Semiconductor failure analysis is a vital part of understanding why semiconductor devices fail while in the hands of the consumer. Only then can manufacturers learn how to improve the design of these devices to prevent failure again in the future. To figure out how a device failed, scientists use a variety of techniques from simple electrical testing all the way to highly sophisticated equipment like the Scanning Electron Microscope.
Each method of failure analysis focuses on different failure mechanisms to find the root cause that led to the failure. With this knowledge, leading manufacturers in the semiconductor industry can improve their design and production processes. Keep reading to find out what all these failure analysis techniques are, and how they help identify the cause of failure.
What Is Semiconductor Failure Analysis
Modern electronic devices like mobile phones, laptops, and smart devices are made up of many different semiconductor devices or integrated circuits (ICs) all working together. While in operation, they are subject to various internal factors from the system itself and external factors from its operating environment. When a device becomes faulty in the consumer’s hands, the first level of troubleshooting is done by technicians who try to pinpoint which component or module can be replaced to repair it.
The failed component is then sent to a lab where a trained failure analyst will take over and try to further isolate the problem to the smallest level on the integrated circuit through a series of steps. This process is called semiconductor failure analysis, and it is necessary to understand the cause of failure and what can be done to prevent it in the future.
Semiconductor Failure Analysis Terminology
It’s important to understand the correct terminology in semiconductor failure analysis. Firstly, there has to be an acceptable definition of what exactly constitutes the failure of a device. If the device is not conforming to its electrical and/or mechanical specifications, this is known as device failure. This can happen in two ways for electrical failures.
When a device is unable to perform its intended function, it is called a functional failure.
Sometimes the device may be functional and even capable of performing as intended. However, it might be unable to meet the electrical specifications for one or more of its measurable characteristics like leakage current. This type of failure is known as a parametric failure.
The failure mode is what describes how the device is failing. This can be done by indicating how much the device is deviating from its specification during operation. Excessive supply current or voltage are good examples of this.
This describes the physical condition that led to the failure of the component like the corrosion of metals or electrostatic discharge.
This is the most important and the purpose of why semiconductor failure analysis is done in the first place. Root causes are what trigger failure mechanisms to kick in. In the case of electrostatic discharge, this could be due to improper grounding. Tracing it even further back, the grounding issue may have arisen due to the use of an incorrect mask during the wafer processing stage. Having identified the root of the problem, the manufacturer can proceed with preventing it from happening again.
Semiconductor Failure Analysis Process
Failure verification is the first step in failure analysis. This step is essential so as not to waste valuable resources on failure analysis which can be very costly and time-consuming. Afterward, a step-by-step process is undertaken with non-destructive techniques first, followed by destructive ones. Let’s see what these techniques are.
Non-Destructive Semiconductor Failure Analysis Techniques
Failure analysis methods that can cause an irreversible change in the sample are called destructive. They are only performed after non-destructive methods are completed for obvious reasons.
Electrical Failure Analysis
At this first step of electrical failure analysis, lab technicians will subject the failed device to multiple electrical tests. One of the reasons is to verify the semiconductor devices are indeed faulty. The second reason is to identify the failure mode which can make the entire process of failure analysis more efficient by pointing analysts toward the next logical step.
Electrical failure analysis involves a comparison of the device’s parameters during operation with its specifications. Any deviation can point towards the mode and semiconductor devices are tested at different temperatures. Of course, there are cases where verification is difficult if the failure had caused too much damage or if the device continues to function when tested.
Verification of semiconductor devices can be done using specially designed with parameter analyzers and Automatic Test Equipment (ATE). This type of equipment include multimeters, oscilloscopes, frequency counters, curve tracers, and power supplies. They can provide data logs for each parameter, making it easier for analysts to identify the problem.
Analysts should take care during this process to avoid secondary damage during the removal of components from a circuit board. Components must also be handled with care and lab technicians need to be aware of necessary precautions to avoid electrostatic discharge.
In this step, failed semiconductor devices that are tiny, to begin with, are put under a microscope for inspection at a higher magnification than what is possible with the human eye. Magnification can go from low-power microscopes at 5X to higher-power microscopes that can magnify up to 1000X. The magnified image is often captured by a camera and displayed on a computer monitor.
There are different types of illumination modes on a microscope like brightfield, darkfield, and interference contrast. Each mode will make it easier to observe some features above others. Brightfield is great at observing the topography of the sample surface, while darkfield will bring out details like scratches and contaminants. Interference contrast mode will make defects like etch pits and cracks look more visible.
Contaminants on the surface which are located between leads can cause shortages. Other issues that can be discovered by an optical microscope are broken dielectrics, fractures in weld seams, signs of chemical damage, and fractures in glass-to-metal seals.
