# Scanning Electron Microscopy: A Practical Guide

Source: https://www.verbalexperiment.com/blog/scanning-electron-microscopy
Published by Verbex. How these guides are written: https://www.verbalexperiment.com/how-we-write

A sharp SEM image can be the least trustworthy object in the lab.

Scanning electron microscopy rasters a focused electron beam across a specimen, usually in vacuum, and builds an image from secondary electrons, backscattered electrons, or X-rays. The electron wavelength is far shorter than the detail a routine image resolves, so real resolution comes from the probe, the sample, and the signal, a gap that most magnification advice ignores.

Voltage, preparation, coating, and charging can invent apparent structure or erase real structure while the picture still looks clean. The conditions recorded beside the micrograph therefore carry as much weight as the micrograph itself. For a biologist or chemist, the useful question is whether the image supports the intended claim.

## Table of Contents

-   [What Scanning Electron Microscopy Can Resolve](#what-scanning-electron-microscopy-can-resolve)
-   [How the Microscope Builds Each Pixel](#how-the-microscope-builds-each-pixel)
    -   [Following the beam](#following-the-beam)
    -   [A signal map, not a conventional photograph](#a-signal-map-not-a-conventional-photograph)
-   [Choosing a Signal and Chamber Mode](#choosing-a-signal-and-chamber-mode)
    -   [Match the operating choice to the question](#match-the-operating-choice-to-the-question)
    -   [Choose the chamber around the specimen](#choose-the-chamber-around-the-specimen)
-   [Preparing the Sample Before Imaging](#preparing-the-sample-before-imaging)
    -   [A preparation sequence with a reason for each step](#a-preparation-sequence-with-a-reason-for-each-step)
    -   [Preserve a comparison specimen](#preserve-a-comparison-specimen)
-   [When a Sharp Image Is Still Wrong](#when-a-sharp-image-is-still-wrong)
    -   [Test the mechanism before the morphology](#test-the-mechanism-before-the-morphology)
    -   [Treat condition-sensitive structure as provisional](#treat-condition-sensitive-structure-as-provisional)
-   [Questions SEM Can Answer in the Lab](#questions-sem-can-answer-in-the-lab)
    -   [Fracture surfaces and particles](#fracture-surfaces-and-particles)
    -   [Biological surfaces and layered materials](#biological-surfaces-and-layered-materials)
-   [Checklist and Session Record Before You Leave](#checklist-and-session-record-before-you-leave)
    -   [A checklist for acquisition and interpretation](#a-checklist-for-acquisition-and-interpretation)
    -   [A copyable session record](#a-copyable-session-record)
    -   [Automation still needs an audit trail](#automation-still-needs-an-audit-trail)

## What Scanning Electron Microscopy Can Resolve

Two particles can be far enough apart to matter and still merge into one bright patch. The image may look crisp, with clean edges and strong contrast, while the gap between the particles is filled in by signal generated deeper in the specimen. Resolution is the smallest separation the full measurement chain can distinguish under the conditions recorded.

**Operating numbers that limit what an SEM image can claim**

| Quantity | Typical value | What it does and does not mean |
| --- | --- | --- |
| Electron wavelength at 20 keV | About 0.0087 nm; roughly 100,000 times shorter than visible light | Sets a theoretical scale; not the practical resolution |
| Everyday feature size | Hundreds of micrometers down to about 1 to 10 nm | Accessible detail is controlled by probe, sample, and signal |
| Field-emission SEM | About 1 nm, with specialized instruments below 1 nm | Requires a suitable specimen and operating conditions |
| Accelerating voltage | About 0.1 to 30 kV | Higher voltage usually means more penetration and a larger interaction volume |

Typical ranges follow the [IEEE overview of scanning electron microscopy](https://technav.ieee.org/topic/scanning-electron-microscopy/).

The short wavelength explains why SEM can outperform light microscopy, but the image is limited elsewhere. The probe has finite size, electrons scatter inside the material, and the detector only sees the signal that escapes from part of that excited volume. Enlarging the display stretches whatever information is already present. If two points are merged in the signal, a bigger export makes a bigger merged patch.

> **Practical rule:** Treat accelerating voltage, signal type, coating, tilt, and charging behavior as part of the result, not as settings to mention after the image looks good.

This is why a specification sheet describes capability under defined conditions rather than detail promised for every specimen. A conductive, stable surface at one voltage can support a claim that a beam-sensitive, charging sample cannot support at the same nominal magnification.

