- Email: [email protected]
Effect of roughness on electrical contact performance of electronic components
Microelectronics Reliability 74 (2017) 100–109
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Effect of roughness on electrical contact performance of electronic components Xin-long Liu, Zhen-bing Cai ⁎, Shan-bang Liu, Jin-fang Peng, Min-hao Zhu ⁎ Tribology Research Institute, Traction Powder State Key Laboratory Southwest Jiaotong University, Chengdu 610031, China
a r t i c l e
i n f o
Article history: Received 9 March 2017 Received in revised form 17 May 2017 Accepted 18 May 2017 Available online xxxx Keywords: Fretting Contact resistance Contact spot Roughness Electric contact
a b s t r a c t This study investigated the effects of electrical contact resistance (ECR) on pogo pins used in mobile phones, chargers, digital cameras, Bluetooth headsets, medical equipment, and other electronic products with different surface roughness. Experimental results revealed that metallic wear debris is generated by fretting motion and formation of a third body on rough surfaces without removal by fretting motion, thus increasing ECR. Wear debris does not easily form the third body at contact areas of smooth surfaces and causes formation of metal–metal contact pattern. Results showed low ECR with fretting motion. 3D and 2D profiles of contact area verified the definition of contacting high spots, further explaining increases in ECR. © 2017 Published by Elsevier Ltd.
1. Introduction Pogo pins are typically used as connectors in mobile phones, chargers, digital cameras, Bluetooth headsets, automotive parts, medical equipment, and other electronic products (Fig. 1) [1]. Maintenance of stable low-value electrical contact resistance (ECR) in electrical connectors and other contact-containing components is essential to protect circuits that contribute to degradation to facilitate correct selection of contact materials and component designs [2]. Unstable ECR values may lead to electrical degradation or temperature increment at contact areas and eventually result in circuit breaker or battery explosion [3]. ECR between pogo pins/plane contacts strongly depends on surface integrity of contact areas. For connectors subject to vibration, fretting is considered a primary degradation mechanism; it is a relative cyclic motion with small amplitude occurring between two oscillating surfaces. In another study, fretting was found to cause intermittent electrical contact, wear, and corrosion on contact materials and caused variations in results of ECR [4,5]. Although pogo pins feature low cost and convenience in manufacturing, they raise a major concern due to their high interfacial contact resistance (CR) caused by mechanical polishing of their surfaces. CR is generally governed by electrical properties of interface layer between contacting surfaces [6]. As a result of surface roughness at microscopic scale, current flows only through these asperities, which occupy a
⁎ Corresponding authors. E-mail addresses: [email protected] (Z. Cai), [email protected] (M. Zhu).
http://dx.doi.org/10.1016/j.microrel.2017.05.024 0026-2714/© 2017 Published by Elsevier Ltd.
small fraction of areas of nominal contacting surfaces in theory. Ra ≤ 1 μm is a processing requirement for surface roughness of commonly used electrical contact connectors [7,8]. ECR depends on materials, diffusion media, compression pressure, surface roughness of contacting materials, and other conditions used during measurements [9,10]. Clearly, studies should research and develop low-cost and reliable solutions for reducing ECR and improving performance of electrical connectors. Further studies must also determine the effects of ECR of pogo pins/ plane contact burring by roughness. Fig. 1 presents the common working state of pogo pins. Pre-tightening force usually exists between pogo pins/contactors. Contactors utilize copper as material. For this study, red copper was the preferred material for contact manufacturer owing to its unique combination of conductivity, strength, stiffness, formability, and low cost [11]. Red copper contacts are usually plated with noble metals to improve their electrical contact performance [12,13]. Pogo pin connectors are the most critical links of electronic and power systems and are needed to provide paths for electronic signals and/or power connections. ECR between pogo pins and copper plates in electronic components is governed by multi-scale surface topography of contacting pairs [14]. The roughness featured at contacting surfaces decreases actual contact areas, leading to a voltage drop across interface [15]. Assembling relationship between electrical components applies pressure at interfaces, leading to increases in contact areas between electrical components and subsequent decreases in ECR [9]. However, excessively large pre-tightening forces may cause difficulty in assembly and deformation of pogo pins [16,17]. Thus, an optimum clamping pressure exists and trades off between competing requirements.
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
101
Fig. 1. (a) Camera connector, (b) direction of movement of contactor, (c) matched contactor.
2. Experimental 2.1. Experimental details As shown in Fig. 2(a), ECR fretting test system was constructed for pogo pins and plane contacts. Fig. 2(b–c) shows the schematic diagram for a measuring apparatus for ECR distribution; the four-terminal method is one of the most commonly used methods in determining contribution of resistance present at interfaces between two conductors [6]. Table 1 provides the conditions for measurement of ECR. Fretting experiments were carried out using the ECR fretting test system manufactured by Tribology Research Institute, Southwest Jiaotong University. The detailed structure of the test is as follows;
displacement was measured by a laser sensor (Figs. 2(a)-7). The sensor (Fig. 2(a)-3) under the lower sample was used to measure friction forces and normal loads and to transmit data to collection cards manufactured by NI Company. The upper sample comprised pogo pins (C3604: Cu, 57.0%–61.0%; Fe ≤ 0.5%; Pb, 1.8%–3.7%; Sn: Fe + Sn ≤ 1.2, and balanced total impurities). Pogo pins employed standard parts to obtain same initial roughness. Red copper was used as plane sample material (C11000: Cu, 99.9%; P, 0.0116%; Fe, 0.016%; Pb, 0.0019%; S,0.0047%; Zn,0.0216%; Sn, 0.0034%, and balanced total impurities). Pogo pins and plane sample contacts were degreased with alcohol and coal oil using an ultrasonic cleaner, dried, and carefully mated in fretting test assembly to create a point contact in pogo pin/plane geometry (Fig. 2(c)).
Fig. 2. (a) Electrical contact resistance (ECR) fretting test system (1. Piezoelectric ceramic actuator. 2. Upper fixture 3. Force transducer. 4. Flat sample. 5. Ball sample. 6. Precision lead screw 7. Displacement sensor). (b) Operating principle of fretting equipment and resistance measurement mechanism. (c) Size of upper and lower samples.
102
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
0.646, 0.268, and 0.116 μm, and they met processing requirements of pogo pin connectors.
