- Email: [email protected]
Effects of salt spray test on lead-free solder alloy
MR-12057; No of Pages 6 Microelectronics Reliability xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/mr
Effects of salt spray test on lead-free solder alloy A. Guédon-Gracia a,⁎, H. Frémont a, B. Plano a, J.-Y. Delétage a, K. Weide-Zaage b a b
Laboratoire IMS, Université de Bordeaux, Talence, France RESRI Group, Institute of Microelectronic Systems (IMS-AS), Leibniz Universität Hannover, Hannover, Germany
a r t i c l e
i n f o
Article history: Received 21 June 2016 Accepted 6 July 2016 Available online xxxx Keywords: Solder joints Electronic assembly Corrosion
a b s t r a c t This paper starts with a bibliographic survey about solder corrosion and experimental results of the corrosion on lead-free solder balls during salt spray tests. Focus is made on the SnAgCu solder alloy. Ball Grid Array assemblies and “Package on Package” components were put up to 96 h in a salt spray chamber at 35 °C with 5% sodium chloride (NaCl) aqua according to the ASTM B117-09 standard. The weight is measured during the test. The solder alloys are observed and analysed along the ageing with optical microscope and scanning electron microscope equipped with an energy-dispersive x-ray system. The solder alloy deterioration is visible after 48 h. The microstructure is analysed in order to determine the corroded residues found on the surface solder balls after the salt spray test. Tin oxychloride (Sn(OH)Cl) is found on BGA solder joints after reflow and on PoP solder balls before reflow. The size of the solder balls has an influence on the corrosion state. Finally a method is developed in order to measure the corrosion product growth on the same sample during the salt environment exposure. © 2016 Published by Elsevier Ltd.
1. Introduction Salt mist environment, commonly found on snowy or oceanfront roads, induce galvanic corrosion in cars, boats, naval surveillance systems or tour helicopters, to name a few. Integrated circuit corrosion is a major cause of electronic system failures - it is responsible for about 20% of them. The main cause is surrounding humidity that speeds up electrolytic and galvanic corrosion, especially in biased circuits. Corrosion is even more accelerated by the presence of chemical pollutants in the humid air, like sodium chloride. Although corrosion of dense materials and large industrial equipment has been widely described in the literature, few data are available on thin-film materials and microelectronic assemblies. After a bibliographic survey, this paper presents an experimental study of lead-free assemblies aged in a saline environment. In the first part of our study, printed circuits boards (PCB) populated with BGA were aged in a salt spray oven. Copper traces as well as SnAgCu (SAC) solder balls were analysed during the ageing. In a second part of the study, focus is on the solder alloy: PoP (package on package) components were aged in saline conditions and the evolution of corrosion was observed and analysed. 2. Bibliographic survey Few papers relate the behaviour of SnAgCu solder alloy in presence of NaCl. There are two kinds of tests to assess the corrosion of material: ⁎ Corresponding author. E-mail address: [email protected] (A. Guédon-Gracia).
the polarization measurements, which provide information on a given metal, and the accelerated ageing tests to estimate the corrosion of systems. 2.1. Polarization measurements The electrode potentiostatic polarization enables the determination of the behaviour of metal in salt water. The principle lies in the measurement of the electrical potential as a function of the current in order to determine the anodic and cathodic areas of this material. Most papers present results of electrochemical corrosion of bulk material in a NaCl solution measured through polarization curves, as in [1,2] or [3] for instance. In [1], the corrosion behaviour of SAC330 alloy in a NaCl solution is compared with that of the SAC305 by potentiodynamic polarization and impedance spectroscopy measurements. The presence of tin oxychlorides or oxyhydroxychlorides was detected at the surface of both alloys investigated after the electrochemical tests. In [2], PCBs with SAC307 were investigated in NaCl solutions and dendrite growth has been observed and correlated with the NaCl concentration. In [3], SAC305 material is studied. In this paper the time to fracture and the critical stress for fracture under mechanical stress in a saline solution are investigated. The authors highlight that the reliability of lead-free solder alloys should be evaluated considering the effects of the combination of environment and applied mechanical stresses. 2.2. Ageing tests in a marine environment The ageing tests in a humidity chamber and in a salt spray chamber enable to observe the corrosion phenomena on electronic assemblies.
http://dx.doi.org/10.1016/j.microrel.2016.07.034 0026-2714/© 2016 Published by Elsevier Ltd.