Apparent depth needs the same caution. SEM images often look richly three-dimensional because topography changes signal collection, but a convincing surface view does not by itself establish true height, undercut shape, or hidden geometry. A micrograph is evidence only when the conditions that produced it are recorded well enough to repeat or challenge it.

## How the Microscope Builds Each Pixel

Consider a fractured polymer mounted on a specimen stub. Before any crack becomes visible, the instrument must generate electrons, focus them, move the probe, and connect the detector output to positions on the screen.

The beam resembles a flashlight moved across a dark object, with the returning signal determining the brightness painted onto the display. The analogy explains scanning, but the detected electrons don't behave like ordinary reflected light.

### Following the beam

-   **Electron gun:** A thermionic or field-emission source produces electrons.
-   **Condenser lenses:** Electromagnetic lenses shape the beam and control the balance between probe current and probe size.
-   **Objective lens:** The final lens focuses the probe onto the specimen.
-   **Scan coils:** Controlled deflection moves the probe through a raster pattern.
-   **Detectors:** Electronics convert the collected signal into brightness at corresponding image positions.

The beam path and detector relationship are illustrated below.

![A diagram illustrating the process of how a scanning electron microscope builds an image pixel by pixel.](https://cdnimg.co/cab61d54-2a92-4be8-97b1-fd84150a185b/22022f02-7e93-46ed-ae0d-41c55195fc21/scanning-electron-microscopy-microscope-diagram.jpg)

The column and chamber operate under controlled vacuum conditions so gas molecules don't excessively scatter the beam. Readers familiar with optical instruments can use this explanation of [microscope parts](https://www.verbalexperiment.com/blog/microscope-parts) to distinguish shared functions, such as focusing, from the different hardware used to perform them.

At the specimen, beam interactions generate **low-energy secondary electrons**, **higher-energy backscattered electrons**, and **characteristic X-rays**. Selecting a detector changes the information assigned to each pixel, even when the beam scans the same area.

### A signal map, not a conventional photograph

Progress required improvements throughout that chain: the [Wiley microscopy introduction](https://catalogimages.wiley.com/images/db/pdf/9780470065402.excerpt.pdf) describes Knoll's **1935** scanning system resolving about **100 micrometers**, compared with Cambridge work reaching about **10 nm by 1960**, approximately a **10,000-fold** reduction in resolvable feature size.

For the fractured polymer, a bright point might result from an exposed edge, detector-facing geometry, or a different material response. Brightness isn't a direct measurement of height.

SEM surfaces look three-dimensional because directional contrast and a large depth of field provide strong depth cues. The result remains a **map of detected signal**, not an ordinary photograph or an automatic surface-height measurement.

## Choosing a Signal and Chamber Mode

For surface shape, secondary-electron imaging is the sensible starting point. For compositional differences, backscattered electrons are often more informative. For elemental identity, the instrument needs an X-ray detector and an EDS acquisition, not merely a visually convincing image.

The [University of Duisburg-Essen SEM guidance](https://www.uni-due.de/imperia/md/content/wet/scanning_electron_microscopy.pdf) describes field-emission systems offering approximately **1 nm resolution** and accelerating voltages of about **0.1 to 30 kV**, while emphasizing that probe size, interaction volume, and signal-to-noise determine practical performance.

**SEM signals and chamber modes**

| Mode | Signal origin | Best used for | Main compromise |
| --- | --- | --- | --- |
| Secondary electrons | Low-energy electrons from a shallow surface region | Topography and fine surface relief | Sensitive to charging and edge brightness |
| Backscattered electrons | Higher-energy electrons scattered by nuclei | Composition and atomic-number contrast | Less surface detail; needs enough landing energy |
| EDS X-rays | Characteristic X-rays from the interaction volume | Element identity and rough amounts | Larger sampling volume and real detection limits |
| Low-voltage SEM | Same detectors at low landing energy | Surface localization on beam-sensitive samples | Weaker signal; slower scans or better detectors |
| Variable-pressure or ESEM | Imaging in a gas environment | Insulating, hydrated, or lightly prepared samples | Often less resolution than optimized high vacuum |

### Match the operating choice to the question

For fine morphology, a **short working distance** and a **small probe current that still provides adequate signal** are useful starting choices. Shortening working distance generally improves probe performance, but changes depth of focus, collection geometry, and practical imaging flexibility.

For atomic-number contrast, sufficient landing energy is needed to produce a useful backscattered signal. Increasing voltage beyond what the question requires can introduce information from deeper material rather than improving the answer.

