Search the Web
 
 

Home > Technical Library > Technical Features

Tin Whisker Application Specific Risk Assessment Algorithm
David Pinsky

Background
The transition to lead-free electronics has reached the stage where changes are occurring in the standard materials and finishes offered by component manufacturers and metal finishers. Most major suppliers of components are in the midst of transitioning away from tin-lead. For this reason, the occurrence of pure tin plating as a standard finish on components used in electronics is on the increase, and will continue to increase for the foreseeable future.

The formation of “tin whiskers” on the surface of tin coatings has been observed for many decades. These whiskers are comprised of nearly pure tin, and are therefore electrically conductive. Pure tin-plating on a wide variety of component types has previously grown tin whiskers. Tin whiskers have been found to form under a range of environments including: space, missile, airborne, shipboard, ground, medical implant, and office.

This has caused, and continues to cause, reliability problems for electronic systems that employ components that are plated with tin. Field failures attributable to tin whiskers have cost individual programs many millions of dollars each, and resulted in significant customer dissatisfaction.

The addition of lead to tin has been shown to suppress the growth of whiskers. Therefore, the use of tin-lead alloy coatings in lieu of pure tin coatings has been a standard procedure. Unfortunately, the push to reduce and eliminate the use of lead has driven many metal finishers to switch from tin-lead to pure tin, resurrecting the specter of tin whiskering.

As a result of this situation, manufacturers of high-reliability systems are frequently forced to decide whether the use of tin poses an acceptable risk in a given application. Issues of cost, schedule and performance are then weighed against these risks. Engineers are therefore asked to assess the severity of the tin whisker risk, and if this risk is deemed unacceptable, consider the relative merits of various mitigation strategies.

Performing Risk Assessments – Current Practice and Suggested Improvement
The current practice in most organizations takes one of two forms. In one approach, an engineer who is not familiar with all of the latest information on tin whiskering is asked to perform the assessment. Unfortunately, the factors that contribute to tin whisker growth are varied and significant confusion and mis-understanding pervades the industry. This creates a situation where mis-leading risk assessments often result. The other approach is to funnel all tin whisker risk assessments to a designated senior “tin whisker guru”. As the volume of pure tin applications continues to explode, the size of this effort is likely to overwhelm a single senior contributor. This commitment of senior engineering talent to what is becoming routine work cheats the organization of a scarce, valuable resource.

In response to this situation, I determined that there was a need for a user-friendly tool that would leverage the knowledge and experience of the “guru” so that a larger pool of less senior personnel could perform quick and meaningful risk assessments. The existence of a common tool would also provide other benefits, such as standardization of information format among collaborating organizations considering tin use. Also, results obtained in different organizations would be directly comparable, without influence by feelings of the particular “gurus” involved on each end.

Algorithm Development
The intent of the algorithm is to assess the risk that for a given application of tin plating, that tin whiskers will bridge between conductors. The term “overall mechanical risk” is used to describe this risk of whisker bridging. This algorithm does not address the consequences of the formation of such a bridge (electrical risk). Experience indicates that for many applications to be assessed that the risk of a whisker bridging is so negligible that further assessment of the consequences are unnecessary. Experience has also shown that in a sizable fraction of the assessments where the mechanical risk is high, the consequences of a bridge are so evident that no further risk assessment need be performed. For these reasons it is anticipated that the vast majority of risk assessments need only consist of the evaluation of the mechanical risk.

The approach taken in formulation of the algorithm is that the mechanical risk is a product of the probability that whiskers will form, and the probability of these whiskers bridging between conductors. The factors that affect whisker growth relate to the properties of the plating and substrate onto which it is plated. The factors that affect the bridging risk relate to geometry of the assembly and the presence or absence of insulating coatings on the conductors.

Note: This algorithm is based upon the premise that failure only occurs if a whisker bridges the entire gap between conductors. This premise applies to most applications, but not to high voltage applications where arcing across gaps is a common failure mode. Therefore, this algorithm may produce misleading results if applied to assess applications where high voltages are present.

The output of the algorithm is a numerical index of relative risk of whisker bridging, and as the levels of risk are anticipated to range over several orders of magnitude, the numerical index will be reported on a log-10 scale. Scaling factors have been selected so that the range of the numerical factor falls between zero and ten. Higher output numbers indicated higher degrees of risk. Since our predictive abilities are quite crude at this juncture, no definitive probabilities can be associated with the numerical output of this algorithm.