Table 1 Details of experimental conditions. Roughness Frequency Amplitude Normal load Current load
0.989, 0.646, 0.268, and 0.116 μm 2 Hz ±10, ±20, and ±45 μm 2N 0.02 A
2.2. Samples Fig. 3 presents surface roughness of plane samples prior to fretting test. Four samples with same material but with different surface roughness were lined up. Preparation for samples involved selection of different specifications of sandpaper to grind sample surface (details in Fig. 3) and inclusion of a polished sample. Afterward, four different surface roughness were obtained. A contour GT-X 3D optical microscope [18] was used to measure surface roughness; roughness measured 0.989,
3. Results and discussion 3.1. Electrical response as a function of different surface roughness Each experimental condition was performed thrice to determine three characteristic curves with the same changing trend and to select the test with median value of ECR. When no value was selected, fretting tests were conducted again, and middle value of all tests was finally selected. For data analysis, the following steps were used. First, resistance data from a large number of original data were extracted to redraw ECR curves. As described above, ten original resistance (three kinds of displacement amplitude and four kinds of roughness) data during each fretting cycles were recorded, and 200,000 resistance data were collected in 104 cycles. Original curve was extracted to obtain understandable
Fig. 3. Sample preparation and method of roughness measurement.
Fig. 4. (a) ECR of contact pairs. (b)–(e) Wear morphology of contact pairs (Ra1 = 0.989, Ra2 = 0.646, Ra3 = 0.268, and Ra4 = 0.116).
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
103
Fig. 5. SR (average value) and instantaneous maximum of ECR of contact pairs.
resistance curve and to quantitatively analyze data. Trials were conducted on average (Rv), maximum (Rmax), and minimum (Rmin) values of every 100 cycles in the original resistance data. Rv was obtained according to the following Eq. (1) [19]: Rv ¼
ðRmax þ Rmin Þ 2
ð1Þ
After obtaining 100 valid algorithm data, ECR curve was constructed. Fig. 4 shows ECR and all selected contact pairs during 104 cycles. As shown in the figure, different surface roughness remarkably influenced extent of fretting tests in all contact pairs. ECR increased with increasing surface roughness. Contact pairs with larger roughness showed higher ECR until the end of fretting, but ECR curve was insufficient for evaluation of properties of all selected samples. Average value, that is, statistical resistance (SR), which was extracted from ECR (average of every 1000 cycles and obtained 10 points), increased at 8000 cycles (Fig. 5). For polished samples, the effect to SR by different roughness was insignificant at initial fretting test as polished samples contained maximal contact areas [20]. As a result of fretting, ECR of materials may suddenly enlarge and then reduce to low values in complete tests. To make the results more intuitive and abundant with selected contact pairs, instantaneous maximum (Rcmax) and average value of ECR (Rcaverage) during completed cycles were introduced in analysis (Fig. 5(b)). As expected, Rcmax of Ra1 (0.989) reached 2722.8 mΩ, which is the highest compared with other roughness values. On the contrary, the lowest roughness was that of Ra4 (0.116). Overall, Rcmax and Rcaverage increased with increasing roughness values. 3.2. Electrical response as a function of sliding regime condition
Fig. 6. Evolution of ECR as a function of different displacement amplitudes.
To obtain more intuitive results for comparison, two extreme values of roughness (Ra1 and Ra4) were selected to analyze electrical response as a function of sliding regime condition. Representative shapes of fretting loops from the beginning (103 cycle) and end of tests (104 cycle) were included to indicate fretting regime. Fig. 6 presents the corresponding friction force/displacement (F-D) curve, which was filtered from original data and redrawn based on 103 and 104 cycles. Displacement contributing to ECR exhibited the same trend whether under rough or polished surfaces. As long as displacement amplitude permits a stabilized partial slip condition, which can be assessed from the F-D curve, a sticking zone is maintained during this condition. Direct metal/metal contacts were generated with lower and stable ECR. This
104
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
Fig. 7. (a) Instantaneous maximum value of ECR (b)–(g). Wear morphology of contact pairs ((b) D1 = ±10 μm, Ra1 = 0.989; (c) D2 = ±20 μm, Ra1 = 0.989; (d) D3 = ±45 μm, Ra1 = 0.989; (e) D1 = ±10 μm, Ra1 = 0.989; (f) D2 = ±20 μm, Ra1 = 0.989; (g) D3 = ±45 μm, Ra1 = 0.989, Ra4 = 0.116).
phenomenon can explain why low ECR is observed under stabilized partial slip situations. When displacement amplitude becomes higher (±45 μm) than transition value (almost ±20 μm) [12,21], then the fretting condition is gross slip; with high value of ECR, the generalized gross slip condition activates debris formation over rough contact areas. For further illustration, polished surface (Ra4) exhibited lower value of ECR while under three displacement amplitudes in comparison with rough surface (Ra1). With the motion of fretting, contact area became less smooth and with the rise of ECR (from the 3.72 to 78.80 mΩ). As the fretting close to the end, the contact area of contact became larger and eventually leads the contact resistance dropped to 62.58 mΩ after 8000 cycles [22,23]. When a greater displacement amplitude imposed a stabilized gross slip condition, ECR increased and destabilized. This tendency was clearly illustrated by plotting instantaneous maximum of ECR (Rcmax) as a function of imposed displacement amplitude (see Fig. 7). 3.3. SEM (scanning electronic microscopy)/EDX (X-ray energy dispersive spectroscope) analysis Fig. 8 shows SEM and EDX of typical samples (F = 2 N, D = ± 45 μm). Contact areas of selected samples showed severe delamination and wear debris. Plough and material removal was also observed in
contact areas. Contact surfaces were covered with black oxide debris. This oxide debris at contact surface possibly caused a steep increase in ECR after a few thousand cycles of fretting. EDX analysis of all tested samples are plotted in the middle of Fig. 8. No evidence proves that different loads or roughness yielded different EDX results.
3.4. Relationship between wear mechanisms and CR analysis For theoretical arithmetic of CR to nominal flat rough surfaces, the following assumptions of the Greenwood and Williamson model were adopted [15,24]: (a) Rough surfaces are isotropic. (b) Bulk deformation is absent; only asperities (a-spots) deform during contact. (c) a-spots summits are spherical. (d) All a-spots summits feature the same radius r, but their height distribution varies statistically. (e) Asperities are distant from one another, and no interaction exists among them. For the pogo pin/flat contact model shown in Fig. 9, both contacting asperities and pin deformed to generate an actual contact area that supported normal load F when the pogo pin was pressed against rough flat surfaces. The contact area of pogo pin is flattened, forming a nominal contact area with radius “a” (Fig. 9: vertical view), over which a uniform separation “Ra” is assumed between the pogo pin and mean of surface heights of the rough flat surface.
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
105
Fig. 8. EDX analysis and wear track morphology of all contact pairs.
Physical micro-contact differs from macro-contact due to influence of surface roughness and smaller contact force available in microswitches. For rough flat surfaces, only high points on each surface participate in contact. Effective contact areas, namely, contact spots or a-spots [1], are significantly smaller than apparent ones but participate in transporting electrons.
Manner of electrons transport through electrical connections must be determined to evaluate CR RC: RC ¼
ρ ρ þ 2r 2na
Fig. 9. Schematic of a contact model for pogo pin/flat contact.
ð2Þ
106
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109 Table 2 Characteristics of used materials.
