ANSYS DfR reversed_white

The Effect of Improper Conformal Coating on SnPb and Pb-free BGA Solder Joints during Thermal Cycling: Experiments and Modeling

Maxim Serebreni1*, Ross Wilcoxon2 , Dave Hillman2 , Nathan Blattau1 , Craig Hillman1

1DfR Solutions, 2Rockwell Collins, *Maxim Serebreni

 

Download PDF

 

Abstract

Application of chip scale packages (CSPs) and bottom terminated components (BTCs) in harsh use environments often requires the use of conformal coatings to meet reliability requirements. In certain coating application methods, the conformal coating materials can flow underneath the component and cause solder joint failure during thermal expansion and contraction of the electronic assembly. In this study, Ball Grid Array (BGA) components were coated with an acrylic conformal coating materials using two application methods and subjected to two thermal cycling profiles to assess the integrity of SnPb and Pb-free BGA components. To better understand the observed failure modes, Finite Element Analysis (FEA) was performed on the conformally coated BGA packages. Material characterization was performed using Dynamic Mechanical Analysis (DMA) and Thermal Mechanical Analysis (TMA) to capture the temperature dependent properties of the conformal coating to better correlate simulation and experimental results. Failure modes were found to greatly depend on the conformal coating material properties around the glass transition temperature (Tg) rather than temperature range. Significant difference in the failure mode was found between the Pb-free and SnPb BGA components with acrylic conformal coating materials and temperatures profiles.

Keywords: Conformal Coating, Thermal Cycling, Ball Grid Arrays, Solder fatigue, Finite Element Modeling

Introduction

Conformal coatings are used on Printed Circuit Boards (PCB) with the intent of providing protection from harsh environments containing moisture and contamination such as dust and metallic debris that could cause shorts in electronic components. In addition, some conformal coatings are designed to provide thermal insulation, shock vibration attenuation and electrical insulation for high voltage components at high altitudes [1]. Conformal coatings are also being used to help mitigate the risk of tin whiskers on pure tin surface finishes [2]. These attributes make conformal coatings especially attractive for high reliability application environments of avionic and aerospace electronics. Previous studies have shown that the application of conformal coating to surface mount resistors and CSPs can reduce the thermal strain in solder joints and thus extend the thermo-mechanical fatigue life of components [3-5].

Potential concerns regarding solder joint integrity arise when the conformal coating material is allowed to flow underneath the package. Recent investigations have show that letting conformal coating flow underneath plastic quad-flat no lead package (PQFN) dramatically reduces the thermomechanical fatigue (TMF) life of the device by changing the equivalent plastic strain of solder joints from predominantly shear to an axial loading mode [6]. In such a condition, the conformal coating expands and contracts in the vertical direction. As thermal cycling progresses, failure in solder joints could result from the lifting of the component that causes excessive tensile and compressive stresses. The amount of axial stress will greatly depend on the leverage of the conformal coating on the particular component and the variation of coefficient of thermal expansion (CTE) and elastic modulus (E) of the coating with temperature. In addition, conformal coating materials with a Tg that is within the thermal cycle range could drastically reduce fatigue life of solder interconnects. As temperature approaches the materials Tg, large expansion occurs along with reduction of material stiffness. In cases when the conformal coating material expansion occurs prior to adequate softening, large stresses would be applied to solder joints. This behavior is inherent in thermoset polymers since materials expansion tends to bedriven by the changes in the free volume while changes in the modulus tends to be driven by the increases in movement of the polymer chains. Due to the intrinsic properties of conformal coating and the large variety of coating materials used by the electronics industry, the effect conformal coatings on solder joint TMF life needs to be investigated.

This research aims to investigate the impact conformal coatings and their application method can have on Pb-free and SnPb area array components under thermal cycling condition. Experimental procedure follows that of previous study conducted by authors in which components with only SnPb solder were used [7]. Results from mechanical characterization of the acrylic conformal coating were used in finite element simulation to capture the effect of acrylic conformal coating and their application method on ball grid (BGA) components and provide further insight into the associated failure mode in each of the tested solder alloys under various thermal cycling ranges.

Experimental Approach

Simulations in the current study were based on experimental conditions performed in a previous investigation that included ball grid array (BGA) components, with tin-lead (63Sn37Pb) solder balls, that were subjected to thermal cycling. Figure 1 shows an assembled test vehicle from that study that included 60 components with half of them conformally coated. In the experimental study of reference [7], an Anatech event detector system continuously monitored the continuity of the daisy-chain BGA components to determine the number of thermal cycles at which solder joint failure occurred.

