With the introduction of chip-on-board technologies, a new set of inspection challenges confronts electronic manufacturers today. These new devices create a unique range of manufacturing and part anomalies.
Micro voids and cracks can result in an early field failure. Over time, a device will overheat as a flaw gets larger and the active substrate loses it heat-sink properties. Attempting to identify this flaw via traditional screening methodologies only increases the size of the delamination during the low cycle fatigue induced by the thermal excitation.
An infrared (IR)-thermal image analysis solution that integrates an IR camera with a repeatable testing system can address these challenges without increasing costs. Thermal image analysis compares a board’s powered-up thermal signature to a statistically sound standard to determine whether the board is acceptable, and if not, why not.
To be effective, a thermal image analysis solution requires five basic elements:
An integrated system with a thermally stable, nonreflective, and repeatable testing environment.
A sensitive (high quantum efficiency) yet robust data acquisition engine (the IR radiometer or camera).
Thermal noise filtering algorithms to render the data useful.
A computer-based, comparative process to grade tested boards as good or bad.
A way to examine the thermal signature of tested boards in detail to permit root cause analysis and board rework.
Using high-resolution IR imaging to sense heat-flow variations within a device provides a consistent and accurate means of viewing workmanship or manufacturing defects. Heat-flow analysis supplies a precise measure of the true internal interconnection integrity.
Heat-flow analysis also is fast, repeatable, and able to identify cracks or connections that exhibit pressure contact with little or no bonding. The image capture and analysis can be accomplished at a pace sufficiently fast to allow real-time inspection within the throughput rates of typical automated production.
Thermal analysis works on the principle of thermal conduction within the device and the circuit card itself. Approximately 99% of all devices and interconnections are mechanical in nature.
If a device has a mechanical flaw such as a void, the thermal path of conduction is lost, and an elevation in temperature will occur at that point for a short period of time. Heat will flow from hot to cold, and the path through a bonded or attached part or interconnection will flow by means of conduction.
Conduction is the most efficient means for heat to flow. If the attachment has a slight void, the heat will increase because it must radiate across the void, the most inefficient means of thermal conduction. This thermal event may only be observable for 120 to 240 ms. Without the aid of computer analysis, the event will be missed. After a period of time, it will appear to be comparable to an acceptable part.
To overcome this problem, a computer must precisely time the screening process and acquire data to identify these small transitory events. Each temperature reference on a board is defined by passing a sample of 30 typically known-good assemblies through the system and computing the mean and standard deviation of the relative temperature for each connection. Unless modified by the operator, the control limits are set at ± three sigma.
A value of ± three sigma statistically will find 95% of the failures. This model can be stored for use later. If a change is made to the board, it is simple to create a new model to represent this evolution. The original model can be saved for analysis and repair of returned boards (bone pile).
To perform a test, a production board is placed in a positioning fixture and functionally powered-up. An image is captured and compared to the model. Then, the pass/fail decision is made.
In the case of a failure, the differential image from the model shows where the tested board is outside the acceptable parameters. This information is the basis for subsequent engineering analysis and repair and rework.
Boards can be tested (including handling) in less than 30 s. Virtually all patent and latent anomalies, including cold solder joints, power to ground shorts, solder bridges, voids, delaminations, weak components, missing components, misaligned components, and lifted leads, can be detected by thermal imaging analysis.
Benefits
Thermal imaging analysis provides four key benefits:
Replaces screening and inspection tools with one detection station.
Reduces warranty and service costs by catching defects, especially latent defects, before they leave the factory.
Decreases bone-pile or dead inventory costs by rapidly identifying defects for repair.
Quickens new product to market by providing design verification of thermal budgets and board operation prior to production.
To appreciate these benefits, let’s look at a typical example. Data was acquired from a device in 0.5-s intervals over 6 s. A normal device displayed a signature shown in Figure 1. Using the same time periods, a device with a micro delamination of the substrata showed the results found in Figure 2.
The thermal signature is 30% higher in the defective device than the acceptable device and is identified even though this defective part successfully passed in-circuit test (ICT), functional testing, and burn-in. Similarly, a void in the underfill of a ball grid array device would have a thermal signature that is quite different from the signature of a device that has been properly attached.
Conclusions
With a systems approach, thermal imaging analysis presents an opportunity to handle the limited real estate and accessibility problems posed by new technologies. It also increases the detection-type coverage to include infant mortality or latent defects in one operation. Using thermal analysis at the functional-test level in the production process acts like a pre-screener and can reduce lengthy functional testing and, quite possibly, the amount of ICT and environmental stress screening.
Jerry Schlagheck is the science and technology advisor at ISIS. Before joining the company, he was technical director of the CELECT Commercial Division of Cincinnati Electronics. From 1978 to 1985, Mr. Schlagheck was an adjunct professor of computer science at Wilmington College and established the Department of Computer Science. ISIS Infrared Screening and Inspection Solutions, 2300 Alfred Nobel Blvd., Ville Saint-Laurent, Quebec, Canada H4S 2A4, (514) 832-0777.
Copyright 1998 Nelson Publishing Inc.
December 1998