Overall Mechanical risk = Rtotal
Total geometric risk factor = Rgeom
Overall whisker growth risk factor = Rgrowth
Scaling constant = K

                                 Equation 1            Rtotal = K + log10 (Rgeom • Rgrowth)

A simplification that I will make to formulate the risk that whiskers will form is to assume that there are four, independent driving mechanisms of concern:

  1. Stress induced during initial tin deposition
  2. Stress developed in the tin as a result of inter-diffusion with the material below during time/temperature exposure
  3. Stress developed over time due to differential CTE between the tin and the controlling substrate, and
  4. Stress induced as a result of externally applied forces.

Initial stress risk factor = Ri
Diffusion stress risk factor = Rd
CTE stress risk factor = Rcte
External risk factor = Rex

                                 Equation 2            Rgrowth = (Ri + Rd + Rcte + Rex)

Whiskers can form bridges in one of two ways. The first is by growing from one conductor and reaching across to a conductor adjacent to the tin-plated conductor, which I call “direct bridging”. The second is for a whisker to from and then break off from its growth site, and then later form a bridge between two other conductors elsewhere, which I call “secondary bridging”.

Geometric risk factor for bridging from site of whisker growth = Rgd
Geometric risk factor for dislodged whiskers = Rgs

                                 Equation 3            Rgeom = Rgd + Rgs

Combining equations 1-3

                                 Equation 4            Rtotal = K + log10 [(Rgd + Rgs)•(Ri + Rd + Rcte + Rex)]

Each of the six Rx values in equation 4 will be calculated based upon attributes of the application.

Rgp = f {r1, r8}
Rgs = g {r10, r11, r12}
Ri = h {r2, r3, r4, r7}
Rd = l {r2, r5, r7}
Rcte = m {r2, r6}
Rex = n {r2, r9}

Functions f, g, h, l, m,and n, are functions. For now these functions are simple products. These functions could be redefined later if data indicates a different type of relationship applies.

With:

r1 = f1(conductor spacing)
r2 = f2(Pb content in plating)
r3 = f3(Sn deposition process)
r4 = f4(Sn deposit thickness)
r5 = f5(composition of material directly beneath Sn deposit)
r6 = f6(substrate controlling the CTE imposed on Sn deposit)
r7 = f7(reflow of Sn deposit)
r8 = f8(type of conformal coating applied directly over Sn deposit)
r9 = f9(use of mechanical hardware that applies stress to the surface of the Sn deposit)
r10 = f10(vulnerability of the assembly to contamination related failure, as indicated by imposed environmental controls during assembly)
r11 = f11(use of conformal coating on conductors throughout assembly)
r12 = f12(airflow within assembly)

Where the functions fx are as defined by the table below.

The Scale factor has been set to K = 8.9, based upon the maximum and minimum values produced by the functions defined below, to set the range of the numerical output to range from zero to ten.



Notes for selection of rx factors:

Conductor Spacing – r1
Minimum spacing between tin-coated surface and nearest conductor that could be at a different electrical potential as measured in units of mils (0.001 inches). Nearby conductors that are covered by insulation are not considered as a possible short destination for a whisker. If both conductors are tin-coated, multiply the separation by 0.6 for entry into this factor.

Pb content (wt%) – r2
This is the percentage by weight of lead (Pb) that is present as an alloying element with the tin. Other elements are not considered for this risk factor.

Process - r3
This is a description of the process by which the tin was deposited. Electro-deposits are typically described as either “bright” or “matte”, which relates to the resultant appearance. Immersion tin is deposited by an electroless plating process. Hot dip involves submerging of the part into a bath of molten tin. If the deposition process is unknown, assume “bright”, as this is the worst-case for whiskering propensity.

Tin thickness - r4
This is based upon the thickness of the deposit in microinches (0.000001”). If a range of thickness is present or may be present, choose the highest possible risk factor. For example, if a plating is known to range between 100 and 300 microinches in thickness, r2 should be set to 1.0, rather than to 0.7.

Material directly beneath the tin- r5
This material is often underplating that is different from the basis material, although tin is also deposited directly onto some basis materials. If the material is a copper alloy termed “brass” or “bronze”, or contains less than 95% copper by wt, use the “Brass of bronze” factor. If the material is a low copper alloy not termed “brass” or “bronze” use the “copper” factor. Low expansion Fe-Ni or Fe-Ni-Co alloys such as alloy 42 or Kovar should be given the risk factor of “ferrous”. The “nickel” factor should be use with a nickel underplate or with any low alloy nickel.

