This work describes the Silveron™ GT-820 Silver-Tin as a robust process for the electrodeposition of silver-tin alloys of composition equal to Ag₃Sn and Ag₄Sn. A design of experiment method was used to study the influence of process parameters on the deposit performance. It was found that the alloy composition is controlled by organics components of the bath. It is independent on the metal concentration ratio and the applied current density inside the specified process window. The silver-tin deposit is semi-crystalline directly after plating but can be converted to a fully crystalline material by a short annealing. This coating is not only ductile and tarnish resistant but also displays a good insertion and retention force needed for press-fit interconnections. The product was tested successfully under serial production conditions and shown to be suitable for tin and tin-lead replacement in solderless interconnections.

1 Introduction

In recent years, silver-tin alloys have become very attractive because of their broad range of applications in electronics. Alloys with composition close to Ag₃Sn are very promising for decorative applications due to their tarnish resistance [1], or for electronic applications because of the high ductility, hardness and abrasion resistance [2, 3]. One of the most important demonstrations of silver-tin deposits for contact finishing is described in the patent of Degner et al. [4] where silver-tin was plated at very low current density using barrel techniques. It was disclosed that the coefficient of friction of a silver-tin alloy decreases with increasing tin content and tends to a value of 0.3 for Ag₄Sn intermetallic. Furthermore, the wear performance of a silver-tin deposit with less than 15 wt% tin was as poor as a pure silver deposit. Improved wear resistance was observed above 15 wt% of tin in the alloy. The contact resistance was shown to increase with the tin content of the alloy but remains below 10 milliohms for Ag₄Sn/Ag₃Sn intermetallic [4]. It is also known that the microhardness of a silver-tin deposit is a function of tin content; it reaches a maximum when the silver lattice is saturated with tin, then decreases due to the presence of a free tin phase in the matrix [3].

All of these studies have been carried out using unstable processes (from either cyanide or cyanide-free baths) which have a limited current density range and are not suitable for mass production. The main difficulty is finding a suitable complexing agent which will sufficiently stabilize silver to avoid spontaneous reaction with tin. This problem was solved recently with Silveron™ GT-820 Silver-Tin, a cyanide-free silver-tin electrolytic process developed by DuPont Electronics and Industrial to produce a silver-tin alloy with a composition close to the ζ-Ag₄Sn and the ε-Ag₃Sn intermetallic phases.

The aim of this work is to study the influence of the process parameters on alloy composition. Advanced properties of the electrolytic silver-tin deposit such as the behavior under thermal treatment, the passivation and the copper diffusion reaction will be described. The pre-industrialization work done in collaboration with several industrial partners such as ROBERT BOSCH GmbH and ept GmbH will be discussed.

2 Experimental

All plating baths for silver-tin electroplating were prepared by using the Silveron™ GT-820 Silver-Tin Makeup solution, the Solderon™ Tin Concentrate and the Solderon™ AO-52 Antioxidant available from DuPont Electronics and Industrial. Silveron™ Silver Solution and Silveron™ GT-820 Silver-Tin Booster  (also available from DuPont Electronics and Industrial) were used respectively to adjust the silver concentration and the free complexing agent (FCA) concentration to any desired value. The electrical conductivity of the electrolyte was achieved by adding 120 ml/L Solderon™ Acid HC, a special methane sulfonic acid available at DuPont Electronic and Industrial. This also provides the low pH (out of measurement limit) needed for the stability of tin(II) ions in the solution.

A “Design of Experiment” (DoE) method was used to setup different test matrices to study the impact of different process variables on the alloy composition or on the deposit properties.

Table 1 displays the DOE matrix used to evaluate the stability of the Silveron™ GT-820 Silver-Tin process. For each run, the Solderon™ AO-52 Antioxidant concentration was 12 ml/L. In additional runs not listed in the table, the silver concentration, the free complexing agent (FCA) concentration, the acid content, or the temperature was lower or higher than those in table 1.

A classical 2 Ampere Hull cell method was used to evaluate the impact of variables on the operating window. This method provides information on the aspect of the deposit under a broad current density range such as 0.1 to 10 Ampere per square decimeter (ASD) for 2 A Hull cell. For each hull cell test, the plating time was 2 min and the agitation was provided by 3 cm magnetic stirrer at 500 rotation per minute (rpm). The thickness on the hull varied with the current density and was generally above the target thickness (0.5 µm) for press-fit interconnections.

