ARTICLE IN PRESS

Optics and Lasers in Engineering 44 (2006) 12831296

Effect of pulsed laser parameters on the corrosion limitation for electric connector coatings
C. Georges, N. Semmar, C. Boulmer-Leborgne
ans, 14 rue dIssoudun, BP6744, GREMI-UMR 6606, CNRS-Universite dOrle ans Cedex 2, France 45067 Orle Received 1 September 2005; received in revised form 20 January 2006; accepted 23 January 2006 Available online 20 March 2006

Abstract Materials used in electrical contact applications are usually constituted of multilayered compounds (e.g.: copper alloy electroplated with a nickel layer and nally by a gold layer). After the electro-deposition, micro-channels and pores within the gold layer allow undesirable corrosion of the underlying protection. In order to modify the gold-coating microstructure, a laser surface treatment was applied. The laser treatment suppressing porosity and smoothing the surface sealed the original open structure as a low roughness allows a good electrical contact. Corrosion tests were carried out in humid synthetic air containing three polluting gases. SEM characterization of cross-sections was performed to estimate the gold melting depth and to observe the modications of gold structure obtained after laser treatment. The effects of the laser treatment were studied according to different surface parameters (roughness of the substrate and thickness of the gold layer) and different laser parameters (laser wavelength, laser uence, pulse duration and number of pulses). A thermokinetic model was used to understand the heating and melting mechanism of the multilayered coating to optimize the process in terms of laser wavelength, energy and time of interaction. r 2006 Elsevier Ltd. All rights reserved.
Keywords: Excimer laser; Pulsed laser treatment; Atmospheric corrosion; Gold coatings; Heat propagation; Melting process

Corresponding author. Tel.: +33 238 494 885; fax: +33 238 417 154.

E-mail address: nadjib.semmar@univ-orleans.fr (N. Semmar). 0143-8166/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.optlaseng.2006.01.008

ARTICLE IN PRESS
1284 C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

1. Introduction Materials used in electrical contact applications are usually constituted of a copper alloy (brass or bronze) interesting for its excellent electrical conductivity and an inexpensive materials cost. However, a major difculty in the use of copper contacts is its high reactivity to atmospheric environment. The classical solution is to use a protective coating of noble metals like gold or palladium. In our study, the copper surface is rstly nickel-plated and then protected by a thin gold coating (0.41 mm thick). The nickel layer is used as a diffusion barrier between the copper and the gold, and the gold layer is used as a corrosion barrier. The nickel and gold coatings are deposited by an electrochemical process. Electrodeposition of these metals usually leads to columnar structures, which favor the concurrent formation of microchannels. In addition, the thickness of the gold alloy coatings must be limited to less then 1 mm for economical considerations. In this case, one cannot completely eliminate the inux of corrosion to the substrate [1]. In order to enhance the protective role of the gold coatings, an additional surface treatment must be applied. Two positive effects are desired: reduction of porosity inside the gold thin layer, and smoothing of the surface to achieve a highly reective surface, an efcient electrical contact, and hence a good product for industry. In this paper a laser heat treatment is tested for this surface modication. The surface treatment for electrical contacts depends on various restraints closely linked to the manufacturing process. It must be fast and usable in air atmosphere. In considering the multilayer compound at the micron level, one should treat only the gold layer without damaging the nickel sub-layer. Moreover, the surface of the material must be heated to the gold melting temperature (without reaching the boiling one) and the temperature should be lower than the nickel melting. Thus, the process requires a low and controlled penetration depth of the heat wave. Pulsed laser re-melting technology, using a pulsed nanosecond lasers, appears to be a soft and fast alternative method. Furthermore, a desired laser treatment could be achieved in a few pulses, that remains interesting and competitive for industrial applications. To check the efciency of our laser treatment, samples were submitted to corrosion tests, microscopy characterization (scanning electron microscopy (SEM), optical microscopy) and chemical analyses (energy dispersive X-ray (EDX), Rutherford backscattering spectroscopy (RBS)) for both treated and untreated samples. In turn, laser parameters were adjusted incrementally until achieving the best improvement in corrosion resistance. To understand the physics of heat melting, the dynamics of laser interaction were modeled via heat conduction equation with phase change. The simulation deals with thermal data including conductivity and thermal diffusivity. Because the situation is different between bulk and thin lm materials, the modeled thermal data were altered with respect to data from the literature to discuss the simulation results. By comparison with cross-sections observed by SEM, realistic values are proposed for thermal data and melting kinetics. These results are shown for gold lms with thicknesses from 200 to 800 nm deposited on nickel substrates.

