Influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear

Influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear

Accepted Manuscript Influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear F. Pompanon, S. Fouvry, O...

10MB Sizes 0 Downloads 22 Views

Accepted Manuscript Influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear

F. Pompanon, S. Fouvry, O. Alquier PII: DOI: Reference:

S0257-8972(18)30967-8 doi:10.1016/j.surfcoat.2018.07.109 SCT 23773

To appear in:

Surface & Coatings Technology

Received date: Revised date: Accepted date:

15 April 2018 23 July 2018 24 July 2018

Please cite this article as: F. Pompanon, S. Fouvry, O. Alquier , Influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear. Sct (2018), doi:10.1016/j.surfcoat.2018.07.109

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear

F.Pompanon1, S. Fouvry1*, O. Alquier2 1

LTDS, CNRS UMR 5513, Ecole Centrale de Lyon,

PT

36 av Guy de Collongue, 69134 Ecully Cedex,

2

PSA Groupe,

RI

France

SC

78943 Vélizy – Villacoublay Cedex,

NU

France

MA

*Corresponding author: [email protected]

Abstract—The use of connectors in electrical devices for automotive has significantly

D

increased during the last decades. These connectors need to keep a low and stable electrical

TE

contact resistance (ECR) otherwise disconnects may occur, inducing critical failures. Close to

AC CE P

the engine, these connectors are subjected to vibrations inducing fretting in the contact (i.e. wear damage induced by small oscillating sliding).

This phenomenon induces surface wear and the formation of oxide debris (third body) which, being trapped within the interface, can drastically increase the electrical contact resistance.

The aim of this study is to investigate the effects of the relative humidity (RH) on the fretting wear rate and the Electrical Contact Resistance (ECR) of a silver plated electrical contact. The analyses show that an increase of RH tends to increase the ECR fretting endurance Nc related

1

ACCEPTED MANUSCRIPT to the failure condition (ΔR > ΔRth=4 mΩ). However, a discontinuous evolution is observed. Below a RH threshold, RH=50%, the endurance increase is rather slow and can be related to a decrease of the friction work at the interface. Above RH=50%, the high humidity conditions modify the rheological properties of the debris layer. The wear rate is becoming smaller and a larger wear volume is required to reach the ECR failure. This induces a fast linear increase of

PT

Nc. Whatever the wear processes, the investigation confirms that ECR failure is reached when

RI

the silver concentration in the inner part of the contact is becoming lower than [Ag]th=5%.

AC CE P

TE

D

MA

NU

SC

Keywords— Fretting wear, electrical contact resistance, silver coating, humidity

2

ACCEPTED MANUSCRIPT 1. Introduction

The use of connectors in electrical devices for automotive application has significantly increased over the past few decades. Generally speaking, connectors need to keep a low and stable electrical contact resistance (ECR) otherwise disconnects may occur, inducing critical failures [1], [2]. However, in automotive, connectors implemented next to the car engine are

PT

subjected to vibrations inducing fretting micro-displacements in the electrical contact. Fretting

RI

damage promotes the formation of an oxide debris layer leading to electrical failures. Recent

SC

investigations suggest that near 15% of the electrical breakdowns are induced by fretting wear damages [3].

NU

Since the first studies realized by Antler [4], [5], the degradation of electrical contacts has

MA

been extensively investigated [6-15] in order to determine the damage mechanisms and to identify viable solutions. The most efficient method to reduce the fretting wear damage

D

consists in applying noble coatings. Gold layers are currently the most used coatings due to

TE

their high electrical conductivity and high corrosion resistance [5], [16-18]. However, due to the uprising price of gold, silver is now considered as a promising alternative [19-22].

AC CE P

Various investigations, led by Hannel et al. [23] and Kassman-Rudolphi and Jacobson [24], [25], shed light on the influence of sliding regimes on electrical connector performances. Hence, as long as partial slip fretting conditions are maintained, an inner metal/metal stick zone operates, promoting stable and low ECR and infinite electrical endurance (Fig. 1). In contrast, above a threshold displacement amplitude (δ*t), a gross slip condition operates, activating a wear process and the formation of low-conduction oxide debris which increase the ECR and lead to finite electrical endurance Nc. Moreover, the ECR endurance depends on the nature of materials. For non-noble layers such as Sn, the electrical failure is almost instantaneous, whereas for noble layers such as Ag and Au, the electrical failure occurs when most of the noble top layer has been worn out. Comparing various Ag/Ag, Au/Au and Sn/Sn 3

ACCEPTED MANUSCRIPT interfaces for a given contact configuration and normal force, Fouvry and al. [22] showed that the ECR endurance can be expressed as a power law function of the fretting sliding amplitude. This quantitative analysis was extended by Laporte and al. [26] who transposed a local friction energy density approach to predict the ECR endurance combining the effects of sliding amplitude, normal force and coating thickness in a single formulation. It was also

PT

demonstrated that the electrical failure can be characterized by threshold concentrations of oxygen and silver inside the fretting scar:

RI

Nc when [Ag]d < [Ag]th ≈ 5 At%

(2)

SC

And [O]d > [O]th ≈ 45 At%

(1)

Where Nc is the ECR endurance related to the electrical failure when ΔR > ΔRth=4 mΩ.

NU

Few investigations focused on the effects of humidity. Park and al. investigating tin-plated

MA

electrical contacts [27] found that humidity increases the performance of the contacts assuming a condensation effect of water vapor lubricating in the interface. Sung and al. also

D

investigated the influence of humidity on copper electrical contacts [28] confirming an

TE

increase of the ECR endurance with the relative humidity. They made the hypothesis that the water molecules prevent the pileup of wear debris by capillarity.

