ARTICLE IN PRESS
Tribology International 40 (2007) 1526–1530 www.elsevier.com/locate/triboint
Microtribological analysis of gold and copper contacts J. Barrigaa,, B. Ferna´ndez-Diaza, A. Juarrosa, S.I.-U. Ahmedb, J.L. Aranac a
Tekniker, Tribology Unit, Otaola 20, 20600 Eibar, Spain Technische Universita¨t Ilmenau, 98693 Ilmenau, Germany c Faculty of Engineering, 48013 Bilbao, Spain
Received 21 July 2006; received in revised form 20 November 2006; accepted 4 January 2007 Available online 1 March 2007
Abstract There are millions of electrical contacts subject to relative motion: connectors, relays, chips in cards, switchesy Friction and wear in this kind of devices are still a source of concern. Even more, recent developments in MEMS RF switches make compulsory a further understanding of tribological processes in order to get higher operating life (1011 cycles). In this work two commonly used electrical conductors are characterized and compared: copper and gold. Both materials have been deposited by PVD sputtering on silicon wafers and plano-convex lenses. Surfaces were characterized by AFM and roughnesses around 1 nm are obtained. Tribological testing with normal loads in the range 1–20 mN have been carried out. Gold presents quite a constant friction (0.20) over a wide range of relative humidity values. However, copper presents lower friction (0.10) at 33% RH and higher friction when humidity is increased. Contact angle measurements have been performed on both surfaces (Au and Cu) using two different liquids: water (polar) and diiodomethane (non-polar). Surface energy and interfacial energy calculations show that energy in the gold–water interface (23 mN/m) doubles that of copper–water interface. Capillary forces play a key role generating friction in these contacts and water absorption capability of both materials determine their frictional properties in the analyzed range of relative humidity. r 2007 Elsevier Ltd. All rights reserved. Keywords: Microtribology; Gold; Copper; Capillary forces
1. Introduction There are millions of electrical contacts subject to relative motion: connectors, relays, chips in cards, switchesy Even more, with the development of RF MEMS switches there is a need of fundamental knowledge on the microtribological properties of conducting materials contacts. Gold and copper are frequently used conductors [1,2]. The contacts in electrical connectors are also commonly gold plated or gold ﬂashed. Contrary to popular belief, this is not done because gold is a better conductor; it is not. Instead, it is done because gold is very resistant to the surface corrosion that is commonly suffered by copper or silver . Currently, maximum RF MEMS switch lifetimes are around 1010 cycles. It is believed that by developing a set of tribological design rules switches could operate in excess of Corresponding author. Tel.: +34 943 206744; fax: +34 943 202757.
E-mail address: [email protected]
(J. Barriga). 0301-679X/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.triboint.2007.01.009
1011 cycles . Stiction of the moving components is the main factor that produces working failures [5,6]. That is why some attempts have been made to develop hydrophobic layers to reduce the absorption of water by the surfaces [7,8]. 2. Materials In this work two different conducting materials have been considered: copper and gold. At 20 1C electrical resistivity of copper is 16.78 nO m and gold presents higher resistance: 22.14 nO m . Both materials have been deposited using physical vapor deposition (PVD) sputtering technology in a clean room (see Fig. 1). Coatings have been deposited over two different substrates: silicon wafer (1 0 0) and borosilicate lenses. The choice of coated lenses was performed in order to be able to make pin on disc tribological tests with very good surface ﬁnishing. Lenses present a curvature radius of 7.78 mm and a diameter of 3 mm (see Fig. 2).
ARTICLE IN PRESS J. Barriga et al. / Tribology International 40 (2007) 1526–1530
Approximately the thickness of the coatings is 150 nm. After the deposition processes, topography of coated surfaces has been characterized using an AFM (NTMDT, Solver Pro). Roughness in 3D has been measured in semicontact mode with a silicon tip (curvature radius 10 nm; k ¼ 11,5 N/m) scanning 3 3 mm2 at a frequency of 1 Hz. Projected (2D) topographies can be seen in Fig. 3. Ra values are 0.90 and 0.80 nm for gold and copper, respectively. Rq values are identical for both surfaces: 1.15 nm. Thus, roughness of both Au and Cu surfaces are in the range of the nanometer. 3. Experimental details Wettability and frictional properties of gold and copper surfaces have been characterized. Contact angles are parameters of interest in systems where the interface is
Fig. 1. Friction results for Au–Au, Cu–Cu and Cu–Au contacts at 33% RH.
