Semiconductor devices are almost always part of a larger, more complex piece of electronic equipment. These devices will operate in concert with other circuit elements and be subject to system, subsystem, and environmental influences.
As so often happens after an equipment failure, a technician will troubleshoot the unit and determine that a particular device is at fault. The device then will be removed, often using less-than-controlled thermal and mechanical stresses, after which the device will be submitted to a laboratory for analysis. While this is not the optimal failure analysis path, it generally is what happens.
Isolating the Failure
When the subsystem responsible for the failure has been identified, the failed section must be further isolated to the board or smallest mechanical structural level containing all related parts and components. The ultimate goal in failure analysis is to arrive at an accurate determination of the cause of failure.
In semiconductor failure analysis, destructive testing usually is necessary in a large portion of the analytical effort. Tasks such as decapsulation, scribing metal, and cross sectioning are acceptable techniques. Performed prematurely, however, these procedures can result in irreversible damage and a ruined analysis. As a result, the analyst must consider the potential damage and purpose of each task and remember the rule followed by carpenters: measure twice and cut once.
Electrical test verification is an essential step in characterizing the suspect device and establishing its role in the circuit malfunction. Electrical testing also is performed to compare the device’s present condition with specified parameters and operation at different temperature extremes.
Often, electrical testing or verification may not be possible because the device is badly damaged. At other times, validation of the failure may not be possible because the device performed within its electrical specifications. Then, a determination must be made about the appropriate direction of testing and examining the failed device.
Care must be observed in removing electronic components to avoid introducing secondary damage. If the device cannot be patched up to perform an electrical check, optical examination may be the only alternative.
Noted anomalies must be analyzed with cause-and-effect in mind. An electrical overstress site on a cracked die in a broken IC package might have occurred prior to lid removal and could be the cause of failure. In all of this activity, it is essential that repair personnel be educated in the importance of ESD precautions and general component handling and care.
Low-Power Stereo Zoom Optical Microscopy
Failure modes readily identified by a low-power stereo microscope include the following:
Contaminants on the package surface, often located between the leads that can cause electrical leakage or shorts. Fractured or broken dielectrics or glass seals. Fractures in weld seams, blowholes, or voids. Broken leads or loose feed-through pins.
Arc-over or burns across the dielectrics.
Medium- to High-Power Optical Microscopy
The medium- to high-power optical microscope, generally used to accentuate a failure mode already detected by other means, is especially effective for examining the following:
Fractures in leads, plating, and glass-to-metal seals. Chemical damage to nonconductive glasses.
Small defects in weld zones.
Particle Impact Noise Detection
Particle impact noise detection (PIND) test systems detect loose particles within a device that has an unfilled internal cavity (Figure 1). The analyst typically would perform a PIND test on a device whenever the suspected failure mode was high leakage, intermittents, or a short. An open circuit usually would not be caused by loose particles but may be detected if the open is caused by a loose bondwire. However, if the part fails the PIND test, it must be decapped to verify the nature of the particle causing the failure since the PIND test cannot differentiate between conductive and nonconductive particles.
X-ray radiography is a good method for nondestructive analysis of most device types. The analyst should inspect for encapsulated foreign material, internal opens and shorts, and changes in alignment due to the encapsulation process. When properly performed, X-ray analysis will not alter or affect either the device or its failure mode.
This technique can provide two major benefits for the failure analyst:
A graphic representation of internal characteristics of the device or sample under inspection prior to any cutting, depotting, or handling.
A graphic representation of how the device is constructed. This enables the analyst to formulate effective disassembly procedures.
Scanning Acoustic Microscopy
Scanning acoustic microscopy (SAM) uses the absorption and reflection of ultrasonic waves in a sample. This technique is especially sensitive to any change in acoustic impedance such as debonds or delamination that can occur in a plastic encapsulated device.
These types of faults are very difficult to resolve using X-ray techniques. Typical applications for SAM include the evaluation of die-attach integrity, detection of voids in the molding compound, characterization of wire bonds, and identification of cracks in the die or molding compound.
Hermeticity testing determines the integrity of the device’s encapsulation. The purpose of the encapsulation is to seal gases or fluids inside the device’s package and prevent gases or fluids from leaking into the device.
Solid-state devices must be protected from the outside environment. The thin-film nichrome resistor is a good example of what can happen from moisture intrusion. These devices are attacked by high moisture and gases on internal surfaces that have no protective plating, resulting in damage from corrosion. To determine the package integrity of these devices and other components, failure analysts may use a gross leak tester or a fine (helium tracer gas) leak tester. Gross leak, or bubble leak, testers use fluorocarbon liquids to analyze the devices.
Fine leak testing is performed using digital helium mass spectrometers. The device being analyzed is pressurized in helium gas, then tested for outgassing helium in the detector.
A leak test, however, may not always be required. Hermeticity testing is a secondary failure analysis tool and generally used to aid the analyst in postulating the probable cause of failure.
Residual Gas Analysis
Knowledge of the gas type inside a package can be extremely important. Many electrical components are hermetically sealed in dry nitrogen. If a salt moisture atmosphere is found inside the package, then corrosion and electrical leakage failures are quite possible.