Material Properties

Acrylic conformal coating is widely used in a large variety of electronic assemblies. To understand the temperature dependent mechanical behavior of the material bulk samples of the acrylic conformal coating were prepared and characterized using TMA and DMA to measure the materials CTE and modulus as a function of temperature as shown in Figure 1 and are implemented in FE modeling. Characterization identified the Tg of the material to be around 15°. The glass transition temperature corresponds to the midpoint of a region in which the material’s modulus decreases and material expansion rapidly increases rather than a single data point in which an abrupt transition occurs.

imp fig 1

Test Vehicle

Test vehicles comprised of daisy chained BGA components with a package size of 17mm x 17mm, 1mm pitch, 256 IO. Each test board was populated with 60 individual components. Test boards selected with FR4 material laminate with 0.081 inch thickness and 8 dummy inner layers. BGA components assembled using 0.005 inch thick stencil. Two conformal coating configuration were applied to the two halves of each test vehicle. The left half of each board was coated using “standard” production spray process and the right half of the boards coated by a manual process using pneumatic syringe to completely fill the conformal coating material under each BGA referred to as “thick” application. This approach represents a worst-case scenario of conformal coating application that can occur in dipping method and heavy spray coating applications.

imp fig 2

Thermal Cycle Testing

Two thermal cycle conditions were used in testing, one with a temperature range of -55°C to +125°C and the second with –20°C to +80°C as shown in Figure 7. Both profiles have a minimum of 15 minute dwell time at each temperature extreme and a ramp rate of 5-10°C/minute according to IPC9701.

imp fig 3

Test Results

Reliability data of BGA component were analyzed using regression analysis to determine the Weibull shape factor (β) and characteristic life (η) for each configuration. The Weibull function correlates the cumulative failure distribution F(t) to the number of thermal cycles at which failure occurs shown in equation (1).

imp fig 4

The reported characteristic life corresponds to number of cycles at which 63.2% of the population is expected to have failed.

Failure Data Analysis

Failure data for each of the component configuration is analyzed for the two thermal cycles and displayed in Table1. Cycles to first failure are indicated for configuration in which only sufficient thermal cycles progressed to cause an “early failure” event; however, in slots indicated by “N/A”, the characteristic life occurs past the experimental time frame. Control samples with SAC305 alloy exhibit shorter fatigue life than the SnPb components under equivalent thermal profiles. As standard acrylic conformal coating is introduced, characteristic life of the two alloys under temperature profile with -20°C to +80°C falls within a narrower range.

imp table 1

The differences between standard and control samples are not consistent from the SAC305 to SnPb alloys but are shown to match under temperature profile 2 with a range of –20°C to +80°C. With an increase in the thermal cycling range, a larger discrepancy between the two solder alloy characteristic life is evident from the cycles to first failure. First failure with standard conformal coating surpassed the initial 1600 thermal cycles for the -55°C to 125°C profile for the SnPb alloy with a much lower first failure observed for the SAC305 alloy.

Figure 4 represents the cumulative failure distribution for the SAC305 components with thick and standard coating for the two temperature profiles. It is evident that the application method impacts fatigue life significantly more than the temperature range. A similar trend is evident for the SnPb components.

imp fig 5

Figure 5 displays the cumulative failure distribution for the SnPb components performed in the previous study along with the standard acrylic coating performed in the current study. Failure rates for the standard and thick coating under profile 1 show close resemblance to that of the SAC305 components with earlier failures in the thick coated components. Initial failures of the standard coated components are displayed for profile 2 and illustrate the drastic increase in fatigue life with standard coating between the two solder alloys and temperature profiles.

imp fig 6

Cross-sectional Failure Analysis

Metallographic cross section assessment performed on both Pb-free and SnPb BGA components post thermal cycling to reveal damage and crack location in solder jonits with and without conformal coating.

imp fig 7

Control components with Pb-free solder show cracking along the solder joint/board pad interface as shown in Figure 6; however; SnPb control components exhibit cracks along the upper solder joints/component pad interface. This initial difference in failure site between the control samples indicates that additional factors such as component warping, pitch size, ball height, local CTE mismatch, package configuration and intermetallic layer all affect fatigue life of components and are inherent factors in each solder alloy and component selection.