Material
Electrical resistivity ρ(20°) (×10–8 Ω•m)
Yield strength (MPa)
Young′s modulus (GPa)
Poisson′s ratio
Resistance temperature coefficient (aR/°C–1)
Density (g/cm3)
Linear expansivity
1.75
239.689
100
0.33
0.00426(20 °C)
8.5
12
1.678
258.646
110
0.37
0.00393(20 °C)
8.9
16.6
C3604 (Pogo pin) C11000 (flat sample)
where r indicates radius of an apparent contact area, n represents the number of a-spots, ρ corresponds to electrical resistivity, and a indicates a-spot radius. Greenwood released the following CR approximation Eq. (3) for multiple distinct contacts by considering interfaces between a-spots. ai a j ρ ρ þ RC ¼ ∑∑ 2∑ai π i≠ j Sij
!
1 2
ð∑ai Þ
ð3Þ
To investigate CR of the pogo pin/flat rough surface contact, actual contact areas must be identified, and contact modes for single a-spots must be studied (Fig. 9). To calculate a of a single a-spot according Hertz contact theory, measured Rai was used as mean Hertz contact radius for single a-spots. According to this assumption, a can be calculated as follows: a¼
1=3 3ðR1 Rai Þ 1−μ 1 2 1−μ 2 2 F þ E1 E2 4ðR1 þ Rai Þ
ð4Þ
where μi indicates Poisson's ratio, and Ei represents modulus of elasticity. Actual values of material characteristics in Table 2 were used to solve Eq. (4). As shown in Eqs. (3)–(4), Rc was determined by contact resistivity, contact area for a single a-spot, and the number of a-spots within an apparent contact area. As distribution of a-spots was random and scattered, the actual number of contact spots was then measured by contour GT-X 3D optical microscope. Measurement was repeated thrice to minimize deviation. Notably, as threshold value, measured roughness Rai was used to determine the number of contact spots that participate
Fig. 10. Theoretical Rc theoretical and measured average RECR average.
in contact and conduction. In this paper, polished samples were considered to contain no contact spot but only apparent contact areas. Calculated CR Rc theoretical are shown in Fig. 10 (RECR average indicates actual average value of measured original resistance data). Results illustrate higher calculated (Rc theoretical) than actual values (RECR average). This observation resulted from deformation of asperities and pogo pins; such deformation can be elastic, elastic–plastic, fully plastic, or any combination of these deformation types. All procedures mentioned above are only applicable to static-state contact but not to evolution of fretting motion. ECR changes during fretting by various processes involve transfer, material removal, and chemical transformations at contact areas and cause difficulty in evaluating ECR of different surface roughness. Many of these processes were cited in preceding sections, but treatment of contact property can be unified by summarizing them [2]. The relationship between ECR and wear mechanisms is summarized as follows. Fretting at contact areas of copper generated oxide debris at interface. This oxide debris showed high resistance and acted as insulator [21]. Accumulation of oxide debris at interface resulted in increased ECR, although some debris were pushed out of the interface by fretting motion [6]. Therefore, to obtain lower CR under fretting, accumulating oxide debris must be removed from contact areas to ensure metal-tometal contact at interface [25]. Evolution processes of wear mechanism are plotted in Fig. 11(a)–(c), which elaborate different roughness of surface topography for trapping oxide debris. Corresponding circuitry diagram is separately shown on right side of the graph. For the copper with rough surface shown in Fig. 11(a), the upper sample is pogo pin, and the lower sample was copper. The path for current passage was less and narrow due to rough surface. Many contact spots, which are also called a-spots, exist between contact areas [14, 26]. As fretting continues, a-spots were gradually worn by pogo pins. However, as wear debris possibly ploughed and contributed to high friction, contact areas of pogo pins also experienced severe wearing. Interface contained trapped wear debris that were expected to plough surfaces [15]. Accordingly, even in the absence of long ploughing grooves, ploughing mechanism may still be the dominant mechanism. Different from smooth surfaces, rough surface contributed to adhesion of wear debris and eventually caused accumulation of oxide particles in the furrow space. ECR of rough surface contacts can be expressed similarly to a network of parallel resistors [15,27]. Equivalent circuitry diagram demonstrates the evolution principle of ECR in every completed cycle. As shown in this circuitry, Rg represents the CR from pogo pin to wear debris, Rd. corresponds to self-resistance of wear debris, and Rp is the CR between wear debris and copper samples. The contact spot between the interface alternate connections in a complete cycle shows similarity to a switch S connected to alternating “a” and “b”. In conclusion, calculation of total resistance RECR is related to Rg, Rd., and Rp. For the plane sample with smooth surface, ECR remained relatively stable during fretting and possibly increased with appearance of aspots (Fig. 11(c)) [14,28]. In the same way, in the corresponding circuitry diagram on the right side of Fig. 11(c), Rg indicates the CR from pogo pin to wear debris, Rd. corresponds to self-resistance of wear debris, Rp
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
107
Fig. 11. Wear mechanisms analysis.
represents the CR between wear debris and plane sample, and Rgp is the CR from pogo pin to plane sample. Given the minimal wear debris adhering to contact area, the contact model was more inclined to form pogo pin/copper contact, that is, metal-to-metal contact and consequently resulted in lower ECR. Fig. 11(b) shows cross-section profiles of samples with roughness of Ra1 and Ra4 after fretting experiment. Cross-section morphology illustrates that plow and wear debris existed in the contact area with rough surface, and smooth surface showed a different phenomenon. Wear debris Stacked on the edge of the contact area. This study can be summarized as follows: (a) ECR of rough surface exhibited higher values as a result of adhesion of wear debris and finally caused accumulation of oxide particles in the furrow space. (b) Wear debris cannot easily accumulate wear particles at contact areas of smooth surfaces, and inclination to form direct contact patterns is the main reason for lower ECR of smooth surface contacts. Fig. 12 shows the 3D profiles of two representative contact pair with different surface roughness (Ra1 and Ra4) and obtained using a contour GT-X 3D optical microscope. Red objects around the edge of fretting area are oxidized debris. Various degrees of material removal were detected at the contact area. 2D profiles were constructed for each contact pair obtained from cross-sections of 3D profiles (Fig. 12b, e). Fig. 12c and f present the last complete cycle of ECR (LCECR) for each contact pair.
Results of surface profile and LCECR confirmed that different surface roughness did not significantly affect wear depth of plane sample and suggested that other factors were responsible for observed changes in ECR. Details of these other factors are as follows. A very high CR (1021.9 mΩ) was detected and is shown in Fig. 12c. High CR was caused by a-spot formed by many contacting high spots from the pogo pin (such as point “a”) and plane sample (such as point “b”), as shown in Fig. 12b. A similar observation was noted for roughness of Ra4 (Fig. 12f). Lower CR (31.0 mΩ) resulted from two relatively well-fitting surfaces (Fig. 12 e). It can be clearly seen from the 3D morphology shown in Fig. 13, wear debris accumulated at the edge of wear area with polished surface (Ra4) compared with rough surface (Ra1). Overall, samples with relatively rough surface generated a surface with more plows during fretting motion. Profiles of contact pair with polished surface preferentially enlarged the actual contact area through numerous point contacts during fretting motion. In this study, the sample with roughness value of Ra4 presented remarkable electrical contact performance. 4. Conclusions Metallic wear debris is generated by fretting motion and forms a third body that stay on rough surfaces and is hardly removed when under
108
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
Fig. 12. Relationship between contact profiles and LCECR.