ARTICLE IN PRESS
C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296 1285

2. Experimental procedures To optimize the laser treatment the laser parameters (wavelength, uence, beam distribution and number of shots) have been varied and applied to samples showing different roughnesses and thicknesses of gold lms. 2.1. Selection of the operating lasers When a laser beam irradiates the connector surface (Fig. 1), only a fraction of the incident beam is absorbed depending on the absorption (A) or reection (R) coefcients (e.g. in the case of opaque surface A 1R). Absorbed photons are distributed in depth or conned in the surface (some nanometers for metallic surfaces in the UV range) and lead to create a conned heat source (d0). Due to the short interaction time, the heat wave propagates only to a small thermal zone noticed zT. Typically, for a laser pulse duration tp 25 ns (average value for excimer lasers used in this study), the thermal depth zT is close to 1.0 mm inside pure gold coating and also for several metals [2]. The nanosecond pulsed lasers seem to be adequate for this surface treatment process bearing in mind that temperature decreases from melting on the surface to nearly room temperature at the interface position. Gold reectivity values are given in the literature [3,4] but there is a spread of values for the same material. Then reectivity values for our gold lms (8590 at% Au) have been determined from experimental investigations using an adapted Fourier transformed infra-red (FTIR) spectroscopy to the UV range. From results in Fig. 2, it can be deduced that efcient lasersurface coupling is obtained for wavelengths less than 0.6 mm. Hence, laser processing should be carried out with a pulsed Nd:YAG tuned onto different wavelengths (266, 355 or 532 nm) or with an excimer laser (157351 nm). The spatial beam homogeneity is also an important parameter to produce homogeneous surface treatment. Excimer lasers provide a quasi-homogeneous rectangular top hat energy distribution, in cross-section. Moreover, this beam is appropriate for surface treatments because it allows exposure of the same energy for the entire treated surface. In comparison, with Nd:YAG

Fig. 1. Scheme of the laser beam interaction with material phenomena.

ARTICLE IN PRESS
1286 C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

100% 90%

Reflectivity coefficient (R: %)

80% 70%
Reflectivity coefficient (R: %) Reflectivity coefficient (R: %) 30% 30% 25% 25% 20% 20% 15% 15% 10% 10% 5% 5% 0% 0% 0.25 0.25

60% 50% 40% 30% 20% 10% 0% 0.25

0.3

0.35 0.35

0.4

0.45 0.45

0.5

wavelength (m) wavelength (m)

0.5

0.75

1.25

1.5

1.75

2.25

2.5

2.75

wavelength (m)
Fig. 2. Reectivity coefcient of gold versus light wavelength.

Fig. 3. Optical observation of the homogeneous impact of the KrF excimer laser beam impact after one pulse.

lasers beam energy distribution is inhomogeneous and induces hot points. In Fig. 3, the characteristic appearance of the uniform and shiny laser impact obtained with a KrF excimer is illustrated. Consequently, in this work the major part of treatments were performed with excimer lasers. Two excimer lasers: a Lambda Physics Compex 205 in KrF gas mixing conguration (l (wavelength) 248 nm, t (pulse duration) 25 ns) and a Questek in XeCl gas mixing conguration (l 308 nm, t 28 ns) were selected and compared for this treatment. The laser beam was focused on a sample surface and the most homogeneous part of the beam was selected using a diaphragm. The sample was mounted in a motorized XY stage, allowing sample translation in front of a