Fig. 2. Friction results for Cu–Cu contacts at 33% and 84% RH.
important, as for example, the surface wettability, surface or interfacial energy in the solid, the analysis of the surface in different materials where the problem of the adhesion exists, all those present in the micro/nanotribological phenomena. Through contact angles it is possible to know how much hydrophilic/hydrophobic is a surface when a liquid is in direct contact with it. Contact angle is a quantitative measurement of the wetting of a solid by a liquid. It is deﬁned geometrically as the angle formed by a liquid at the three phase boundary where a liquid, gas and solid intersect in a point. This value depends on the superﬁcial energy of the substrate just as superﬁcial tension of the liquid. If the contact angle is greater than 901, the surface is said to be hydrophobic and if it is less than 901, hydrophilic. Contact angles were determined putting a small drop of the liquid on the surface of the solid substrate. Then, the drop was illuminated with diffuse light in order to get a clearer image in the edges of the drop. The image of the drop was taken with a CCD and processed by a computer to calculate contact angle. In this work we have used a CSM microtribometer in linear reciprocating mode to study the friction properties of the samples. The mode of operation of this device is very simple: A known load is applied through a cantilever (Fig. 4) with a small sample attached to its edge (in this case, the lenses). The sample slides on the surface and the coefﬁcient of friction is determined during the test from the deﬂexions that the cantilever suffers. A steel cantilever was used (Kx ¼ 169 N/m; Kz ¼ 138 N/ m). Testing conditions were selected in order to obtain proper friction plots versus time. In the Fig. 5 we can observe two different friction plots versus time. On the left, there is a typical graph for a reciprocating movement. On the right, testing conditions and cantilever choice are incompatible: the stiffness is too low in the lateral direction and the sample cannot follow properly the imposed movement. Severity index is usually deﬁned as the product of the coefﬁcient of friction, the normal load applied and the speed (SI ¼ m Fn v), which is a measure of the power
Fig. 3. Friction results for Au–Au contacts at 33% and 84% RH.
ARTICLE IN PRESS 1528
J. Barriga et al. / Tribology International 40 (2007) 1526–1530
Fig. 6. Contact angle measurements for gold and copper surfaces with water DI and diiodomethane test liquids.
Fig. 4. Cantilever to apply normal force and measure friction force in the microtribometer.
Fig. 5. Friction plots versus time in reciprocating movement. On the right, testing conditions and cantilever choice are incompatible: the stiffness is too low in the lateral direction.
(energy/time) dissipated in the contact. We have related this magnitude to the lateral stiffness of the cantilever Kx and we deﬁne a new parameter, A ¼ SI/Kx. We noticed that when A is in the range of 108–109 m/s2 proper friction plots are obtained and sliding takes places. With this empirical rule, we selected the testing conditions for the microtribological tests: Sliding speed was 32 mm/s and the stroke was 400 mm. Load was varied between 1 and 20 mN. Tests were performed at room temperature (20 1C). 4. Results and discussion When a polar solvent, as, for example, deionized water, is used as test liquid, the contact angle is mainly a measurement of the polarity of the surface, i.e., the grade of hydrophobia/hydrophilia of the surface. On the other hand, if the liquid tested is a non polar one, as for example diiodomethane, contact angle shows the attractive forces to the surface owing to the van der Waals forces.
Results of contact angle measurements are shown in Fig. 6. We can observe that copper surfaces are more hydrophobic than gold surfaces. This means that gold surface will absorb more water from the environment. For determining the superﬁcial energy of a substrate a method can be used based on the Owens–Wendt model : ﬃ qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃi 2 hqﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 þ cos y ¼ gdsg gdlg þ gpsg gplg (1) glg where y is the contact angle, glg is the liquid surface tension (liquid–gas) and gsg is the solid surface tension (solid–gas), or free energy. The addition of d and p in the subscripts refer to the ‘‘dispersion’’ and ‘‘polar’’ components (Table 1). When contact angle of a liquid is used to calculate the surface energy of a substrate there are several details that must be taken into account in order to ensure the correct application of Eq. (1): The substrate has to be atomically ﬂat and chemically homogeneous. We assume that these approximations are applicable to our samples. Taking into account the balance of forces (see Fig. 7) we obtain the so called Young’s equation gsg ¼ gsl þ glg cos y
which is necessary to calculate the interface energy between the solid surfaces (gold or copper) and water. Calculations give values of 23 and 12 mN/m for water/gold and water/ copper interfaces, respectively. On the other hand, frictional tests were conducted. In the Figs. 8– 10 there are graphical representations of friction forces versus applied normal forces in the microtribological tests. As there is a linear relationship we can calculate a coefﬁcient of friction for each set of points. In the Fig. 8 there is a comparison of results obtained in Au–Au, Cu–Cu and Cu–Au contacts (in this last case the ﬂat sample was coated with copper and the upper specimen—lens—was gold coated). We observe that in gold contacts friction coefﬁcient is twice the value obtained for copper contacts (0.21 and 0.10, respectively). In the combined Cu–Au contact, we obtain the same value for the coefﬁcient of friction than for Au–Au contact. Gold (more hydrophilic) dominates in the frictional properties. This suggests that
ARTICLE IN PRESS J. Barriga et al. / Tribology International 40 (2007) 1526–1530
Table 1 Surface energy of gold and copper Au
The surface energy is the sum of the polar and dispersive components.