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
109
Fig. 13. Cross section of 3D morphology for two contacts: (a) Ra1 and (b) Ra4.
stabilized gross slip conditions. Formation of a third body transforms contact surface into metal–debris–metal pattern and eventually causes increases and instability of ECR. On the contrary, lower displacement amplitude difficult to generate wear debris and show lower ECR. Wear debris hardly formats debris beds at smoother contactors and causes formation of more metal–metal contacts that maintain relatively lower and stable ECR with fretting motion.
Acknowledgement This research was supported by National Natural Science Foundation of China (Contract numbers (No.51375407, 51575459, U1530136, 51627806), supported by Young Scientific Innovation Team of Science and Technology of Sichuan (No.2017TD0017). References [1] N. Amin, Z.L. Yi, A practical investigation on the root causes of the mechanical damages of pogo pin type test sockets to IC packages in final test, IEEE International Conference on Semiconductor Electronics. IEEE Xplore 2008, pp. 393–397. [2] B. Huang, X. Li, Z. Zeng, et al., Mechanical behavior and fatigue life estimation on fretting wear for micro-rectangular electrical connector, Microelectron. Reliab. 66 (2016) 106–112. [3] M. Esteves, A. Ramalho, F. Ramos, Electrical performance of textured stainless steel under fretting, Tribol. Int. 110 (2017) 41–51. [4] M.J. Kim, H.K. Kim, Effects of temperature and span amplitude on fretting corrosion behavior of tin-plated electrical contacts, Microelectron. Reliab. (2017) http://doi. org/10.1016/j.microrel.2016.12.014. [5] J. Yasuda, K. Iida, Y. Saitoh, S. Sawada, Y. Hattori, Y. Hattori, Effect of Intermittent Time on Fretting Corrosion of Tin Plated Coupon, 2011 45–49. [6] E. Mengotti, L.I. Duarte, J. Pippola, et al., Fretting corrosion: analysis of the failure mechanism for low voltage drives applications, Microelectron. Reliab. 54 (9–10) (2014) 2109–2114. [7] D. Qiu, L. Peng, P. Yi, X. Lai, A micro contact model for electrical contact resistance prediction between roughness surface and carbon fiber paper, Int. J. Mech. Sci. 124–125 (2017) 37–47. [8] E. Crinon, J.T. Evans, The effect of surface roughness, oxide film thickness and interfacial sliding on the electrical contact resistance of aluminium, Mater. Sci. Eng. A 242 (1–2) (1998) 121–128. [9] K. Pereira, T. Yue, Multiscale analysis of the effect of roughness on fretting wear, Tribol. Int. (2017) http://dx.doi.org/10.1016/j.triboint.2017.02.024.
[10] B. Avasarala, P. Haldar, Effect of surface roughness of composite bipolar plates on the contact resistance of a proton exchange membrane fuel cell, J. Power Sources (2009) 225–229. [11] I.H. Sung, J.W. Kim, H.J. Noh, H. Jang, Effect of displacement and humidity on contact resistance of copper electrical contacts, Tribol. Int. (2016) 256–261. [12] S. Fouvry, P. Jedrzejczyk, P. Chalandon, Introduction of an exponential formulation to quantify the electrical endurance of micro-contacts enduring fretting wear: Application to Sn, Ag and Au coatings, Wear (2011) 1524–1534. [13] T.S.N.S. Narayanan, Y.W. Park, Y.L. Kang, Fretting corrosion of lubricated tin plated copper alloy contacts: effect of temperature, Tribol. Int. (2008) 87–102. [14] D. Hill, et al., Characterising the surface roughness of AFM grown SiO2, on Si, Microelectron. Reliab. (2001) 1077–1079. [15] J. Molina, R. Valderrama, C. Zuniga, et al., Influence of the surface roughness of the bottom electrode on the resistive-switching characteristics of Al/Al2O3/Al and Al/ Al2O3/W structures fabricated on glass at 300 °C, Microelectron. Reliab. 54 (12) (2014) 2747–2753. [16] T. Yue, M.A. Wahab, Finite element analysis of fretting wear under variable coefficient of friction and different contact regimes, Trib. Int. (2017) 274–282. [17] Z. Ye, H. Moon, H.L. Min, A. Martini, Size and load dependence of nanoscale electric contact resistance, Tribol. Int. (2014) 109–113. [18] N. B. D. O. M, New Benchmark 3D Optical Microscope System, http://embeddedcomputing.com/news/bruker-benchmark-optical-microscope-system/. [19] X.Y. Lin, L.J. Xu, Y.C. Shao, G.P. Luo, Research on fretting resistance and fretting wear property of Ni-Au contact pair, IEEE, Holm Conference on Electrical Contacts. 2011 Vol.22 (2015) 1–6. [20] B. Arrazat, et al., Discrete Analysis of Gold Surface Asperities Deformation under Spherical NanoIndentation Towards Electrical Contact Resistance Calculation, 2011 1–8. [21] M. Esteves, A. Ramalho, F. Ramos, Fretting behavior of the AISI 304 stainless steel under different atmosphere environments, Tribol. Int. 88 (2015) 56–65. [22] M. Antler, Survey of contact fretting in electrical connectors, IEEE Trans. Comp. Hybrids Manuf. Technol. (1985) 87–104. [23] N. Saka, J.L. Ming, N.P. Suh, The role of tribology in electrical contact phenomena, Wear 100 (1–3) (1984) 77–105. [24] J.A. Greenwood, J.B.P. Williamson, Contact of nominally flat surfaces, Proc. R. Soc. Lond. 295 (1442) (1966) 300–319. [25] M. Park, B.A. Cola, T. Siegmund, J. Xu, M.R. Maschmann, T.S. Fisher, et al., Effects of a carbon nanotube layer on electrical contact resistance between copper substrates, Nanotechnology 17 (9) (2006) 2294–2303. [26] T.S.N.S. Narayanan, Y.W. Park, Y.L. Kang, Fretting-corrosion mapping of tin-plated copper alloy contacts, Wear 262 (1–2) (2007) 228–233. [27] I. Minowa, M. Nakamura, Contact resistance between plated conductors and current density distribution in a contact spot, Ieice Tech. Rep. Emd 93 (10) (1994) 13–18. [28] D.J. Dickrell, M.T. Dugger, The effects of surface contamination on resistance degradation of hot-switched low-force MEMS electrical contacts, IEEE Holm Conference on IEEE 2005, pp. 255–258.