AC CE P

The aim of this work is to study the influence of humidity on electrical contacts. This is an essential aspect for car industries in order to insure a proper functioning of connectors in various climates. The effects of humidity on the coefficient of friction will be first considered to clarify the lubrication hypothesis, and then the third body behavior and the wear rate will be studied in order to understand and characterized the electrical failure. 2. Experimental details 2.1 Materials The samples used in the present study were crossed-cylinders with a homogenous Ag/Ag contact (i.e. similar upper and lower cylinders) (Fig. 2). The interface structure (Fig. 3) consisted in a 37 wt.% zinc (CuZn37) brass alloy substrate onto which a 2 μm electrolytic 4

ACCEPTED MANUSCRIPT nickel interlayer was deposited in order to limit copper diffusion. On this standard structure, 2 µm of pure silver was deposited using a dedicated electrolytic process [22], [23], [26].

2.2 Contact Configuration and Electrical Resistance Measurement The sample system used in this study consisted of two 90° crossed-cylinders with a 2.35mm

PT

cylinder radius (Fig. 2). According to the Hertzian theory [29], this contact configuration is

RI

equivalent to a sphere/plane contact with R=2.35 mm. During the test, the electrical contact resistance was measured using a 4-wire method [23]. A current source applied I=5mA with

SC

10V voltage compliance while a μvoltmeter system measured the contact voltage at a

NU

resolution of 0.01μV. This system makes it possible to measure the electrical resistance from 10-6 to 103 Ω.

MA

As previously mentioned, under gross slip condition, fretting wear damage reduces the ECR endurance. To quantify this aspect, the number of fretting cycles needed to reach the ΔRth=4

2.3 Experimental Setup

TE

D

mΩ ECR threshold inducing electrical failure is defined as Nc ECR endurance (Fig. 4).

AC CE P

A specific fretting set-up was designed and built for this study to simulate environmental conditions similar to those in a car engine. Fig. 5 shows a diagram of the machine. The fretting displacement (δ) was applied to the upper specimen using an electromagnetic shaker and measured using a high-resolution laser extensometer. The upper specimen holder was fixed to the shaker using flexure strips to guarantee the application of the normal force. The lower sample was fixed on the bottom specimen holder above which a piezoelectric load sensor allows the tangential force recording during the test. A dead mass was placed on the upper holder to assure the normal load. Tests are performed in a controlled chamber in which the ambient parameters (temperature and humidity) are monitored on a climatic generator (Fig. 6). 5

ACCEPTED MANUSCRIPT

The dedicated software, programmed on a LabView platform, was developed to provide all necessary recording and monitoring. The test parameters which are recorded and controlled during each test are: the fretting displacement amplitude (δ*) and related fretting sliding amplitude (δ*0) [from ±0.5 to ±40 μm], the tangential force amplitude (Q*) [from 0 to ±10 N],

PT

the frequency (f) [from 0.005 to 50 Hz], the normal force (P) [from 1 to 6 N], the relative

RI

humidity (RH) [from 1 to 99%], and the temperature (T) [from 20 to 150 °C].

SC

2.4 Fretting loop analysis

The analysis of the fretting loop allows the determination of the tangential force and

NU

displacement amplitudes (Q*- δ*). However, the estimation of the real displacement in the contact is very difficult. Indeed, the displacement amplitude (δ*) is measured outside the

MA

interface and includes the contact displacement (𝛿𝑐∗ ) but also the test system accommodation

(3)

TE

𝛿 ∗ = 𝛿𝑐∗ + 𝛿𝑠∗ = 𝛿𝑐∗ + 𝐶 × 𝑄 ∗

D

(𝛿𝑠∗ ) [26]:

With C the tangential compliance of the specimens and test system.

AC CE P

An alternative strategy to avoid the test signature is to consider the fretting cycle aperture (δ*0) measured when Q=0 during the fretting cycle. This variable is not affected by the tangential accommodation of the fretting apparatus since that it is measured when Q=0. When the cyclic plastic accommodation of the contact is small, δ0 is quasi equivalent to the effective sliding amplitude so that (δ*g ≈ δ*0) [26] (Fig. 7). With the (𝛿0∗ ) variable all the results obtained using different test machines (displaying various tangential stiffness), can be compared directly without any machines stiffness corrections.

6

ACCEPTED MANUSCRIPT The friction energy dissipated during a fretting cycle, Ed (J), was determined by integrating the δ - Q loop (i.e. area of the fretting cycle). An averaged description of the friction behavior during the fretting cycle was achieved using an “energy friction coefficient” [30]: µ𝑒 =

𝐸𝑑

(4)

4 × 𝛿0 ×𝑃

PT

2.5 Test conditions

RI

The normal force was kept constant at P=3N as well as the temperature at T=25°C. A constant

SC

fretting sliding amplitude was applied, δ*0=±9µm, at a 30 Hz frequency. The relative humidity (RH) is the only variable parameter, controlled by the climatic generator in a range from 10%

NU

to 90%. Tests were performed at various levels of relative humidity: 10%, 30%, 50%, 60%, 70%, 80% and 90%.

MA

3. Results

D

3.1 Influence of the relative humidity on the ECR

TE

Fig. 8 (a) plots the evolution of the ECR versus the fretting cycles for distinct relative

AC CE P

humidity from 10% to 90%. When the RH increases, a smoother rising of ECR is observed inducing longer Nc endurances. By plotting the evolution of the ECR versus N/Nc, two behaviors are observed: -

Below RH=50%, the ECR displays a quasi linear increase until the failure (Fig. 8 (b)).

-

Above RH=50% RH, the ECR shows a plateau evolution between 1 and 3 mΩ before displaying a sharp increase until the ECR failure (Fig. 8 (c)).