imp fig 8

Cross-sections of failed Pb-free solder joints are shown in Figure 7 for both the standard and thick coated components. Standard Pb-free components exhibit similar cracking location as those in the control components with corner joints cracking before joints at the center due to distance to neutral point effect. Since a similar characteristic life between the control and standard coated components was observed, the failure location was as expected to be comparable. Introduction of the thick conformal coating was found to have a different failure mode than the standard coating. Since the acrylic conformal coating represents a much larger area than the effective area of the solder joints between the PCB and substrate, the mechanical strains placed on the solder joints were imposed by the thermal expansion of the conformal coating. Figure 8b) shows severe plastic deformation in the Pb-free solder joints and no evident distance to neutral effect. Similar cracking at the board pad is observed for the -20°C to 80°C thermal cycle with less plastic deformation as in Figure 7. Large compressive strains caused solder joint to be extruded outwards. This failure mode is also attributed to the lack of adhesion and interaction between the conformal coating and solder alloy during the low and high temperature extremes. At the high temperature dwell, solder joints creep and conformal coating softens and eliminates any possible adhesion between solder and the acrylic material. At the cold temperature, extreme, the solder is compresses and squeezed outward in the normal direction to the applied load. The lack of hydrostatic stress on solder joints during low temperature dwell implies that no physical constraint is being placed on the solder and allows for the deformation to occur. This solder/conformal coating interaction is greatly dependent on the conformal coating CTE and modulus at the dwell temperature. It is important to note that the cross-section shown in Figure 7 is that after 1000 thermal cycles, and electrical failure was detected in a fraction of the number of cycles took to severely deform the Pb-free solder alloy to the one shown.

Failure analysis of the Pb-free solder joints exhibits similar failure location on the upper solder/component pad interface as the control components. Figure 8 displays SnPb BGA with standard acrylic coating post 1600 thermal cycles with slight cracking and grain coarsening under temperature profile 2.

imp fig 9

Cross-sections of the SnPb components reveals that the magnitude of damage closely correlate to the characteristic life variation found in the BGA solder joints under both thermal cycles. A different failure mechanism with the thick acrylic coating between the SnPb and the Pb-free solder joints was observed along with a phenomenon that explains the difference in cycles to failure primarily between the thick coated components. Figure 9 displays the SnPb solder joints with thick acrylic coating post thermal cycling for a) corner joint and b) joints at the center of the row for the -20°C to 80°C thermal cycles. Cracking along the diagonal of the SnPb solder is found along with grain coarsening at the crack area and similar extrusion as seen in the Pb-free components. Whoever; unlike the Pb-free components, electrical failures in the SnPb components were detected as cracking along the diagonal of the solder joints propagated prior to the severe compressive deformation took place. Unlike the Pb-free solder, distance to neutral point in the thick coated SnPb components remains dominant and is evident from the lack of damage in solder joints at the center of the row with progressively lower damage in solder joints closer to the neutral point of the component. This phenomenon can be attributed to the lower stiffness of the SnPb solder joints which allow for more components warpage and load sharing between neighboring solder joints.

imp fig 10

Variation in the characteristic life between the Pb-free and SnPb components is also strongly dependent on the local CTE mismatch between solder alloy and component. Due to a different microstructure and mechanical properties, the stress distribution in SnPb corner joints is subjected under lower axial strain but with larger shear strains than the Pb-free solder. In conjunction with the axial compression due to the thick acrylic coating, solder joints are placed under mixed-mode thermomechanical loading which progresses cracking along the diagonal of the solder joint.

In a few instances, no failure was recorded in SnPb components with thick acrylic coating. Figure 10 displays corner and middle row SnPb components with thick coating cycles from -20°C to 80°C for 1600 cycles with not an evident change in resistance. Both locations along the row show severely compressed solder joints with cracks running along the diagonal of the component. It is believed that in this instance the large compressive stresses caused crack closure to overate the rate of crack propagation and avoid an open to occur. This process is not fully explained and its exact contribution to the observed increase in cycles to failure in SnPb components is not yet known.