ARTICLE IN PRESS
C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296 1287

xed laser. In order to compare the inuence of the laser pulse number at the same location, 1100 shots on the same impact site were performed. 2.2. Samples characteristics Samples were constituted of a substrate (copper alloybrass or bronze) of 0.2 mm thickness covered by two successive coatingsnickel layer (2 mm) and gold layer (0.75 mm). Tested samples had different roughnesses (mirror or laminate polished (Ra o0:1 mm) or laminate (Ra X0:15 mm)) in order to study their inuence on the laser treatment efciency with respect to the corrosion tests. 2.3. Corrosion tests Several atmospheric corrosion tests were performed using three or four gases (Cl2, NO2, SO2 and H2S) known to be aggressive with electrical contacts. The corrosion rates used in this study were xed by the Bellcore norm [5]. The tests were carried out at 25 1C, in humidied air (relative humidity 85%) containing three polluting gases: NO2, SO2 with concentrations of 0.2 vpm (volume per million, i.e. 0.2 106 liters per air liter) and Cl2 with concentration of 0.01 vpm. The test duration varied between 3 and 12 days. At regular intervals of time, the samples were characterized to check the evolution of the corrosion.

3. Results Three major parameterslaser uence, substrate roughness and laser beam homogeneity have been investigated. 3.1. Effect of laser uence Different laser uences were used to determine the energy range needed for gold melting in the case of rough substrates (Ra 40:15 mm) commonly used in industrial applications (0.75 mm gold thickness). For uences lower than 400 mJ/cm2 laser impacts were not observed. For uences ranging between 400 and 700 mJ/cm2 the laser impacts are gloss (as wanted for connector market) and over this uence value, the laser impact induces a surface aspect modication (impact zones become white). In Table 1, the treatment results on laser impact appearance and corrosion behavior are classied. There was no signicant improvement in corrosion resistance observed for samples treated with one laser shot at uences lower than 700 mJ/cm2. Corrosion aggregates were always very numerous in the low uence treated zones (for more details, please see Ref. [6] corresponding to a previous study). Samples treated at uences higher than 750 mJ/cm2 exhibited more promising corrosion behavior. The tests have been achieved for few laser shots from one to ve. Optical microscopy photos in Fig. 4 (top) show the comparison between the untreated area (a) and treated areas, respectively with one (b) and ve (c) laser shots at 850 mJ/cm2 after

ARTICLE IN PRESS
1288 C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

Table 1 Experimental laser conditions and corrosion behavior for samples treated between 400 and 950 mJ/cm2 Excimer laser KrF uence (mJ/cm2) 400/450/500/550/600/650/ 700 700/750/800/850/900/950/ Number of laser pulse/impact 1 1,2,5,10 Impact appearance Corrosion behavior

Gloss Opaque (uences4750 mJ/cm2)

Bad Good (number of pulses 45)

Fig. 4. Optical microscopy (top) and SEM (bottom) images of samples after corrosion tests (a) as received, (b) treated with one laser pulse at 850 mJ/cm2, and (c) with ve laser pulses at 850 mJ/cm2.

corrosion tests. The corrosion germs for a sample treated by one laser pulse (b) are smaller than the corrosion germs observed in the untreated zone (a). For samples treated by ve laser pulses (c), the corrosion germs in the laser impact area are much less numerous. SEM observations (Fig. 4 bottom) at high magnication ( 10,000) of the three zones ((a) untreated area, (b) treatment with one laser pulse, (c) treatment with ve laser pulses) show a surface modication increasing roughness versus the pulse number. This last observation has been veried using an optical prolometer. The roughness measurements corresponding to 900 mJ/cm2 laser uence are listed in Table 2. In order to explain the surface aspect and roughness changes of these samples (Ra 40:15 and 0.75 mm gold coating) different analyses were achieved. SEM, RBS and EDX analyses have been done inside the impact zones. On samples treated at