These two typical evolutions suggest two different wear processes below and above RH=50%. For a better interpretation of the RH dependency, the Nc endurances are compared (Fig.9).

7

ACCEPTED MANUSCRIPT For every relative humidity, the tests were duplicated at least 3 times. The ECR endurance increases following a bilinear evolution with a transition at RH=50%. Two endurance domains are observed: -

Domain (I) RH ≤ 50% : the ECR endurances are low and the increase with RH is slow. A rather low scattering is observed. The ECR endurance can be approximated by a linear

PT

relation : (5)

RI

𝑁𝑐 = 𝐾𝑁𝑐(𝐼) × [𝑅𝐻] + 𝑁𝑐0%

Domain (II) RH > 50% : the ECR endurance displays a fast linear increase. Nc is

NU

-

SC

With 𝐾𝑁𝑐(𝐼) = 1 900 𝑐ycles/%RH and 𝑁𝑐0% = 84 000 cycles.

multiplied by a factor 4 from RH=50% to RH=80%. However, the scattering tends to

MA

continuously increase with RH. The ECR endurance is expressed using the following

D

basic expression :

𝑁𝑐 = 𝐾𝑁𝑐(𝐼𝐼) × [𝑅𝐻 − 50%] + 𝑁𝑐50%

TE

𝑁𝑐90% −𝑁𝑐50% 40

= 19 000 cycles/%RH and 𝑁𝑐50% = 180 000 cycles.

AC CE P

With 𝐾𝑁𝑐(𝐼𝐼) =

(6)

3.2 Tribological analysis of the ECR failure evolution Fig. 10 plots the evolution of the coefficient of friction until the ECR failure for various relative humidity levels. It is interesting to note that below RH=50%, the decrease of the coefficient of friction can be related to an increase of Nc. However, above RH=30%, the coefficient of friction stabilizes around 0.7 whereas Nc displays a sharp increase from Nc=2.105 cycles when RH=50% up to Nc=7.105 cycles when RH=80%. Such independent evolution of Nc versus µ contradicts the Park’s lubrication hypothesis. Indeed, if the ECR endurance increase was only related to a lubrication process, similar endurances should be observed above RH=50% when the coefficient of friction stabilizes at 0.7. 8

ACCEPTED MANUSCRIPT This conclusion is confirmed by comparing Fig. 9 and Fig. 11. In domain (I), the increase of Nc matches well with a reduction of µe. But in domain (II) (i.e. RH > 50%), µe stabilizes at 0.65 whereas Nc displays a sharp increase. Besides, µe stabilizes at 0.65 for RH=30% approximatively, far below the RH=50% threshold related to Nc transition. Hence it definitively confirms that the ECR endurance is not

PT

exclusively linked to the friction response but mainly related to the wear process.

RI

Again, a bilinear formulation can be considered to express the evolution of the coefficient of

SC

friction:

MA

With µ𝑒0% = 1.1 and 𝐾µ𝑒 = −1.5 × 10−2 For RH ≥ 30% : µ𝑒 = µ𝑒 (>30%) = 0.65

(8)

TE

D

3.3 Wear analysis

(7)

NU

For RH < 30% : µ𝑒 = 𝐾µ𝑒 × 𝑅𝐻 + µ𝑒0%

After each test, specimens were cleaned in an ultra-sonic ethanol bath to remove most of the

AC CE P

wear debris remaining in the fretting scar. Then 3D surface profiles were obtained to establish the wear volume on lower and upper specimens. The total wear volume VNc (i.e. the sum of upper and lower wear volumes) at the ECR failure is plotted versus the applied relative humidity in Fig.12.

The mean values of the maximum wear depths measured on the two counterparts (hmax) are compared in Fig.13 The wear volume shows a similar evolution to the ECR endurance with a discontinuity at RH=50%. In domain (I), both VNc and hmax remain constant. By contrast in domain (II), above RH=50%, a linear increase is observed.

9

ACCEPTED MANUSCRIPT In domain (I), the wear volume required to reach the ECR failure is constant but tends to increase above RH=50%. This indirectly suggests a change in the electrical properties of the debris layer (the larger RH the higher the third body conductivity). Regarding the scattering, as for the Nc endurance, VNc displays a lower scattering in domain (I). However, unexpectedly, an opposite tendency is observed for hmax. To interpret such a

PT

mismatching scattering it must be underlined that VNc corresponds to the total net missing

RI

volume (upper and lower specimens) whereas hmax characterizes a local degradation of the interface. The scattering of hmax is more affected by adhesive wear mechanisms than the total

SC

wear volume since the wear volume transferred to a counterpart is compensated by a

NU

symmetrical removal of material on the opposite side. Alternatively it can be supposed that the larger the RH, the lower the adhesive wear process and the lower the scattering as

MA

illustrated in Fig. 13.

Again VNc and hmax evolutions versus RH can be approximated using simple linear

TE

RH=50%.