For alloy composition measurements, silver-tin was plated on connector pins or on flat nickel-coated brass substrate using a beaker containing silver and platinized titanium mixed anodes. Two current rectifiers were used, one to supply 80% of the total current to the silver anode and the second provide 20% of the total current to the insoluble anode; the cathode of each of rectifier was connected to the sample. The plating time needed for 0.5 µm target Silver-tin thickness was 60 sec and 6 sec respectively for 1 and 10 A/dm². This shows a linear correlation between the plating time and the current density.

A Fischerscope X-Ray Model XDV-SD (X-ray fluorescence (XRF) device from Helmut Fischer AG) was calibrated with different silver-tin standards provided by the supplier and used to measure the alloy composition of silver-tin deposits. The texture and the influence of annealing on the recrystallization of silver-tin were studied with the X’Pert High Score PW 3209 (from Philips) x-ray diffractometer operating under reflection mode with the Copper Kα radiation.

The ductility of silver-tin deposit was measured using a Bend Tester device according to the ASTM B489-85 specification.  For diffusion and recrystallization tests, coatings were heated in a Binder force convection oven model FD 56 operating under ambient air conditions. The tarnishing of the silver-tin deposit was studied immersing coupons in a 2 wt% potassium polysulfide solution for 10 min at room temperature.

The bath composition was also varied under serial production conditions (at ept GmbH) to see the impact on the quality of the deposit. First, a bath of table 2 with low concentration of tin was made and tested. This was further adjusted stepwise to the desired design by adding extra chemicals. The plating speed of silver-tin process is 0.5µm/min at 1 A/dm². The number of plating cells was modified according to the line speed and the applicable current density to get the time need for 0.5 µm (required thickness for Press-fit application) Silver-tin. A pure silver strike (activation layer) of about 10 to 30 nm was coated on nickel to promote the adhesion silver-tin deposit.  All plated samples were subjected to quality control tests, e.g. appearance, alloy composition using XRF device and adhesion test by pin bending method, to final assembly and testing to study the impact of variables on the desired response. The insertion and the retention force were measured at Robert Bosch GmbH and ept GmbH according to IPC-9797 specification.

3 Results and discussion

3.1 Influence of the bath composition and parameters on silver-Tin deposit

Hull cell is a fast and simple method to evaluate the appearance and the applicable current density range of a process. Bath compositions displayed in table 1 were tested by using both Hull cell and direct electroplating on connector pins. Figure 1 shows the impact of extreme variation of the bath composition on the operating window of the silver-tin deposit. A uniform bright deposit (dark area in figure 1) is generally obtained in the current density range from 0.1 to 6 ASD independent of the metal concentration ratio in the bath. Higher applicable current density such as 8 ASD can be achieved with a high silver concentration.

To confirm the stability and the robustness of the process window, samples prepared from various DOE matrices were measured and all data were analyzed using a statistical analysis software. Figure 2 displays the alloy composition of silver-tin plated on connector pins as function of the process variables. For the same sample, measurements were taken at different positions of the pin indicated by the red spots in figure 2b. This gives a general view on the alloy composition and the thickness distribution on complex geometries. The mean diamond in figure 2, shows that at constant metal concentration, temperature, and FCA, the silver content decreases slightly with the increasing current density. In general, the silver content of the deposit is between 78 and 85 wt% with the mean value at 81.5wt%. This value is relatively higher than the exact alloy composition due to the presence of pure silver strike layer between the nickel and AgSn₂₀. Since the thickness of the AgSn₂₀ deposit was around 0.5 micron (specification for press-fit application), a pure silver strike layer (about 10 to 30 nm) at the nickel interface can impact the final alloy composition measurement.

Within the same plating parameter, the variation of the alloy composition shown in Fig.2a is due to the different plating conditions at the individual locations on the strip, the round pin geometry, surface roughness of stamped pins, the measurement position (different current distribution) on the pin, or the irregular distribution of the silver strike layer at the interface. For the lab DOE, samples were plated on a strip with many pins in a glass beaker without any shielding, this is not the ideal condition for uniform thickness distribution. For those measurements, the XRF standard for thin silver strike was not available from the supplier and the strike thickness could not be considered in the measurement. With 30 nm pure silver strike bottom layer, the measured silver concentration in 0.5 µm AgSn₂₀ is about 1.2 wt% higher than in the real AgSn₂₀ composition. The absolute XRF measurement error is about ± 0.35 wt%. These factors account for the variability observed in the alloy composition. More stable and uniform alloy composition with around 79% silver was measured when AgSn₂₀ is plated on flat Hull cell coupons. Because of the above variability, another DOE was done on a serial production line to validate the alloy composition and the thickness distribution.