ARTICLE IN PRESS
C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296 1289

Table 2 Average (Ra) and maximum (Rmax) measured roughnesses versus laser pulse number for 900 mJ/cm2 As received zone Laser impact with one laser pulse 0.16 0.21 Laser impact with two laser pulses 0.18 0.25 Laser impact with ve laser pulses 0.22 0.66 Laser impact with ten laser pulses 0.30 1.10

Ra value (mm) Rmax (mm)

0.15 0.2

14000

12000

treated zone (10 laser pulses)

untreated zone

number of counts (a.u) nu

10000

8000

6000

4000

2000

Gold thickness 0.8 m

0 100 200 300 400 500 600 700 800 900 1000

0 channel number

Fig. 5. RBS spectra of sample treated with 900 mJ/cm2, ten laser pulses (2 MeV alpha beam).

900 mJ/cm2, RBS spectra obtained in the treated and untreated zones are given in Fig. 5. In the untreated case, the gold thickness is represented by the wide and distinct range from channels 350 to 950. In the treated case, the gold/nickel interface is not dened and tends to expect the formation of a gold/nickel alloy. However, SEM observations (Fig. 6a) show a clearly distinguished gold/nickel interface, but with irregular gold thickness. As illustrated on Fig. 6b, for 10 mm width zone, the local gold thickness varies from 0.55 to 2 mm since the average thickness is 0.75 mm. During the RBS analysis, ion beam focuses onto 1 mm2 zone. So its spectra results from the averaged thicknesses and consequently it could not be well dened. Additional EDX analyses were performed in the laser impact zone and revealed only gold presence. Consequently, SEM observations coupled to EDX analyses are evidences that there is no formation of goldnickel alloy. With high uences (4750 mJ/cm2) and a number of laser pulses higher than ve, an improvement in corrosion resistance was achieved. In the same time, roughness is

ARTICLE IN PRESS
1290 C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

Fig. 6. SEM observations of cross-sections of a sample treated with 900 mJ/cm2, ten laser pulses.

increased leading to bad electrical contacts undesirable for the industrial process. One should conclude that lower uences (o 700 mJ/cm2) are necessary for treating these samples. However, in this case (high roughness and low uence) no corrosion improvement was achieved as shown previously [6]. Hence polished samples (Ra o0:1 mm) will be considered in the following part. 3.2. Observation of the surface smoothing and melting On Fig. 7, results of corrosion tests on a polished sample treated by 550 mJ/cm2 are presented for one laser pulse. The laser impact is delimited by the white dots. The corrosion germs are much smaller than in the untreated area. As detailed in a previous study [7], it is shown that the laser treatment improved the corrosion resistance by reducing the size of the germs. As a general rule, the best results are obtained for a number of pulses ranging from one to ve and laser treatment is more efcient on polished samples. These results are justied by the fact that the melting occurs on a small surface zone. When the roughness is high, some zones could not be recovered during the melting process. SEM observations of two zones with some cracks and scratches on the gold layer are presented in Fig. 8a. Observations of the same zone after laser processing (one pulse Questek XeCl at 550 mJ/cm2) show the evolution of surface defects according to their size Fig. 8b. The laser treatment eliminates scratches into the thickness of the gold layer and smoothes the surface (the supercial gold aggregates are melted). For more signicant and major defects as the cracks, there was an improvement but not total elimination (bottom). To study the laser treatment effect on the gold melting thickness, SEM characterizations on cross-sections of samples treated with the KrF laser at different uences and with one laser pulse (Fig. 9) have been performed. For sample treated at 430 mJ/cm2 (Fig. 9a) any evident melting zone in the gold thickness cannot be observed. For sample treated at 750 mJ/cm2 (Fig. 9b), the formation of cavities at 0.15 mm in depth can be observed, i.e. the laser melting treatment only acts on

ARTICLE IN PRESS
C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296 1291

Fig. 7. Optical observation of a polished sample (Ni: 2 mm, Au: 0.75 mm) after corrosion test (192 h); the treated area is surrounded by doted lines (550 mJ/cm2, one laser pulse).