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Effect of roughness on electrical contact performance of electronic components Xin-long Liu, Zhen-bing Cai ⁎, Shan-bang Liu, Jin-fang Peng, Min-hao Zhu ⁎ Tribology Research Institute, Traction Powder State Key Laboratory Southwest Jiaotong University, Chengdu 610031, China
a r t i c l e
i n f o
Article history: Received 9 March 2017 Received in revised form 17 May 2017 Accepted 18 May 2017 Available online xxxx Keywords: Fretting Contact resistance Contact spot Roughness Electric contact
a b s t r a c t This study investigated the effects of electrical contact resistance (ECR) on pogo pins used in mobile phones, chargers, digital cameras, Bluetooth headsets, medical equipment, and other electronic products with different surface roughness. Experimental results revealed that metallic wear debris is generated by fretting motion and formation of a third body on rough surfaces without removal by fretting motion, thus increasing ECR. Wear debris does not easily form the third body at contact areas of smooth surfaces and causes formation of metal–metal contact pattern. Results showed low ECR with fretting motion. 3D and 2D profiles of contact area verified the definition of contacting high spots, further explaining increases in ECR. © 2017 Published by Elsevier Ltd.
1. Introduction Pogo pins are typically used as connectors in mobile phones, chargers, digital cameras, Bluetooth headsets, automotive parts, medical equipment, and other electronic products (Fig. 1) [1]. Maintenance of stable low-value electrical contact resistance (ECR) in electrical connectors and other contact-containing components is essential to protect circuits that contribute to degradation to facilitate correct selection of contact materials and component designs [2]. Unstable ECR values may lead to electrical degradation or temperature increment at contact areas and eventually result in circuit breaker or battery explosion [3]. ECR between pogo pins/plane contacts strongly depends on surface integrity of contact areas. For connectors subject to vibration, fretting is considered a primary degradation mechanism; it is a relative cyclic motion with small amplitude occurring between two oscillating surfaces. In another study, fretting was found to cause intermittent electrical contact, wear, and corrosion on contact materials and caused variations in results of ECR [4,5]. Although pogo pins feature low cost and convenience in manufacturing, they raise a major concern due to their high interfacial contact resistance (CR) caused by mechanical polishing of their surfaces. CR is generally governed by electrical properties of interface layer between contacting surfaces [6]. As a result of surface roughness at microscopic scale, current flows only through these asperities, which occupy a
⁎ Corresponding authors. E-mail addresses: [email protected] (Z. Cai), [email protected] (M. Zhu).
http://dx.doi.org/10.1016/j.microrel.2017.05.024 0026-2714/© 2017 Published by Elsevier Ltd.
small fraction of areas of nominal contacting surfaces in theory. Ra ≤ 1 μm is a processing requirement for surface roughness of commonly used electrical contact connectors [7,8]. ECR depends on materials, diffusion media, compression pressure, surface roughness of contacting materials, and other conditions used during measurements [9,10]. Clearly, studies should research and develop low-cost and reliable solutions for reducing ECR and improving performance of electrical connectors. Further studies must also determine the effects of ECR of pogo pins/ plane contact burring by roughness. Fig. 1 presents the common working state of pogo pins. Pre-tightening force usually exists between pogo pins/contactors. Contactors utilize copper as material. For this study, red copper was the preferred material for contact manufacturer owing to its unique combination of conductivity, strength, stiffness, formability, and low cost [11]. Red copper contacts are usually plated with noble metals to improve their electrical contact performance [12,13]. Pogo pin connectors are the most critical links of electronic and power systems and are needed to provide paths for electronic signals and/or power connections. ECR between pogo pins and copper plates in electronic components is governed by multi-scale surface topography of contacting pairs [14]. The roughness featured at contacting surfaces decreases actual contact areas, leading to a voltage drop across interface [15]. Assembling relationship between electrical components applies pressure at interfaces, leading to increases in contact areas between electrical components and subsequent decreases in ECR [9]. However, excessively large pre-tightening forces may cause difficulty in assembly and deformation of pogo pins [16,17]. Thus, an optimum clamping pressure exists and trades off between competing requirements.
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
101
Fig. 1. (a) Camera connector, (b) direction of movement of contactor, (c) matched contactor.
2. Experimental 2.1. Experimental details As shown in Fig. 2(a), ECR fretting test system was constructed for pogo pins and plane contacts. Fig. 2(b–c) shows the schematic diagram for a measuring apparatus for ECR distribution; the four-terminal method is one of the most commonly used methods in determining contribution of resistance present at interfaces between two conductors [6]. Table 1 provides the conditions for measurement of ECR. Fretting experiments were carried out using the ECR fretting test system manufactured by Tribology Research Institute, Southwest Jiaotong University. The detailed structure of the test is as follows;
displacement was measured by a laser sensor (Figs. 2(a)-7). The sensor (Fig. 2(a)-3) under the lower sample was used to measure friction forces and normal loads and to transmit data to collection cards manufactured by NI Company. The upper sample comprised pogo pins (C3604: Cu, 57.0%–61.0%; Fe ≤ 0.5%; Pb, 1.8%–3.7%; Sn: Fe + Sn ≤ 1.2, and balanced total impurities). Pogo pins employed standard parts to obtain same initial roughness. Red copper was used as plane sample material (C11000: Cu, 99.9%; P, 0.0116%; Fe, 0.016%; Pb, 0.0019%; S,0.0047%; Zn,0.0216%; Sn, 0.0034%, and balanced total impurities). Pogo pins and plane sample contacts were degreased with alcohol and coal oil using an ultrasonic cleaner, dried, and carefully mated in fretting test assembly to create a point contact in pogo pin/plane geometry (Fig. 2(c)).
Fig. 2. (a) Electrical contact resistance (ECR) fretting test system (1. Piezoelectric ceramic actuator. 2. Upper fixture 3. Force transducer. 4. Flat sample. 5. Ball sample. 6. Precision lead screw 7. Displacement sensor). (b) Operating principle of fretting equipment and resistance measurement mechanism. (c) Size of upper and lower samples.
102
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
0.646, 0.268, and 0.116 μm, and they met processing requirements of pogo pin connectors.
Table 1 Details of experimental conditions. Roughness Frequency Amplitude Normal load Current load
0.989, 0.646, 0.268, and 0.116 μm 2 Hz ±10, ±20, and ±45 μm 2N 0.02 A
2.2. Samples Fig. 3 presents surface roughness of plane samples prior to fretting test. Four samples with same material but with different surface roughness were lined up. Preparation for samples involved selection of different specifications of sandpaper to grind sample surface (details in Fig. 3) and inclusion of a polished sample. Afterward, four different surface roughness were obtained. A contour GT-X 3D optical microscope [18] was used to measure surface roughness; roughness measured 0.989,
3. Results and discussion 3.1. Electrical response as a function of different surface roughness Each experimental condition was performed thrice to determine three characteristic curves with the same changing trend and to select the test with median value of ECR. When no value was selected, fretting tests were conducted again, and middle value of all tests was finally selected. For data analysis, the following steps were used. First, resistance data from a large number of original data were extracted to redraw ECR curves. As described above, ten original resistance (three kinds of displacement amplitude and four kinds of roughness) data during each fretting cycles were recorded, and 200,000 resistance data were collected in 104 cycles. Original curve was extracted to obtain understandable
Fig. 3. Sample preparation and method of roughness measurement.