D

correlations. The wear damage parameters display a similar discontinuous evolution to Nc at

AC CE P

For RH ≤ 50% (domain (I)) :

𝑉𝑁𝑐 = 𝑉𝑁𝑐(𝐼) = 𝑉𝑁𝑐(50%) = 2.76 × 105 µ𝑚3

(9)

For RH > 50% (domain (II)) : 𝑉𝑁𝑐 = 𝐾𝑉𝑁𝐶(𝐼𝐼) × (𝑅𝐻 − 50%) + 𝑉𝑁𝑐(𝐼)

With 𝐾𝑉𝑁𝐶(𝐼𝐼) =

𝑉𝑁𝑐(90%) − 𝑉𝑁𝑐(50%) 40

(10)

= 6.6 × 103 µ𝑚3 /%𝑅𝐻

An equivalent formulation can be extracted from hmax so that : For RH ≤ 50% (domain (I)) : 10

ACCEPTED MANUSCRIPT ℎ𝑚𝑎𝑥 = ℎ𝑚𝑎𝑥(𝐼) = ℎ𝑚𝑎𝑥(50%) = 5.2 µ𝑚

(11)

For RH > 50% (domain (II)) : ℎ𝑚𝑎𝑥 = 𝐾ℎ𝑚𝑎𝑥(𝐼𝐼) × (𝑅𝐻 − 50%) + ℎ𝑚𝑎𝑥(𝐼)

(12)

PT

With 𝐾ℎ𝑚𝑎𝑥(𝐼𝐼) = 5 × 10−2 µ𝑚/%𝑅𝐻 3.4 Wear rate analysis

RI

Assuming that ECR endurance is ruled by surface wear, it appears essential to explicit the

SC

energy wear rate evolution versus RH.

𝑉𝑁𝑐 ∑ 𝐸𝑑𝑁𝑐

(13)

MA

𝛼=

NU

For every test condition the so called α energy wear rate [30] is established:

D

With ∑ 𝐸𝑑𝑁𝑐 the accumulated friction energy dissipated at the Nc ECR failure.

TE

Fig. 14 plots the evolution of experimental α values versus RH.

AC CE P

As for VNc and Nc a discontinuous evolution is observed at RH=50% which confirms a drastic change of the wear process above RH=50%. A quasi constant evolution is observed in domain (I) (RH < 50%) followed by a very sharp drop right after RH=50%. Then a smoother linear decrease is observed after RH=60%. The very sharp decrease between 50% and 60% requires a complex exponential formulation to fully express the α evolution. Our objective is to explicit the wear rate evolution using simple linear extrapolations. Therefore the 50% - 60% discontinuous transition is neglected and the global wear rate is expressed using two separate linear segments extrapolating the linear decreasing response in domain (II) until RH=51%. Hence 11

ACCEPTED MANUSCRIPT For RH ≤ 50% (domain (I)) : 𝛼 = 𝛼(𝐼) = 2.4 × 104 µ𝑚3 /𝐽

(14)

For RH > 50% (domain (II)) : 𝛼 = 𝐾𝛼(𝐼𝐼) × (𝑅𝐻 − 51%) + 𝛼(51%)

(15)

PT

With 𝛼(51%) = 1.26 × 104 µ𝑚3 /𝐽 and 𝐾𝛼(𝐼𝐼) = −1 × 102 (µ𝑚3 . 𝐽−1 )/%𝑅𝐻

RI

This non continuous description may be physically questionable but it simplifies the ECR

SC

formulation.

NU

3.5 Prediction of the Nc ECR endurance

MA

The previous investigations suggest that the Nc ECR endurance is mainly controlled by the fretting wear rate and the wear volume required to achieve the electrical failure. Assuming a

TE

VNc – α linear correlation.

D

linear wear damage evolution, the theoretical Nc endurance may be extrapolated from a basic

AC CE P

Hence :

∑ 𝐸𝑑 𝑁𝑐 ≈ 𝑁𝑐 × 4 × 𝛿0 ∗ × µ𝑒 × 𝑃

(16)

From the combination of Eq. 13 with Eq. 16 we can extrapolate a theoretical endurance 𝑁𝑐𝑡ℎ based on a surface wear description: 𝑁𝑐𝑡ℎ =

𝑉𝑁𝑐

(17)

𝛼 × 4 × 𝛿0 ∗ × µ𝑒 × 𝑃

VNc, α and µe parameters display significant fluctuation versus RH so the previous global relationship can be declined in several expressions as a function of RH domains : (I) - A : RH < 30% :

12

ACCEPTED MANUSCRIPT 𝑁𝑐𝑡ℎ =

𝑉𝑁𝑐(𝐼) ∗

𝛼(𝐼) × 4 × 𝛿0 × 𝑃 × (𝐾µ𝑒 × 𝑅𝐻 + µ𝑒0% )

(I) - B : 30% ≤ RH ≤ 50% :

𝑁𝑐𝑡ℎ =

𝑉𝑁𝑐(𝐼)

(18)

𝛼(𝐼) × 4 × 𝛿0 ∗ × 𝑃 × µ𝑒 (>30%) )

𝐾𝑉𝑁𝐶(𝐼𝐼) × (𝑅𝐻−50%) + 𝑉𝑁𝑐(𝐼) (𝐾𝛼(𝐼𝐼) × (𝑅𝐻−51%)+ 𝛼(51%) )× 4 × 𝛿0 ∗ × 𝑃 × µ𝑒 (>30%) )

RI

𝑁𝑐𝑡ℎ =

PT

(II) : RH > 50% :

SC

Fig. 15 compares the Nc endurances extracted from the ECR analysis and the predicted 𝑁𝑐𝑡ℎ

NU

values.

A very good correlation is observed which a posteriori demonstrates that the endurance may

MA

be expressed using an adequate formulation of silver wear rate. Such surface wear correlation involves the coefficient of friction which affects the friction dissipation, the VNc wear volume

TE

D

required to reach the ECR failure and the energy wear rate (α). Using this global fretting wear rate description, some individual evolutions can be described.

AC CE P

For instance the smooth rising of Nc in domain (I) (Fig.9) may be explained by the linear decreasing of the coefficient of friction whereas both VNc and α remain nearly constant. Alternatively the fast rising of Nc in domain (II) is the conjunction of a linear increase of VNc (Fig. 12) and a quasi linear decrease of α (Fig. 14). 3.3 Tribo-chimical interpretation of the Nc-RH evolution Fig. 17 compares the fretting scars at the ECR failure for RH=10% (domain I) and RH=90% (domain II). Only the bottom specimens are illustrated, as similar morphologies were observed on the upper conterfaces.