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
2
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
Corrosion is accelerated by the presence of chlorine. Some papers relate the behaviour of SAC solder joints during salt spray ageing. In [4], the effect of pre-ageing in a saline environment on the longterm reliability of wafer-level-chip-scale-package (WLCSP) was investigated. WLCSPs assembled with SAC 305 were thermally cycled to failure, with and without 96 h pre-treatment in a 5%NaCl salt spray environment. Pre-ageing resulted in strong (43%) life-time reduction. Moreover, the authors correlated the corrosion path and Sn lattice basal plane by orientation imaging microscopy. The surface morphology and microstructural changes in Sn-4.0%Ag0.5%Cu lead-free solder balls subjected to salt spray (5%NaCl solution) tests were investigated by Song and Lee [5]. They investigated the effects of corrosion on the mechanical strength of solder balls. Compared to the Sn-37%Pb solder balls, Sn-4.0%Ag-0.5%Cu (SAC405) lead-free solder balls are more easily corroded in the salt spray test (see Fig. 1). Two kinds of corroded residues are found on the surface of solder balls after the salt spray test. The presence of Ag3Sn in SAC solders accelerates the corrosion of tin during the salt spray test because of the galvanic corrosion mechanism. The galvanic corrosion forms a brittle Ag3Sn structure at the corroded regions. As for SAC a passivation layer (made of oxides and hydroxides most of the time) appears on the alloy as a protection against chloride ions attacks (see Fig. 2). The alloy is thus only a little corroded on the surface and remains unaltered in depth. The copper percentage increases the corrosion resistance. In [7] the creep corrosion of SnAg (Sn-3.0Ag) and SnCu (Sn-0.5Cu) solders under NaCl solution is investigated at room temperature. The propagation of surface cracks was studied with the result that creep corrosion cracking should be incorporated in the reliability evaluation of lead-free solder alloys. In a harsh environment, with high chloride content, SnCu alloy shows a corrosion similar to that of SnAgCu. It is corrosion-resistant thanks to the formation of a thin passivation layer that protects the alloy against corrosion sub-products. In SnAg solder, the presence of a
Fig. 1. Surface morphology changes of solder balls (diameter 760 μm) after the salt spray test [5].
Fig. 2. Cross-section of Sn-Ag-Cu solders after potentiodynamic polarization test [6].
noble corrosion-resistant metal, namely silver, provides good corrosion resistance. 3. Experimental study on assemblies The test vehicle was a 10.2 × 14 cm FR4 board with 6 daisy chained components in BGAs. Solder balls (256 per BGA: 16 × 16) and solder alloy are in SAC305 (see Fig. 3). The board was cut into 6 parts containing a measurable daisy-chain structure and some copper pads without connections on the other board side. Each part was submitted to different ageing durations. Ageing was performed in a salt spray chamber at 35 °C (Salt Spray chamber Ascott CC450iP). The saline fog was generated from a 5% sodium chloride (NaCl) solution, as defined by ASTM-B117 [8], one of the most widely used tests for evaluating corrosion resistance of materials and assemblies to an aqueous salt atmosphere. The experimental procedure was detailed in [9]. Corrosion was mainly observed on the metallic compounds, that is to say the copper traces and the solder alloy. On the free copper pads,
Fig. 3. BGA assembly (cut along the red lines). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
3
the attack is fast: after 24 h, it is clearly visible by optical microscope observation; after 48 h, some pads and tracks are totally eroded (see Fig. 4). The microstructural changes of Sn-Ag0.5Cu solder joints exposed to salt spray test were investigated, and partly presented in [9]. The following conclusions have been drawn after 96 h of ageing at 35 °C with 5% sodium chloride (NaCl) aqua: - no electrical failures are detected in BGA assemblies, - the change of assembly weight is not significant, - the corrosion of the free copper pads lead to the disappearance of some pads (see Fig. 4), - the corrosion products observed in solder joints are consist of tin chloride and tin oxychloride (see Fig. 5).
In addition to the corrosion evolution during ageing, this study also aims at evaluating the storage of corroded samples in air afterward. Actually the samples can be stored between the test end and the sample observations. That is why the influence of the storage after the test must be assessed. So, after the salt spray test, two techniques are used to clean the samples [9] and the sample corrosion evolution is observed during one week in air depending on the way of cleaning. As a result, the behaviour is similar in both cases whatever the ageing test duration. So, on the one hand, the corrosion products do not evolve during the one week which follows the corrosion test. On the other hand, the cleaning technique does not influence the sample corrosion evolution. These first investigations on solder joints show the importance of understanding the corrosion phenomena of the lead-free solder alloy in salt and humid environment. In order to analyse the solder alloy corrosion before the reflow process, ageing tests on PoP components have been conducted before their assembly.
Fig. 5. Corroded area after 48 h: presence of tin chloride.