Despite this variation within individual sample, the alloy composition remains substantially stable when changing the bath formulation, the applied current density, or the operating temperature. For alloy electroplating, changing the metal concentration ratio usually impacts the alloy composition. In the current process, a strong variation of the metal concentration ratio in the bath has almost no impact on the alloy. The alloy composition is controlled by the additive present in Silveron™  GT-820 Process bath rather than the concentration gradient of metallic ions. Such behavior is due to the complexation reaction which modified the plating potential of silver and tin ions. This is vital for the industrialization of an electrolytic silver-tin process because minor variations of the process parameters will not affect the quality.

3.2 Effect of thermal treatment on electrolytic silver-tin deposit

3.2.1 Recrystallization of silver-tin

Properties of electrolytic materials depend on factors such as the amount of organic co-deposition, the stress created by the hydrogen evolution, or the crystallinity of the material. Generally, directly electroplated materials are not in thermodynamic equilibrium, therefore is either amorphous or semi-crystalline. The diffraction pattern of the silver-tin coating was evaluated on a  2.5 µm AgSn₂₀ deposit plated using the optimal bath formulation. Thick AgSn₂₀ such as 2.5 µm was selected to minimize measurement inference from the base material and to improve the signal-to-noise ratio. Figure 3a compares the texture of the silver-tin deposit as plated and for different annealing times at 150 °C. As plated, the deposit has a broad peak with low intensity at 2 Theta around 35 to 40 degrees. After annealing at 150 °C for 2 min, three peaks appear at low diffraction angle with the intensity ratio close to ε-Ag₃Sn intermetallic compound. Further annealing could not change the intensity of the respective peaks. This suggests that the material is at the most stable crystalline phase. At 150 °C, the recrystallization of thin AgSn₂₀ is complete after 2 min.

Figure 3b displays the texture of AgSn₂₀ deposit as a function of the annealing temperature and time. For clarity, only the low 2 Theta diffractions which contain the main characteristic peaks of Ag₃Sn /Ag₄Sn are displayed in 3b. At 100 °C, the maximum XRD peak intensity is achieved for annealing times between 10 and 30 minutes, further annealing did not change the intensity. At intermediate temperatures such as 150 – 180 °C, two minutes are enough to achieve the maximum intensity which corresponds to the complete recrystallization. For higher temperatures (210 – 240 °C) stable peaks are obtained after 1 min.

Table 2 summarizes the recrystallization time of 2.5 µm AgSn₂₀ deposit as a function of the annealing temperature. The time required decreases with the increasing temperature.  High temperature will provide more activation energy for the preferred crystal orientation. The recrystallization may have occurred earlier than the time provided in the table, but this could not be identified clearly because the test was performed under static heating rather than a dynamic temperature ramping follow by in-situ XRD measurement. Minimum time of 30-60 seconds  is required for temperature stabilization after placing the sample in the oven. Therefore, results of table 2 should be considered as the approximated recrystallization time rather than real values.

The short recrystallization time observed at high temperature is very interesting because this could be easily implemented at the production line if a crystalline AgSn₂₀ deposit is desired. Such annealing can be achieved by increasing the temperature in the drying furnace or by extra furnace installation. Based on the above data, full recrystallization could be expected within 5 to 15 sec at 250 °C. Such a short time range is typical for a reel-to-reel production line.

3.2.2 Recrystallization and ductility of silver-tin

Samples plated using the DOE test matrix in table 1 were cut into two parts. The first part was tested directly after plating while the second half was heated at 180 °C for 1 hour for recrystallization before testing.  Figure 4 displays the effect of the recrystallization on the properties of silver-tin deposit. The ductility of a 1.5 µm deposit depends on the plating current density and the post-treatment. With the Bend Tester device, the maximum measurable ductility value is 11.8%; this corresponds to about 180 degree bending of the coupon.

For a fresh silver-tin deposit, the ductility is around 7.8 % at low current densities such as 1 to 5 A/dm² and for high tin concentration in the bath. The value drops to about 3.5 to 4 % at high current density. Ductility value of about 7% meets the requirement for connector applications. Samples plated around the limiting current density of the process (~10 A/dm²) and low temperature were burned at some positions and may account for the low ductility values. In general, the silver- to- tin concentration ratio in the bath also affects the ductility of fresh deposits. From low to high current density, the decrease in ductility is more pronounced for baths containing high silver-to-tin concentration ratios.

