Head in Pillow - BGA Soldering Problem

Introduction

Modern electronic assemblies very often use advanced integrated circuits in packages such as Ball Grid Array (BGA/uBGA) or Chip Scale Package (CSP). Less common forms include Package on Package (PoP), Wafer-Level Chip Scale Packages (WLCSP), and Land Grid Array with Solder Balls (LGA-SB). These package types feature solder ball terminations distributed across the bottom side of the device. This enables a very high number of signals (I/O) to be placed within a small footprint, which is critical for high I/O count devices like: processors (CPUs), advanced microcontrollers (MCUs), memory devices, and various FPGA chips.

Despite their many advantages, so-called "solder ball" packages also have drawbacks. One of the key limitations is the restricted ability to visually inspect solder joints - simply put, they cannot be visually assessed (except.. in some cases, for solder balls located at the package edges.. it is possible).

As a result, X-ray inspection is commonly used for solder joint quality control. It is reasonably effective at detecting certain issues, such as solder bridges and excessive voiding. However, anomalies like Head-In-Pillow (HiP) are difficult to clearly identify using conventional 2D X-ray systems. In practice, this means that an HiP defect may pass unnoticed through process inspection and electrical testing, only to manifest later during device operation.

In this article, we discuss the Head-In-Pillow problem and the typical causes behind this defect.

..and now let's dive in..

Head-In-Pillow

BGA devices with so-called "collapsing balls" are commonly used components in SMT technology. The balls are made of lead-free alloy or, less frequently, lead-based alloy. During the reflow soldering process, the balls melt and collapse onto the already molten solder paste printed on the PCB pads. Under correct process conditions, the alloy from the ball and the alloy from the paste mix together, forming a uniform solder joint with a characteristic barrel like shape profile.

The Head-In-Pillow occurs when the solder ball fails to form a metallurgical bond with the molten solder paste during reflow phase. Instead, the ball rests on the molten paste just like a "head on a pillow" creating a weak mechanical joint and intermittent electrical connection.

Below is a BGA solder ball after a Dye-and-Pry test, clearly showing the Head-In-Pillow effect. The red color is the dye (penetrant), that I used during the test. The ball has a visible indentation filled with the red dye, which is the final confirmation of a lack of contact between the ball and the solder alloy on the PCB.

The Head-In-Pillow defect is difficult to detect in mass production. Typical X-ray (2D) inspection often does not allow for a clear distinction between a good solder joint and a HiP defect. Of course, high-end X-ray inspection systems make HiP detection much easier. Identifying this anomaly requires experienced operator, image analysis from multiple viewing angles, and what is always in short supply -- the time :)

A Head-In-Pillow joint may be electrically conductive, but it is intermittent. Such a defect may not be detected during ICT or functional testing and may appear later, during the field operation.

From the perspective of soldering quality standards such as IPC-A-610, the Head-In-Pillow anomally is considered a defect for all IPC product classes.

Causes of Head-In-Pillow

There are many potential causes of the Head-In-Pillow defect[1,2], which makes its elimination challenging. Below are the most common sources of this problem:

• Component warpage. Excessive warpage of a BGA/CSP/PoP component during reflow can lift the solder balls above the solder paste. If the ball and the paste melt without phisical contact, and the ball then collapses onto an already solidified paste deposit and HiP defect is formed. This is caused by differences in the coefficients of thermal expansion (CTE) of the package materials or by an incorrect reflow soldering profile.
• PCB warpage. A similar mechanism occurs when the PCB laminate bends. Poor copper balance, a very thin laminate + insufficient board support in the reflow oven - all promote PCB deformation. The mechanism is therefore similar as above. It causes lack of simultaneous contact between the solder ball and the paste during reflow.
• Solder ball oxidation. An excessive oxide layer on the surface of BGA balls makes oxide removal by the flux (from the SMT paste) difficult. This is most commonly caused by improper BGA component storage, i.e., failure to follow MSL (Moisture Sensitivity Level) handling requirements. It is also worth noting that baking reduces moisture but at the same time increases balls oxidation level.
• Insufficient solder paste volume. If SMT solder paste is printed in an insufficient amount, it limits proper contact between the paste (and its flux) and the solder ball. This condition reduces the effectiveness of the flux action and the wetting of the ball. Root causes may include solder paste printing issues (stencil design, apertures, paste transfer) or non-optimal PCB design (pads with too small diameter, via-in-pad issues, etc.).
• Incorrect reflow profile. The reflow profile is essentially a compromise between requirements of the solder paste, the components and the PCB itself. Many factors within the profile can contribute to HiP. For example excessive flux evaporation during preheat, which promotes re-oxidation of the solder balls, etc.

There are many other contributing factors[1,2] that may influence the occurrence of this problem, such as too low flux activity in the SMT paste, interactions between the paste and the BGA ball alloy (material), silver segregation, component placement accuracy, and more.

Summary

Head-In-Pillow (HiP) is one of the most problematic soldering defects in BGA assemblies because it can remain undetected during the production process. As a result, failures often appear only during operation in the field, which is, of course, a major quality issue.

Effective HiP reduction requires a solid understanding of the root causes and the implementation of appropriate corrective actions. Such actions (as always) should primarily focus on eliminating the root causes themselves (stencil aperture changes, reflow profile optimization, etc) with detection methods addressed as a secondary priority.