Water vapor content and residual gas analysis (RGA) are performed using the same basic technique. Outgassing of lubricants that condense on relay contacts can increase contact resistance on the relays at low currents and voltages. Outgassing of epoxies in hybrid circuits, which subsequently condense on die surfaces, can lead to electrical leakage failures. Depending on the materials and systems involved, RGA should only be performed by those certified in the technique, and even then there are concerns about inaccuracies and inconsistencies.1
Low-temperature vacuum drying of a package can be performed to dry out suspected external moisture or to evaporate volatile contaminants without exposing internal components to elevated temperatures. Recovery of the device after a low- temperature vacuum drying would strongly suggest absorbed or trapped moisture on the external surfaces.
A low-temperature bake with a hole punctured in the package removes moisture and volatile gases from the package interior. Recovery after this procedure could indicate trapped internal moisture or volatile contaminants.
A bake at high temperature can heal or reverse degraded electrical characteristics when the failure is caused by ionic contamination or bound charge leakage paths by dispersing the charges on the die. This would indicate the device failed as a result of a manufacturing fault rather than electrical damage from external means. Electrical parts irreversibly damaged by external overstress conditions do not heal by a simple bake.
Depending on the situation, external washing during device failure analysis may be advisable. The types of washes typically used are deionized or distilled water wash, acetic acid wash, solvent washes, and plasma cleaning. Solvent washes are recommended when minimal effects on metals and glasses are desired. The use of ultrasonics, however, might cause damage. Fluorine-based plasma cleaning is recommended for glasses, while oxygen-based plasma cleaning is used for organics.
Package Opening, Decapsulation
The goal in decapsulation is to expose the failure and internal construction of the device without altering the failure mode. Depending on the suspected failure mode, the appropriate techniques for opening a device generally are mechanical and chemical.
Mechanical techniques usually apply to metal, glass, and ceramic packages and typically require jewelers’ tools. Chemical techniques generally are used on plastic or epoxy encapsulated devices. Chemicals, for example, are the logical choice for removing die coatings.
Acid etching involves the use of hot sulfuric acid, fuming nitric acid, hydrofluoric acid, phosphoric acid, or hydrochloric acid. Most acids, however, generally are not very selective because they will attack materials indiscriminately. Commercial depotting chemicals are available for more selective decapsulation.
Internal Visual Examination
Following decapsulation, optical microscopes or a scanning electron microscope (SEM) can be used to evaluate physical anomalies, damaged areas, or electrically overstressed areas. SEMs also can perform voltage contrast and electron beam induced current (EBIC) analysis of devices under biased conditions.
Biased devices also can be examined with high-magnification infrared (IR) thermography. This technique identifies hot spots as possible failure sites. An ultraviolet (UV) microscope can evaluate organic contamination.
An energy dispersive X-ray analysis (EDXA) attachment for the SEM can use the generated X-rays of the SEM to analyze the material composition. When the sample under observation is bombarded by a high-energy beam of electrons, X-rays are given off which impinge on the silicon surface of the EDXA detector.
The penetration depth of X-rays into the silicon is a direct function of the energy of the X-ray. Along the penetration track, interaction occurs between the X-ray and silicon atoms, creating hole-electron pairs. The currents generated are sampled, and the magnitude of pulses usually is a multichannel spectral output with peaks at specific energies, representing X-rays for the various elements. The major advantage of the EDXA system is the simultaneous detection of the entire energy/elemental spectrum.
Other more specialized contaminate surface analysis techniques include auger electron spectroscopy (AES), energy spectroscopy chemical analysis (ESCA), secondary ion mass spectroscopy (SIMS), wavelength dispersive X-ray (WDX) analysis, and electron microprobe (EMP) analysis. The determination of oxides and nitrides is a significant advantage of the EMP over EDXA. AES may involve ion etching of the surface, resulting in a depth profile on the contaminant being analyzed.
ESCA uses valence state information of the material present on the surface of the device and provides excellent resolution of the various carbon compounds. It is used for determining the molecular structure of polymer coatings and identifying chemical states. SIMS is the most sensitive of all techniques and the only instrument capable of directly measuring dopant profiles in a semiconductor.
During the course of failure analysis, there are occasions when it is appropriate or mandatory to cross-section the sample. The goal of cross sectioning or microsectioning in any electrical failure analysis is to expose internal features of components and their packaging.
Size, complexity, and target tend to differentiate the cross sectioning of semiconductor devices from the more typical metallurgical samples. Many of the physical features and subcomponents are quite small, while the complexity lies in the wide variety of materials that can be encountered in a single specimen. A specified target could be a shorted junction area with a feature size of 1 micron. The experience required to hit such targets is largely acquired through practice (Figure 2).
Analyzing the Evidence
After a certain amount of evidence has been accumulated and preliminary conclusions formulated, the pattern and extent of subsequent investigation should be directed toward confirmation of the probable cause and the elimination of other possibilities. As new facts modify first impressions, different hypotheses of failure will develop and be retained or abandoned as dictated by the findings.
During the analysis, it is important to recognize work that does not produce useful results. Certain types of negative evidence may be helpful in dismissing some causes of failure from consideration.
About the Author
Perry L. Martin developed the Electronic Failure Analysis Laboratory at National Technical Systems, formerly McClellan Air Force Base. He also has taught at Loyola University and California State-Sacramento and conducts in-plant failure analysis seminars through Technology Seminars. Mr. Martin is the author of the Electronic Failure Analysis Handbook. National Technical Systems, 5743 Smithway St., City of Commerce, CA 90040, (800) 434-7431, e-mail: [email protected].
Copyright 2000 Nelson Publishing Inc.