Cornerstone guide · Technique

SEM Sample Preparation: A Practical Guide

Nine-tenths of image quality is decided before the chamber closes. This is the part of microscopy that never makes the brochure.

1The rule that governs everything

The microscope images the surface you give it. Not the surface the sample had in service, not the surface you meant to prepare — the one that actually goes under the beam, fingerprints, solvent residue, smeared metal and all. Every artifact you introduce during prep will be rendered at five thousand times magnification with beautiful clarity, and the instrument will not distinguish your artifacts from your science.

So the guiding principle is subtractive: at every step, ask what this step could add to or remove from the surface, and whether you would recognize that change later. The best prep is usually the least prep that solves the actual problem.

2Four questions before you touch the sample

Is it conductive? Metals and carbons image as-is. Oxides, polymers, glass, ceramics, and biological material will accumulate charge under the beam and repay you with drift, streaks, and glowing edges. Your answer determines whether you coat, drop the voltage, or run low vacuum — section 4.

Will it survive vacuum? Anything wet, oily, or volatile is a problem — for the image and for the pump. Damp samples must be dried (properly, not just left out; see biological prep below). Outgassing samples like fresh epoxy mounts need full curing time. When in doubt, weigh the sample, desiccate overnight, weigh again.

Will it survive the beam? Polymers, pharmaceuticals, and delicate biological structures can melt, crack, or shrink under a focused beam at high voltage. If your material has a glass transition temperature below your coffee's temperature, plan for low kV and low current from the start.

What do you actually need to learn? Topography tolerates rougher prep than composition. If EDS is the goal, a flat, uncoated-or-carbon-coated surface matters far more than cosmetic polish. If fracture analysis is the goal, the fracture surface is sacred — see the fractography note below. Decide the question first; it changes the prep.

3Mounting well

The mount has three jobs: hold the sample still at nanometer scale, connect it electrically to ground, and present it at a useful height and orientation. Most "mystery" image problems — drift that outlasts settling time, banding, intermittent focus wander — are mounting problems wearing a disguise.

Carbon tape, used with respect

Double-sided conductive carbon tape is the default mount for good reason, and it earns its bad reputation only when misused. Its failure modes are predictable: it outgasses when fresh from a long-sealed roll (press it down, give it a few minutes in air, or pre-pump it); heavy samples creep on it, appearing as slow unidirectional drift at high magnification; and small particles sink into the adhesive, half-burying the very features you wanted. For particles below a few microns, prefer a polished stub, a silicon chip, or a filter mount over bare tape.

Silver paint and other commitments

Colloidal silver (or carbon) paint gives a rigid, well-grounded mount for bulk samples: a dot under the sample, a thin bridge from the sample's top edge down to the stub if the sample is thick or coated only on top. Let it dry fully — wet paint in the chamber is a pump-oil-and-regret situation. For samples that must come back off intact, remember paint is semi-permanent; plan removal before you commit.

Cross-sections and awkward geometry

Mount cross-sections in spring clips or purpose-made holders rather than improvising with tape at right angles. For polished mounts in epoxy or Bakelite, the mount itself is an insulator: bridge the sample surface to the stub with a stripe of paint or foil tape, or the beautifully polished specimen will charge in the middle of its beautifully insulating puck.

4The conductivity problem

An electron beam delivers charge; the sample must get rid of it. When it can't, incoming electrons pile up, the local surface potential climbs, and the image degrades in characteristic ways — abnormal brightness, sudden image shifts, distortion, and in bad cases the beam is effectively deflected by the sample itself. You have three tools, used alone or in combination.

Tool one: a conductive coating

Sputtering a few nanometers of metal gives the surface a path to ground and a generous supply of secondary electrons. Thickness discipline matters: aim for the thinnest coat that stops charging — commonly 3–10 nm. Too thick and you are imaging your coating, not your sample; fine surface texture disappears under a metal blanket, and sputter grain becomes visible at high magnification.

Tool two: lower the voltage

At low kV (roughly 5 kV and below, sample-dependent) the balance between electrons in and electrons out can approach equilibrium, and many insulators image acceptably uncoated. The cost is resolution and signal; the benefit is a pristine, uncoated surface. Modern field-emission columns make this a far better trade than it used to be.