For EDS, the operating conditions must excite the elements of interest. The compromise is that analytical signal may originate from a larger region than the visible surface feature.

The main surface-imaging and composition-imaging choices are summarized visually below.

![An infographic titled Choosing a Signal and Chamber Mode comparing surface shape imaging and atomic-number contrast methods.](https://cdnimg.co/cab61d54-2a92-4be8-97b1-fd84150a185b/4bc87c7a-af79-448c-81cf-d44642a8562f/scanning-electron-microscopy-imaging-modes.jpg)

### Choose the chamber around the specimen

Lower landing energy generally improves surface localization and can limit charging or damage in biological and polymeric specimens. Its cost is weaker signal, potentially requiring slower acquisition or more sensitive detection.

For charging-prone biological material, the [Oxford Instruments acquisition guide](https://www.oxinst.com/learning/view/article/optimising-data-acquisition-for-biological-sem) notes that backscattered imaging may remain usable when secondary-electron imaging is compromised. Variable-pressure and environmental operation offer another choice: chamber gas helps limit charge, potentially reducing preparation or coating requirements.

ESEM isn't conventional high-vacuum imaging with no preparation. Hydrated specimens still require an appropriate instrument, chamber conditions, and handling route.

A [lab equipment list](https://www.verbalexperiment.com/blog/lab-equipment-list) can support session planning, but access to a detector or chamber mode should be confirmed before specimen preparation.

> **Record the conditions:** Accelerating voltage, spot setting or beam current, working distance, detector, chamber mode, and coating belong with the image. Magnification alone cannot reproduce it.

## Preparing the Sample Before Imaging

Preparation should begin with the feature that must survive. A route suitable for identifying inorganic particles may be unsuitable for preserving a hydrated extracellular matrix.

Cleaning, drying, and coating aren't neutral steps. Each can remove, distort, or cover the feature under investigation.

### A preparation sequence with a reason for each step

1.  **Define the claim and clean selectively.** Remove loose debris only when it isn't the target. Record solvents and cleaning methods, because a rinse that removes contamination may also extract a component of interest.
    
2.  **Mount with electrical continuity.** Secure the specimen to a stub using conductive adhesive or suitable clips. Contact must connect the specimen to ground, not merely hold it mechanically.
    
3.  **Expose and orient the relevant face.** Position a fractured or polished surface so the region of interest is accessible to the beam and detector. Handling tools should avoid the measurement area.
    
4.  **Describe biological preservation explicitly.** Record fixation, dehydration, and drying as separate operations. A dried cell exterior may remain recognizable even after its dimensions or surrounding matrix have changed.
    
5.  **Choose coating around the analysis.** Carbon, gold, or platinum/palladium can provide conductivity. Carbon often avoids introducing metal-coating peaks into EDS, but it complicates interpretation when carbon itself is an analyte.
    
6.  **Document the deposited layer.** Record coating material, thickness, deposition method, and elapsed time before imaging. If thickness is only an instrument estimate, label it as an estimate.
    
7.  **Consider an uncoated route before committing.** Low-voltage, variable-pressure, or environmental imaging may better suit some insulating, polymeric, or hydrated specimens. Once a coating has obscured fine surface texture, an uncoated image of that same surface is no longer available.
    

### Preserve a comparison specimen

Where material permits, retaining an untreated portion makes preparation effects easier to test. For example, a gel intended for morphology measurements could be divided between the planned drying route and an alternative preservation route.

The comparison doesn't establish that either route is artifact-free. It reveals whether the claimed structure survives a meaningful preparation change.

> Preparation should preserve the feature needed for the claim, not merely produce the easiest specimen to image.

A useful preparation record includes failed attempts and deviations. An extra rinse, extended drying interval, or unexpected adhesive contact may explain an image more convincingly than a later adjustment of brightness or contrast.

## When a Sharp Image Is Still Wrong

On an insulating polymer, clean horizontal stripes can look like banded microstructure until the same field drifts brighter between frames. That pattern points to charge building faster than it can leave the specimen, so the first move is grounding and lower dose before any morphology claim.

A [NIST analysis of electron–specimen interactions](https://nvlpubs.nist.gov/nistpubs/jres/092/jresv92n3p205_A1b.pdf) explains that signals arise from an interaction volume rather than strictly from the beam's impact point. Accelerating voltage, composition, density, and geometry determine where the detected information originates.