After recrystallization, the ductility of the silver-tin deposit increases to above 11% for all bath compositions and most current densities. These good values correspond to the maximum elongation which can be measured using the Bend Tester device. Only few samples plated at 10 A/dm², which were partially burned, showed a low ductility value after recrystallization.

3.2.3 Recrystallization and tarnishing of silver-tin

The tarnishing of the silver-tin deposit was studied without surface modification with a self-assembled monolayer. A pure silver sample was added for comparison. Figure 5 compares the tarnish test result of different silver-tin samples with the pure silver reference.

Without anti-tarnish protection, pure silver displays the characteristic dark blue discoloration after immersion in polysulfide; this confirms the effectiveness of the polysulfide test solution. Under the same testing conditions, a fresh silver-tin alloy shows a light brown discoloration which indicated a limited reaction of the deposit with the polysulfide. Interestingly, silver-tin samples which were pre-heated at 150 °C remain fully bright after 10 min immersion. There is no reaction between the polysulfide solution and the recrystallized silver-tin deposit.  Further XPS investigation was done to understand why fresh silver-tin deposit behaves differently to the recrystallized sample.

In the case of a fresh silver tin deposit, the XPS results of figure 6 show a high intensity of the silver 3d peak followed by a low tin 3d intensity. This indicates a greater silver-to-tin atomic ratio on the surface. After annealing at 150 °C or after high-temperature / high-humidity test, silver 3d signal intensity is smaller than the tin 3d peaks. The surface is richer in tin atoms than silver atoms. Those ratios are confirmed by the atomic percentage of elements displayed in table 3. Data of table 3 indicates a high ratio between tin oxide and the metallic tin on the surface of those samples. Under humidity or annealing conditions, a tin-hydroxide or tin-oxide layer is formed on the surface of silver-tin deposit. This surface enrichment with a thin layer of tin-oxide can account for the best tarnish resistance of the silver-tin alloy after recrystallization. It is known that tin oxide doesn’t form a tarnish layer under sulfide exposure.

3.3. High temperature adhesion of silver-tin deposit on nickel

There has been demand for silver and silver-based coatings for high temperature applications, typically at temperatures above 150 °C. At such high temperatures, a nickel barrier layer is used to prevent a rapid diffusion of copper into silver. In addition, oxygen can diffuse through the silver layer at high temperature, resulting in the oxidation of the nickel surface followed by adhesion failure between the nickel and silver. A typical example of pure silver adhesion failure on nickel after 1000 h annealing at 200 °C is shown in figure 6. The silver layer can be easily removed by hand or by tape test, leading to a dark nickel surface on the remaining substrate.

In contrast, the silver-tin deposit did not show any adhesion failure on nickel after 1000 h at 200 °C and the layer could not be removed by cross hatch tape test. Similar high temperature adhesion of a silver-tin deposit on nickel was recently obtained by Tarutani Yoshie from Mitsubishi Materials [5] by using a flash layer of copper or palladium between the nickel deposit and the silver-tin. In the present case, the surface tin atoms in silver-tin deposit could have absorbed oxygen leading to the surface oxidation and passivation of the deposit, thereby preventing further migration of oxygen to the interface between the silver-tin layer and the nickel. On the other hand, inter-diffusion between tin atoms and nickel may occur at the interface, leading to the formation of a small NiSn intermetallic which strengthens the adhesion between both layers.Very small Kirkendall voids were observed on some high magnification SEM images of the interface not shown in this report. These voids did not impact the adhesion during the actual investigation time.

3.4. Copper diffusion in silver-tin deposit

Copper diffusion into the silver-tin deposit depends strongly on the storage temperature. From room temperature to 100°C, there was no obvious diffusion observed after 500 h storage. At 125 °C, a small intermetallic inclusion appears at the interface after 500 h. Increasing the storage temperature will change the kinetics of copper diffusion. At 180 °C, the intermetallic phase appears at the interface after only 48 h; this grows with time and occupies more than 50% of the silver-tin layer after 500 h storage.

The data of figure 7 also shows that copper does not migrate directly through the grain boundary. Rather it reacts with tin and silver to form a possible ternary AgSnCu metallic compound. This Silver-tin-copper compound grows gradually with time and will come to the surface of AgSn₂₀ deposit. At that time, a brown or tarnish-type color is observed on the silver-tin deposit. The contact resistance of silver-tin will increase due to the presence of copper and tin oxide on the surface. Such material may not provide reliable electrical connection after pin insertion into the printed circuit board. For press-fit interconnections, such a high annealing temperature is not applicable.

