Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e i n f o a b s t r a c t
Article history: Pressure-free bonding using mixed Cu–NiO nanoparticles was examined. The effect of mixing NiO nano-
Received 27 October 2014 particles with Cu nanoparticles on bonding strength was identified for a NiO content between 0.005 and
Received in revised form 12 December 2014 2 wt%. The bonding layer, which presented a high bonding strength, was thin and dense. The Ni in the
Accepted 27 December 2014
bonding layer was segregated at the surface and at grain boundaries of crystalline Cu grains as a metallic
Available online 9 January 2015
Cu–Ni alloy. These results suggest that the effect of the NiO addition is caused by the enhancement of
sintering by surface diffusion in the bonding layer, resulting from sufficient removal of surface oxides
Keywords:
on the Cu nanoparticles through their reaction with the NiO nanoparticles.
Metals and alloys
Nanostructured materials
Ó 2015 Elsevier B.V. All rights reserved.
Sintering
Diffusion
Microstructure
http://dx.doi.org/10.1016/j.jallcom.2014.12.224
0925-8388/Ó 2015 Elsevier B.V. All rights reserved.
T. Satoh, T. Ishizaki / Journal of Alloys and Compounds 629 (2015) 118–123 119
3. Results
than that of the Cu only bonding layer, and it had a gradually and (b). This result indicates that there are few oxide phases in
increasing tendency with increasing NiO quantity. Because the the bonding layer. As shown in Fig. 4(c), Cu was distributed in all
dependence of the thickness on the NiO content is opposite to that regions where there was any sort of substance in the STEM image
of the bonding strength shown in Fig. 1, it is believed that the (Fig. 4(a)). In contrast, the Ni segregated at the grain boundaries
bonding strength and bonding layer thickness have an inverse rela- and on the crystalline Cu surface without maintaining a particulate
tionship with the NiO content. shape, as shown in Fig. 4(d). Fig. 5 shows EDS spectra taken from
Cross-sectional SEM images taken near the interface between the grain boundary (point A) and the intragranular (point B) region
the mixed Cu–NiO nanoparticle bonding layer and the Cu base in Fig. 4(a). O was scarcely detected in each spectrum. On the other
plate are shown in Fig. 3. The Ni layer was so thin that it could hand, the intensity corresponding to Ni Ka at the grain boundary
not be identified in the images. The bonding layer directly con- region (Fig. 5(a)) was much greater than that at the intragranular
nected to the base plate without producing any additional interfa- region where Ni was scarcely detected. Ni/Cu atomic ratios at the
cial phases in any of the interfaces. Additionally, structural defects grain boundary and the intragranular region calculated from the
including cracks and micro-voids were observed inside all bonding EDS spectra were 0.040 and 0.0015, respectively. It is inferred that
layers. Major cracks were observed in the mixed Cu–NiO bonding the metallic state of Ni segregated at the grain boundaries and sur-
layers of 0.01–1 wt% NiO content (Fig. 3(b), (c), and (d)). On the face of the crystalline Cu.
other hand, the micro-voids were larger and more frequently
observed in the Cu only and Cu-10 wt% NiO bonding layers 4. Discussion
(Fig. 3(a) and (e)) over the others. These results indicate that the
bonding layer exhibiting a higher bonding strength and decreased The improved bonding strength and changes in the bonding
thickness has a larger amount of cracks and lower amount of layer through mixing the NiO and Cu nanoparticles were described
micro-voids. in Section 3. The mechanism of this bonding strength improvement
Fig. 4 shows the STEM image and EDS maps of the Cu-2 wt% NiO will be discussed here.
nanoparticle bonding layer. The distribution of O was uniform, The bonding strength increased as the bonding layer thickness
with a limited dependence on morphology, as shown in Fig. 4(a) decreased, as described in Section 3. Therefore, the bonding
Fig. 3. Cross-sectional SEM images near the interface between the mixed Cu–NiO nanoparticle bonding layer and the Cu base plate; the NiO contents are (a) 0, (b) 0.01, (c) 0.1,
(d) 1, and (e) 10 wt%.