Moreover each component has 152 balls for the top and 353 balls for the bottom. That allows observing the corrosion on a large number of samples using a few components. The test specimens are placed in the cyclic corrosion test chamber at 35 °C and exposed to a continuous indirect spray of a fog generated from a 5% sodium chloride (NaCl) solution. The samples are removed at 24, 48, 72 and 96 h. Previous to the ageing, all specimens are dried in an oven at 125 °C during 24 h. Just after removing them from the chamber, the samples are washed in deionized water and dried with a stream of clean, compressed air. 4.2. Results
4. Corrosion of the lead-free solder alloy 4.1. Experimental procedure PoP components (see Fig. 6) are similar to BGA components. Package on package is a configuration where two BGAs are stacked on top of each other (Fig. 6.a). Solder joints are between the top package (Fig. 6.b) and the bottom package and between the bottom package and the PCB. The top package is generally thicker, containing multiple or stacked dies, while the bottom package is thinner, with either smaller or thinner die. As the corrosion of the solder balls must be observed in this experiment, the top and bottom packages are studied separately. The material of the solder balls is Sn3Ag0·5Cu solder alloy and the ball size are 420 μm for the top and 287 μm for the bottom.
Fig. 4. Temporal evolution of the copper pad corrosion. (a) t = 0 h (b) t = 24 h (c) t = 48 h.
During the salt spray test, the relative weight gain is less than 1%. The sample weight is measured using a precision weighing balance, OHAUS Discovery with a precision of 0.01 mg. As the weight values are precise, the relative weight gain is effectively due to the salt spray test but it is very low. No corrosion is observed on balls before the test (see Fig. 7). After 48 h of salt spray exposure, some corroded pits are visible on the surface of the solder balls (Fig. 8). Besides the ball surface is not uniformly attacked. The corrosion is confined to a few localized pits. After 96 h of salt spray exposure, the corrosion of the solder balls becomes very
Fig. 6. PoP assemblies (a) OMAP35xx Processor (bottom device) and PoP Memory (top device) stack up on PCB [10]. (b) Top and (c) Bottom packages [11].
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
4
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
Fig. 7. Top solder balls before the test.
Fig. 10. Bottom solder balls after 96 h in salt spray chamber.
The presence of Sn, O and Cl in the external layer is confirmed by a local EDX analysis (see Fig. 13). 4.3. Method of corrosion product measurement
Fig. 8. Top solder balls after 48 h in salt spray chamber.
serious (see Fig. 9). This kinetic is similar as the results obtained in [5] and recalled in Fig. 1. The bottom solder balls seem to be less eroded than the top solder balls (compare Fig. 10 to Fig. 9). The top balls are larger and more spherical than the bottom balls. So the ball geometry has an influence on the corrosion phenomenon. A possible explanation for this phenomenon is that a protecting passivation layer forms more quickly on a smaller ball than on a larger one. After 96 h, some local pits have been observed on the solder surface (see Fig. 11). The concentration of Cl and Na elements is higher in these areas than elsewhere on the ball surface. EDX analyses were also conducted (see Fig. 12). They enable to determine the presence of various elements (Sn, O and Cl) and the type of solder corrosion products after 96 h in salt spray test. The external layer contains tin and oxygen and a small quantity of chlorine. This compound probably forms the above-mentioned passivation layer. The largest corroded area is also composed of tin, oxygen and chlorine, but with a higher proportion of Cl: it is tin oxychloride (Sn(OH)Cl).
Fig. 9. Top solder balls after 96 h in salt spray chamber.
The corrosion product is measured at the end of the test on the cross-section. But as this method is destructive, it is not possible to measure the corrosion product growth on the same sample during a test. Therefore, a method must be developed in order to measure this growth on the same sample during the ageing. With the profilometer shown Fig. 14 these measurements could be done on the same series of solder balls at various exposure times. Fig. 15 gives an example of solder ball profile. The ultimate goal of our studies is to simulate the corrosion phenomenon of the solder alloy. However, the simulation programs enabling the corrosion prediction in solder alloys used in microelectronic
Fig. 11. Solder ball surface: (a) SEM picture (b) Sn mapping (c) O mapping (d) Cl mapping (e) Na mapping.
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
5
Fig. 12. Top corroded solder ball: (a) SEM picture (b) Sn mapping (c) Cl mapping (d) O mapping.
assemblies are not available yet. We presented an approach for calculating the corrosion similarly to IMC-growth calculation in [12] and a possibility to fit in corrosion models into finite element simulations in [13]. To enable this modelling, the corrosion evolution must be analysed in order to develop the corresponding model. After model implementation into finite element systems, the simulation results should be compared to these profilometer measurements for validation. Eventually, simulation can be used to predict corrosion behaviour in different environments.
Fig. 14. Profilometer.
5. Discussion and conclusion This paper has presented an experimental study of the corrosion of microelectronic assemblies in a NaCl humid atmosphere. BGA assembled on a copper based PCB, as well as PoP components were studied.
Fig. 13. Local analysis of corrosion after 96 h ageing (SEM image and EDX analyse).