13

ACCEPTED MANUSCRIPT The 3D and 2D surface profiles display a homogeneous wear shape for RH=10% whereas a rather periodic wave distribution is observed for RH=90%. These two fretting scar morphologies are well illustrated examining the SEM pictures. For RH=10% the third body is formed of large platelet structures made of agglomerated nano oxide debris. This powder structure can easily be ejected from the interface inducing higher wear rates. At RH=90%, the

PT

debris appear more compacted and trapped within the inner part of the contact. The higher magnification SEM observations suggest the formation of rolls which can explain the wave

RI

shape observed in the surface profiles. The wear debris ejected from the interface appear

SC

thinner and never display large platelets as observed for the dry condition.

NU

Hence the higher cohesive properties of the third body particularly in the inner part of the contact tend to reduce the debris ejection flow and therefore can explain the lower wear rate

MA

under high humidity conditions.

To better quantify the chemical structure of the worn interface, the Laporte and al.

D

methodology [26] is applied. It consists in averaging the chemical composition in a square

TE

domain with a diagonal length about 20% of the final fretting scar diameter (Fig. 16). This

AC CE P

averaging procedure is applied on upper and lower fretting scar and the averaged value is considered.

Fig. 18 compares such averaged chemical composition at the ECR failure for RH=10% and RH=90%. As previously observed in [22] and [26], the ECR failure is reached when the concentration of silver (noble element) is getting lower than [Ag]th ≈ 5 at% and the oxygen concentration is reaching [O] ≈ 45 at%. The relative humidity does not seem to influence the threshold [Ag]-[O] criterion. The ECR failure is achieved when a homogeneous oxide layer is covering the interface and when most of the noble Ag element is removed from the fretting scar.

14

ACCEPTED MANUSCRIPT The effect of humidity is better illustrated by comparing the concentration of elements present in the Ni sublayer and Cu-Zn brass substrate. It is interesting to note that at RH=10%, the Ni concentration is nearly two times larger than Cu. At RH=90%, a reverse evolution is observed, the Ni concentration is becoming very low (4 at%) whereas the Cu concentration reaches 30 at%. This is consistent with the fact that at higher RH, ECR failure is reached for

PT

larger wear volumes and wear depths, involving in consequence a larger proportion of

RI

substrate elements.

SC

3.4 Evolution of the fretting scar degradation

To interpret the relationship between the ECR evolution and the surface degradation,

NU

interrupted tests were achieved and SEM and EDX analyses of the fretting scar were performed. For dry condition, RH=10% (short Nc), the following tests were analyzed: N=500;

MA

1,000; 5,000; 10,000; 15,000; 20,000; 35,000; 50,000; 62,000; 83,000 and 97,300 [26]. For RH=90% (longer endurance) the studied tests were: N=1,000; 6,000; 50,000; 100,000;

TE

D

400,000; 700,000; 875,000 and 900,000. Fig. 18 illustrates the EDX chemical mapping evolution for RH=90%. It can be observed that

AC CE P

the Nickel sublayer is reached during the first 1 000 cycles and the oxide debris layer appears after less than 6,000 cycles ([O] concentration). Copper and Zinc, substrate elements, are observed after 50,000 fretting cycles. Thus, the fretting scar is damaged and covered with oxide debris a long time before the ECR failure (Nc=900,000 cycles). To quantify these analyses, the evolution of ECR (R), and the averaged chemical concentrations are compared in Fig. 20 for RH=10% and RH=90% versus fretting cycle. Various X scale axis (i.e. linear, logarithmic and N/Nc normalized representation) are considered.

15

ACCEPTED MANUSCRIPT Such a global comparison is very interesting in order to identify the key element controlling the ECR failure. Indeed, the analyses for RH=90% show that the oxygen concentration exceeds the 50 % threshold around 50,000 fretting cycles, long before Nc=900,000. Besides, [O] at N=50,000 cycles is even larger than at the failure. This suggests that the oxygen concentration is not a so pertinent criterion to quantify the ECR failure.

PT

Comparing the evolutions of the copper concentration, it is interesting to note that for

RI

RH=90% the copper concentration is above 30 at% at Nc, while it falls below 10% when RH=10%. This mismatching correlation at Nc is clearly illustrated by comparing [Cu] versus

SC

N/Nc (Fig. 20 e.3). A similar conclusion can be derived by comparing the concentrations of

NU

Nickel. Indeed, the [Ni] concentration at Nc is very different whether RH=10 or 90%. This suggests that Ni and Cu elements and related oxides are not directly controlling the ECR

MA

endurance.

Finally the best correlation is observed for silver material. For any relative humidity, a

D

monotonic decrease is observed. The ECR failure is systematically achieved when [Ag] <

TE

[Ag]th=5 at%. By comparing the [Ag] versus N/Nc it is surprising to see similar tendencies.

AC CE P

[Ag] drops below 20 at% after less than 5% of the Nc total life. Then a plateau evolution is observed until the ECR failure when [Ag] < [Ag]th. Such a plateau evolution is consistent with the flat evolution of the ECR (R), particularly when RH=90%. A similar tendency is also observed at RH=10% (Fig. 20 a.3). However, the plateau response was very long and characterized by a very low resistance value (≈ 2 mΩ) when RH=90% whereas it becomes very short and related to a rather high resistance (≈ 3.5 mΩ) when RH=10%. 3.5 Micro Raman analysis To complete the structural analysis, micro-Raman spectra of the debris layer were performed for the two RH=10% and 90% conditions at Nc and Nc/2 for RH=90% (Fig. 21).