T. Satoh, T. Ishizaki / Journal of Alloys and Compounds 629 (2015) 118–123 121
Fig. 4. (a) STEM image and EDS maps of the Cu-2 wt% NiO bonding layer for (b) O, (c) Cu, and (d) Ni.
strength was plotted against the thickness of the bonding layer in boundary phases of metallic Ni itself, even if electron beam spread-
Fig. 6. The previously described relationship between the bonding ing within the sample is considered. On the other hand, the Cu–Ni
strength and bonding layer thickness was confirmed. Since the Cu system is isomorphous [23]. Ni is therefore considered to form a
nanoparticle bonding layer without any NiO particles follows the Cu–Ni solid solution, although the existence of an extremely thin
same correlation that is indicated by a region enclosed with dashed Ni layer at the grain boundary cannot be denied. If this hypothesis
line in Fig. 6, the NiO nanoparticles contribute minimally to the is correct, the Ni atoms originating from the NiO nanoparticles dif-
bonding strength. fused to the Cu nanoparticle surfaces in addition to alloying and
The bonding layer displaying the higher bonding strength and reducing. The NiO forms a solid solution with CuO rather than
decreased thickness has a larger number of cracks and lower Cu2O [24], and along with CuO, it is reduced more easily than
amount of micro-voids, as shown in Fig. 3. The cracks are caused Cu2O [25–27]. Therefore, the alloying and reduction of the NiO
by the bonding layer shrinkage associated with the sintering pro- and Cu2O, described in Eqs. (1) and (2), were proposed in our pre-
cess. In contrast, the decreased quantity of micro-voids originated vious article [19].
from the crystal growth of the Cu nanoparticles. It is therefore
likely that the sintering improved the bonding layer with the NiO þ xCu2 O ! NiðCux ÞO1þx þ xCu ð1Þ
higher bonding strength and lower thickness.
NiðCux ÞO1þx þ ð1 þ xÞH2 ! NiðCux Þ þ ð1 þ xÞH2 O ð2Þ
Next, the effect of the NiO nanoparticles mixed with the Cu
nanoparticles on the layer’s sintering behavior will be examined.
Sintering of the nanoparticle bonding layer generally starts with When the Cu2O phase on the Cu nanoparticles is removed by these
the surface diffusion of metallic atoms originating in the nanopar- reactions, the metallic Cu and Ni–Cu solid solution produced by
ticles. However, the nanoparticles’ surface oxides inhibit an active these reactions are expected to actively diffuse from the Cu nano-
surface diffusion. Therefore, promoting a reduction of the surface particles, promoting the bonding layer sintering. The distribution
oxides stimulates the sintering. In the case of the Cu nanoparticle of Ni in the bonding layer, shown in Fig. 4(d), supports the above
bonding, reducing the Cu monovalent oxide phase (Cu2O) is con- model. Moreover, the effect of mixing the Ni and Ni-based alloy
sidered important because its activation energy for reduction by nanoparticles with the Cu nanoparticles is expected to be caused
H2 is higher than that of CuO [21,22]. On the other hand, the metal- by the reactions in Eqs. (1) and (2) as well. Although the surface
lic state of Ni segregated at the grain boundaries and surface of the oxide of the nanoparticles contributes to this reaction, the volume
crystalline Cu, as described in Section 3. Cu existed at the region ratio of the surface oxide to the nanoparticle would be small
where Ni segregated as shown in Fig. 4(c), and the Ni/Cu atomic [17,18]. In contrast, the entire volume of the NiO nanoparticles
ratio was 0.040 at the grain boundary region. This value does not mixed into the Cu nanoparticles could be involved in the reaction.
seem to be large enough to indicate that that there are any grain As a result, it is believed that the effective NiO nanoparticle content
122 T. Satoh, T. Ishizaki / Journal of Alloys and Compounds 629 (2015) 118–123
Fig. 6. The shear strength of Cu plates bonded using the mixed Cu–NiO nanopar-
ticles plotted relative to bonding layer thickness.
5. Conclusions