The distinction between the apparent surface and the sampled volume is illustrated below.

![A diagram comparing a crisp scanning electron microscope image with the subsurface electron interaction volume it generates.](https://cdnimg.co/cab61d54-2a92-4be8-97b1-fd84150a185b/c3a0236a-5266-4f03-805d-9908fb54a1d9/scanning-electron-microscopy-interaction-volume.jpg)

**SEM artifacts and the first check**

| What appears | Likely cause | First check |
| --- | --- | --- |
| Horizontal stripes or drifting brightness | Charge buildup on an insulator | Check specimen-to-stub grounding, lower kV or current, try BSE or low vacuum |
| Feature size changes with voltage | Interaction volume, not just the surface | Repeat at lower landing energy |
| Bright rims on every edge | Secondary-electron edge effect | Compare with a backscattered image |
| Unexpected Au, Pd, or Pt peaks | Conductive metal coating | Record the coat; consider carbon if compatible with the analytes |
| Collapsed cells or cracked gels | Drying and dehydration | Document the drying route; consider a gentler method |
| Pits or growing holes during scanning | Beam damage | Lower dose, scan faster, or inspect a fresh field |

### Test the mechanism before the morphology

If striping changes with scan direction, improves after better grounding, or fades at lower current, charging is the stronger explanation than a real repeating structure. A feature whose apparent width grows at higher voltage needs a repeat at lower landing energy, because the beam may be sampling material below the surface. A bright patch should then be treated as a volume signal until a shallower acquisition supports the surface reading.

The [SEM sample-preparation teaching guide](https://www.hou.usra.edu/meetings/modernsemaug2025/curriculum-resources/Session2_SamplesandSamplePrep.pdf) identifies charging and preparation as major sources of misleading images. Fixation, dehydration, drying, polishing, cleaning, and coating can preserve some features while altering others.

### Treat condition-sensitive structure as provisional

A dendritic-looking network in a dried gel can sharpen after coating and still collapse further under a longer dry or a slower scan. If that network appears only after aggressive drying and high dose, the defensible claim concerns the prepared specimen, not the hydrated material.

An unexpected feature earns confidence by surviving comparison across independent fields and one changed operating or preparation condition. If it disappears after changing detector, voltage, vacuum mode, or coating, its interpretation remains conditional until the cause is understood.

A less dramatic image acquired with gentler settings can be the stronger scientific record. Keep the conditions beside the micrograph: landing voltage, detector, chamber mode, scan speed, coating, and drying route are often what make a sharp image believable.

## Questions SEM Can Answer in the Lab

SEM is most useful when each acquisition has a bounded question. The following examples separate an observable feature from a conclusion that requires additional evidence.

### Fracture surfaces and particles

**Fracture surface:** Secondary electrons can show crack paths and surface relief in a metal or polymer. Backscattered imaging can help distinguish phases with different average atomic numbers, but a fracture image alone doesn't establish the mechanism that initiated failure.

**Particle or powder:** Images can support measurements of visible particle shape and projected dimensions. EDS can assess elements in the probed region, but analyzing selected particles doesn't establish the bulk composition of an entire container.

EDS sensitivity must be stated with its conditions. The [NIST review of SEM/EDS trace-analysis limits](https://www.nist.gov/publications/semeds-trace-analysis-limits-imposed-fluorescence-detector) reports silicon-drift-detector detection limits commonly below **1% by mass**, below **0.1% in the best cases**, and relative precision generally better than **1–5%**; optimized acquisitions of approximately **500 seconds** can reach below **0.001 mass fraction**, or **1,000 ppm**, with about **0.0005 mass fraction** achievable for many elements.

That review also identifies carbon, aluminum, and silicon as exceptions because fluorescence from detector construction materials can worsen detection limits, often by an order of magnitude or more. Those best-case analytical figures shouldn't be assigned to a routine spectrum without checking acquisition conditions, background, peak overlap, and specimen geometry.

**SEM-EDS isn't a trace-analysis method by default.** Elemental identification also doesn't identify crystal structure; diffraction-based evidence is needed for that question.

### Biological surfaces and layered materials

**Biological surface:** A dried biofilm, cell exterior, or tissue specimen can reveal surface organization if charging is controlled. The permitted claim concerns the prepared specimen, not necessarily its native dimensions or hydrated architecture.

**Coated or layered material:** Low-voltage imaging can emphasize the outer surface. A higher-energy acquisition or a prepared cross-section may address underlying material, but contrast through a coating doesn't by itself establish layer thickness or an abrupt interface.

