Professional Documents
Culture Documents
To cite this article: H. Yumoto , S. Shimada , C. Minami , T. Inoue , S.J. Li & M. Ishihara
(1997) Effect of oxides on the adhesion of Cu films deposited onto stainless steel by
electron shower and thermal evaporation methods, Journal of Adhesion Science and
Technology, 11:5, 665-677, DOI: 10.1163/156856197X00651
Taylor & Francis makes every effort to ensure the accuracy of all the information
(the “Content”) contained in the publications on our platform. However, Taylor
& Francis, our agents, and our licensors make no representations or warranties
whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the
opinions and views of the authors, and are not the views of or endorsed by Taylor
& Francis. The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and Francis
shall not be liable for any losses, actions, claims, proceedings, demands, costs,
expenses, damages, and other liabilities whatsoever or howsoever caused arising
directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan,
sub-licensing, systematic supply, or distribution in any form to anyone is
expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
Effect of oxides on the adhesion of Cu films deposited onto
Abstract-When Cu films were deposited by thermal evaporationonto stainless steel substratesat 30°C,
the oxygen gas in the vacuum chamber (1.5 x 10-3 Torr) caused the adhesion of Cu films to increase
from 3 to 5 MPa. Moreover, it increased further from 13 to 16 MPa when deposited at 300°C. The
Cu film was not peeled off when deposited by the electron shower method and the epoxy resin failed
(20 MPa), and this was independentof the addition of oxygen gas. As the chemical shift of Cu 2p3/2
was observed at the interface between the Cu film and the substrate when oxygen gas was added, it is
concluded that the adhesion is mainly determinedby the chemical bonding, such as CuO and Cu2O. The
depth profile of Cu 2p3/2 measured by X-ray photoelectronspectroscopy(XPS) using Ar etching showed
apparent thermal diffusion of Cu into the substrate. But the Ar etching rate was decreased by Cu oxides
at the interface. The amount of oxides depended on the substrate temperatureand the deposition method
for Cu film. Therefore, the depth profile of Cu measured by XPS did not represent the thermal diffusion
of Cu into the substratecorrectly. When the etching rate was modified,the diffusionof Cu was almost the
same for different samples deposited at the same temperature, and the effect of the thermal diffusion on
the adhesion was small. The adhesion on hydrated [Cr(OH)3.0.4H2O]and hydroxide [Cr(OH)3]surfaces
was lower than that on the oxide (Cr2O3)surface. In other words, the pretreatmentof the substrates was
very important to the adhesion.
1. INTRODUCTION
As the electrical and thermal conductivities of stainless steel are low, they are generally
increased by coating with Cu or Au. In order to use Cu-coated stainless steel for
switches and electrical terminals, strong adhesion of the Cu film is required.
We found that the adhesion of a Cu film deposited on wet polished stainless steel
by an electron shower was at least 100 times higher than when it was deposited
by conventional thermal evaporation [1, 2]. The adhesion mechanism changed from
2. EXPERIMENTAL
Figure 1 shows a schematic drawing of the electron shower apparatus. The base
pressure in the vacuum chamber was 2 x 10-5 Torr. A potential of 1 kV was applied
between the heated tungsten (W) filament (190 A) and the ring-shaped anode (20 cm
in diameter), so that thermal electrons were accelerated towards the stainless steel
substrates [type 304 (10 x 10 x 1 mm)]. The electron current and electron density of
the electron shower were 200 mA and 0.1 mA/cm2, respectively.
The surface of the substrate was irradiated with the electrons so that the surface
was activated to enhance chemical reaction. The substrate temperature was 300°C,
which was due to the heat from the W filament for the electron shower. After 5 min,
Cu (99.9%) was evaporated from a heated Mo boat. The evaporated Cu atoms passed
through the electron shower to become ionized or electronically excited, and deposited
on the substrate. The electron shower continued during Cu film deposition. The
thickness of the Cu films was about 0.15 pm. Oxygen gas (99.9%, 1.5 x 10-3 Torr)
was sometimes introduced into the chamber from an injection pipe (0 4.2 mm). The
gap between the injection pipe and the substrate was about 10 cm.