Substrate controlling the CTE – r6
This may be the basis metal of the component in question, but often is not. Some judgement will be necessary to determine which material in a complex stack-up will dominate the CTE that is imposed onto the tin deposit. The term “low expansion alloy” is used to describe metals such as Alloy 42 or Kovar that have been formulated to exhibit a low CTE that is compatible with ceramic and glass. All other alloys where the majority constituent is Fe should be classed under “ferrous”.

Plating reheated – r7
This factor relates to thermal treatments the tin was subjected to after deposition. If the deposit was fully melted and re-solidified, use the “fused” factor. (Note: solder re-flow operations will not necessarily fuse a pure tin deposit. Use the “fused” rating only if full melting of the plating is known to have occurred.) Some manufacturers utilize a special annealing process as a means to mitigate tin whisker risk. If the deposit is known to have been subjected to a treatment whose express purpose is such mitigation, use the “annealed” rating. Normal solder reflow processing should not be considered as “annealing” unless there is specific data to support such a classification.

Conformal Coat – r8
This refers to organic coating applied directly over the tin deposit. If a coating is known to be urethane in excess of 1 mil thick, or silicone in excess of 1 mil thick, use the appropriate ratings. If Parylene is used, apply the rating (no minimum thickness). If a different coating type is used, or if a urethane or silicone coating of less than 1 mil is used, apply the “other” rating. If no coating at all is use, apply the “none” rating (not the “other” rating).

Use of Mechanical HWD – r9
This factor is used to rate the amount of mechanical force that is applied to the surface of the deposit. If any mechanical component is in contact with the tin surface such that compression of the tin could occur, use the “fasteners” rating. Components soldered onto the surface do not count for this risk factor. If no such components bear on the tin surface, use the “none” rating. (Note: the factor for “none” is not zero because some mechanical damage is assumed to always be present on the surface due to normal handling, etc.)

Where was assembly performed? – r10
This factor is used to assess the overall vulnerability of the system to dysfunction as a result of the presence of small pieces of conductive contaminants. The concept behind this risk factor is that a device that is assembled in a clean room is likely to be very susceptible to contamination-related failure (or the expense of clean room operations would not be justified). Conversely, an assembly that is made under field conditions is likely to be fairly immune to conductive contamination (or it would never function). Another way to view is to consider how the addition of a few loose whiskers affects the total amount of conductive contamination present. For a clean-room build assembly, this would be a significant increase, while the same number of whiskers could represent a negligible increase for a system that is assembled in the field.
If the assembly that contains the tin-coated part is assemble in a clean room of any rating, use the “clean-room” factor. If the assembly occurs in a special “cleaner” area that has no specific rating (like a closed room with laminar-flow benches) use the “special clean area’ rating. If the assembly occurs in a normal factory environment (indoor, temperature-humidity controlled, workers in street clothes) use the “typical factory” rating. If the assembly is performed in an uncontrolled location (outdoor, open hangar, garage, etc.) use the “field assembly” rating.

Use of CC on conductors in enclosure – r11
The purpose of this factor is to help determine the risk of failure due to a loose whisker causing a “secondary” short. To determine the proper setting one must consider which electronics whiskers could possibly reach. In general, all electronics with a path through the air from the tin plated surface should be considered. For example, if the tin surface is within a sealed box, only those conductors within the box would be at risk for secondary shorting. For the purposes of this factor, conformal coat of all types and thickness are equivalent. If all exposed conductors are coated, apply the “all” value. If no coatings are used, apply the “none” value. Often, some but not all, conductive surfaces will be coated. In this case select either the “most” or “some” applies. Use the “most” value if all but a few conductors are known to be coated, otherwise apply the “some” value.

Air flow within assembly – r12
The purpose of this factor is to rate the risk that whiskers will break off and migrate to regions of the assembly at a distance from their site of growth. If air is forced over the tin-coated component by use of fans, etc. then use the “forced air” setting. If the assembly is used in a dynamic environment such as flight or ground vehicles, or large stationary machines with many moving parts, use the “dynamic environment” setting (unless the higher “forced air” setting applies). If cooling is achieved by passive means only and the application is in a fairly static environment, select the “none” setting.