Mode selection should follow specimen behavior:

-   **Stable, conductive, or appropriately coated material:** Conventional high-vacuum SEM can support detailed morphology.
-   **Insulating material needing reduced coating:** Variable-pressure SEM may offer a more suitable compromise.
-   **Hydrated or minimally prepared material:** ESEM may be appropriate when the instrument and chamber conditions support the intended state.

The defensible result is a claim tied to a defined specimen state and sampling region, not a general conclusion inferred from an attractive field.

## Checklist and Session Record Before You Leave

A micrograph saved by itself records what the specimen looked like under one combination of voltage, detector, and vacuum. It says nothing about whether those settings created the structure a reader sees. The session record exists to close that gap: it should let another scientist understand why an image was accepted, not just locate the file.

### A checklist for acquisition and interpretation

1.  **State the question.** Identify whether the acquisition addresses shape, composition contrast, or elements.
    
2.  **Confirm specimen identity and preparation.** Include mounting, drying or polishing route, and coating material and thickness. Record deviations rather than substituting the intended protocol.
    
3.  **Check the scale bar.** Retain calibration and acquisition metadata. Display magnification changes with reproduction size and isn't a sufficient measurement record.
    
4.  **Save operating conditions.** Include accelerating voltage, working distance, detector, vacuum mode, and spot setting or beam current.
    
5.  **Screen for artifacts before measuring.** Check charging, drift, contamination, edge brightness, and damage during repeated exposure.
    
6.  **Challenge unexpected features.** Compare separate fields and change an informative condition. Record whether the feature persisted and what alternative explanation remains.
    
7.  **Document EDS acquisition.** Save acquisition time, elements sought, processing choices, and limitations, including special caution for carbon, aluminum, and silicon.
    
8.  **Separate observation from interpretation.** A bright region is an observation. Greater density, added thickness, or a biological structure is an interpretation requiring support.
    

The essentials can stay taped beside the acquisition workstation.

![A checklist for scanning electron microscopy procedures including stating questions, confirming samples, and recording scale bars.](https://cdnimg.co/cab61d54-2a92-4be8-97b1-fd84150a185b/5cac9ab6-4c8d-4d1d-9d5f-887c96f7fadd/scanning-electron-microscopy-checklist.jpg)

### A copyable session record

| Record field | Information to retain |
| --- | --- |
| Scientific objective | Feature, comparison, and permitted claim |
| Specimen and preparation | Identity, mounting, preservation, polishing, coating |
| Acquisition | Voltage, current or spot setting, working distance, detector, vacuum mode |
| Measurement context | Scale, calibration, image filenames, representative fields |
| Interpretation | Observations, suspected artifacts, accepted and excluded regions |
| Follow-up | Changed conditions, remaining uncertainty, next comparison |

Each row guards against a different failure. The acquisition row matters because two fields imaged at different voltages are separate measurements with different sampling depths, not interchangeable copies of the same surface. The interpretation row forces the distinction that item 8 makes: what the detector produced versus what the scientist believes it means. The follow-up row turns a lone observation into a comparison someone else can repeat.

### Automation still needs an audit trail

The [critical review of AI-enhanced SEM and EDS](https://pure.qub.ac.uk/en/publications/critical-review-of-recent-advances-in-ai-enhanced-sem-and-eds-tec/) describes segmentation, object detection, and related analysis methods. Their outputs still require specimen-aware validation: an algorithm can consistently count coating texture or charging contrast instead of the target structure.

Representative fields, excluded regions, calibration, labels, and uncertainty should accompany automated results. A practical [science experiment documentation workflow](https://www.verbalexperiment.com/blog/science-experiment-documentation) keeps those decisions connected to the original observations.

A record can store conditions; judgment stays with the operator. Scientists remain responsible for interpreting micrographs and for the validity of any feature they report. Before session closure, the record should hold the voltage, detector, working distance, vacuum mode, coating, and the evidence supporting that interpretation.

* * *

A voice lab notebook such as Verbex gives a scientist whose hands are busy at the bench a way to capture preparation changes and acquisition notes while they happen, with each note timestamped and organized into a reviewed lab record that exports to PDF, Word, or an ELN, all processed on the device. Scientists preparing for an SEM session can visit [Verbal Experiment or Verbex](https://www.verbalexperiment.com) and pair the checklist above with a contemporaneous record that later readers can trust as much as the micrographs themselves.