3.5. Industrial plating using Silveron™ GT-820 Silver-Tin Process

The industrial test and validation of the operating window is a key step for successful industrialization of a process. The influence of factors such as the agitation and line speed, the metal concentration ratio in the bath, and the applied current densities were studied under real production conditions at ept GmbH. Samples plated under different conditions were submitted to quality control tests to check the adhesion on nickel base material, appearance, color uniformity, thickness distribution, and the alloy composition. The test matrix and results are summarized in table 4.

Under controlled-immersion or reel-to-reel selective plating conditions at the maximum line speed, Silveron™ GT-820 Silver-Tin can provide a uniform and bright deposit in the current density range from 2 to 10 A/dm². Decreasing the line speed to the minimal value will reduce the upper applicable current density to about 8 A/dm². For controlled-immersion plating, the agitation of the electrolyte is minimized so that the surface of the fluid remains flat enough for the adjustment of the pin immersion depth. The main contribution for the transport of metallic ions to the cathode surface is provided by the line speed; therefore, decreasing the speed will reduce the agitation and the applicable current density range.

The results in table 4 also demonstrate that the variation of the silver-to-tin concentration ratio in the bath has almost no impact on the quality and the alloy composition of silver-tin deposit. Comparable results were obtained in the laboratory. This is crucial for industrialization because small variation in the metal concentration in the bath will not affect the final material. In general, the thickness distribution on pin met the expectation and all samples passed the quality specifications for different types of pin designs.

Samples were used for qualification for different press-fit interconnections. Figure 8 displays examples of insertion and retention forces of a silver-tin pin during the press-in and -out in the printed circuit board (PCB) with organic solderability preservative (OSP) finish. The forces are very constant over the complete DOE matrix.  For the actual pin and the plated through-hole (PTH) design, the value of the insertion and the retention force is within the specification for these products.

Changing the metal concentration ratio, the line speed, or the applicable current density does not affect the insertion and the retention force. Comparable results were obtained after insertion in other PTH finish materials. At this stage of investigation, Silveron™ GT-820 Silver-Tin is stable and suitable for industrialization. It provides a consistent coating under a broad range of operating parameters.

Conclusion

This study shows that Silveron™ GT-820 Silver-Tin is a robust process for the electrodeposition of silver-tin alloys of composition equal to Ag₃Sn and Ag₄Sn. The process is shown to be very stable under laboratory and production conditions.  The alloy composition is controlled by the organic components of the electrolyte rather than the concentration ratio of metallic ions providing a consistent deposit across process parameters.

The silver-tin deposit is semi-crystalline directly after plating; it can be converted into fully crystalline material by short annealing after plating. If needed, the annealing can be done online by adjusting the temperature of the drying oven after plating. The recrystallization of silver-tin improved the ductility and tarnish resistance of the material. The deposit is thermally stable on nickel at high temperature such as 200 °C where the adhesion remains good after 1000 h. On copper and copper alloys, there is a temperature threshold at around 100 °C for copper diffusion. Such interesting properties could open the possibility of using silver-tin deposit for high temperature interconnection where adhesion failure is observed for pure silver coating on nickel.

The product was tested successfully under production conditions. It was confirmed that changing the metal concentration ratio in the bath or the operating parameters such as the line speed doesn’t affect the quality of the deposit. Silver-tin coatings displayed a good insertion and retention force, positioning this material as suitable candidate for tin-lead replacement in solderless interconnections. In addition, the low number of makeup components to be analyzed and the easy processing are among other advantages of the product. Nevertheless, full industrialization will be confirmed only after extended production to verify process performance.  

Reference

1. H. Leidheiser and A.R.P. Ghuman, J. Electrochem. Soc: Electrochemical Science and Technology, 120 (4) (1973) 484.
2. A. Hrussanova, I. Krastev, J. Appl. Electrochem. 39 (2009) 989
3. A. Hrussanova, I. Krastev, G. Beck, A. Zielonka, J. Appl. Electrochem.40 (2010) 2145.
4. W. Degner, J. Fiedler, R. Nutsch & H. Liebscher, DDR 265182A1 (1987).
5. T.Yoshie, K. Kenji  JP2021025086A (2021).