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
6
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
Finally, the kinetics of the corrosion is similar for balls of different diameters: some corroded areas appear on the lead-free solder balls after 48 h and the corrosion becomes very strong after 96 h. An experimental procedure for modelling this evolution using a profilometer has been proposed. References Fig. 15. Profile in z-axis along 6 solder balls of the Top component.
Even if free copper is rapidly corroded, the whole assembly remains electrically functional along the entire test duration. In fact, when assembled, the copper is not the first part to be attacked by the corrosive atmosphere. Hence, the main focus was on the solder balls, made of SnAgCu. Bibliography has shown that corroded samples are weakened: preaged components have a shorter lifetime when submitted to mechanical or thermal-mechanical stresses [5]. Our study revealed that in assembled samples, the corroded areas are preferentially located around the interfaces between the solder and the pads, more particularly on the component side in the case under study. This can explain the results presented in [5], as the corroded areas are more fragile. Moreover, these parts are the most constrained ones; [14] have correlated Sn anisotropy with stress, and [15] corrosion path and Sn grain orientation. Our results are linking these two findings. Physical analyses reveal Sn(OH)Cl in corroded zones either before or after reflow. In neither case, Na is present. On the BGA SAC balls (after reflow), pits of SnCl4 are also present. On the PoP SnAgCu balls (before reflow), two layers are present on the surface; the external one appears as tin-rich, the second one is Sn(OH)Cl. One other difference between soldered and non-soldered SAC balls is that in free balls, no preferential location for corrosion exists. All these results do not depend on the cleaning procedure used after ageing.
[1] F. Rosalbino, E. Angelini, G. Zanicchi, R. Carlini, R. Marazza, Electrochemical corrosion study of Sn–3Ag–3Cu solder alloy in NaCl solution, Electrochim. Acta 54 (28) (2009) 7231–7235. [2] N.K. Othman, K.Y. Teng, A. Jalar, F.C. Ani, Z. Samsudin, Electrochemical migration behaviours of low silver content solder alloy SAC 0307 on printed circuit boards (PCBs) in NaCl solution, Trans Tech Publications Ltd, 2016. [3] K.I. Yokoyama, D. Tsuji, J.I. Sakai, Fracture of sustained tensile-loaded Sn–3.0 Ag–0.5 Cu solder alloy in NaCl solution, Corros. Sci. 53 (10) (2011) 3331–3336. [4] T.K. Lee, B. Liu, B. Zhou, T. Bieler, K.C. Liu, Correlation between Sn grain orientation and corrosion in Sn-Ag-Cu solder interconnects, J. Electron. Mater. 40 (9) (2011). [5] F. Song, S.W. Ricky Lee, Corrosion of Sn-Ag-Cu lead-free solders and the corresponding effects on board level solder joint reliability, ECTC Proceedings, 2006. [6] D. Li, P.P. Conway, C. Liu, Corrosion characterization of tin-lead and lead-free solders in 3,5 wt.% NaCl solution, Corros. Sci. 50 (2008) 995–1004. [7] K. Yokoyama, A. Nogami, J. Sakai, Creep corrosion cracking of Sn–3.0Ag and Sn– 0.5Cu solder alloys in NaCl solution, Corros. Sci. 86 (2014) 142–148. [8] ASTM, B117-11 Standard Practice for Operating Salt Spray (Fog) Apparatus, 2011. [9] A. Guédon-Gracia, H. Frémont, J.-Y. Delétage, K. Weide-Zaage, Corrosion Study on BGA Assemblies, SMTA Pan Pacific 2016, Hawaï, 2016. [10] K. Gutierrez, G. Coley, PCB assembly guidelines for 0.4 mm package-on-package (PoP) packages, part II, TEXAS Intruments Application Report SPRAAV2A– November, 2013. [11] AMKOR, PoP datasheet, www.amkor.com/go/Package-on-Package. [12] L. Meinshausen, K. Weide-Zaage, H. Frémont, Dynamical IMC-growth calculation, Microelectron. Reliab. 55 (9) (2015) 1832–1837. [13] K. Weide-Zaage, A. Moujbani, H. Frémont, A. Guédon-Gracia, Harsh marine environment-toward corrosion simulation, Pan Pacific Microelectronics Symposium, 2015. [14] E.W.C. Coenen, Correlation between Thermal Fatigue and Thermal Anisotropy in Pure Sn and a Pb-free Solder Alloy, September 2005. [15] T.K. Lee, B. Liu, B. Zhou, T. Bieler, K.C. Liu, Correlation between Sn grain orientation and corrosion in Sn-Ag-Cu solder interconnects, J. Electron. Mater. vol. 40 (9) (2011) 1895–1902.