16

ACCEPTED MANUSCRIPT The analysis at Nc/2 for RH=90% was performed because for this test duration the concentration of Ni, Cu and Zn were comparable to those observed at Nc for RH=10%. The Raman spectra suggest that the oxides formed at RH=10% and RH=90% are similar: NiO [31], [32], CuO [33-35] and ZnO [32], [36]. When [Ni] ≥ 20 at% and [Cu] < 10 at% which is observed at Nc/2 for RH=90% and at Nc for

PT

RH=10% (Fig. 20 d.3 & e.3) the Raman analysis suggest a dominating concentration of NiO.

RI

It is only when a significant wear of the substrate is achieved, so that [Cu] ≥ 25 at% (Fig. 20

SC

e.3), that ZnO oxides can be identified from the micro Raman spectrum.

NU

Despite the numerous analyses done, no hydrated structure can be observed for RH=90%.

MA

4. Discussion

The current analyses demonstrate that the key factor controlling the Nc ECR endurance is the

D

concentration of silver element embedded in the debris layer. ECR failure is reached when the

TE

atomic concentration of silver is becoming lower than [Ag]th=5 at% (Fig. 20). This confirms

AC CE P

the conclusion developed by Fouvry and al. [22]. The difference in term of ECR endurance (Nc) may be explained by the significant variation of wear rate with the relative humidity. The debris layer displaying a powder structure under dry condition becomes more cohesive above RH=50% (Fig. 16). The silver element embedded in the third body is less easily ejected from the interface which implies a larger wear volume to reach the ECR failure (Fig. 12). Moreover the higher cohesive properties of the third body require a larger friction energy to eject the debris from the interface which promotes a discontinuous decrease of the friction energy wear rate above RH=50% (Fig. 14). Then by considering that RH=50% corresponds to a cohesive transition of the third body layer, the fast increase of Nc ECR endurance under wet conditions (Fig. 9) can be explained

17

ACCEPTED MANUSCRIPT by the increase of VNc combined with the reduction of α. Hence using such a third body description, the evolution of Nc as a function of the relative humidity can be explained. Another aspect concerns the value of the contact resistance in the plateau period before the ECR failure. Fig. 20 (a.3) confirms that the contact resistance is reduced by a factor 2 when RH=90% compared with dry RH=10% condition.

PT

Various hypotheses can be considered to explain the lower resistivity of the plateau evolution

RI

under wet condition. Such a high conductivity could be induced by the formation of hydrate

SC

structures which are potentially more conductive than the powder oxide debris generated under dry conditions. However micro Raman investigations never revealed the formation of

NU

hydrate structures in the third body layer even for very wet RH=90% conditions.

MA

Another explanation may be related to the relative proportion of Ni and Cu oxides. As deduced from the wear volume investigation, the fact that a larger wear volume is required to

D

reach the ECR failure when RH > 50% induces a larger proportion of Copper oxide in the

TE

third body compared to Ni one. However, this result contradicts the usual conclusion stating that copper oxides are much less conductive than Ni oxides. According to this it may be

AC CE P

expected to observe a higher conductivity at RH=10% ([Ni]=40 at% versus [Cu]=3 at%) than at RH=90% ([Ni]=20 at% versus [Cu]=11 at%). The analyses show the exact opposite. The plateau resistivity at RH= 90% is nearly two times lower than the one observed at RH= 10% (Fig. 20 a.3). Hence, the relative proportion of Ni and Cu oxides is not the controlling factor of the pre-failure third body resistivity. Finally the best correlation seems to be observed with the evolution of the silver concentration. Previous comparisons confirm that the ECR failure is driven by a threshold concentration of silver element remaining in the third body layer so that N=Nc when [Ag] ≤ [Ag]th=5%. The observation of figure 20 (a.3) where the atomic concentration of silver is plotted versus the normalized test endurance (N/Nc) suggests that the pre-failure resistivity is 18

ACCEPTED MANUSCRIPT also controlled by the silver concentration. Indeed, it is interesting to note that during the plateau evolutions, the silver concentration is 2 times superior at RH=90% than at RH = 10% respectively [Ag]plateau

(RH=90%)

≈ 16% and

[Ag]plateau

(RH=10%)

≈ 8%. This tendency is fully

consistent with the ratio close to 2 observed between R[RH=10%] ≈ 3.5 mΩ and R[RH=90%] ≈ 2 mΩ during the pre-failure ECR plateau responses. The higher the silver concentration in the

PT

oxide debris layer, the lower the contact resistivity before the ECR failure.

SC

The following conclusions can be drawn from this study.

RI

5. Conclusion

- A dedicated micro fretting test, allowing representative experiments of fretting wear on

NU

automotive connectors, was developed and connected with a climatic generator controlling

MA

the relative humidity.

- The investigation of an Ag/Ag interface confirms a reduction of the electrical contact

TE

D

resistance (ECR) with the applied relative humidity (RH). - This investigation also confirms an increase of the ECR endurance with the relative

AC CE P

humidity (i.e. Nc when ΔR ≥ ΔRth = 4 mΩ). - A bilinear evolution of Nc versus RH is observed marking a discontinuous evolution at RH=50%. Below RH=50% the rise is smooth and mainly related to a reduction of the friction dissipation (i.e. lubrication effect). Above RH=50% a fast rise is observed which is induced by a simultaneous increase of the wear volume required to reached the ECR failure and a sharp drop of the wear rate. - SEM, EDX and micro Raman investigations confirm that Nc ECR endurance is controlled by the silver atomic concentration embedded in the oxide third body layer. Thus, whatever the relative humidity and surface wear : ECR failure so that N=Nc when [Ag] = [Ag]th =5%

19

ACCEPTED MANUSCRIPT The relative proportion of Ni versus Cu elements controlled by the surface wear extension is not influencing this “silver concentration” criterion. Besides, debris oxidation is effective from the very beginning of the fretting test and therefore cannot be considered as a driving factor for the ECR failure. - SEM, EDX and micro Raman investigations also conclude that the pre-failure plateau

PT

evolution of the contact resistivity is controlled by the concentration of silver embedded in the

RI

oxide debris layer: the higher the silver concentration the smaller the third body resistivity.

