Scanning Electron Microscopy (SEM) is one of the most important imaging techniques used in semiconductor failure analysis. While optical microscopes remain extremely useful for initial inspection, many semiconductor defects are simply too small to characterize adequately with visible-light microscopy. SEM uses a focused beam of electrons rather than light, allowing engineers to examine semiconductor surfaces, interconnects, particles, cracks and cross-sections at much higher resolution.
SEM is particularly valuable because it combines high-resolution imaging with:
The Microelectronics Failure Analysis Desk Reference describes SEM as one of the most versatile instruments in a failure-analysis laboratory. Although techniques such as TEM and AFM can achieve higher ultimate resolution in certain applications, SEM is often more practical because large areas can be examined with relatively little sample preparation. Under suitable conditions, SEM resolution can approach approximately 1 nm.
In semiconductor failure analysis, SEM is often used after an electrical failure has already been characterized and the suspected failure location has been narrowed down.
A typical workflow might be:
Electrical failure detected
↓
Electrical Failure Analysis
↓
Fault localization
↓
Sample preparation / delayering / FIB cross-section
↓
SEM inspection
↓
EDS or TEM if required
↓
Physical defect identified
↓
Failure mechanism and root cause determined
SEM therefore provides an important bridge between locating an electrical problem and physically understanding the defect responsible for it.
A Scanning Electron Microscope creates an image by scanning a finely focused electron beam across the surface of a sample. As the incident electrons interact with atoms in the sample, several types of signals are produced.
These may include:
Different signals reveal different information about the semiconductor. For conventional SEM imaging, two of the most important are:
Secondary Electrons (SE)
and
Backscattered Electrons (BSE).
Characteristic X-rays can additionally be analyzed using Energy Dispersive X-ray Spectroscopy (EDS/EDX) to provide elemental information about a defect or contamination.
The SEM contains an electron source that generates electrons. Those electrons are accelerated through the electron column and focused using electromagnetic or electrostatic lenses. Scanning coils then move the beam across the sample in a raster pattern.
A simplified SEM process is:
Electron source
↓
Electron acceleration
↓
Electron lenses focus the beam
↓
Scanning system moves beam across sample
↓
Beam interacts with sample
↓
Detectors collect emitted signals
↓
Signal intensity creates the image
The quality of the final SEM image depends on several interacting parameters, including:
Understanding these parameters is important because simply increasing magnification does not automatically provide more useful information.
Several types of electron sources have historically been used in scanning electron microscopes.
The uploaded reference discusses four principal types:
Traditional tungsten filaments are relatively simple and inexpensive. However, their brightness and achievable resolution are lower than modern field-emission sources, and filament life is comparatively limited.
Lanthanum hexaboride has a lower work function than tungsten and therefore can produce greater electron brightness.
This can provide more electron current in a smaller beam spot.
The trade-off is that LaB6 sources require a better vacuum because the source is more sensitive to contamination.
High-performance SEM systems commonly use field-emission electron sources.
A very sharp tungsten tip concentrates the electric field sufficiently to produce electrons through field emission.
These sources provide very high brightness, enabling small electron-beam spot sizes.
Cold field emission offers excellent low-voltage performance but can suffer from beam-current instability because the source is highly sensitive to contamination.
Schottky field-emission systems use thermal assistance and provide a more stable beam, making them widely useful for high-performance semiconductor SEM analysis.
The incoming electron does not simply strike the surface and stop. Once the primary electron beam enters the sample, the electrons scatter both laterally and vertically while gradually losing energy. The result is often represented as an interaction volume beneath the point where the beam entered the sample. Different signals originate from different parts of this interaction volume.
This is extremely important in semiconductor failure analysis because the signal being measured determines what part of the sample is actually being observed.
For example:
This means that SEM image resolution and EDS spatial resolution are not necessarily the same.
Secondary Electron (SE) imaging is the most common SEM imaging mode for high-resolution surface examination. Secondary electrons are low-energy electrons. Because they can travel only a very short distance through material before losing their energy, only secondary electrons created close to the surface can escape and reach the detector.
This makes SE imaging:
The amount of secondary-electron emission also changes with the local surface angle. Edges, slopes and raised features therefore produce contrast that makes the SEM image appear three-dimensional.
This high depth perception is one of the reasons SEM images are relatively intuitive to interpret.
In semiconductor failure analysis, SE imaging can be useful for examining:
The technique is especially useful when the physical shape of the defect matters.