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
Contents lists available at ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/mr
Effects of salt spray test on lead-free solder alloy A. Guédon-Gracia a,⁎, H. Frémont a, B. Plano a, J.-Y. Delétage a, K. Weide-Zaage b a b
Laboratoire IMS, Université de Bordeaux, Talence, France RESRI Group, Institute of Microelectronic Systems (IMS-AS), Leibniz Universität Hannover, Hannover, Germany
a r t i c l e
i n f o
Article history: Received 21 June 2016 Accepted 6 July 2016 Available online xxxx Keywords: Solder joints Electronic assembly Corrosion
a b s t r a c t This paper starts with a bibliographic survey about solder corrosion and experimental results of the corrosion on lead-free solder balls during salt spray tests. Focus is made on the SnAgCu solder alloy. Ball Grid Array assemblies and “Package on Package” components were put up to 96 h in a salt spray chamber at 35 °C with 5% sodium chloride (NaCl) aqua according to the ASTM B117-09 standard. The weight is measured during the test. The solder alloys are observed and analysed along the ageing with optical microscope and scanning electron microscope equipped with an energy-dispersive x-ray system. The solder alloy deterioration is visible after 48 h. The microstructure is analysed in order to determine the corroded residues found on the surface solder balls after the salt spray test. Tin oxychloride (Sn(OH)Cl) is found on BGA solder joints after reflow and on PoP solder balls before reflow. The size of the solder balls has an influence on the corrosion state. Finally a method is developed in order to measure the corrosion product growth on the same sample during the salt environment exposure. © 2016 Published by Elsevier Ltd.
1. Introduction Salt mist environment, commonly found on snowy or oceanfront roads, induce galvanic corrosion in cars, boats, naval surveillance systems or tour helicopters, to name a few. Integrated circuit corrosion is a major cause of electronic system failures - it is responsible for about 20% of them. The main cause is surrounding humidity that speeds up electrolytic and galvanic corrosion, especially in biased circuits. Corrosion is even more accelerated by the presence of chemical pollutants in the humid air, like sodium chloride. Although corrosion of dense materials and large industrial equipment has been widely described in the literature, few data are available on thin-film materials and microelectronic assemblies. After a bibliographic survey, this paper presents an experimental study of lead-free assemblies aged in a saline environment. In the first part of our study, printed circuits boards (PCB) populated with BGA were aged in a salt spray oven. Copper traces as well as SnAgCu (SAC) solder balls were analysed during the ageing. In a second part of the study, focus is on the solder alloy: PoP (package on package) components were aged in saline conditions and the evolution of corrosion was observed and analysed. 2. Bibliographic survey Few papers relate the behaviour of SnAgCu solder alloy in presence of NaCl. There are two kinds of tests to assess the corrosion of material: ⁎ Corresponding author. E-mail address: [email protected] (A. Guédon-Gracia).
the polarization measurements, which provide information on a given metal, and the accelerated ageing tests to estimate the corrosion of systems. 2.1. Polarization measurements The electrode potentiostatic polarization enables the determination of the behaviour of metal in salt water. The principle lies in the measurement of the electrical potential as a function of the current in order to determine the anodic and cathodic areas of this material. Most papers present results of electrochemical corrosion of bulk material in a NaCl solution measured through polarization curves, as in [1,2] or [3] for instance. In [1], the corrosion behaviour of SAC330 alloy in a NaCl solution is compared with that of the SAC305 by potentiodynamic polarization and impedance spectroscopy measurements. The presence of tin oxychlorides or oxyhydroxychlorides was detected at the surface of both alloys investigated after the electrochemical tests. In [2], PCBs with SAC307 were investigated in NaCl solutions and dendrite growth has been observed and correlated with the NaCl concentration. In [3], SAC305 material is studied. In this paper the time to fracture and the critical stress for fracture under mechanical stress in a saline solution are investigated. The authors highlight that the reliability of lead-free solder alloys should be evaluated considering the effects of the combination of environment and applied mechanical stresses. 2.2. Ageing tests in a marine environment The ageing tests in a humidity chamber and in a salt spray chamber enable to observe the corrosion phenomena on electronic assemblies.
http://dx.doi.org/10.1016/j.microrel.2016.07.034 0026-2714/© 2016 Published by Elsevier Ltd.