Fig. 4. (a) ECR of contact pairs. (b)–(e) Wear morphology of contact pairs (Ra1 = 0.989, Ra2 = 0.646, Ra3 = 0.268, and Ra4 = 0.116).
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
103
Fig. 5. SR (average value) and instantaneous maximum of ECR of contact pairs.
resistance curve and to quantitatively analyze data. Trials were conducted on average (Rv), maximum (Rmax), and minimum (Rmin) values of every 100 cycles in the original resistance data. Rv was obtained according to the following Eq. (1) [19]: Rv ¼
ðRmax þ Rmin Þ 2
ð1Þ
After obtaining 100 valid algorithm data, ECR curve was constructed. Fig. 4 shows ECR and all selected contact pairs during 104 cycles. As shown in the figure, different surface roughness remarkably influenced extent of fretting tests in all contact pairs. ECR increased with increasing surface roughness. Contact pairs with larger roughness showed higher ECR until the end of fretting, but ECR curve was insufficient for evaluation of properties of all selected samples. Average value, that is, statistical resistance (SR), which was extracted from ECR (average of every 1000 cycles and obtained 10 points), increased at 8000 cycles (Fig. 5). For polished samples, the effect to SR by different roughness was insignificant at initial fretting test as polished samples contained maximal contact areas [20]. As a result of fretting, ECR of materials may suddenly enlarge and then reduce to low values in complete tests. To make the results more intuitive and abundant with selected contact pairs, instantaneous maximum (Rcmax) and average value of ECR (Rcaverage) during completed cycles were introduced in analysis (Fig. 5(b)). As expected, Rcmax of Ra1 (0.989) reached 2722.8 mΩ, which is the highest compared with other roughness values. On the contrary, the lowest roughness was that of Ra4 (0.116). Overall, Rcmax and Rcaverage increased with increasing roughness values. 3.2. Electrical response as a function of sliding regime condition
Fig. 6. Evolution of ECR as a function of different displacement amplitudes.
To obtain more intuitive results for comparison, two extreme values of roughness (Ra1 and Ra4) were selected to analyze electrical response as a function of sliding regime condition. Representative shapes of fretting loops from the beginning (103 cycle) and end of tests (104 cycle) were included to indicate fretting regime. Fig. 6 presents the corresponding friction force/displacement (F-D) curve, which was filtered from original data and redrawn based on 103 and 104 cycles. Displacement contributing to ECR exhibited the same trend whether under rough or polished surfaces. As long as displacement amplitude permits a stabilized partial slip condition, which can be assessed from the F-D curve, a sticking zone is maintained during this condition. Direct metal/metal contacts were generated with lower and stable ECR. This
104
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
Fig. 7. (a) Instantaneous maximum value of ECR (b)–(g). Wear morphology of contact pairs ((b) D1 = ±10 μm, Ra1 = 0.989; (c) D2 = ±20 μm, Ra1 = 0.989; (d) D3 = ±45 μm, Ra1 = 0.989; (e) D1 = ±10 μm, Ra1 = 0.989; (f) D2 = ±20 μm, Ra1 = 0.989; (g) D3 = ±45 μm, Ra1 = 0.989, Ra4 = 0.116).
phenomenon can explain why low ECR is observed under stabilized partial slip situations. When displacement amplitude becomes higher (±45 μm) than transition value (almost ±20 μm) [12,21], then the fretting condition is gross slip; with high value of ECR, the generalized gross slip condition activates debris formation over rough contact areas. For further illustration, polished surface (Ra4) exhibited lower value of ECR while under three displacement amplitudes in comparison with rough surface (Ra1). With the motion of fretting, contact area became less smooth and with the rise of ECR (from the 3.72 to 78.80 mΩ). As the fretting close to the end, the contact area of contact became larger and eventually leads the contact resistance dropped to 62.58 mΩ after 8000 cycles [22,23]. When a greater displacement amplitude imposed a stabilized gross slip condition, ECR increased and destabilized. This tendency was clearly illustrated by plotting instantaneous maximum of ECR (Rcmax) as a function of imposed displacement amplitude (see Fig. 7). 3.3. SEM (scanning electronic microscopy)/EDX (X-ray energy dispersive spectroscope) analysis Fig. 8 shows SEM and EDX of typical samples (F = 2 N, D = ± 45 μm). Contact areas of selected samples showed severe delamination and wear debris. Plough and material removal was also observed in
contact areas. Contact surfaces were covered with black oxide debris. This oxide debris at contact surface possibly caused a steep increase in ECR after a few thousand cycles of fretting. EDX analysis of all tested samples are plotted in the middle of Fig. 8. No evidence proves that different loads or roughness yielded different EDX results.
3.4. Relationship between wear mechanisms and CR analysis For theoretical arithmetic of CR to nominal flat rough surfaces, the following assumptions of the Greenwood and Williamson model were adopted [15,24]: (a) Rough surfaces are isotropic. (b) Bulk deformation is absent; only asperities (a-spots) deform during contact. (c) a-spots summits are spherical. (d) All a-spots summits feature the same radius r, but their height distribution varies statistically. (e) Asperities are distant from one another, and no interaction exists among them. For the pogo pin/flat contact model shown in Fig. 9, both contacting asperities and pin deformed to generate an actual contact area that supported normal load F when the pogo pin was pressed against rough flat surfaces. The contact area of pogo pin is flattened, forming a nominal contact area with radius “a” (Fig. 9: vertical view), over which a uniform separation “Ra” is assumed between the pogo pin and mean of surface heights of the rough flat surface.
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
105
Fig. 8. EDX analysis and wear track morphology of all contact pairs.
Physical micro-contact differs from macro-contact due to influence of surface roughness and smaller contact force available in microswitches. For rough flat surfaces, only high points on each surface participate in contact. Effective contact areas, namely, contact spots or a-spots [1], are significantly smaller than apparent ones but participate in transporting electrons.
Manner of electrons transport through electrical connections must be determined to evaluate CR RC: RC ¼
ρ ρ þ 2r 2na
Fig. 9. Schematic of a contact model for pogo pin/flat contact.
ð2Þ
106
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109 Table 2 Characteristics of used materials.
