyang Institute
of Metals Research, Chinese Academy of Sciences, in
the preparation of the rapidly-solidified brazing foils.
References
[1] A.P. Xian, Z.Y. Si, Interlayer design for joining pressureless
sintered sialon ceramic and 40 Cr steel brazing with Ag57Cu38Ti5
filler metal, J. Mater. Sci. 27 (1992) 1560.
[2] F.H. Bao, J.L. Ren, Y.H. Zhou, Properties of vacuum brazed
Si3N4/steel joint using active brazing filler metal, Trans. China
Welding Inst. 11 (1990) 200.
[3] C. Wan, H. Xiong, Z. Zhou, Wettability of CuNiTi alloys on
Si3N4 and interfacial reaction products, J. Mat. Sci. Tech. 12
(1996) 219.
[4] C. Wan, H. Xiong, Z. Zhou, Joining of Si3N4/Si3N4 with Cu-
NiTiB paste brazing filler metals and interfacial reactions of the
joints, J. Mater. Sci., 1996 (submitted).
[5] Y.C. Kim, K. Saida, Y.H. Zhou, H.N. Tom, in: K.J. Chen
(Ed.), Selection of Insert Layer Materials in Si3N4/steel Joining,
Welding, Joining, Coating and Surface Modification of Ad-
vanced Materials, Proc. Pre-Assembly Symp. of 47th Annual
Assembly of IIW, Vol. II, Dalian, China, 1 – 2 Sept 1994, pp.
336 – 341.
 H. Xiong et al. / Journal of Materials Processing Technology 75 (1998) 137–142
Fig. 5. X-ray diffraction pattern of the fracture surface of a Si3N4/Si3N4 joint made using the paste CuNiTiB brazing alloy (brazed at 1353 K for
10 min).
4. Conclusions
The following conclusions can be drawn.
Joining of Si3N4 to Si3N4 was conducted using
Fig. 6. Microstructures of the CuNiTIB brazing alloy: (a) before and
(b) after the rapid-solidification treatment (etched by HF + HNO3
solution).
141
rapidly-solidified CuNiTiB brazing filler foils of differ-
ent thickness. It was found through micro-analysis that
during the brazing process, both Ni and Ti in the
brazing alloy diffused to the Si3N4 and segregated
there. However, the segregation zone of Ti is wider
than that of Ni, thus Ti segregated more closely to the
Si3N4, which signifies that only Ti in the brazing alloy
participated originally in the interfacial reactions at the
interface close to the Si3N4, and then both Ni and Ti
were involved in the reactions at the interface a little
further away from the Si3N4. The reaction products of
TiN, Ti5Si3, TiSi and NiTiSi were believed to be pro-
duced at the interface of the joints.
Fig. 7. Photographs of the Si3N4/Si3N4 joints (upper) and the corre-
sponding fracture samples (lower) using: (a) paste CuNiTiB brazing
alloy; (b) rapidly solidified CuNiTiB brazing foil of 40 mm thickness
(brazed at 1353 K for 10 min).