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
2
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
Corrosion is accelerated by the presence of chlorine. Some papers relate the behaviour of SAC solder joints during salt spray ageing. In [4], the effect of pre-ageing in a saline environment on the longterm reliability of wafer-level-chip-scale-package (WLCSP) was investigated. WLCSPs assembled with SAC 305 were thermally cycled to failure, with and without 96 h pre-treatment in a 5%NaCl salt spray environment. Pre-ageing resulted in strong (43%) life-time reduction. Moreover, the authors correlated the corrosion path and Sn lattice basal plane by orientation imaging microscopy. The surface morphology and microstructural changes in Sn-4.0%Ag0.5%Cu lead-free solder balls subjected to salt spray (5%NaCl solution) tests were investigated by Song and Lee [5]. They investigated the effects of corrosion on the mechanical strength of solder balls. Compared to the Sn-37%Pb solder balls, Sn-4.0%Ag-0.5%Cu (SAC405) lead-free solder balls are more easily corroded in the salt spray test (see Fig. 1). Two kinds of corroded residues are found on the surface of solder balls after the salt spray test. The presence of Ag3Sn in SAC solders accelerates the corrosion of tin during the salt spray test because of the galvanic corrosion mechanism. The galvanic corrosion forms a brittle Ag3Sn structure at the corroded regions. As for SAC a passivation layer (made of oxides and hydroxides most of the time) appears on the alloy as a protection against chloride ions attacks (see Fig. 2). The alloy is thus only a little corroded on the surface and remains unaltered in depth. The copper percentage increases the corrosion resistance. In [7] the creep corrosion of SnAg (Sn-3.0Ag) and SnCu (Sn-0.5Cu) solders under NaCl solution is investigated at room temperature. The propagation of surface cracks was studied with the result that creep corrosion cracking should be incorporated in the reliability evaluation of lead-free solder alloys. In a harsh environment, with high chloride content, SnCu alloy shows a corrosion similar to that of SnAgCu. It is corrosion-resistant thanks to the formation of a thin passivation layer that protects the alloy against corrosion sub-products. In SnAg solder, the presence of a
Fig. 1. Surface morphology changes of solder balls (diameter 760 μm) after the salt spray test [5].
Fig. 2. Cross-section of Sn-Ag-Cu solders after potentiodynamic polarization test [6].
noble corrosion-resistant metal, namely silver, provides good corrosion resistance. 3. Experimental study on assemblies The test vehicle was a 10.2 × 14 cm FR4 board with 6 daisy chained components in BGAs. Solder balls (256 per BGA: 16 × 16) and solder alloy are in SAC305 (see Fig. 3). The board was cut into 6 parts containing a measurable daisy-chain structure and some copper pads without connections on the other board side. Each part was submitted to different ageing durations. Ageing was performed in a salt spray chamber at 35 °C (Salt Spray chamber Ascott CC450iP). The saline fog was generated from a 5% sodium chloride (NaCl) solution, as defined by ASTM-B117 [8], one of the most widely used tests for evaluating corrosion resistance of materials and assemblies to an aqueous salt atmosphere. The experimental procedure was detailed in [9]. Corrosion was mainly observed on the metallic compounds, that is to say the copper traces and the solder alloy. On the free copper pads,
Fig. 3. BGA assembly (cut along the red lines). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
3
the attack is fast: after 24 h, it is clearly visible by optical microscope observation; after 48 h, some pads and tracks are totally eroded (see Fig. 4). The microstructural changes of Sn-Ag0.5Cu solder joints exposed to salt spray test were investigated, and partly presented in [9]. The following conclusions have been drawn after 96 h of ageing at 35 °C with 5% sodium chloride (NaCl) aqua: - no electrical failures are detected in BGA assemblies, - the change of assembly weight is not significant, - the corrosion of the free copper pads lead to the disappearance of some pads (see Fig. 4), - the corrosion products observed in solder joints are consist of tin chloride and tin oxychloride (see Fig. 5).
In addition to the corrosion evolution during ageing, this study also aims at evaluating the storage of corroded samples in air afterward. Actually the samples can be stored between the test end and the sample observations. That is why the influence of the storage after the test must be assessed. So, after the salt spray test, two techniques are used to clean the samples [9] and the sample corrosion evolution is observed during one week in air depending on the way of cleaning. As a result, the behaviour is similar in both cases whatever the ageing test duration. So, on the one hand, the corrosion products do not evolve during the one week which follows the corrosion test. On the other hand, the cleaning technique does not influence the sample corrosion evolution. These first investigations on solder joints show the importance of understanding the corrosion phenomena of the lead-free solder alloy in salt and humid environment. In order to analyse the solder alloy corrosion before the reflow process, ageing tests on PoP components have been conducted before their assembly.
Fig. 5. Corroded area after 48 h: presence of tin chloride.