Backscattered Electrons (BSE) are primary electrons that have been scattered back out of the sample. They generally retain substantially more energy than secondary electrons. Because backscattered electrons originate from deeper in the sample and can undergo significant lateral scattering, BSE imaging is generally less surface sensitive than SE imaging. However, it provides an important advantage:
Backscattered-electron intensity depends strongly on atomic number.
Materials containing heavier elements generally produce a stronger backscatter signal and appear brighter. This means BSE imaging can help distinguish materials that have similar surface shapes but different compositions. The uploaded reference gives an example involving an aluminum interconnect with a tungsten cap damaged by electromigration. The secondary-electron image reveals detailed surface morphology, while the backscatter image makes the tungsten regions significantly easier to distinguish because of their higher atomic number.
| SEM Signal | Main Information | Strength in Semiconductor FA |
|---|---|---|
| Secondary Electrons (SE) | Surface morphology and topography | High-resolution inspection of cracks, particles, interconnects and exposed defects |
| Backscattered Electrons (BSE) | Atomic-number / material contrast | Distinguishing materials and highlighting high-Z structures |
| Characteristic X-rays | Elemental composition | Identifying contamination and material composition using EDS |
| Specimen/electrical current | Electrical behavior | Specialized fault-isolation methods such as voltage contrast and EBIC |
It is common for a failure analyst to examine the same region using more than one detector. SE and BSE images can therefore provide complementary information.
SEM is not normally the first step of a semiconductor failure investigation. A modern FA workflow should ideally narrow down the location of the failure before destructive sample preparation begins.
After deprocessing, metallization and thin-film structures can be examined with SEM until the physical defect is identified. This is a crucial concept. SEM is extremely powerful, but randomly searching a complex IC at high magnification is inefficient. The better approach is:
Electrical evidence first → localization → targeted SEM inspection.
SEM can contribute to the investigation of many semiconductor failure mechanisms.
Examples include:
SEM can reveal physical damage such as:
After electrical fault localization and sample preparation, SEM can reveal:
SEM can help identify:
Electrical overstress can create:
SEM can help characterize the morphology after the failure site has been exposed.
SEM provides detailed images of corrosion morphology and, when combined with EDS, can help characterize associated elemental contamination.
Particles and residues can be imaged at high magnification.
SEM combined with EDS is particularly valuable when the analyst must determine whether a particle is metallic, inorganic or associated with a process contaminant.
SEM can reveal:
One of the most important limitations of conventional SEM is that it primarily provides information about exposed surfaces.
Many semiconductor defects are buried beneath:
This is why SEM works so closely with sample-preparation techniques such as:
SEM may provide the image, but careful sample preparation often provides the access.
One of the most powerful semiconductor FA combinations is:
Focused Ion Beam + SEM
A suspected failure location can first be identified electrically. FIB can then remove material precisely at the region of interest. The newly exposed cross-section is examined with SEM.
A typical workflow might be:
Electrical open detected
↓
Fault localization identifies suspect via
↓
FIB mills a cross-section
↓
SEM reveals via void or structural defect
This is far more efficient than mechanically sectioning a large region and hoping to intersect the exact defect.
For more information, see:
Modern failure-analysis laboratories frequently use dual-column FIB-SEM instruments. The FIB performs the material removal while the SEM provides high-resolution electron imaging.
This allows the analyst to:
without removing the sample from the tool. FIB-SEM therefore combines site-specific preparation and characterization in a single platform.
SEM is also valuable beyond the semiconductor die itself. After non-destructive techniques such as:
have narrowed the defect location, package structures can be exposed for SEM analysis.
Package structures that may be examined include:
Mechanical cross-sectioning can introduce artifacts when analyzing soft materials such as solder or gold. Polishing can smear these materials over the real defect and make interpretation difficult. The uploaded reference describes the use of FIB ion milling to clean a manually sectioned cracked solder bump, removing the smear so that the true failure structure could be examined by SEM. This illustrates an important FA principle:
Sample-preparation quality directly affects SEM interpretation.
A beautiful SEM image does not necessarily show the original defect if the preparation process itself has altered the sample.
One of the most useful additions to an SEM is an Energy Dispersive X-ray Spectroscopy detector, often called:
When the electron beam interacts with atoms in the sample, inner-shell electrons can be removed. Electrons from higher energy levels then fall into the vacancies. This process produces characteristic X-rays associated with the elements in the material. By measuring the energy of these X-rays, an EDS system can help determine the elemental composition of the analyzed region.
Suppose SEM reveals an unusual particle bridging two conductors. The image alone may show: There is a particle.
