The short answer: use SEM (scanning electron microscopy) when you need a 3D-looking picture of a surface, and use TEM (transmission electron microscopy) when you need to see what is inside a sample at near-atomic resolution. The two techniques are not competing options for the same job — they detect different signals, need differently prepared specimens, and answer different questions. Many core facilities run both, and materials/biology labs routinely image the same specimen on each instrument for complementary information.
Quick answer table
| Parameter | SEM (Scanning Electron Microscopy) | TEM (Transmission Electron Microscopy) |
|---|---|---|
| Signal detected | Secondary electrons (SE, topography) and backscattered electrons (BSE, compositional/Z-contrast) emitted from or reflected off the surface | Electrons transmitted through the specimen, including diffracted beams; energy-dispersive X-rays (EDS) in analytical TEM modes |
| Specimen thickness limit | No real limit for imaging — bulk solids up to several centimetres, constrained only by chamber size and stage travel | Must be electron-transparent, typically under 100 nm; routine ultrathin sections run 50–100 nm, cryo-EM specimens are often thinner |
| Typical resolution | Roughly 1–20 nm depending on instrument, accelerating voltage and detector; high-end field-emission SEM (FE-SEM) can approach 1 nm | Roughly 0.1–0.2 nm; aberration-corrected instruments resolve individual atomic columns |
| What the image shows | Surface topography rendered with strong depth of field, plus compositional contrast from BSE | Internal ultrastructure as a 2D projection through the whole specimen thickness; diffraction contrast reveals crystal lattice and defects |
| Sample prep burden | Light to moderate: fix, dry (often critical-point or freeze-drying for soft/biological material), mount on a stub, sputter-coat with a thin conductive layer if the sample is not already conductive | Heavy: fixation, dehydration, resin infiltration and embedding, ultramicrotomy to cut sections under 100 nm, heavy-metal staining (uranyl acetate, lead citrate) — or, for cryo-EM, plunge-freezing/vitrification instead of chemical fixation |
| Magnification range | Roughly 10× to 500,000× in routine use, higher on specialised FE-SEM instruments | Roughly 1,000× to well over 1,000,000×; routine biological ultrastructure work is often done in the 5,000–100,000× range |
| Sample format | Bulk solids, powders, fracture surfaces, whole small organisms — largely unrestricted in geometry besides chamber size | Must be mounted on a small (~3 mm) TEM grid as an ultrathin section, particle suspension, or replica |
| Vacuum requirement | High vacuum is standard; variable-pressure/environmental SEM modes exist for hydrated or non-conductive samples without coating | High vacuum throughout the column is required for conventional TEM |
| Cost and access per session | Lower capital and operating cost; more widely available in core facilities; faster turnaround because sample prep is simpler | Higher capital cost, specialised staffing for both prep and operation, and longer turnaround — sectioning and staining alone can take a full day or more before the microscope session even starts |
Exact resolution and magnification figures depend on the specific instrument, accelerating voltage, detector configuration and specimen — treat the ranges above as typical, not universal, and check your facility’s instrument specification sheet before quoting a number in a methods section.
How each technique actually works
SEM: scanning a focused beam across a surface
A field-emission or thermionic electron gun produces a beam that is focused to a fine probe and rastered (scanned) across the specimen surface, point by point, line by line. As the beam hits each point, it knocks loose secondary electrons from near-surface atoms and also generates backscattered electrons from deeper elastic scattering events. Detectors collect these signals and the intensity at each scan position builds up a pixel in the final image. Because the signal comes from the surface interacting with a beam that never has to pass through the whole specimen, SEM tolerates thick, bulk, irregularly shaped samples — the image is essentially a map of how the surface responds to the beam at each point, which is why SEM images have the strong, almost photographic sense of depth and topography that TEM images lack.
TEM: transmitting a beam through a thin section
TEM uses a higher-energy, broader electron beam that illuminates the whole field of view at once and passes through the specimen. Regions of the sample that scatter or absorb more electrons (denser material, heavy-metal stain, thicker regions) appear darker; regions that transmit more electrons appear lighter. Because the image is formed by transmission rather than surface reflection, the specimen must be thin enough for enough electrons to get through — hence the sub-100 nm thickness requirement. The trade-off for that demanding prep is far higher resolution: TEM’s shorter effective wavelength and direct transmission geometry let it resolve internal structures — organelles, viral capsids, dislocations in a crystal lattice, individual atomic columns in a well-prepared material sample — that SEM cannot access at all, because SEM never sees past the surface.