Moreover each component has 152 balls for the top and 353 balls for the bottom. That allows observing the corrosion on a large number of samples using a few components. The test specimens are placed in the cyclic corrosion test chamber at 35 °C and exposed to a continuous indirect spray of a fog generated from a 5% sodium chloride (NaCl) solution. The samples are removed at 24, 48, 72 and 96 h. Previous to the ageing, all specimens are dried in an oven at 125 °C during 24 h. Just after removing them from the chamber, the samples are washed in deionized water and dried with a stream of clean, compressed air. 4.2. Results
4. Corrosion of the lead-free solder alloy 4.1. Experimental procedure PoP components (see Fig. 6) are similar to BGA components. Package on package is a configuration where two BGAs are stacked on top of each other (Fig. 6.a). Solder joints are between the top package (Fig. 6.b) and the bottom package and between the bottom package and the PCB. The top package is generally thicker, containing multiple or stacked dies, while the bottom package is thinner, with either smaller or thinner die. As the corrosion of the solder balls must be observed in this experiment, the top and bottom packages are studied separately. The material of the solder balls is Sn3Ag0·5Cu solder alloy and the ball size are 420 μm for the top and 287 μm for the bottom.
Fig. 4. Temporal evolution of the copper pad corrosion. (a) t = 0 h (b) t = 24 h (c) t = 48 h.
During the salt spray test, the relative weight gain is less than 1%. The sample weight is measured using a precision weighing balance, OHAUS Discovery with a precision of 0.01 mg. As the weight values are precise, the relative weight gain is effectively due to the salt spray test but it is very low. No corrosion is observed on balls before the test (see Fig. 7). After 48 h of salt spray exposure, some corroded pits are visible on the surface of the solder balls (Fig. 8). Besides the ball surface is not uniformly attacked. The corrosion is confined to a few localized pits. After 96 h of salt spray exposure, the corrosion of the solder balls becomes very
Fig. 6. PoP assemblies (a) OMAP35xx Processor (bottom device) and PoP Memory (top device) stack up on PCB [10]. (b) Top and (c) Bottom packages [11].
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
4
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
Fig. 7. Top solder balls before the test.
Fig. 10. Bottom solder balls after 96 h in salt spray chamber.
The presence of Sn, O and Cl in the external layer is confirmed by a local EDX analysis (see Fig. 13). 4.3. Method of corrosion product measurement
Fig. 8. Top solder balls after 48 h in salt spray chamber.
serious (see Fig. 9). This kinetic is similar as the results obtained in [5] and recalled in Fig. 1. The bottom solder balls seem to be less eroded than the top solder balls (compare Fig. 10 to Fig. 9). The top balls are larger and more spherical than the bottom balls. So the ball geometry has an influence on the corrosion phenomenon. A possible explanation for this phenomenon is that a protecting passivation layer forms more quickly on a smaller ball than on a larger one. After 96 h, some local pits have been observed on the solder surface (see Fig. 11). The concentration of Cl and Na elements is higher in these areas than elsewhere on the ball surface. EDX analyses were also conducted (see Fig. 12). They enable to determine the presence of various elements (Sn, O and Cl) and the type of solder corrosion products after 96 h in salt spray test. The external layer contains tin and oxygen and a small quantity of chlorine. This compound probably forms the above-mentioned passivation layer. The largest corroded area is also composed of tin, oxygen and chlorine, but with a higher proportion of Cl: it is tin oxychloride (Sn(OH)Cl).
Fig. 9. Top solder balls after 96 h in salt spray chamber.
The corrosion product is measured at the end of the test on the cross-section. But as this method is destructive, it is not possible to measure the corrosion product growth on the same sample during a test. Therefore, a method must be developed in order to measure this growth on the same sample during the ageing. With the profilometer shown Fig. 14 these measurements could be done on the same series of solder balls at various exposure times. Fig. 15 gives an example of solder ball profile. The ultimate goal of our studies is to simulate the corrosion phenomenon of the solder alloy. However, the simulation programs enabling the corrosion prediction in solder alloys used in microelectronic
Fig. 11. Solder ball surface: (a) SEM picture (b) Sn mapping (c) O mapping (d) Cl mapping (e) Na mapping.
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
5
Fig. 12. Top corroded solder ball: (a) SEM picture (b) Sn mapping (c) Cl mapping (d) O mapping.
assemblies are not available yet. We presented an approach for calculating the corrosion similarly to IMC-growth calculation in [12] and a possibility to fit in corrosion models into finite element simulations in [13]. To enable this modelling, the corrosion evolution must be analysed in order to develop the corresponding model. After model implementation into finite element systems, the simulation results should be compared to these profilometer measurements for validation. Eventually, simulation can be used to predict corrosion behaviour in different environments.
Fig. 14. Profilometer.
5. Discussion and conclusion This paper has presented an experimental study of the corrosion of microelectronic assemblies in a NaCl humid atmosphere. BGA assembled on a copper based PCB, as well as PoP components were studied.
Fig. 13. Local analysis of corrosion after 96 h ageing (SEM image and EDX analyse).
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034
6
A. Guédon-Gracia et al. / Microelectronics Reliability xxx (2016) xxx–xxx
Finally, the kinetics of the corrosion is similar for balls of different diameters: some corroded areas appear on the lead-free solder balls after 48 h and the corrosion becomes very strong after 96 h. An experimental procedure for modelling this evolution using a profilometer has been proposed. References Fig. 15. Profile in z-axis along 6 solder balls of the Top component.