SC

- SEM and optical observations suggest that the effect of relative humidity is indirectly controlled by the rheological properties of the debris layer so that the surface wear and related

NU

ECR endurance evolution may be explained by the third body theory developed by Berthier

MA

and co-authors [37-39].

Under dry conditions (RH≤ 50%), a powder debris structure is observed inducing an easy

D

debris ejection from the interface. This can explain the faster elimination of silver from the

TE

contact and the rather small related wear volume required to reach the ECR failure. Besides, it

AC CE P

also explains the rather high wear rate observed for such dry conditions. Under wet conditions (RH>50%) the third body layer appears more cohesive. The silver debris are less easily ejected from the interface which can explain the larger wear volume required to reach ECR failure. This higher cohesive property of the debris reduces the ejection rate and in consequence the global wear rate. The synergic interaction between an increase of the threshold wear volume and a reduction of the wear rate promotes a sharp linear increase of Nc versus RH. This investigation suggests that the beneficial effect of relative humidity is not controlled by a lubrication process as suggested by Park and al. but can rather be explained by an evolution of the rheological properties of debris and the related wear processes. Indeed the friction analysis

20

ACCEPTED MANUSCRIPT shows an asymptotic decreases which stabilizes around µe=0.65 above RH=30%. This value is far below the endurance transition observed at RH=50%. Moreover, although the coefficient of friction is stable above RH=50%, a sharp increase of Nc is still observed. Many other aspects should be investigated to better assess the influence of relative humidity on ECR endurance, as for instance the effect of variable humidity condition. Besides more

PT

fundamental aspects must be considered like the quantification of the rheological properties of

RI

debris layer using new micro mechanical technics [40]. Finally it holds a crucial interest to explain the threshold RH=50% value which appears to be the driving factor of debris

AC CE P

TE

D

MA

NU

SC

rheological properties, wear rate and in fine the ECR endurance.

21

ACCEPTED MANUSCRIPT 6. References

[1]

Quasi-freestanding graphene on Ni(110): A graphene/metal contact with suppressed interface states, Nano Res. 9 (2016) 1795.

[2]

Y. Zhou, Q. Xie, H. Wang, and X. Wei, Investigation on the Thermal Failure of MicroUSB Connectors in Mobile Phones, IEEE Transactions on Components, Packaging and Manufacturing Technology (2017) 1–10 J. C. Guézel, Comment fiabiliser les connecteurs, Ind. Technol. 846 (2003) 56–58.

[4]

M. Antler, Electrical effects of fretting materials: a review*, Wear 106 (1985) 5–33.

[5]

M. Antler and M. H. Drozdowicz, Fretting corrosion of gold-plated connector contacts, Wear 74 (1981) 27–50.

[6]

N. Benjemaa and J. Swingler, Correlation between Wear and Electrical behaviour of contact interfaces during Fretting vibration, Proc. 23rd ICEC Conf. (2006) 215–219.

[7]

Y. Zhou, B. Yao, S. Ge, C. Hong, and J. Zhang, Failure Mechanism of Sliding Electrical Contacts with Various Plated Materials, Proc. 27th ICEC Conf. (2014) 108– 113.

[8]

R. Hombo, N. Kato, T. Nozu, N. Inayoshi, T. Takeno, H. Miki, T. Takagi, J. Fontaine, and M. Belin, Electrical and Tribological Characteristics of Copper Containing Diamond-like Carbon Nanocomposite Coating on Brass Substrate Sliding against Brass Ball, Proc. 27th ICEC Conf. (2014) 154–157.

[9]

R. D. Malucci, Accelerated testing of tin-plated copper alloy contacts, IEEE Trans. Components Packag. Technol. 22 (1999) 53–60.

TE

D

MA

NU

SC

RI

PT

[3]

AC CE P

[10] R. D. Malucci, Characteristics of Films Developed in Fretting Experiments on Tin Plated Contacts, Proc. 45th IEEE HOLM (1999) 175–185. [11] M. Braunovic, Effect of intermetallic phases on the performance of tin-plated copper connections and conductors, Proc. 49th IEEE HOLM (2003) 124–131. [12] S. Noël, N. Lécaudé, S. Correia, P. Gendre, and A. Grosjean, Electrical and tribological properties of tin plated copper alloy for electrical contacts in relation to intermetallic growth, Proc. 52nd IEEE Holm (2006) 1–10. [13] S. Noël, D. Alamarguy, L. Baraton, and P. Laurat, Influence of Contact Interface Composition on the Electrical and Tribological Properties of Nickel Electrodeposits during Fretting Tests, Proc. 26th ICEC Conf. (2012). [14] L. Xueyan, L. Guoping, and X. Liangjun, Fretting Property of Asymmetrical Metal Contact Pairs, Proc. 26th ICEC Conf. (2012). [15] X. Liu, Z. Cai, S. Liu, J. Peng, and M. Zhu, Effect of roughness on electrical contact performance of electronic components, Microelectronics Reliability 74 (2017) 100–109 22

ACCEPTED MANUSCRIPT [16] H. Essone-Obame, L. Cretinon, B. Cousin, N. Ben Jemaa, E. Carvou, and R. El Abdi, Investigation on friction coefficient evolution for thin-gold layer contacts, Proc. 26th ICEC Conf. (2012) 411–416. [17] W. Ren, L. Cui, J. Chen, X. Ma, and X. Zhang, Fretting Behavior of Au Plated Copper Contacts Induced by High Frequency Vibration, Proc. 58th IEEE Holm (2012) 204– 210.