STEM: a hybrid worth knowing about
Scanning transmission electron microscopy (STEM) rasters a focused probe like SEM does, but through a thin specimen like TEM, collecting the transmitted and scattered signal at each scan point. It sits on the same instrument platform as many modern TEMs and is commonly used alongside EDS for atomic-resolution compositional mapping. If your facility mentions STEM, treat it as a TEM-family mode — it still needs an electron-transparent specimen.
Choosing between SEM and TEM for your sample
- You need to see a surface, fracture, coating or particle shape — SEM. This is the default choice for topography, surface roughness, powder morphology, fracture-surface analysis and most routine “what does this look like” imaging.
- You need internal ultrastructure — organelles, viral particles, membranes, crystal defects — TEM. If the biological or structural question is inside the specimen, SEM cannot answer it; only TEM’s transmission geometry can.
- You need elemental composition mapped across a surface — SEM with an EDS (energy-dispersive X-ray spectroscopy) detector is the standard combination, and is far less prep-intensive than TEM-EDS for the same question.
- You need near-atomic resolution of a crystal lattice or nanoparticle internal structure — TEM (or STEM), with no SEM substitute at that resolution scale.
- You are prep-constrained — limited sample, limited time, or a sample that cannot survive ultramicrotomy — SEM’s lighter prep burden makes it the practical choice even when TEM would technically answer a deeper question.
- You are imaging a virus, protein complex or macromolecular assembly at near-native resolution — this is TEM territory specifically via cryo-EM, where the specimen is vitrified rather than chemically fixed and stained. Raw cryo-EM datasets are commonly archived in EMPIAR; see CASRAI’s EMPIAR guide for how that data-deposit workflow works.
It is common, not exceptional, to need both: a materials scientist might survey a fracture surface by SEM first to find a region of interest, then extract and thin a specific site with a focused ion beam (FIB) for TEM analysis of the same feature at higher resolution. Budget for both the SEM screening time and the TEM prep/session time if your project genuinely needs that workflow.
Sample preparation compared, step by step
Preparation is where the two techniques diverge most in practice, and it is usually the deciding factor for turnaround time and cost, more than the imaging session itself.
Typical SEM workflow: fix the specimen (chemical fixation for biological material, or none needed for a dry inorganic sample) → dehydrate through a graded ethanol or acetone series if biological → dry (critical-point drying or freeze-drying preserves surface structure better than air-drying, which causes collapse from surface tension) → mount on an aluminium stub with conductive adhesive → sputter-coat with a few nanometres of gold, gold-palladium or carbon if the specimen is not intrinsically conductive, to prevent charging under the beam.
Typical TEM workflow (conventional, resin-embedded): fix (often a two-step aldehyde then osmium tetroxide fixation for biological tissue) → dehydrate through a graded solvent series → infiltrate and embed in an epoxy or acrylic resin → polymerise/cure → cut ultrathin sections (50–100 nm) on an ultramicrotome with a diamond or glass knife → collect sections on a TEM grid → stain with heavy metals (uranyl acetate and lead citrate are standard for biological contrast) to make otherwise nearly transparent biological material scatter electrons differently across structures.
Cryo-TEM workflow (an alternative to chemical fixation): apply the specimen to a grid → blot excess liquid → plunge-freeze in liquid ethane fast enough to vitrify water rather than let it crystallise → image at cryogenic temperature, preserving near-native structure without chemical fixatives or stains distorting it. This is the standard approach for cryo-EM structural biology.
For materials samples (metals, semiconductors, ceramics) TEM prep instead typically involves mechanical thinning followed by ion milling, or site-specific extraction and thinning with a focused ion beam (FIB) — a different workflow from the resin-and-microtome route used for soft biological tissue, but the same underlying requirement: get the sample under ~100 nm thick.