Even if free copper is rapidly corroded, the whole assembly remains electrically functional along the entire test duration. In fact, when assembled, the copper is not the first part to be attacked by the corrosive atmosphere. Hence, the main focus was on the solder balls, made of SnAgCu. Bibliography has shown that corroded samples are weakened: preaged components have a shorter lifetime when submitted to mechanical or thermal-mechanical stresses [5]. Our study revealed that in assembled samples, the corroded areas are preferentially located around the interfaces between the solder and the pads, more particularly on the component side in the case under study. This can explain the results presented in [5], as the corroded areas are more fragile. Moreover, these parts are the most constrained ones; [14] have correlated Sn anisotropy with stress, and [15] corrosion path and Sn grain orientation. Our results are linking these two findings. Physical analyses reveal Sn(OH)Cl in corroded zones either before or after reflow. In neither case, Na is present. On the BGA SAC balls (after reflow), pits of SnCl4 are also present. On the PoP SnAgCu balls (before reflow), two layers are present on the surface; the external one appears as tin-rich, the second one is Sn(OH)Cl. One other difference between soldered and non-soldered SAC balls is that in free balls, no preferential location for corrosion exists. All these results do not depend on the cleaning procedure used after ageing.
[1] F. Rosalbino, E. Angelini, G. Zanicchi, R. Carlini, R. Marazza, Electrochemical corrosion study of Sn–3Ag–3Cu solder alloy in NaCl solution, Electrochim. Acta 54 (28) (2009) 7231–7235. [2] N.K. Othman, K.Y. Teng, A. Jalar, F.C. Ani, Z. Samsudin, Electrochemical migration behaviours of low silver content solder alloy SAC 0307 on printed circuit boards (PCBs) in NaCl solution, Trans Tech Publications Ltd, 2016. [3] K.I. Yokoyama, D. Tsuji, J.I. Sakai, Fracture of sustained tensile-loaded Sn–3.0 Ag–0.5 Cu solder alloy in NaCl solution, Corros. Sci. 53 (10) (2011) 3331–3336. [4] T.K. Lee, B. Liu, B. Zhou, T. Bieler, K.C. Liu, Correlation between Sn grain orientation and corrosion in Sn-Ag-Cu solder interconnects, J. Electron. Mater. 40 (9) (2011). [5] F. Song, S.W. Ricky Lee, Corrosion of Sn-Ag-Cu lead-free solders and the corresponding effects on board level solder joint reliability, ECTC Proceedings, 2006. [6] D. Li, P.P. Conway, C. Liu, Corrosion characterization of tin-lead and lead-free solders in 3,5 wt.% NaCl solution, Corros. Sci. 50 (2008) 995–1004. [7] K. Yokoyama, A. Nogami, J. Sakai, Creep corrosion cracking of Sn–3.0Ag and Sn– 0.5Cu solder alloys in NaCl solution, Corros. Sci. 86 (2014) 142–148. [8] ASTM, B117-11 Standard Practice for Operating Salt Spray (Fog) Apparatus, 2011. [9] A. Guédon-Gracia, H. Frémont, J.-Y. Delétage, K. Weide-Zaage, Corrosion Study on BGA Assemblies, SMTA Pan Pacific 2016, Hawaï, 2016. [10] K. Gutierrez, G. Coley, PCB assembly guidelines for 0.4 mm package-on-package (PoP) packages, part II, TEXAS Intruments Application Report SPRAAV2A– November, 2013. [11] AMKOR, PoP datasheet, www.amkor.com/go/Package-on-Package. [12] L. Meinshausen, K. Weide-Zaage, H. Frémont, Dynamical IMC-growth calculation, Microelectron. Reliab. 55 (9) (2015) 1832–1837. [13] K. Weide-Zaage, A. Moujbani, H. Frémont, A. Guédon-Gracia, Harsh marine environment-toward corrosion simulation, Pan Pacific Microelectronics Symposium, 2015. [14] E.W.C. Coenen, Correlation between Thermal Fatigue and Thermal Anisotropy in Pure Sn and a Pb-free Solder Alloy, September 2005. [15] T.K. Lee, B. Liu, B. Zhou, T. Bieler, K.C. Liu, Correlation between Sn grain orientation and corrosion in Sn-Ag-Cu solder interconnects, J. Electron. Mater. vol. 40 (9) (2011) 1895–1902.
Please cite this article as: A. Guédon-Gracia, et al., Effects of salt spray test on lead-free solder alloy, Microelectronics Reliability (2016), http:// dx.doi.org/10.1016/j.microrel.2016.07.034