But it may not establish what the particle is.
EDS may indicate whether the material contains elements associated with:
The combination provides both:
SEM → morphology
and
EDS → elemental composition
The package-failure chapter in the uploaded reference gives an example where SEM imaged contamination and EDS analysis identified the contaminant as tin.
A typical contamination investigation might proceed as follows:
Electrical leakage / short detected
↓
Defect location identified
↓
SEM images foreign material
↓
EDS spectrum collected on particle
↓
Reference spectrum collected nearby
↓
Elemental differences compared
↓
Possible contamination source investigated
This can help connect the physical defect to a manufacturing or environmental root cause. However, EDS must be interpreted carefully. The X-ray signal can come from a larger interaction volume than the high-resolution SEM image, which means nearby material can contribute to the spectrum.
The apparent resolution of an SEM image should not be confused with the spatial resolution of EDS. An SEM may create an image from a very small beam spot, but high-energy electrons can penetrate deeply and scatter laterally through the sample. Characteristic X-rays may therefore be generated from a much larger volume. The uploaded reference notes that at high accelerating voltages electrons can penetrate several micrometers into bulk material, which can significantly degrade the spatial resolution of SEM-based EDS. This means the analyst should not automatically assume that an EDS spectrum represents only the tiny feature visible underneath the beam.
Accelerating voltage is one of the most important SEM operating parameters.
A common misconception is that higher voltage is always better because the electron beam can often be focused into a smaller spot. High voltage does provide advantages. But it also causes electrons to penetrate deeper into the sample and spread over a larger interaction volume. Low beam voltage can therefore provide more surface-sensitive information.
For example, the reference compares SEM images of an aluminum interconnect at 20 kV, 10 kV and 1 kV. At high voltage, buried tungsten plugs become visible through the aluminum because of the greater penetration depth.
At lower voltage, more surface detail—including thin contamination features—becomes visible. This demonstrates that the correct voltage depends on the question being asked.
Can provide:
But may also cause:
Can provide:
But may offer:
Field-emission systems are particularly valuable because their high brightness can support good performance at relatively low accelerating voltage.
One of the biggest practical problems in SEM is sample charging. A conductive sample connected to ground can remove excess electrical charge created by the incident electron beam.
An insulating sample cannot. Charge can therefore accumulate on the surface.
This may create:
Semiconductor samples often contain combinations of conductive and insulating materials, so charging can be especially relevant after deprocessing or package analysis.
Several approaches may be used.
Good electrical contact between conductive regions and the sample holder is important.
Lower voltage often reduces charging because the electron interaction with the sample becomes more surface localized.
At certain beam energies, the balance of incoming and outgoing electrons may reduce net charging.
A very thin conductive film can dissipate charge.
Common coating materials discussed in the reference include:
However, coating should not automatically be used.
A conductive coating modifies the sample. For semiconductor failure analysis this can create several problems.
The coating may:
The uploaded reference warns that analysts should avoid sputter coating where possible, particularly when high-resolution surface analysis or subsequent elemental analysis is required. At extremely high magnification, the microscope can end up imaging the grain structure of the coating rather than the actual semiconductor surface.
Carbon coating offers one particular advantage. Unlike many metallic coating materials, carbon generally introduces less interference into EDS analysis of semiconductor samples.
It can therefore be useful when charging must be controlled during elemental mapping. However, the reference notes that carbon does not improve secondary-electron emission in the same way as some metal coatings and may not provide the best morphology image. The choice of coating should therefore depend on what analysis will follow.
Tilting the semiconductor sample can significantly change the appearance of an SEM image.
A tilted image can provide a more intuitive three-dimensional view of:
However, tilt also foreshortens one dimension of the sample image, so it can complicate accurate dimensional measurements. For semiconductor metrology, the analyst must therefore understand the image geometry.
Obtaining a high-quality SEM image involves more than adjusting focus. Electron optics can produce astigmatism, where the beam is focused differently in different directions. SEM instruments provide stigmator adjustments that compensate for this effect. The reference emphasizes that focus and astigmatism correction are among the most important skills required to obtain true high-resolution images.
An apparently blurred semiconductor defect may therefore be caused by:
The analyst should eliminate imaging problems before interpreting the defect.
Working distance is the distance between the sample and the final SEM lens. Shorter working distances can often improve high-resolution performance because the electron optics can form a better focused probe.
However, short working distances also increase the risk of the sample or sample holder colliding with the objective lens. Large semiconductor packages, tilted specimens and irregular samples therefore require careful mounting.