Fig. 7. Drop of liquid on a solid surface: Schematic drawing of forces in the interface.
4.5 4 3.5 3 2.5 2 1.5 1 0.5 0
Au-Au contacts 33% RH
µ = 0.21 µ = 0.18
Fig. 10. Friction results for Au–Au contacts at 33% and 84% RH.
5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0
Au-Au Cu-Cu Cu-Au
µ = 0.21 µ = 0.10 µ = 0.21
Fn (mN) Fig. 8. Friction results for Au–Au, Cu–Cu and Cu–Au contacts at 33% RH.
observe that in both cases (33% and 84% RH) similar values of friction are obtained (0.2). If capillary force (through water condensation on the surfaces) is the main mechanism generating friction in our system, it seems that at 33% RH the saturation value is already reached. Note that friction values for Cu–Cu contacts at 84% RH are identical to Au–Au contacts. This indicates that water condensation reached its saturation value at high humidity but not at 33% RH. Even more, we conclude that friction values are independent of material surface.
5. Summary and conclusions 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0
Cu-Cu contacts 33% RH 84% RH
µ = 0.10 µ = 0.21
Fn (mN) Fig. 9. Friction results for Cu–Cu contacts at 33% and 84% RH.
capillary forces are the main mechanism involved in generating resistance to the movement. In Fig. 9 there is a comparison of frictional forces for Cu–Cu contacts in different humidity values. We observe that an increase of 50% in RH involves an increase of 100% in the coefﬁcient of friction. In the Fig. 10 there is a comparison of frictional forces for Au–Au contacts under different humidity values. We
From the work presented in this paper the following conclusions can be extracted: Surface energy and interfacial energy calculations show that energy in the gold–water interface (23 mN/m) doubles that of copper–water interface. Thus gold surface will absorb more water from the environment. An increase of 50% of relative humidity produces an increase in the friction coefﬁcient in Cu–Cu contacts. Capillary forces (typically in the range of mN) produce this raise in the resistance to movement: the effective load is the sum of the applied load and the capillary force and so the apparent coefﬁcient of friction increases. However, in Au–Au contacts no difference was observed in the studied range. Under ambient conditions copper contacts produce lower friction than gold surfaces (0.1 and 0.2, respectively). Friction values for Cu–Cu contacts at 84% RH are identical to Au–Au contacts. This indicates that water condensation reached its saturation value at high humidity but not at 33% RH.
ARTICLE IN PRESS 1530
J. Barriga et al. / Tribology International 40 (2007) 1526–1530
In the combined Cu–Au contact, we obtain the same value for the coefﬁcient of friction than for Au–Au contact. Gold (more hydrophilic) dominates in the frictional properties. This suggests that capillary forces are the main mechanism involved in generating friction in our system. Once saturation is obtained in the water condensation on copper and gold surfaces capillary forces play a key role producing friction and friction values are independent of surface material. Acknowledgments Authors would like to thank the Basque Government for ﬁnancial support of this work through SAIOTEK programme. Thanks also to Pamela Dickrell (University of Florida) for her suggestions and to Ferran Xinxo (CNMBarcelona) for the gold coatings. References  Hsu T-R, Materials for MEMS and microsystems, MEMS and microsystems: design and manufacture. McGraw-Hill, 2002, [Chapter 7] ISBN: 007-239391-2.
 Patton ST, Zabinski JS. Fundamental studies of Au contacts in MEMS RF swtches. Tribol Lett 2005;18(12).  Holmberg K, Matthews A. Coatings tribology. ISBN: 0-444-88887051994. Elsevier Science; 1994. p. 155–71.  Brown C, Krim J, Morris A. Analysis of cycle lifetimes and failure modes for RF MEMS switches, WTC2005-63733. Washington: World Tribology Congress; 2005.  Zhuang YX, Menon K. Capillary induced stiction and adhesion of MEMS materials, WTC2005-63203. Washington: World Tribology Congress; 2005.  Borruto A, Crivellone G, Marani F. Inﬂuence of surface wettability on friction and wear tests. Wear 1998;222:57–65.  Perry SS, Laboriante I, Yan X. Vapor phase lubrication of gold/gold interfaces, WTC2005-63773. Washington: World Tribology Congress; 2005.  Kiely JD, Houston JE, Mulder JA, Hsung RP, Zhu XY. Adhesion, deformation and friction for self assembled monolayers on Au and Si surfaces. Tribol Lett 1999;7(2–3):103–7.  Ashcroft NW, Mermin ND. Solid state physics. Ed: Harcourt, 1976. ISBN: 0-03-083993-9.  Owens DK, Wendt RC. Estimation of the surface energy of polymers. J Appl Polym Sci 1969;13:1741–7.