PT

[18] Å. Kassman Rudolphi and S. Jacobson, Stationary loading, fretting and sliding of silver coated copper contacts — influence of corrosion films and corrosive atmosphere, Tribol. Int. 30 (1997) 165–175.

RI

[19] B. H. Chudnovsky, Degradation of Power Contacts in Industrial Atmosphere : Silver Corrosion and Whiskers, Proc. 48th IEEE Holm (2002) 140–150.

SC

[20] T. Imrell, The importance of the thickness of silver coating in the corrosion behaviour of copper contacts, Proc. 37th IEEE Holm (1991).

NU

[21] M. Sun, M. Rong, Q. Wang, and D. Chen, Modification of Ag-plated contacts by nitrogen ion implantation, Proc. 42th IEE Holm (1996) 467–471.

MA

[22] S. Fouvry, P. Jedrzejczyk, and 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 271 (2011) 1524–1534.

D

[23] S. Hannel, S. Fouvry, P. Kapsa, and L. Vincent, The fretting sliding transition as a criterion for electrical contact performance, Wear 249 (2001) 761–770

TE

[24] A. Kassmann Rudolphi and S. Jacobson, Gross plastic fretting mechanical deterioration of silver coated electrical contacts, Wear 201 (1996) 244–254.

AC CE P

[25] Å. Kassman Rudolphi and S. Jacobson, Gross plastic fretting — examination of the gross weld regime, Wear 201 (1996) 255–264. [26] J. Laporte, O. Perrinet, and S. Fouvry, Prediction of the electrical contact resistance endurance of silver-plated coatings subject to fretting wear, using a friction energy density approach, Wear 330-331 (2015) 170-181. [27] Y. W. Park, T. S. N. Sankara Narayanan, and K. Y. Lee, Fretting corrosion of tinplated contacts, Tribol. Int. 41 (2008) 616–628. [28] I.H Sung, J.W Kim, H.J Noh and H. Jang, Effects of displacement and humidity on contact resistance of copper electrical contacts, Tribol. Int. 95 (2016) 256-261. [29] K.L Johnson, Contact Mechanics, Cambridge University Press, 1985. [30] S.Fouvry, Ph Kapsa and L. Vincent, Quantification of fretting damage, Wear 200 (1996) 186-205 [31] N. Mironova-Ulmane, A. Kuzmin, I. Steins, J. Grabis, I. Sildos, and M. Pärs, Raman scattering in nanosized nickel oxide NiO, Journal of Physics: Conference Series. 93 (2007) 23

ACCEPTED MANUSCRIPT [32] T. Tangcharoen, W. Klysubun, and C. Kongmark, Synthesis of nanocrystalline NiO/ZnO heterostructured composite powders by sol-gel auto combustion method and their characterizations, Journal of Molecular Structure 1156 (2018) 524–533 [33] S. Dolai, R. Dey, S. Das, S. Hussain, R. Bhar, and A. K. Pal, Cupric oxide (CuO) thin films prepared by reactive d.c. magnetron sputtering technique for photovoltaic application, Journal of Alloys and Compounds 724 (2017) 456–464. [34] M. Verma, V. Kumar, and A. Katoch, Sputtering based synthesis of CuO nanoparticles and their structural, thermal and optical studies, Materials Science in Semiconductor Processing 76, (2018) 55–60

RI

PT

[35] Q. Zhang et al., CuO nanostructures: Synthesis, characterization, growth mechanisms, fundamental properties, and applications, Progress in Materials Science 60 (2014) 208–337

SC

[36] R. Raji and K. G. Gopchandran, ZnO nanostructures with tunable visible luminescence: Effects of kinetics of chemical reduction and annealing, Journal of Science: Advanced Materials and Devices 2 (2017) 51–58

NU

[37] Y. Berthier, L. Vincent, and M. Godet, Fretting fatigue and fretting wear, Tribology International 22 (1989) 235–242

MA

[38] Y. Berthier, Experimental evidence for friction and wear modelling, Wear 139 (1990) 77–92

D

[39] N. Fillot, I. Iordanoff, and Y. Berthier, Wear modeling and the third body concept, Wear 262 (2007) 949–957

AC CE P

TE

[40] A. Viat et al., Brittle to ductile transition of tribomaterial in relation to wear response at high temperatures, Wear 392–393 (2017) 60–68

24

ACCEPTED MANUSCRIPT Highlights

PT

RI

SC NU MA



D



TE



We studied the influence of humidity on the endurance of silver-plated electrical contacts subjected to fretting wear thanks to a dedicated micro fretting test connected with an automatic climatic generator. The investigations show a bilinear evolution of the endurance (i.e. Nc when ΔR ≥ ΔRth = 4 mΩ) versus RH is observed marking a discontinuous evolution at RH=50%. Below RH=50% the rising is smooth and mainly related to a reduction of the friction dissipation (i.e. lubrication effect). Above RH=50% a fast rising is observed induced by a simultaneous increase of the wear volume required to reached the ECR failure and a sharp drop of the wear rate. Chemical analyses show that the ECR endurance is controlled by the silver atomic concentration embedded in the interface. SEM and optical observations suggest that the effect of relative humidity is indirectly controlled by the rheological properties of the debris layer.

AC CE P



25

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Figure 19

Figure 20

Figure 21