SEM imaging always involves practical trade-offs between:
SEM images can contain distortions that are unrelated to the sample. At very low magnification, large beam-scan angles can create effects such as:
This is another reminder that SEM images should be treated as analytical measurements rather than simple photographs.
SEM can also be used for electrical fault localization. Voltage Contrast (VC) occurs because electrical potential affects the emission and collection of secondary electrons. Structures with different electrical states may therefore appear with different brightness.
In Passive Voltage Contrast (PVC), the device is not externally powered. Floating structures can charge under the electron beam and appear differently from structures electrically connected to larger conductive networks.
This can help identify:
The failure-analysis reference describes PVC as a widely used fault-localization technique in FIB and SEM.
With Active Voltage Contrast, electrical bias or device activity is used while SEM imaging is performed. This can extend fault localization beyond static connectivity problems. Voltage contrast can therefore transform SEM from a purely physical-imaging tool into an electrical diagnostic tool.
A specialized SEM technique is Electron Beam Induced Current (EBIC). When the electron beam penetrates semiconductor material, it can generate electron-hole pairs. Electric fields at p-n junctions and depletion regions separate these carriers. The resulting electrical current can be measured and converted into an image. EBIC can reveal electrical behavior associated with semiconductor junctions and can help identify defective regions. The uploaded reference describes examples where EBIC was used to image junction structures and identify defective transistor implants after appropriate deprocessing. EBIC is more specialized than conventional SEM imaging but demonstrates the much broader diagnostic capability of electron-beam tools.
As process nodes shrink, optical fault-isolation methods increasingly face spatial-resolution limitations. SEM can provide significantly higher imaging resolution. The uploaded reference describes backside SEM analysis of advanced devices where certain fin-level defects become visible after silicon removal even though they may be obscured from the front side by metal and contact structures.
The FA flow may include:
Backside thinning
↓
Electron-beam fault localization / voltage contrast
↓
Nanoprobing
↓
FIB sample preparation
↓
TEM root-cause analysis
This illustrates how SEM fits into a modern multi-tool analysis workflow rather than replacing every other technique.
A conventional SEM usually examines bulk samples using emitted electrons. However, a sufficiently thin sample can also be examined using transmitted electrons. This technique is often called STEM-in-SEM.
A FIB can prepare a thin specimen, often below approximately 100 nm, which is then placed over a suitable detector. Because the electrons travel through a thin sample rather than scattering deeply through bulk material, transmitted-electron imaging can provide improved spatial localization. The uploaded reference describes STEM-in-SEM as a useful way to obtain TEM-like contrast using an SEM platform, although it does not provide all the capabilities of a dedicated TEM.
Optical microscopy and SEM are complementary.
Advantages include:
Advantages include:
A good failure-analysis workflow does not automatically begin with the highest-resolution tool. Optical inspection should often be performed first because it is fast and non-destructive. SEM is then applied where greater detail is needed.
SEM and FIB perform fundamentally different functions.
Primarily uses electrons to:
Primarily uses ions to:
This is why combined FIB-SEM instruments are so useful.
FIB creates access.
SEM examines what has been exposed.
SEM and TEM also serve different purposes.
Usually provides:
Provides:
However, TEM requires a very thin sample and generally analyzes a much smaller region.
In many difficult FA cases the sequence is:
SEM finds and characterizes the defect
↓
FIB prepares a lamella
↓
TEM performs final nanoscale analysis
SEM is particularly useful when:
SEM is less useful when the relevant defect remains deeply buried and no preparation has exposed it.
A well-controlled SEM investigation might follow this sequence:
Review:
Potential techniques include:
Use appropriate electrical or optical fault-isolation techniques.
Depending on the device:
Select suitable:
Determine whether morphology or material contrast provides additional evidence.
Characterize suspicious materials or contamination.
If SEM cannot resolve the root cause:
may be required.
Higher voltage can increase penetration and hide subtle surface details.
A larger image does not necessarily contain additional physical information.
Conductive coatings may destroy or obscure important FA evidence.
A small particle in the SEM image may generate an EDS spectrum containing signal from surrounding materials.
Polishing, milling or coating may create features that look like real failures.
Searching blindly at high SEM magnification can waste considerable analysis time.
SEM reveals physical structure, but the observed defect must still be correlated with electrical data and manufacturing history to establish the true root cause.