Examples of Numerical Values for various combinations of factors
Example 1 – Copper wire hot dipped into tin-lead solder, attached using a screwed-down lug, with the nearest conductor ¼” away, no conformal coat on any conductors, assembly under normal factory conditions. No forced air cooling.
Example 2 - Copper wire plated with bright tin 250 micro-inches, attached using a screwed-down lug, with the nearest conductor ¼” away, no conformal coat on any conductors, assembly under normal factory conditions. No forced air cooling.
Example 3 – Ceramic chip capacitor plated with nickel and 250 microinches of matte tin with a subsequent annealing process, soldered to a CCA with 100 mils spacing between devices, no conformal coat, normal factory assembly. Forced air cooling, static application.
Example 4 – Fine pitch SMT device 10 mil pitch, with copper leads matte tin plated 100 microinches thick, no conformal coat, normal factory assembly. No forced air cooling, static application.
Example 5 – Kovar cover for a MCM that is plated with copper and then bright tin 200 microinches thick, spacing between lid and internal devices is 150 mils, no coatings used, assemble in a clean room. No forced air cooling, static application.
Example 6 – Brass terminal lug that is plated with 300 microinches of bright tin. Part is screwed down. Nearest non-coated conductor is 200 mils away. Entire assembly including lug is coated with 3 mils of urethane. Normal factory assembly. No forced air cooling, static application.
Example 7 – Circuit card with copper that is coated with immersion tin 30 microinches thick. Tin is fused after application. Design rules specify minimum conductor spacing to be 15 mils. No conformal coat applied. Assembly is performed in a normal factory. No forced air cooling.
Example 8 – Steel component is plated with copper and 150 microinches of matte tin without subsequent fusing or annealing. Part is screwed down internal to a sensor device, which is assembled under a laminar flow hood outside a clean room. Nearest conductor is 100 mils distant. No conformal coating is used. No forced air cooling. Avionic application.
Example 9 – Leads for hermetic devices are made from alloy 42 plated with copper and bright tin 250 microinches thick. Leads are plated on the outside of the device only. Part is assembled onto a CCA in a typical factory with Parylene conformal coat. The nearest conductor are the leads themselves with a pitch of 100 mils. Forced air cooling.
Example 10 - Leads for hermetic devices are made from alloy 42 plated with copper and bright tin 250 microinches thick. Leads are plated internally and externally. Device is assembled in a clean room without internal coatings. Part is assembled onto a CCA in a typical factory with Parylene conformal coat. The nearest conductor are the leads themselves with a pitch of 100 mils. Forced air cooling.



Examples of failures known to have resulted from tin whisker formation.
Example 1 - Bright tin plated 200 microinches thick over brass internal to a sealed device. Minimum conductor spacing to tin was 50 mils. No conformal coating or forced air cooling. Static application. Device was assembled at a laminar flow bench, not in a clean room.
 
Example 2 – Bright tin plated 200 microinches thick on a brass heatsink, which is fastened down with screws at NHA. No conformal coat on this part, but most of the rest of the assembly was coated. Nearest uncoated conductor was 100 mils distant. Assembled in a normal factory. Assembly was cooled using forced air. Static application
 
Example 3 - Bright tin plated 200 microinches thick on a copper housing wall. No fasteners, conformal coat, or air cooling. Airborne application. Nearest uncoated conductor was 100 mils distant. Assembled at a laminar flow bench, not in a clean room.
 
Example 4 – IC packaged in a DIP with copper leads plated with matte tin of unknown thickness. Lead spacing was between 30 and 40 mils. No conformal coat was used, and the assembly took place in a typical factory. Air cooling unknown. Notes: Minimum spacing is multiplied by 0.6 due to tin on adjacent surfaces. Thickness will be assumed to be 200 microinches as this is fairly common. Static application. Air cooling will not be assumed.
 
Example 5 – Power distribution device with internal vias consisting of 200 microinches of bright tin-plated onto copper. Bright tin plated copper pins pass through the vias, with an air gap of around 50 mils between. Normal factory assembly, no focred air through vias. Static application. No secondary reflow or annealing.


PAYMENT OPTIONS
Credit Cards, Purchase Order, Checks
Copyright © 2002 Raytheon Company - All rights reserved.  Legal Notice
 
Login to ReAct Online