When the electron shower was not operating, this apparatus was used for conven-
tional thermal evaporation. Therefore, the Cu films were deposited by two methods:
electron shower and thermal evaporation.
The adhesion of the Cu films was measured by the tensile method (pull rate of
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
0.5 mm/min). Cylindrical rod-shaped brass studs were glued to both sides of the
film-substrate sample using epoxy resin as the glue. The adhesion strength S is
defined as S = P/A, where P is the load when the Cu film pulls off and A is the
contact area. The number of test samples was 3-5 for each condition.
3.1. Adhesion
The surfaces of type 304 stainless steel substrates were polished and anodically oxi-
dized (1-3 A/cm2) in a solution (HN03-water, 4 : 6) to obtain a mirror finish. After
that, they were chemically polished in a HN03 solution to obtain a smoother surface
and to remove Cr hydroxide from the surface. These electropolished stainless steel
samples were received from Nichizo Seimitu Kenma Co., Ltd.
The results of the adhesion measurements are summarized in Table 1. Heating the
substrate at 300°C increased the adhesion from 3 to 13 MPa for films deposited by
thermal evaporation (TE). When the electron shower (ES) was used, the adhesion was
greatly enhanced and the Cu film did not delaminate before failure of the epoxy resin
bond. The strength in this case was about 20 MPa, which corresponds to the adhesion
of the epoxy [4].
When oxygen gas was introduced into the vacuum chamber, the adhesion was
increased. The adhesion strengths for TE+ 30 °C+02 and TE+300°C+02 were 5 and
668
Table 1.
Adhesion of Cu films deposited on anodically oxidized stainless steel
substratesby electron shower (ES) and by thennal evaporation(TE)
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
16 MPa, respectively, as compared with 3 and 13 MPa for TE+30°C and TE+300°C,
respectively. In the case of ES with oxygen introduction (ES + 02 + 300"C), the Cu
film adhesion was too high and the film did not pull off. In the case of glass substrate,
the adhesion strengths of ES + 300°C and ES + 300°C + 02 were 11and > 20 MPa,
respectively [5]. Therefore, the adhesion was increased by the addition of oxygen
gas also in the ES method. It is considered that both thermal diffusion and chemical
bonding at the interface affect the adhesion, but it is not known which effect is more
dominant.
Figure 2 shows the depth profiles of Cu 2p3/2 obtained from XPS. The substrate
temperature of the four samples is the same (300°C), but the depth profile is different
depending on the deposition method. The relative peak intensity of Cu decreases
from 1 to 0 at the interface, which shows apparent diffusion of Cu. This diffusion is
the highest in the case of ES + 02 + 300°C, which is a strange result, because the
addition of oxygen gas enhances the formation of oxides such as CuO and Cu20 at
the interface and the oxides prevent the diffusion of Cu. We think that this is due to
the difference in the thickness of the Cu oxides. The Ar etching rates for the XPS
measurement are different among the samples. The depth profiles do not show the
diffusion of Cu correctly.
Figure 3 shows the depth profiles of Fe 2p3/2 of the substrate. Fe diffuses from the
stainless steel substrate into the Cu film and Cu diffuses into the substrate. However,
as the diffusion coefficient of Fe into Cu (1.01 cm2 / s) is lower than that of Cu into Fe
(9.8 cm2/s) [6], we neglect the diffusion of Fe. The difference in the depth profiles
of Fe 2p3/2 is mainly due to the Ar etching rate for XPS, and not to the diffusion of
Fe. We define te as the Ar etching duration from the appearance of Fe 2 p3 /2 until its
saturation. Short te means that the thickness of the Cu oxides is thin. The value of te
was varied by the preparation conditions:
669
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
The value of te was the same for samples (a) and (b). The formation of the oxides
was increased by the introduction of oxygen and the ES enhanced the formation of
oxides more than the TE. We assume that the true etch rate is inversely proportional
to te. The modified relative etching time tm (dimensionless) is defined as t/te, where
t is the Ar etching time.