Material
Electrical resistivity ρ(20°) (×10–8 Ω•m)
Yield strength (MPa)
Young′s modulus (GPa)
Poisson′s ratio
Resistance temperature coefficient (aR/°C–1)
Density (g/cm3)
Linear expansivity
1.75
239.689
100
0.33
0.00426(20 °C)
8.5
12
1.678
258.646
110
0.37
0.00393(20 °C)
8.9
16.6
C3604 (Pogo pin) C11000 (flat sample)
where r indicates radius of an apparent contact area, n represents the number of a-spots, ρ corresponds to electrical resistivity, and a indicates a-spot radius. Greenwood released the following CR approximation Eq. (3) for multiple distinct contacts by considering interfaces between a-spots. ai a j ρ ρ þ RC ¼ ∑∑ 2∑ai π i≠ j Sij
!
1 2
ð∑ai Þ
ð3Þ
To investigate CR of the pogo pin/flat rough surface contact, actual contact areas must be identified, and contact modes for single a-spots must be studied (Fig. 9). To calculate a of a single a-spot according Hertz contact theory, measured Rai was used as mean Hertz contact radius for single a-spots. According to this assumption, a can be calculated as follows: a¼
1=3 3ðR1 Rai Þ 1−μ 1 2 1−μ 2 2 F þ E1 E2 4ðR1 þ Rai Þ
ð4Þ
where μi indicates Poisson's ratio, and Ei represents modulus of elasticity. Actual values of material characteristics in Table 2 were used to solve Eq. (4). As shown in Eqs. (3)–(4), Rc was determined by contact resistivity, contact area for a single a-spot, and the number of a-spots within an apparent contact area. As distribution of a-spots was random and scattered, the actual number of contact spots was then measured by contour GT-X 3D optical microscope. Measurement was repeated thrice to minimize deviation. Notably, as threshold value, measured roughness Rai was used to determine the number of contact spots that participate
Fig. 10. Theoretical Rc theoretical and measured average RECR average.
in contact and conduction. In this paper, polished samples were considered to contain no contact spot but only apparent contact areas. Calculated CR Rc theoretical are shown in Fig. 10 (RECR average indicates actual average value of measured original resistance data). Results illustrate higher calculated (Rc theoretical) than actual values (RECR average). This observation resulted from deformation of asperities and pogo pins; such deformation can be elastic, elastic–plastic, fully plastic, or any combination of these deformation types. All procedures mentioned above are only applicable to static-state contact but not to evolution of fretting motion. ECR changes during fretting by various processes involve transfer, material removal, and chemical transformations at contact areas and cause difficulty in evaluating ECR of different surface roughness. Many of these processes were cited in preceding sections, but treatment of contact property can be unified by summarizing them [2]. The relationship between ECR and wear mechanisms is summarized as follows. Fretting at contact areas of copper generated oxide debris at interface. This oxide debris showed high resistance and acted as insulator [21]. Accumulation of oxide debris at interface resulted in increased ECR, although some debris were pushed out of the interface by fretting motion [6]. Therefore, to obtain lower CR under fretting, accumulating oxide debris must be removed from contact areas to ensure metal-tometal contact at interface [25]. Evolution processes of wear mechanism are plotted in Fig. 11(a)–(c), which elaborate different roughness of surface topography for trapping oxide debris. Corresponding circuitry diagram is separately shown on right side of the graph. For the copper with rough surface shown in Fig. 11(a), the upper sample is pogo pin, and the lower sample was copper. The path for current passage was less and narrow due to rough surface. Many contact spots, which are also called a-spots, exist between contact areas [14, 26]. As fretting continues, a-spots were gradually worn by pogo pins. However, as wear debris possibly ploughed and contributed to high friction, contact areas of pogo pins also experienced severe wearing. Interface contained trapped wear debris that were expected to plough surfaces [15]. Accordingly, even in the absence of long ploughing grooves, ploughing mechanism may still be the dominant mechanism. Different from smooth surfaces, rough surface contributed to adhesion of wear debris and eventually caused accumulation of oxide particles in the furrow space. ECR of rough surface contacts can be expressed similarly to a network of parallel resistors [15,27]. Equivalent circuitry diagram demonstrates the evolution principle of ECR in every completed cycle. As shown in this circuitry, Rg represents the CR from pogo pin to wear debris, Rd. corresponds to self-resistance of wear debris, and Rp is the CR between wear debris and copper samples. The contact spot between the interface alternate connections in a complete cycle shows similarity to a switch S connected to alternating “a” and “b”. In conclusion, calculation of total resistance RECR is related to Rg, Rd., and Rp. For the plane sample with smooth surface, ECR remained relatively stable during fretting and possibly increased with appearance of aspots (Fig. 11(c)) [14,28]. In the same way, in the corresponding circuitry diagram on the right side of Fig. 11(c), Rg indicates the CR from pogo pin to wear debris, Rd. corresponds to self-resistance of wear debris, Rp
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
107
Fig. 11. Wear mechanisms analysis.
represents the CR between wear debris and plane sample, and Rgp is the CR from pogo pin to plane sample. Given the minimal wear debris adhering to contact area, the contact model was more inclined to form pogo pin/copper contact, that is, metal-to-metal contact and consequently resulted in lower ECR. Fig. 11(b) shows cross-section profiles of samples with roughness of Ra1 and Ra4 after fretting experiment. Cross-section morphology illustrates that plow and wear debris existed in the contact area with rough surface, and smooth surface showed a different phenomenon. Wear debris Stacked on the edge of the contact area. This study can be summarized as follows: (a) ECR of rough surface exhibited higher values as a result of adhesion of wear debris and finally caused accumulation of oxide particles in the furrow space. (b) Wear debris cannot easily accumulate wear particles at contact areas of smooth surfaces, and inclination to form direct contact patterns is the main reason for lower ECR of smooth surface contacts. Fig. 12 shows the 3D profiles of two representative contact pair with different surface roughness (Ra1 and Ra4) and obtained using a contour GT-X 3D optical microscope. Red objects around the edge of fretting area are oxidized debris. Various degrees of material removal were detected at the contact area. 2D profiles were constructed for each contact pair obtained from cross-sections of 3D profiles (Fig. 12b, e). Fig. 12c and f present the last complete cycle of ECR (LCECR) for each contact pair.
Results of surface profile and LCECR confirmed that different surface roughness did not significantly affect wear depth of plane sample and suggested that other factors were responsible for observed changes in ECR. Details of these other factors are as follows. A very high CR (1021.9 mΩ) was detected and is shown in Fig. 12c. High CR was caused by a-spot formed by many contacting high spots from the pogo pin (such as point “a”) and plane sample (such as point “b”), as shown in Fig. 12b. A similar observation was noted for roughness of Ra4 (Fig. 12f). Lower CR (31.0 mΩ) resulted from two relatively well-fitting surfaces (Fig. 12 e). It can be clearly seen from the 3D morphology shown in Fig. 13, wear debris accumulated at the edge of wear area with polished surface (Ra4) compared with rough surface (Ra1). Overall, samples with relatively rough surface generated a surface with more plows during fretting motion. Profiles of contact pair with polished surface preferentially enlarged the actual contact area through numerous point contacts during fretting motion. In this study, the sample with roughness value of Ra4 presented remarkable electrical contact performance. 4. Conclusions Metallic wear debris is generated by fretting motion and forms a third body that stay on rough surfaces and is hardly removed when under
108
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
Fig. 12. Relationship between contact profiles and LCECR.