X-ray microscopy is a great non-destructive method for inspecting the interior of semiconductor devices. The different materials of the device will cause the x-rays to transmit in different ways, creating an image of varying contrasts that make it easy to distinguish these components. Any open circuits or shorts, encapsulated contaminants, voids, cracks, and alignment issues on the interior can be identified. Aluminum wires on the other hand are invisible to x-rays.
It takes trained personnel to operate the equipment, although more advanced models of x-ray microscopes can provide automated sample handling and various other improvements. With an x-ray image providing analysts a way to understand how semiconductor devices are put together, it becomes easier to plan further disassembly for the next step in failure analysis.
Scanning Acoustic Microscopy
Scanning acoustic microscopy (SAM) uses ultrasound waves which are transmitted through the device to get an idea of what the fault is. Some SAMs do this by interpreting the echo sent back by the sample while others check the sound wave that comes out of the other end of the sample. Sound waves in frequencies between 5 to 150 MHz are used in this technique of failure analysis.
Scanning acoustic microscopes can reveal issues in semiconductor devices like debonding and delamination which x-ray microscopes cannot offer. SAMs are great for failure analysis of interfaces and can be used to check for voids or cracks in the molding compound, the integrity of the die-attach material, wirebonds, and cracks in the die. A die is basically a small block of semiconducting material over which a circuit is made.
Semiconductor devices are encapsulated to prevent fluids and gases from outside coming in. If this seal contains leaks, moisture can get in, resulting in corrosion of surfaces that have no protective plating on them. Electrical shorts and parametric failures can also be caused by moisture getting in.
There are two types of hermeticity testing for semiconductor devices. Fine leak testing involves examining the package for small damage or defects that can lead to minor leakages. This is usually done using Helium gas to pressurize the sample and a detector checks for outgassing. Gross leak testing checks for large amounts of damage on the encapsulation by immersing the sample in fluorocarbon liquids and observing for signs of bubbles appearing.
Destructive Semiconductor Failure Analysis Techniques
A technique is called destructive if it causes permanent damage to semiconductor devices under analysis. This damage could be electrical, mechanical, or visual. Some techniques can even alter the chemical composition. There is no going back after this and every step needs to be properly planned to not accidentally compromise the failure analysis.
Opening Of An Integrated Circuit
After all the non-destructive methods have been exhausted, it is finally time to open up the electronic components and check inside. The objective here is to expose the device for inspection and further analysis of its internal structure without affecting its failure mode. There are both chemical and mechanical methods for opening a device.
If the device had been hermetically sealed, a mechanical process called de-lidding or decapping is used. By applying opposite torques to the top and bottom of a ceramic integrated circuit, its lid can be removed. If it’s a metal can type of device, the weld can be cut to open it.
For plastic-encapsulated ICs, a chemical process called decapsulation is used. Acid can be applied to the surface of the package to remove the plastic cover with sulphuric and nitric acids being used for this purpose. An automated process called Jet Etching is used that produces far better results than manual etching since the application of acid is controlled by the machine.
A visual inspection can be performed by using optical microscopes as mentioned before. Furthermore, other methods such as electron microscopy and Auger electron spectroscopy can be utilized as well for the examination. For more on the different types of microscopes used to inspect semiconductor devices, read What Does A Semiconductor Microscope Do?
Cross-sectioning also known as micro sectioning is to expose a cross-section of the sample for further analysis by optical or electron microscopy. Naturally, this involves slicing along a plane parallel to the plane of interest. Precision diamond saws are used for this step followed by grinding and sawing to ensure all scratches are removed from the cross-section surface.
Hot Spot Detection
Hot spot detection is also called micro thermography and involves identifying areas that indicate unnatural amounts of heating. These ‘hot spots’ are usually the result of high current flow and are caused by defects like leaky junctions, metallization shorts, and dielectric ruptures.
Hot spot detection can be done by applying liquid crystal on the die surface, while a bias is applied to the semiconductor device. As the defective area conducts a current, the heat generated will begin to change the optical properties of the liquid crystal. These changes can be observed through an optical microscope by adjusting the polarization settings. Alternatively, infrared cameras can also be used to locate hot spots.
However, it must be noted that the mere presence of hot spots doesn’t indicate a failure site, and the anomaly causing the problem may be located elsewhere. Therefore hot spot detection must be performed with other types of failure analysis to complement these findings.
Microprobing is a form of failure analysis where a needle is precisely aimed and connected to the point of interest on the die circuit to make electrical contact. After contact is made, electrical measurements can be taken. An analyst uses a microprobing station that can precisely align these needles as required and a schematic diagram of the IC can be used to identify which nodes need to be connected for probing.