Figure 4 shows the depth profiles of Cu 2p3/2 for the modified relative etching time
tm. We think that these curves express the thermal diffusion of Cu more correctly,
compared with Fig. 3. It should be noticed that the depth profile obtained directly
from XPS does not always show the thermal diffusion correctly. It was found that
when the substrate temperature was the same (300°C), the depth profiles of Cu 2 p3 /2
followed almost a curved line, which was independent of the preparation condition.
Thus, the differences in the thermal diffusion of Cu due to the preparation condition at
the same deposition temperature cannot be distinguished by XPS, and the difference
in the adhesion cannot be explained by the thermal diffusion. But as the diffusion
length of Cu was increased by heating the substrate from 30 to 300°C, the increase
in adhesion by heating in the TE method without oxygen gas was mainly due to the
thermal diffusion of Cu into the substrate.
Figure 5 shows the Cu 2p3/2 spectra for Cu films prepared previously [1]. Spec-
trum (a) was measured at the interface of the ES + 300 °C sample, and spectrum (b) at
that of the TE + 30°C sample. These films were deposited on wet polished stainless
steel substrates. The peak energy of (b) is 932.9 eV, close to the binding energy of
Cu in the metallic state (932.5 eV) [7]. Spectrum (c) is the residue of (a) - (b),
671
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
Figure 5. Cu 2 p3/2spectra [1]: (a) ES + 300°C, (b) TE + 30°C, and (c) residue of (a) - (b).
which represents the electronically excited net Cu spectrum. The peak value of spec-
trum (c) is 933.5 eV, so that chemical shift occurred in the ES sample. The depth
profile of spectrum (c) was obtained by repeating this spectrum separation at different
Ar etching times. The maximum value of the depth profile of (c) was 933.8 eV,
corresponding to CuO [8]. As the binding energies of Cu metal and Cu20 are the
same, it is difficult to separate them. The formation of Cu20 was, however, confirmed
from the Auger electron spectrum of Cu L3M4.5M4.5 [1]. The high adhesion due to
the electron shower is caused by chemical bonding such as Cu oxides (Cu20, CuO)
with oxygen on the wet polished substrate.
Figure 6 shows the depth profiles of Cu 2P3/2 for the Cu film deposited by TEA+02 +
300°C. When oxygen gas was introduced, the chemical shift (maximum 933.2 eV) of
the Cu peak was observed even in the TE sample. The high adhesion due to oxygen
gas was caused by enhancement of the chemical bonding.
In a previous study [1, 2], Cu films were deposited on wet polished stainless steels.
They were polished by 4000-grit SiC abrasive paper with water as the lubricant, to
672
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
Figure 6. Depth profileof Cu 2 p3 /2for the Cu film preparedby TE+02 + 300 °C on anodicallyoxidized
stainless steel substrate. The peak shifted from 532.8 to 933.2 eV with increasingetching time.
obtain a mirror finish. Their adhesion was lower than those deposited on the anodically
oxidized stainless steels and the values were (a) 0.08 MPa (TE + 30°C), (b) 2 MPa
(TE + 300°C), and (c) > 8 MPa (ES + 300°C). In the case of ES + 300°C, the Cu
film did not delaminate before failure of the epoxy resin bond (8 MPa). But the bond
strength of epoxy resin used in the previous study was lower than that in this work.
The adhesion of the TE + 30°C sample deposited on the wet polished substrate was
very low.
Figure 7 shows the depth profiles of Cr 2 p3 /2 and 0 Is for a Cu film deposited on
a wet polished stainless steel by (a) TE + 30 °C and (b) TE + 300 °C. Figure 7a shows
that there are two peaks in Cr (574.1 and 577.0 eV) and 0 (530.5 and 531.5 eV).
Table 2 shows the chemical shifts of Cr 2p3/2, O 1 s, and Cu 2p3/2, obtained from
the literature [7-9]. The Cr peak at 574.1 eV was the Cr metal state and the oxy-
gen peak at 531.5 eV was Cr(OH)3. The binding energies of 577.0 eV of Cr and
530.5 eV of 0 correspond to Cr(OH)3.0.4H20. The depth profile of oxygen shows
that Cr(OH)3 0.4H20 was formed at the interface and became Cr(OH)3 close to the
673
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
Figure 7. Depth profiles of Cr 2 p3 /2and 0 Is for a Cu film deposited on a wet polished stainless steel:
(a) TE + 30"C and (b) TE + 300 ° C.
substrate. Figure 7b shows that the peak of Cr(OH)3 .0.4H20 disappeared upon heat-
ing the substrate (TE + 300°C), but that Cr(OH)3 existed. The depth profiles of Cr
and oxygen for the ES sample were similar to those for the TE + 300°C sample [2].