Fig. 8. SEM observations of an identical zone before and after laser treatment (550 mJ/cm2, one laser pulse) of a polished sample.

ARTICLE IN PRESS
1292 C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

Fig. 9. SEM observations of cross sections of samples (substrate Ra 40:15 mm) treated with different laser uences, after one laser pulse.

0.15 mm. For sample treated at 900 mJ/cm2 (Fig. 9c) the cavities are initiated at the gold/nickel interface and there is a surface damaging. The gold melting thickness with laser treatment is less than 0.15 mm (for uences lower than 700 mJ/cm2). The cross-section images show that under 500 mJ/cm2 there is no evidence of melting with depth. Above a laser uence of 700 mJ/cm2, there is a damaging of the gold layer, probably at the melting zone limit, and there are also cavity formations. 3.3. Effect of the laser beam space distribution As the efcient uence range necessary to treat the gold surface without damaging is very small (550700 mJ/cm2) it is essential to have a spatially homogeneous laser beam to avoid any hot points. For the two excimer lasers used in this study, the results concerning the corrosion resistance are nearly similar but the use of the XeCl laser seems to give better results maybe because its time duration is slightly longer (28 ns compared to 25 ns for KrF excimer laser). To check this hypothesis, samples were irradiated by one pulse of a new Cilas-XeCl (l 308 nm, t 50 ns) laser beam at 600 mJ/cm2. Corrosion tests on these samples do not show any signicant improvement of corrosion resistance, as this uence was not high enough to melt sufciently the gold layer. Indeed, the instantaneous power density is too weak (E12 MW/cm2 instead of E20 MW/cm2 for Questek XeCl and Lambda KrF.) To complete this study, a thermal modeling has been undertaken to dene the best laser conditions able to predict the needed laser uence depending on the laser pulse time duration and shape that can melt the thin gold lm without melting the nickel layer.

4. Simulation of laser-induced melting A one-dimensional unsteady heat equation was solved using nite difference method [8], with discretization of space (X) and time (G) as reduced variables. The related boundary conditions were already published [9]. To ensure the convergence

ARTICLE IN PRESS
C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296 1293

of the numerical calculation, the number of matrices was selected greater than 1000 for the two variables (time and position). The optical depth (d0) is close to 10 nm when the thermal depth (zT) is about 1000 nm for 25 ns pulse duration (average value considering lasers used in this study), the absorbed laser beam is treated as a surface boundary condition according to the optical depth, and not as a volume heat source [2]. In the case of speed heat diffusion the thermal depth may be easily estimated using p a good approximation [2] given here by the following relation: zT atp , where a is the thermal diffusivity, T the absolute temperature, t the time variable, z the space variable (depth) and tp the pulse duration. Despite the fact that the thermal diffusivity value of pure gold in bulk state is well known (E1 cm2/s) [10], it remains difcult to obtain exact values for gold thin lms used for the coatings on connectors. We have studied the change of the melting depth according to different parametersthermal properties, laser pulse shape and nally gold lm thickness. In the last case it was necessary to consider a thermal contact resistance at the substrate interface. 4.1. Effect of thermal properties The rst simulation is employed to understand the effect of the thermal diffusivity (a) on the melting depth time behavior. When the gold layer is electroplated on the substrate, the heat propagation is strongly modied by the presence of grains (Fig. 8), contrary to the same material in the bulk state. In practice we have no data to correlate the physical properties at a macroscopic level (i.e. thermal and electrical conductivity, mechanical properties) to the elaboration processing of thin layers [11,12]. So to start this simulation we t laser power pulse shapes, measured by HAMAMATSU UV photodiode, for lambda KrF (l 248 nm, t 25 ns) and Cilas XeCl (l 308 nm, t 50 ns) by linear curves (P-KrF and P-XeCl) as shown on Fig. 10. The Questek XeCl laser form is not represented because it is similar to the lambda KrF shape. Considering the tted curves at 750 mJ/cm2 absorbed laser uence (e.g. laser power Pa in W/m2), for 25 ns pulses in the KrF laser at the FWHM, 3.0 1011 W/m2 maximum power is reached in 10 ns and for 50 ns pulses in the XeCl laser at the FWHM, 1.5 1011 W/m2 maximum power is reached in 10 ns. For the bulk state thermal diffusivity (e.g. 1 cm2/s) the simulation reveals no melting. Nevertheless the thermal diffusivity for this calculation may be lower as it is applied to a thin lm. Moreover, the gold microstructure (Fig. 8) has a consequence in the change of the diffusivity values. Curves C1, C2 and C3 on Fig. 11 correspond to melting depths (x) obtained for different diffusivity values (a) of, respectively, 0.4, 0.2 and 0.1 cm2/s. The laser uence considered in this case (750 mJ/cm2) was chosen to be higher than experimental laser uences to enable the study of melting change versus thermal diffusivity. By considering laser uences below this, no melting was predicted with thermal diffusivities equal to 0.2 or 0.4 cm2/s. For a thermal diffusivity value 0.1 cm2/s, the melting threshold is equal to 375 mJ/cm2 in absorbed uence (430 mJ/cm2 in the real case). This result is in agreement with experimental