The troubleshooting methods used for the die circuits are no different than testing a full-sized circuit board. Anomalous voltages and currents detected by accessing these nodes on the die can help in isolating the problem area. A faulty IC can even be compared with a working IC to identify nodes that may have failed. It takes skill to properly align the probe needles with a node, and incorrect probing may cause scratches, permanently damaging the sample.
Scanning Electron Microscopy And Transmission Electron Microscopy
The Scanning Electron Microscope or SEM can be used to inspect samples on the nanometer scale. The sample is illuminated with a beam of electrons instead of light, and the reflected electrons are detected and analyzed. Through image processing of these detected signals, we can obtain a topographical 3D image of the surface of the sample. Physical defects on the die, bonding failures, contaminants, cracks, wire fractures, seal defects and more could be discovered through a scanning electron microscope.
A Transmission Electron Microscope or TEM uses electrons transmitted through the sample to obtain a 2D projection of the inner structure of a sample. This can reveal the morphology, composition, and crystalline structure of semiconductor devices. TEM is a great way to detect the presence of impurities and offers much greater spatial resolution than SEM. However, these microscopes require more power to operate and a complex process to prepare the sample for inspection.
This is done through a technique called EDXA or Energy-Dispersive X-ray Analysis. It requires a scanning electron microscope to bombard the sample with electrons and measure the amounts of energy present in the x-rays that are generated here. X-rays at specific energies correspond to the elements present in the sample, allowing them to be identified. This EDXA method can help analysts find contaminants present in the sample.
EDXA can be complemented with Fourier Transform Infrared (FTIR) Spectroscopy to obtain details on the molecular structure of the sample. Depending on the chemical composition and bonds present in the sample, the amount of infrared radiation it can absorb or reflect will change.
FTIR testing can be done by exposing the sample to infrared radiation and then detecting the frequencies of infrared that are transmitted or reflected. These readings are then plotted on a graph and the patterns can be matched with signatures for known materials. FTIR spectroscopy can detect the smallest amounts of organic contaminants present in the sample.
Focused Ion Beam
This failure analysis method is similar to scanning electron microscopy. A focused ion beam uses Gallium ions (Ga+) instead of a beam of electrons. Some sample material is dislodged when this beam of ions hits it. This material could be atoms, ions, or electrons and is analyzed. A highly magnified image of the surface can be obtained through this process.
A low beam is used when magnification is the only requirement. By using a more powerful beam, more surface material can be removed with die surface milling or cross-sectioning applications possible. The removal of material is called sputtering and is useful for further analysis when deeper inspection of the sample is required.
Surface Analysis With Auger Emission Spectroscopy
Auger Emission Spectroscopy (AES) is a form of surface analysis technique similar to EDAX that uses a special type of electrons called Auger electrons. With excellent resolution, AES can provide compositional analysis of areas less than one micron across. Its sensitivity is so high that it can detect elements that make up less than 1% of the atomic composition of the sample.
Contaminants and very thin oxide layers on the surface can be detected through AES. However, through ion-sputtering, a crater can be made in the sample surface which can allow matter deeper inside to be analyzed. Combined with raster scanning, the elemental distribution of the sample can then be mapped to generate 3D images. This technique is called Scanning Auger Microscopy (SAM).
Determining Failure Mode And Failure Mechanisms
Apart from the failure analysis methods covered here, the semiconductor industry utilizes other techniques like Light Emission Microscopy (LEM) which involves detecting low-level photon emissions from failure sites, Atomic Force Microscopy which measures local properties of the sample surface, and Electron Beam-Induced Current (EBIC) which measures the current induced in a sample when bombarded with electrons from a scanning electron microscope.
The various tests must corroborate each other’s findings, and if there is any inconsistency, this has to be first resolved before heading to the next step in the failure analysis process. As new readings are taken, an analyst must be prepared to accept that preliminary conclusions may have been wrong. If conducted properly, these techniques for failure analysis should collectively point toward the real reason for the failure. They should help locate the site of the failure on the semiconductor device.
After determining the failure mode, mechanisms, and finally the root cause of failure, these findings can help the manufacturer perfect their production processes. This can lead to overall better-performing semiconductor devices that will power our next generation of technologies. To learn more about semiconductor manufacturing, and failure analysis, check out Inquivix Technologies!
A die is a block of semiconducting material like Silicon, and integrated circuit parts are fabricated on top of this.
Semiconductor devices can fail due to several reasons like high temperature, humidity, excessive current or voltage, mechanical stress, manufacturing defects, or contaminants to name a few.
A failure analysis engineer or failure analyst is a person who uses various techniques to determine the root cause of failure for semiconductor devices or integrated circuits.