A critical distinction in failure analysis is the difference between:
For example:
Symptom: IC fails electrically
↓
Failure mode: Open circuit
↓
SEM observation: Void in a metal interconnect
↓
Failure mechanism: Electromigration
↓
Root cause: Excessive current density or process-related weakness
SEM may reveal the void.
It does not automatically prove why the void formed.
Root-cause analysis requires correlation with electrical behavior, process information, reliability history and other evidence.
| Failure / Suspected Defect | SEM Contribution | Often Combined With |
| Metal open | Inspect break, void or damaged interconnect | EFA, FIB |
| Metal short | Identify bridge or conductive material | EFA, FIB |
| Via/contact defect | Inspect cross-section | FIB, TEM |
| Electromigration | Image voids/extrusions and interconnect damage | EFA, FIB |
| EOS damage | Inspect melted or burnt structures | EMMI, thermal imaging |
| Contamination | Image particle/residue | EDS/EDX |
| Corrosion | Examine morphology | EDS/EDX |
| Bond-wire failure | Inspect fracture or bond interface | X-ray, optical |
| Solder-joint failure | Examine crack/void cross-section | X-ray, FIB |
| Die crack | High-resolution crack inspection | Optical, SAM |
| Nanoscale transistor defect | Initial localization/inspection | FIB, TEM, nanoprobing |
SEM stands for Scanning Electron Microscopy or Scanning Electron Microscope, depending on context.
SEM provides high-resolution images of exposed semiconductor and package structures.
It is used to inspect physical defects including cracks, particles, interconnect damage, corrosion and FIB cross-sections.
Optical microscopy uses visible light.
SEM scans the sample with electrons and can achieve substantially higher resolution and greater depth of field.
Optical inspection is usually faster and should often be performed before SEM.
Secondary electrons are low-energy electrons generated near the surface of the sample.
Because only electrons generated close to the surface escape, they provide strongly surface-sensitive, high-resolution images.
Backscattered electrons are primary electrons scattered back out of the sample.
Their intensity depends strongly on atomic number, so BSE imaging can provide useful material contrast.
SEM-EDS combines Scanning Electron Microscopy with Energy Dispersive X-ray Spectroscopy.
SEM shows the morphology of the defect while EDS provides information about its elemental composition.
SEM can image particles or residue, and EDS can help determine their elemental composition.
Additional surface-analysis techniques may be required if more detailed chemical information is needed.
Conventional secondary-electron SEM primarily images exposed surfaces.
Higher beam energies may provide some subsurface contrast, but buried defects generally require delayering, cross-sectioning or FIB preparation.
Electron-beam exposure can affect some semiconductor devices, especially when electrical parameters are sensitive to beam interaction.
The appropriate beam voltage, current and exposure should therefore be selected according to the analysis objective.
Voltage contrast uses differences in electron emission associated with electrical potential or connectivity.
It can help identify open, shorted or floating structures and may be performed in passive or actively biased configurations.
Electron Beam Induced Current uses the SEM beam to generate electron-hole pairs in semiconductor structures.
Electrical current generated at junctions can be measured and converted into an image, providing additional electrical information about the device.
SEM is generally easier to use for large-area surface inspection and requires less sample preparation.
TEM examines electrons transmitted through an extremely thin specimen and provides substantially higher-resolution information about internal semiconductor structures.
SEM primarily images the sample.
FIB physically removes or deposits material.
Dual-beam FIB-SEM systems combine both functions in one tool.
SEM is most effective when it is used as part of a systematic investigation rather than as an isolated imaging tool.
The failure-analysis process should ideally narrow the problem from:
Device
↓
Circuit
↓
Region
↓
Individual structure
↓
Physical defect
SEM becomes particularly valuable near the end of this sequence.
Once the suspect region has been exposed, SEM can provide detailed physical evidence that can be correlated with:
In difficult cases, SEM can then guide further analysis using TEM or other high-resolution techniques.
The strength of Scanning Electron Microscopy therefore lies not only in its resolution.
Its real value in semiconductor failure analysis is its combination of high-resolution imaging, large depth of field, flexible sample access, detector options, elemental-analysis compatibility and integration with FIB preparation.
If you have a failed semiconductor device that requires SEM analysis, AnySilicon can help connect you with semiconductor failure-analysis companies and laboratories with the appropriate capabilities.
Typical services may include:
Find a Semiconductor Failure Analysis Company
When requesting SEM failure-analysis support, provide information about the semiconductor technology, package type, electrical failure signature, suspected defect location, analysis already completed and number of failed samples available. This can help the laboratory select the correct sample-preparation and SEM imaging strategy.