imp fig 11

Finite Element Modeling

BGA package is modeled using commercial finite element software ABAQUS 16.4. Global/local modeling approach is used to create a quarter symmetric finite element model of the BGA package and has proven to provide accurate results in modeling solder in electronic packaging [9]. Copper pads on both the PCB and substrate are modeled without solder mask. Plugs of fine mesh were created to model corner solder joints and plugs with coarse mesh used to model the rest of the solder joints as shown in Figure 11.

imp fig 12

To capture creep and plasticity deformation of solder joints, Schubert’s constitutive model based on a hyperbolic sine function was implemented for both Pb-free and SnPb solder shown in equation (1) [8]. Where 𝜀̇ 𝑐𝑟 is the steady state creep strain rate, 𝐴1 constant, 𝐻1 is the apparent activation energy, k is the Boltzmann’s constant, T is the absolute temperature, 𝜎 is the applied stress, 𝛼 and n prescribe the power law relationship between creep strain rate and applied stress.

imp fig 13

Modeling of the components with standard conformal coating was omitted from the analysis since failure rates of standard coated components matches those with the control components in both solder alloys and thermal cycles. Figure 12 shows the global view of the quarter symmetric BGA model.

imp fig 14

Material properties used in the analysis are presented in Table 2. These values are a culmination of both published and measured quantities and are assumed to be linear elastic except solder alloy

imp table 2

Local BGA plugs and a single quarter symmetric model were merged and tie constrains generated between plug surfaces and global model. To avoid incorrect results in the thick coated model, a 20 µm gap was placed between solder and conformal coating as shown in Figure 13 to avoid potential over constraining of the solder joint during thermal cycling.

imp fig 15

Pb-free BGA

Figure 14 shows the Von Mises stress contour plot for the corner and die shadow joints at the beginning of the hot dwell period. Stresses do not correspond to the directionality of the load but imply that the location of maximum stress is found at the solder/copper pad interface along with a slight

imp fig 16

distance to neutral point effect at the corner joint which is not presented in joints along the die shadow and is characteristic of predominantly axial loading. The location of maximum stress changes from the upper to lower interface during high and low dwell periods, respectively. To illustrate the effect of the thick conformal coating on the stress-strain state of the solder, a 25 µm layer of elements directly above the board copper pad is volume averaged. Figure 15 illustrates the average maximum principal strain for the Pb-free BGA for temperature profile 1 with and without conformal coating. largest average strain value in BGA solder joints occurs at the center and graduate decreases toward the solder/copper interface while average

imp fig 17

stress values are largest at the interface and decrease toward the center of the joint. Since solder displacement is driven by thermal expansion of the conformal coating, solder joints are placed under displacement control loading conditions. With the thick conformal coating, larger axial strains accumulate during cold temperature dwell. This values occurs at a point at which the acrylic material still maintains rigidity with high modulus. More damage is accumulating at the start of the glass transition temperature region rather than after the glass transition. At the onset of the transition region, the elastic modulus of the acrylic has not sufficiently decreased to allow even for a small increase in the CTE.

To illustrate the difference in compressive loading caused by conformal coating contraction during cold temperature dwell the average axial strains are compared. Figure 16 illustrated the axial strains with time for the Pb-free BGA with and without conformal coating for the same layer of elements. It can be seen that the first high temperature dwell, both the control and thick coated joints reach an equivalent state; however, thick coated joints reach a much larger compressive strain than the control. This cause a shift from positive to negative axial mean strain which correlates to the failure mode observed in cross-sectional analysis.

At the start of the subsequent thermal cycle, the strain state at the interface of the Pb-free solder joint is under compressive loading. The existence of a compressive preload has been previously shown to contribute to larger accumulation of plastic work per thermal cycle [10]. In this experimental condition, the compressive preload occurs at the cold temperature dwell and continues up to the glass temperature at which the acrylic material softens. A similar trend is observed at both temperature profiles with higher strain obtained for profile with higher temperature extremes.

imp fig 18

SnPb BGA

The stress-strain behavior of the SnPb solder joints at the interface is similar to that observed in Pb-free components only with noticale difference in magnitude. Figure 17 illustrates the maximum principal strains at the corner SnPb BGA with and without conformal coating at the high temperature dwell.

imp fig 19

The strain distribution confirms the dominance of distance to neutral point effect in SnPb components. Figure 18 illustrates the same SnPb components during the cold temperature dwell in which a noticeable difference is observed due to the thick conformal coating.

imp fig 20

Larger strains are concentrated along the diagonal of the joint with thick coating compared to a more uniform distribution of the control component. This simulation assists in correlating the failure model observed in cross-sectional analysis to the temperature dependent material behavior of the acrylic coating.