674
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
Figure 7. (Continued).
Figure 8 shows the depth profiles of Cr 2 p3/2 and 0 Is for a Cu film deposited
by TE + 30°C on anodically oxidized stainless steel. Many oxide peaks are included
in the 0 Is wide spectrum: Cr203, Cr(OH)3.0.4H20, Cu20, CuO, and Cu(OH)2, but
Cr(OH)3 is not. In this case, the substrate was anodically oxidized and chemically
675
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
Bindingenergy(eV)
Figure 8. Depth profiles of Cr 2 p3 /2and 0 Is for a Cu film deposited by TE + 30°C on anodically
oxidized stainless steel. The peak of 0 Is shifted from 530.5 (Cu20) to 530 eV (Cr203) with increasing
etching time.
polished to remove the hydroxide, so that it was difficult to form Cr(OH)3. As the
adhesion on the anodically oxidized substrate is higher than that on the wet polished
one, it is said that the adhesion on the oxide (Cr203) surface is higher than that on
the hydrated [Cr(OH)3 .0.4H20] and hydroxide [Cr(OH)3] surfaces.
676
Table 2.
Peak positions of Cr 0 ls, and Cu 2 p3 /2[7-9]
Downloaded by [Simon Fraser University] at 19:07 11 November 2014
4. CONCLUSION
The addition of oxygen gas enhanced the formation of Cu oxides such as CuO and
Cu20, and the adhesion strength was mainly due to these chemical bonds. As these
Cu oxides at the interface affected the Ar etching rate for XPS, the depth profile of Cu
measured by XPS did not represent the thermal diffusion of Cu correctly. We proposed
a correction method for the Ar etching rate. This correction showed that the diffusion
of Cu became almost the same among samples deposited at the same temperature,
and the effect of the thermal diffusion on the adhesion was not so important.
Cr oxide (Cr203) was mainly formed on the stainless steel substrates by anodic
oxidation but Cr hydrate [Cr(OH)3.0.4H20] and Cr hydroxide [Cr(OH)3] were formed
by wet polishing. The hydrate was removed by heating the substrate at 300 °C, but
the hydroxide was not. These oxides, hydrate, and hydroxides affected the adhesion
of Cu films to stainless steel. The adhesion of the Cu films deposited on the Cr
hydrate at 30°C was very low (0.08 MPa), and the adhesion on the hydroxide at high
temperature (300°C) was 2 MPa. Although Cu films were deposited on the Cr oxide
at a low temperature (30°C), the adhesion was still high (3 MPa).
In the case of the electron shower method, the adhesion of Cu films to the Cr oxide
was extremely high, and the Cu films did not pull off before failure of the epoxy resin
bond. The adhesion in this case was 20 MPa.
REFERENCES
1. H. Yumoto,K. Watanabe,K. Akasi and N. Igata, Appl. Surface Sci. 48/49, 173-177 (1991).
2. H. Yumoto,K. Higai and T. Ariyama, J. Surface Finish. Soc. Jpn. 46, 553-557 (1995).
3. P. Benjamin and C. Weaver,Proc. R. Soc. London, Ser. A 261, 516-531 (1961).
4. A. Kikuchi, S. Baba and A. Kinbara, J. Vac.Soc. Jpn. 25, 258-262 (1982).
5. H. Yumoto,S. Shimada, T. Inoue, N. Igata and T. Ariyama, in: Proceedings of International Confer-
ence on Microstructuresand Functions of Materials, pp. 205-208. The OrganizationCommittee of
ICMFM '96, Science Universityof Tokyo Press, Tokyo (1996).
6. The Japan Institute of Metals, in: KinzokuDatabook, p. 25. Maruzen, Tokyo (1974).
677