The effect of different post-electroplating surface modification treatments on tin whisker growth

What are tin whiskers?

Tin whiskers are growths of pure tin that form spontaneously from tin electroplated surfaces. Tin whiskers can grow up to a few millimetres in length (but typically only a few 100 μm) and may be straight, kinked, curved or in the form of odd-shaped eruptions. Tin whiskers can cause problems in electronics by growing long enough to bridge between surfaces causing the circuit to short out, a situation worsened by the continuing trend for miniaturisation in electronics and move to Pb-free. The growth of these whiskers can begin very soon after the electroplating of a sample, however the initiation of growth may take years to occur; it is this unpredictability of whiskers that causes concern to reliable operation. At the current time, there is no one widely accepted mechanism for why and how whiskers grow; though there are number of theories and some commonly agreed factors that affect the growth of tin whiskers.

<p>Small tin whisker</p>

What are the intended outcomes for this research?

  • Create an oxide coating that is thicker than those previously produced using electrochemical oxidation.
  • Significantly reduce whisker growth compared with an air-formed oxide.
  • To understand how oxide coatings affect whisker growth.

How are the intended outcomes going to be achieved and analysed?

Conversion coatings will be produced to form a significantly thicker oxide layer compared with electrochemical oxidation.

  • X-ray photoelectron spectroscopy (XPS) will be used to analyse both the thickness and chemistry of the oxide layers.
  • Scanning electron microscopy (SEM) and optical microscopy will be used to analyse the surface morphology of the oxide layers, specifically the conversion coatings to check the quality of the coating.
  • Focused ion beam (FIB) SEM will also be used to analyse the effect of the oxide layer has on the intermetallic compound (IMC) growth and also to measure the thickness of the conversion coatings and correlate the results with those obtained from XPS.

It is thought that by having a thicker oxide layer, whisker growth can further be reduced.
  • Optical microscopy will be used to count filament-type whiskers and large eruptions on all samples periodically over time. Twenty random frames will be counted on each sample and the resulting whisker density will be extrapolated up to 1 cm2.
  • On each frame, the longest apparent whisker length will be measured and the longest for each sample will be plotted over time to analyse how the whisker length increases with time.

Whisker growth and oxide analysis

Whisker growth was evaluated using optical microscopy; three randomly selected samples for the electrochemically oxidised and untreated groups of samples were analysed at each analysis point and the average whisker density calculated. The first graph shows that, for both sets of samples, the whisker density increases progressively with time, however, the presence of the electrochemically formed oxide significantly reduces whisker growth compared with that from the untreated samples that were left to develop a native air-formed oxide.

The second graph shows that the thickness of the oxide on the untreated samples steadily increases over time and becomes similar in thickness to that on the electrochemically oxidised samples after ~84 days of storage; this graph also shows that the thickness of the oxide on the electrochemically oxidised samples remains relatively consistent over time with very little change over a period of 204 days. The increase in oxide thickness for the untreated samples may contribute to the reduced rate of whisker initiation observed after ~84 days. However, the reduced rate of whisker growth may be due to other factors such as reduced rate of intermetallic compound (IMC) growth.

<p>Graph showing whisker growth with time</p>
<p>Graph showing how oxide thickness changes with time</p>

Whisker length analysis

The longest apparent length of whisker in each frame used to determine whisker density was measured at each time interval and is plotted in the top graph. For the untreated samples, the results indicate that the whisker length increases gradually over the 220 days of storage but that the rate at which the whiskers grow progressively reduces with time. Compared with the untreated samples, the length of the longest whisker is significantly reduced for the electrochemically oxidised samples throughout the period analysed, with little apparent increase in whisker length between ~80 and ~220 days. The bottom 2 graphs shows the change in the distribution of the longest whisker per frame for both electrochemically oxidised and untreated samples, between 14 days and 119 days of storage. 

<p>Graphs showing how whisker length changes with time</p>

After 14 days all the measured whiskers were less than 100 µm in length for both the untreated samples and the electrochemical oxidised samples. After 119 days the whisker distribution for the untreated samples has shifted to much greater lengths and each frame analysed contained a whisker greater than 20 µm in length. The whisker length distribution for the untreated samples shows that the majority of the whiskers were greater than 100 µm in length after 119 days, whereas after 14 days, no whiskers were greater than 100 µm in length. In comparison, for the electrochemically oxidised samples, most of the whiskers were still less than 40 µm in length, although the maximum whisker length has increased compared with after 14 days of storage. The bottom left graph shows that the whisker lengths on the electrochemically oxidised samples are greatly reduced compared with the untreated samples, thereby significantly reducing the risk of the whiskers bridging a gap between components or connections and causing a short circuit. The reduction in the number of long filament whiskers may be due to the increased incubation period for the electrochemically oxidised samples.

Alternatively, it may suggest that the electrochemical oxide influences the rate at the whiskers grow once the whisker has penetrated the oxide layer. It might be expected that when a whisker penetrates through the oxide film, the whisker would have a similar rate of growth to the untreated samples; therefore, the electrochemical oxide also appears to be having an influence on the driving force for whisker growth as well as inhibiting whisker initiation. It was shown previously that for Sn-Cu on Cu, IMC growth didn’t appear to be influenced by the presence of the electrochemical oxide [1]. In the absence of an oxide film, the surface of the tin serves as a source for vacancies to facilitate tin diffusion through the coating, to support the growth of whiskers [2]. However, the presence of a sufficiently thick oxide may impede diffusion of tin atoms by reducing the number of available surface vacancies [3], therefore slowing the rate at which whiskers may grow. This may suggest that whisker growth on the electrochemically oxidised samples is diffusion limited.