Conclusions

This study resulted in identifying the effect of conformal coating materials on solder joint fatigue life in Pb-free and SnPb BGA packages. Experimental testing of acrylic conformal coating materials with various temperature dependent CTE and E exhibited failure modes ranging from fatigue to overstress in Pb-free and SnPb solder joints. Finite element simulation proved to be correlate well with associated stress-strain state to the observed failure mechanism. An accurate characterization of the conformal coating temperature dependent properties has shown that the glass transition temperature of the conformal coatings is a critical factor affecting fatigue life. Thermal cycling profiles which crosses the glass transition temperature of the material proven to be more damaging than the temperature range with thick conformal coating application. SnPb BGA components have proven to be more robust to acrylic conformal coating under both temperature extremes and application method. This results is supported by the mechanical behavior inherent to SnPb solder and the package type used in this study. Additional experimentation is required to fully investigate the influence of conformal coating BGA components by altering the conformal coating materials and package type along with thermal cycling conditions.

Acknowledgments

The authors would like to thank Gil Sharon for his fruitful discussion of the modeling results as well as technical staff of DfR Solutions and Rockwell Collins for assisting with sample preparation, test setup and material characterization.

References

  1. Qi, Haiyu, Michael Osterman, and Michael Pecht. "Design of experiments for board-level solder joint reliability of PBGA package under various manufacturing and multiple environmental loading conditions." Electronics Packaging Manufacturing, IEEE Transactions on 32.1 (2009): 32-40.
  2. Woodrow, Tom and Ledbury, Eugene, "Evaluation of Conformal Coatings as a Tin Whisker Mitigation Strategy", IPC/JEDEC 8th International Conference on Lead-Free Electronic Components and Assemblies, San Jose, CA, April 18-20, 2005
  3. Kokko, Kati, Laura Frisk, and Pekka Heino. "Thermal cycling of flip chips on FR-4 and PI substrates with parylene C coating." Soldering & Surface Mount Technology 22.3 (2010): 42-48.
  4. Norman, Jacov, Shlomo Segman, Pei-haw Tsao, and Arkady Voloshin. "Effect of Conformal Coating on Low Temperature Induced Strain in Electronic Packages."Proceedings of the 1992 Joint ASME/JSME Conference on Electronic Packaging 2 (1992): 797-802.
  5. Tong, Ho-Ming, et al. "Effects of parylene coating on the thermal fatigue life of solder joints in ceramic packages." Components, Hybrids, and Manufacturing Technology, IEEE Transactions on 16.5 (1993): 571-576.
  6. Vianco, Paul, and Michael K. Neilsen. "Thermal Mechanical Fatigue of a 56 I/O Plastic Quad-Flat Nolead (PQFN) Package." SMTA International Conference Proceedings (2015): 85-94.
  7. Wilcoxon, Ross, Dave Hillman, Doug Pauls, and Dan White. "The Impact of Improper Conformal Coating Processes on BGA Solder Joint Integrity." SMTA International Conference Proceedings (2015): 870-882
  8. Schubert, A., et al. "Fatigue life models for SnAgCu and SnPb solder joints evaluated by experiments and simulation." Electronic Components and Technology Conference, 2003. Proceedings. 53rd. IEEE, 2003.
  9. Fan, Xuejun, Min Pei, and Pardeep K. Bhatti. "Effect of finite element modeling techniques on solder joint fatigue life prediction of flip-chip BGA packages." 56th Electronic Components and Technology Conference 2006. IEEE, 2006.
  10. Yu, Da, Hohyung Lee, and Seungbae Park. "Reliability assessment of preloaded solder joint under thermal cycling." Journal of Electronic Packaging 134.4 (2012): 041008.

CC fig 2.png

Resource:

CONFORMAL COATING: WHY, WHAT, WHEN, AND HOW

Conformal coating is applied to circuit cards to provide a dielectric layer on an electronic board. This layer functions as a membrane between the board and the environment. With this coating in place, the circuit card can withstand more moisture by increasing the surface resistance or surface insulation resistance (SIR). With a higher SIR board, the risk of problems such as cross talk, electrical leakage, intermittent signal losses, and shorting is reduced.