Reading and troubleshooting: common failure modes
| Symptom | Likely cause | Fix |
|---|---|---|
| SEM: bright flaring, streaking or image drift (“charging”) | Non-conductive sample not fully coated or grounded, allowing charge to build up under the beam | Re-coat with a thicker or more continuous conductive layer, ensure the stub-to-sample conductive path is intact, lower the accelerating voltage, or switch to a variable-pressure/low-vacuum SEM mode |
| SEM: flat, low-contrast image with little apparent depth | Detector mix or working distance not suited to the feature of interest — e.g. relying on BSE alone when topography is the goal | Switch to or add the secondary-electron detector, shorten working distance, adjust the tilt angle |
| SEM or TEM: streaky, elongated features at high magnification | Uncorrected astigmatism in the objective lens | Run the stigmator correction routine on a high-contrast edge feature before capturing final images |
| TEM: sections tear, show chatter marks, or vary unevenly in thickness (visible as uneven staining/contrast bands) | Dull diamond/glass knife, incorrect knife clearance angle, or resin that is too soft/hard for the block | Replace or re-approach the knife edge, adjust cutting angle and speed, verify resin polymerisation time/temperature |
| TEM: holes or bubbling appear in the imaged area during observation | Beam damage from prolonged electron dose, common in biological and polymer specimens | Reduce electron dose/exposure time, image a fresh area for final data, or switch to a low-dose/cryo-EM imaging protocol |
| TEM: low or patchy contrast overall | Insufficient or uneven heavy-metal staining, or precipitate contamination on the stain (lead citrate is especially sensitive to atmospheric CO2, which forms lead carbonate precipitate) | Re-stain with fresh reagent, stain under a CO2-free environment (e.g. with NaOH pellets in the staining dish), check grid cleanliness before staining |
| SEM or TEM: fine granular or web-like contamination building up on the imaged area over time | Hydrocarbon contamination from the vacuum system or sample handling, cracked by the beam and deposited on the surface | Plasma-clean the specimen and/or chamber before imaging, minimise beam dwell time on the same spot during focusing |
Frequently asked questions
Is TEM always higher resolution than SEM?
In practical terms, yes — TEM’s sub-nanometre to atomic-scale resolution exceeds what SEM achieves, because TEM’s transmission geometry and shorter effective wavelength access finer detail than a surface-scanning technique can. The trade-off is that TEM only images an extremely thin slice or particle, while SEM images a whole bulk surface; “higher resolution” does not mean “more informative” for every question.
Can the same sample be imaged with both SEM and TEM?
Often yes, but rarely at the same prep stage. A common workflow is to survey a bulk sample by SEM, identify a region of interest, then extract and thin exactly that region (often via focused ion beam milling) down to TEM-compatible thickness for higher-resolution follow-up.
Do I need to coat every sample for SEM?
Only if it is not already electrically conductive. Metals and other conductive materials can often be imaged uncoated. Biological, polymer and ceramic samples typically need a thin sputter-coated conductive layer (or imaging in a variable-pressure/low-vacuum SEM mode) to avoid charging artifacts.
What does a TEM image actually represent, if it is a 2D projection?
It represents the cumulative electron density the beam encountered passing through the entire specimen thickness at each point — overlapping structures at different depths within that ~50–100 nm section project onto the same 2D pixel. This is a real limitation for interpreting 3D shape from a single TEM image, which is part of why electron tomography (collecting images across a tilt series and reconstructing a 3D volume) exists as a follow-on technique.
Which is cheaper to run, SEM or TEM?
SEM, on essentially every axis — lower instrument capital cost, simpler and faster sample prep, and generally lower core-facility hourly rates. TEM’s cost is driven by both the instrument itself and the specialised, labour-intensive sample preparation (ultramicrotomy, staining, or cryo-vitrification) that has to happen before a session can even begin. Exact rates vary by institution and instrument, so check your core facility’s published fee schedule rather than assuming a figure.
Related CASRAI guides
- Transmission Electron Microscopy: Sample Preparation Workflow and Resolution Limits — a deeper look at the TEM-specific prep workflow summarised above.
- Confocal Microscopy: Principle, Setup, and When to Use It — the light-microscopy alternative when optical (not electron) resolution is sufficient and live/fluorescent imaging is needed.
- Atomic Force Microscopy: Contact, Tapping and Non-Contact Modes Compared — a non-electron, probe-based alternative for surface topography at nanometre resolution.
- Dynamic Light Scattering: Z-Average, PDI and How to Read a DLS Report — a faster, non-imaging alternative for nanoparticle size distribution.
- X-Ray Crystallography: How It Works, From Crystal to PDB Deposit — the alternative route to atomic-resolution structure when a diffracting crystal is available.
- Auger Electron Spectroscopy (AES): How It Works and What It Measures — another electron-beam surface technique, focused on elemental composition of the top few atomic layers.
- EMPIAR: The Electron Microscopy Public Image Archive for Raw Cryo-EM Data — where to deposit or find raw cryo-EM datasets once TEM imaging is complete.
- Gram Stain: Procedure, Reagents and How to Read the Result — a light-microscopy specimen-prep and staining workflow for comparison against the heavy-metal TEM staining discussed above.
- Cell Culture Basics: A Beginner’s Guide for New Lab Members — relevant if your SEM/TEM specimen originates from cultured cells.
- Analytical Balance Calibration and Proper Weighing Technique — for the reagent weighing involved in preparing fixatives and stains.