<p>Photographs showing how to improve tungstate conversion coatings</p>

Tungstate Conversion coatings

The set of the images to the left are photographs of tungstate conversion coatings on electroplated pure Sn. The photographs show that for increased cycles, the quality of the coating decreases. This was due to gas being liberated at the service, causing non-uniformity of coating thickness. This can be overcome by reducing the current density, as shown in the bottom row of photographs. These photographs show that lowering the current density while passing an equivalent charge will improve coating quality; additionally an increased amount of charge can be passed using a lower current density, therefore enabling the formation of thicker coatings. 

Surface analysis of Tungstates

It was found that the tungstate conversion coatings had much finer cracking compared with the molybdate conversion coatings. However, it was observed that for thicker tungstate conversion coatings there were areas that had either flaked off or were thinner. It was also observed that all of the whiskers were growing from these areas. The micrographs here show a) 10 mA cm-2 at 30 cycles, b) 5 mA cm-2 at 60 cycles, c) 10 mA cm-2 at 90 cycles, d) 10 mA cm-2 at 150 cycles and e) 10 mA cm-2 at 90 cycles. They show that for fewer cycles and lower current density there is no cracking, however for more cycles and increased current density, cracking becomes apparent.

<p>Electron micrographs showing the surface of different tungstate coatings</p>

Molybdate Conversion Coatings

The images to right are optical micrographs of four different molybdate coatings after ~2 months of room temperature storage; the top left was produced by simple immersion for 5 min, the top right was produced at -0.45 V vs. Ag/AgCl for 5 min, the bottom left was produced at -0.6 V vs. Ag/AgCl for 5 min, and the bottom right was produced at -0.75 V vs. Ag/AgCl. It was observed no cracks were present in either of the top two samples, whereas the bottom two samples had a high density of cracks; with the samples produced at -0.75 V had the largest amount of cracking. This cracking may act as weak points for whiskers to grow through.


<i></i>Optical micrographs of molybdate
conversion coatings produced using varying parameters</p><p>

<p>Electron micrograph of several whiskers breaking through the cracks of a molybdate conversion coating</p>

Surface analysis of Molybdates

A high density of cracking was observed for molybdate conversion coatings produced at increased cathodic potentials. It was also observed that all the whiskers observed grew from the cracks, as shown in the micrograph. The cracks act as weak points and allow the whiskers to grow, relatively unimpeded compared to trying to grow through the bulk of the coating. However, due to the high density of cracks, there was a high density of whiskers compared with the lower cathodic potenital of -0.4 V vs. Ag/AgCl even though the coating was thicker. 

Conversion Coatings Whisker studies

The first and second graphs show the whisker growth studies for the molybdate conversion coatings and tungstate conversion coatings, respectively. It can be seen from both graphs that both of the conversion coatings will significantly reduce whisker growth compared with a native air-formed oxide. However, the graphs show that the tungstate conversion coating produced at 5 mA cm^-2 for 90 cycles will reduce whisker growth the most. The least effective tungstate coating had a whisker reduction ratio of ~36:1, whereas the least effective molybdate conversion coating had a whisker reuction ration of ~8:1. It should be noted, however, that this is still comparable to the reduction ratio of an electrochemically formed oxide. The reason for the higher whisker density for the -0.6 V molybdate conversion coating, was the fact that it contained a high density of cracks where whiskers were observed to grow from.

<p>Graphs showing whisker growth for conversion coatings with time</p>


The following conclusions can be drawn from the results obtained in the current study:

  • An electrochemical oxidation treatment has been demonstrated to provide continuous whisker mitigation over a period of 220 days for 2 µm Sn-Cu deposits on Cu.
  • An electrochemical oxidation treatment not only reduces whisker density but also significantly reduces whisker length compared with an untreated surface.
  • Conversion coating treatments may be used to develop oxides that are significantly thicker than those that have been achieved to date by an electrochemical oxidation method.
  • Initial observations show that both molybdate and tungstate conversion coatings provide improved whisker mitigation compared with electrochemically oxidised surfaces.

  • For the tungstate conversion coated samples, most whiskers were observed to grow from areas of the coating where the oxide was thinner or completely absent (‘weak points’). Further optimisation of processing parameters (current density, number of cycles) is required to avoid the formation of ‘weak points’ in the oxide coating
  • Compared with the tungstate conversion coatings, an increased density of long, wide, deep cracks are formed on molybdate conversion coated samples produced at -0.6 V and -0.75 V vs. Ag/AgCl, which reduces their ability to mitigate whisker growth.
  • For both molybdate and tungstate conversion coatings, a thicker coating doesn’t necessarily provide improved whisker mitigation since the presence of defects within the coating will strongly influence its ability to mitigate whisker growth.
  • Further studies on the distribution and development of compressive stresses within the tin coatings are suggested as a valuable step towards understanding how post-electroplating surface treatments may affect the development of internal stresses and whisker growth.

Further information

For further information and/or ideas for further work please contact Dan Haspel at PEMC.

This work was part of a paper submitted to the Transactions of the IMF journal which can be found at the following DOI: 10.1080/00202967.2019.1587263