Root Cause of Black Pad Failure of Solder Joints with Electroless Ni/Immersion Gold Plating, The

JOM, Jun 2006 by Zeng, Kejun, Stierman, Roger, Abbott, Don, Murtuza, Masood

This paper reports on a study of the reaction of solder with the electrolessnickel with immersion gold (ENIG) plating system, and the resulting interfacial structures. A focused-ion beam (FIB) was used to polish the cross sections to reveal details of the microstructure of the ENIGplated pad with and without soldering. High-speed pull testing of solder joints was performed to expose the pad surface. Results of scanning-electron microscopy/ energy-dispersive x-ray analysis of the cross sections and fractured pad surfaces support the suggestion that black pad is the result of galvanic hyper-corrosion of the plated electroless nickel by the gold plating bath. Criteria are proposed for diagnosing black pad of ENIG plating.

INTRODUCTION

The plating system of electroless nickel with immersion gold (ENIG) has been widely used to finish solder pads of printed circuit boards (PCBs), as well as ball-grid array (BGA) and flip chip substrates. It wets well by solder,1-3 provides a flat and uniform surface, and shows high via strength, an important design consideration for thick PCBs with high aspect ratio vias.4,5 Electroless nickel plating Ni(P) often has a lower total cost of ownership than electrolytic nickel plating. The most attractive advantage of ENIG over electrolytic Ni/Au plating is that it can be applied to fine-pitch BGA substrates without complicating the design layout.6,7 Any electrolytic process requires electric connection to each pad. If the pitch is too small, the electric connection (bussing) is difficult, and processing costs become prohibitive. Therefore, electrolytic Ni/Au is used only for substrates with a sufficiently large conductor pitch to permit busses to each pad. Another disadvantage of electrolytic Ni/Au plating is thickness variation. The thickness of electrolytic plating is sensitive to current density, the voltage drop over the conductors, and the geometry of the metal surface. On some designs, thickness variations can be as much as �1 �m in the nickel (for a nominal 5 �m nickel thickness specification) and � 0.2 �m in the gold (for nominal 0.7 �m gold thickness). The upper end of this gold thickness may cause gold embrittlement in line-pitch BGA joints.8,9 For ENIG, the thickness of both nickel and gold is much better controlled. Usually, thickness is 5�0.5 �m for electroless Ni(P) and 0.1 �0.02 �m or less for immersion gold.

However, ENIG finishes have exhibited a black pad detect that can cause brittle fracture at the interface between the solder and metal pad.4,5,10-16 The failure typically occurs during mechanical or thermal-mechanical testing. The worst cases are BGA package solder joint failure during a customer's surface mount assembly process, or in the product's final use by a consumer. To the unaided eye, a solder joint that fails from black pad shows a flat pad where the solder hall separated from the pad. Under an optical microscope, the flat pad surface is observed to have little or no solder remaining on it. In a scanningelectron microscope (SEM), some small crystals of tin-bearing intermetallic compounds (IMCs) may be found on the pad surface. However, no evidence for the ductile fracture of the solder can be observed. In cross sections of the failed joint, Ni^sub 3^Sn^sub 4^ (for SnPb solder joints) or Cu^sub 6^Sn^sub 5^ (for SnAgCu solder joints) is found on the solder side, but a phosphorous content higher than that of the Ni(P) plating is detected on the pad side. Because of this observed high phosphorous content, many in the industry hold that the ENIG black pad defect solder joint failure is caused by the phosphorous content of the Ni(P) plating.

The purpose of this paper is to clear the contusion about the solder joint failure caused by ENIG black pad detect. The authors will demonstrate that a high phosphorous content by itself cannot be taken as evidence for black pad. and the origin of black pad is not in the solder or soldering process. Criteria will be defined for identification of black pad failure.

See the sidebar for experimental procedures.

RESULTS

Figure I presents an ENIG-plated pad where the solder ball fell off after thermal cycling. Though there were some bright IMC crystals on the pad. there was virtually no solder residue. In a tilt view, the pad appeared Hat. At higher magnification, a feature that looked like the boundaries of the plating nodules was observed in a top view (Figure 2a). In the 30-degree tilt view in Figure 2b, the boundary-like features appeared as separations from the plating nodules, hereafter called mud-cracks because of their appearance. The pad surface was rather clean. Energy-dispersive x-ray (EDX) analysis of the area in Figure 2a. which was about 765 �m^sup 2^. found 81.0Ni. 5.2Sn. and 13.8P(wt.%). Assuming that all the signals of tin were from the few bright Ni^sub 3^Sn^sub 4^ crystals, and neglecting this solder residue on the pad surface, the atomic ratio of Ni:P in the pad surface was calculated from these data to be 75.1:24.8, very close to the stoichiometry of Ni^sub 3^P. This indicates that the joint was broken between Ni^sub 3^P and Ni^sub 3^Sn^sub 4^, and the mud-cracks were in the Ni^sub 3^P layer. In a cross section of the pad, the mud-cracks shown in Figure 2 appeared as spikes in the pad surface (Figure 3). This is the conclusive evidence that the ENIG plating had the black pad detect.