Tool three: low-vacuum mode

Variable-pressure and low-vacuum modes leak a controlled amount of gas into the chamber; gas molecules ionize and neutralize surface charge, letting you image insulators with no coating at all. Full-size variable-pressure columns from JEOL and Hitachi offer this, and it has become a headline feature of desktop instruments like the SNE-Alpha, where it pairs naturally with quick-turnaround work. The costs: secondary-electron imaging gives way to backscatter or specialized detectors, and the gas skirt scatters the beam — fine for imaging, but it degrades the spatial fidelity of EDS, since scattered electrons excite X-rays well away from the point you think you're analyzing.

5Prep by material

Metals and alloys

For microstructure: section (abrasive saw with coolant, or slow diamond saw), mount, grind through successive grits, polish to the finish your question requires, clean scrupulously — solvent wash, ultrasonic bath, dry with clean compressed gas — and etch only if you need grain contrast in SE. For BSE work, a well-polished unetched surface often shows the microstructure by atomic-number contrast alone, with no etchant artifacts to argue about.

For fractography, the rule is different and absolute: never touch, clean-with-abrasives, or "improve" a fracture surface. Cut well away from the fracture, protect the surface during sectioning, and remove oils with solvent only. The fracture surface is the evidence; everything you do to it is evidence tampering.

Polymers and soft materials

Polymers combine the two hard problems: they insulate and they are beam sensitive. Coat them (Au/Pd or Pt; err thin), image at low kV (1–5 kV) and modest probe current, and don't park the beam — focus and stigmate on a sacrificial area, then move to fresh ground for the real image. Melting, bubbling, or a spreading dark rectangle means the beam is winning: drop kV, drop current, scan faster, or spread the dose over a larger field. For internal structure, cryo-fracture (snap after liquid-nitrogen immersion) gives cleaner cross-sections than cutting, which smears.

Powders and particles

The whole game is dispersion: individual particles, well separated, firmly attached. The dry route — touch a clean stub gently to a small, spread sample, or puff a whisper of powder over adhesive from a distance, then tap and blow off everything loose — beats dipping a stub into the jar, which produces avalanche terrain no one can measure. The wet route: suspend a few milligrams in a volatile carrier (isopropanol, ethanol; water only with patience), sonicate briefly, drop-cast onto a polished stub or silicon chip, let it dry undisturbed. Loose particles are not a cosmetic problem — they end up in the column. If it can be blown off, it was going to come off.

Biological material

Biology is mostly water, and vacuum is mostly not. Air-drying delicate specimens collapses them — surface tension at the receding water line flattens cells like tents in a storm. The classical route: chemical fixation (buffered glutaraldehyde), graded dehydration into ethanol (30 → 50 → 70 → 90 → 100%), then either critical-point drying, which sidesteps surface tension entirely, or HMDS as the quicker bench substitute for sturdier specimens. Then mount, sputter-coat, and image gently at low kV. Hard biological material — bone, shell, teeth, insect cuticle, plant seeds and awns — is far more forgiving: often desiccation, a coat, and modest expectations will do.

Semiconductors and cross-sections

Silicon cleaves; use that. A scribe and a controlled snap along a crystal axis gives a mirror cross-section through many device stacks with zero smearing — often better than a polished section for a quick look. Handle wafers by edges with clean tweezers; every fingerprint is a carbon contamination map at EDS time. For site-specific cross-sections through a particular feature, mechanical polishing with progressively finer media works down to a point; past it lies focused ion beam territory, which is its own guide.

Fibers, filters, and membranes

Mount filters flat and whole — punch a disc, tape the perimeter, paint a grounding bridge to the stub. Fibers want to move under the beam: secure both ends. Loaded filters (air sampling, water filtration) are particle prep and filter prep at once: coat lightly and image before anything has a chance to migrate.

6Troubleshooting charging

Charging announces itself in dialects. Learn to read them:

7Quick reference

8Further reading

Goldstein et al., Scanning Electron Microscopy and X-Ray Microanalysis (4th ed., Springer) remains the reference the field argues from. For free, structured coursework, myScope (Microscopy Australia) and the University of Cambridge's DoITPoMS teaching packages are excellent. The Microscopy Society of America maintains community resources and local affiliate societies where, it must be said, most real prep knowledge actually lives.