ARTICLE IN PRESS
1294
3.5E+11 P-KrF 3.0E+11 0.5

C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

0.6

2.5E+11 0.4

Pa (W/m)

2.0E+11

D-KrF P-XeCl 0.3

1.5E+11 D-XeCl 1.0E+11 0.2

5.0E+10

0.1

0.0E+00 0 10 20 30 40 50 60 70 80 90 100

t (ns)

Fig. 10. Laser pulse shape effect on the melting depth.

3.5E+11

0.5 0.45

3.0E+11 0.4
P-KrF

2.5E+11

0.35 0.3 0.25

Pa (W/m)

1.5E+11

C3

0.2
C2

1.0E+11

0.15 0.1

5.0E+10

C1

0.05 0

0.0E+00 0 10 20 30 40 50 60

t (ns)

Fig. 11. Melting depth versus time for different diffusivity values.

results which have shown that the laser uence must be at least equal to 400 mJ/cm2. It can be concluded that the thermal diffusivity of the gold layer is very low in comparison with the value of the handbook data for the bulk state.

(m)

2.0E+11

(m)

ARTICLE IN PRESS
C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296 1295

To melt 0.25 mm of the layer, with the Lambda KrF laser beam, the surface must absorb 750 mJ/cm2 for 0.1 cm2/s thermal diffusivity value. Under the same conditions, 0.2 mm of the gold layer is melted at 600 mJ/cm2. The lifetime of this melting bath is then close to 20 ns. As a general rule, when the diffusivity value is low the melting depth and its lifetime are larger, and the critical time of melting initiation becomes shorter. 4.2. Effect of laser pulse shape To study the laser shape effect on the melting kinetics, a second simulation was performed using a Cilas XeCl pulse shape (l 308 nm, t 50 ns) and considering a xed value of thermal diffusivity (0.1 cm2/s). Illustrations on Fig. 10 show that for the same laser uence (e.g. 750 mJ/cm2) the melting depth (x) is different between curves D-lambda KrF and D-Cilas XeCl (D for melting depth curves and P for laser power intensity curves). In the case of P-lambda KrF laser pulse, the maximum of instantaneous power is two times higher than the P-Cilas XeCl laser pulse, and it is reached in nearly the same time (10 ns). So, the maximum melting depth ratio (e.g. D-lambda KrF to D-Cilas XeCl) seems to be proportional to the ratio of maximum laser power. Conversely, the lifetime of the melting zone is more important in the case of P-Cilas XeCl pulse. Despite the weak value of the melting depth, the heat induced after 50 ns laser interaction is sufcient to maintain the liquid state. The P-Cilas XeCl laser pulse shape gives the maximum melting lifetime, which is more advantageous to recover the best surface homogeneity. However, as we have seen previously the laser space inhomogeneity remains a limiting factor. 4.3. Effect of gold layer thickness and interface roughness To understand the effect of lm thickness on the melting kinetics, we have extended our modeling to study the thermally thin metallic coatings [8]. The thermal contact resistance is introduced in the computation procedure when the heat wave propagation reaches the gold/nickel interface. It is then possible to calculate the critical temperatures and the melting threshold uence for high and low contact resistance values (e.g. high and low substrate roughness). Under these conditions, the temperature prole and melting depth were plotted considering different gold layer thickness (200, 500 and 800 nm). When we consider thermal diffusivity a 0:1 cm2 =s and laser uence 750 mJ/cm2, the most important results are:

 

Larger the layer thickness (200, 500 and 800 nm), higher the maximum melting zone (100, 220 and 340 nm, respectively). But the ratio of melting zone to layer thickness decreases steadily. Smaller the layer thickness, higher the effect of increasing thermal resistance at the interface on the melting zone. Typically, for 0.8 mm gold layer no change occurs on the melting zone (340 nm) whatever the value of the thermal contact resistance (from 1.7 1011 to 1.7 107 K m2/W) [8].

ARTICLE IN PRESS
1296 C. Georges et al. / Optics and Lasers in Engineering 44 (2006) 12831296

5. Conclusion This study demonstrated that laser processing could improve the corrosion resistance in gold-plated contacts for several laser conditions. From the observed results compared to modeling data, it appears that the most important laser parameter limiting the processing efciency is the space beam homogeneity. Concerning the material parameter, the roughness substrate is the most limiting one. As a general rule, the best results for laser melting treatment against corrosion can be obtained using a pulsed laser emitting in the visible or UV wavelength range. It has been observed that the laser treatment is highly efcient on polished samples (Ra o0:1 mm), nevertheless the treatment stays available for a substrate roughness reaching a value as high as 0.15 mm. The number of laser pulses has to range from 1 to 5 at the same location on the sample and the laser uence must be over 550 mJ/cm2 to be efcient in corrosion tests but has to be limited to 750 mJ/cm2 to prevent the sample from a roughness increase of its surface. The modeling has been considerably improved during the study by adapting the thermal diffusivity values. The melting zone is calculated close to the range 100300 nm in the case of 0.40.2 cm2/s thermal diffusivity values, and is in a good agreement with the SEM observations. The role of thermal contact resistance is not always important except for coatings less than 200 nm. References
[1] Odvenall I, Leygraf C. J Electrochem Soc 1997;144(10):351825. [2] Prokhorov AM, Konov VI, Ursu I, Mihailescu IN. Laser heating of metals, Adam higher series. Bristol: IOP Publishing; 1990. [3] Touloukian YS, DeWitt DP. Thermal radiative properties: metallic elements and alloys, vol. 7. New York: IFI, Plenum; 1972. [4] Hottel HC, Sarom AF. Radiative transfer. New York: McGraw-Hill; 1967. [5] Norm Bellcore Generic requirements for separable electrical connectors used in telecommunications hardware GR-1217-CORE, Issue 1. November 1995. [6] Georges C, Sanchez H, Semmar N, Boulmer-Leborgne C, Perrin C, Simon D. Appl Surf Sci 2002;186(14):11723. [7] Georges C, Semmar N, Boulmer-Leborgne C, Perrin C, Simon D. Proc SPIE 2002;4760:98693. [8] Semmar N, Boulmer-Leborgne C. J Phys IV 2004;120:41320. [9] Semmar N, Georges C, Boulmer-Leborgne C. Microelectron J 2002;33:70510. [10] Touloukian YS, Makita T. Thermal properties of matter. New York: Plenum; 1970. [11] Orain S, Scudeller Y, Brousse T. Seventh therminic workshop. 2001. 2569. [12] Orain S, Scudeller Y, Garcia S, Brousse T. Int J Heat Mass Transfer 2001;44:397384.