X. Liu et al. / Microelectronics Reliability 74 (2017) 100–109
109
Fig. 13. Cross section of 3D morphology for two contacts: (a) Ra1 and (b) Ra4.
stabilized gross slip conditions. Formation of a third body transforms contact surface into metal–debris–metal pattern and eventually causes increases and instability of ECR. On the contrary, lower displacement amplitude difficult to generate wear debris and show lower ECR. Wear debris hardly formats debris beds at smoother contactors and causes formation of more metal–metal contacts that maintain relatively lower and stable ECR with fretting motion.
Acknowledgement This research was supported by National Natural Science Foundation of China (Contract numbers (No.51375407, 51575459, U1530136, 51627806), supported by Young Scientific Innovation Team of Science and Technology of Sichuan (No.2017TD0017). References [1] N. Amin, Z.L. Yi, A practical investigation on the root causes of the mechanical damages of pogo pin type test sockets to IC packages in final test, IEEE International Conference on Semiconductor Electronics. IEEE Xplore 2008, pp. 393–397. [2] B. Huang, X. Li, Z. Zeng, et al., Mechanical behavior and fatigue life estimation on fretting wear for micro-rectangular electrical connector, Microelectron. Reliab. 66 (2016) 106–112. [3] M. Esteves, A. Ramalho, F. Ramos, Electrical performance of textured stainless steel under fretting, Tribol. Int. 110 (2017) 41–51. [4] M.J. Kim, H.K. Kim, Effects of temperature and span amplitude on fretting corrosion behavior of tin-plated electrical contacts, Microelectron. Reliab. (2017) http://doi. org/10.1016/j.microrel.2016.12.014. [5] J. Yasuda, K. Iida, Y. Saitoh, S. Sawada, Y. Hattori, Y. Hattori, Effect of Intermittent Time on Fretting Corrosion of Tin Plated Coupon, 2011 45–49. [6] E. Mengotti, L.I. Duarte, J. Pippola, et al., Fretting corrosion: analysis of the failure mechanism for low voltage drives applications, Microelectron. Reliab. 54 (9–10) (2014) 2109–2114. [7] D. Qiu, L. Peng, P. Yi, X. Lai, A micro contact model for electrical contact resistance prediction between roughness surface and carbon fiber paper, Int. J. Mech. Sci. 124–125 (2017) 37–47. [8] E. Crinon, J.T. Evans, The effect of surface roughness, oxide film thickness and interfacial sliding on the electrical contact resistance of aluminium, Mater. Sci. Eng. A 242 (1–2) (1998) 121–128. [9] K. Pereira, T. Yue, Multiscale analysis of the effect of roughness on fretting wear, Tribol. Int. (2017) http://dx.doi.org/10.1016/j.triboint.2017.02.024.
[10] B. Avasarala, P. Haldar, Effect of surface roughness of composite bipolar plates on the contact resistance of a proton exchange membrane fuel cell, J. Power Sources (2009) 225–229. [11] I.H. Sung, J.W. Kim, H.J. Noh, H. Jang, Effect of displacement and humidity on contact resistance of copper electrical contacts, Tribol. Int. (2016) 256–261. [12] S. Fouvry, P. Jedrzejczyk, P. Chalandon, Introduction of an exponential formulation to quantify the electrical endurance of micro-contacts enduring fretting wear: Application to Sn, Ag and Au coatings, Wear (2011) 1524–1534. [13] T.S.N.S. Narayanan, Y.W. Park, Y.L. Kang, Fretting corrosion of lubricated tin plated copper alloy contacts: effect of temperature, Tribol. Int. (2008) 87–102. [14] D. Hill, et al., Characterising the surface roughness of AFM grown SiO2, on Si, Microelectron. Reliab. (2001) 1077–1079. [15] J. Molina, R. Valderrama, C. Zuniga, et al., Influence of the surface roughness of the bottom electrode on the resistive-switching characteristics of Al/Al2O3/Al and Al/ Al2O3/W structures fabricated on glass at 300 °C, Microelectron. Reliab. 54 (12) (2014) 2747–2753. [16] T. Yue, M.A. Wahab, Finite element analysis of fretting wear under variable coefficient of friction and different contact regimes, Trib. Int. (2017) 274–282. [17] Z. Ye, H. Moon, H.L. Min, A. Martini, Size and load dependence of nanoscale electric contact resistance, Tribol. Int. (2014) 109–113. [18] N. B. D. O. M, New Benchmark 3D Optical Microscope System, http://embeddedcomputing.com/news/bruker-benchmark-optical-microscope-system/. [19] X.Y. Lin, L.J. Xu, Y.C. Shao, G.P. Luo, Research on fretting resistance and fretting wear property of Ni-Au contact pair, IEEE, Holm Conference on Electrical Contacts. 2011 Vol.22 (2015) 1–6. [20] B. Arrazat, et al., Discrete Analysis of Gold Surface Asperities Deformation under Spherical NanoIndentation Towards Electrical Contact Resistance Calculation, 2011 1–8. [21] M. Esteves, A. Ramalho, F. Ramos, Fretting behavior of the AISI 304 stainless steel under different atmosphere environments, Tribol. Int. 88 (2015) 56–65. [22] M. Antler, Survey of contact fretting in electrical connectors, IEEE Trans. Comp. Hybrids Manuf. Technol. (1985) 87–104. [23] N. Saka, J.L. Ming, N.P. Suh, The role of tribology in electrical contact phenomena, Wear 100 (1–3) (1984) 77–105. [24] J.A. Greenwood, J.B.P. Williamson, Contact of nominally flat surfaces, Proc. R. Soc. Lond. 295 (1442) (1966) 300–319. [25] M. Park, B.A. Cola, T. Siegmund, J. Xu, M.R. Maschmann, T.S. Fisher, et al., Effects of a carbon nanotube layer on electrical contact resistance between copper substrates, Nanotechnology 17 (9) (2006) 2294–2303. [26] T.S.N.S. Narayanan, Y.W. Park, Y.L. Kang, Fretting-corrosion mapping of tin-plated copper alloy contacts, Wear 262 (1–2) (2007) 228–233. [27] I. Minowa, M. Nakamura, Contact resistance between plated conductors and current density distribution in a contact spot, Ieice Tech. Rep. Emd 93 (10) (1994) 13–18. [28] D.J. Dickrell, M.T. Dugger, The effects of surface contamination on resistance degradation of hot-switched low-force MEMS electrical contacts, IEEE Holm Conference on IEEE 2005, pp. 255–258.