Mounting, sputter coating, drying biological specimens and cutting cross sections, plus a diagnostic table that traces the common image faults back to the preparation step that caused them.
Most service calls that begin with the words the microscope is not resolving properly end at the sample. The column is aligned, the vacuum is good, the detector is fine, and the specimen is a poorly mounted powder with a coating that never reached the stub. Sample preparation is where an expensive instrument is either used or wasted, and it is the part of the workflow that receives the least training and the smallest budget.
What follows is the preparation sequence in the order it happens, with the failure mode attached to each step.
Electrons arriving at the sample must leave it. On a conductor they flow to earth through the stub and the stage. On an insulator they accumulate, the surface takes up a negative potential, and that potential deflects the incoming beam. What you see on screen is not subtle once you know it: bright glaring patches that flare and fade, dark bands, whole scan lines displaced sideways, features that jump between frames, and an apparent drift that is not drift at all.
There are five ways out and you will usually combine them. Apply a conductive coating. Reduce the accelerating voltage, because every material has a voltage at which the electrons leaving balance the electrons arriving, and working near it reduces charging without any coating. Use variable pressure mode if the instrument has it, so gas ions neutralise the surface. Scan faster and average many frames instead of one slow scan, since charge builds during the dwell. And check the conduction path, which is the step most often skipped.
The mount has one job: hold the specimen still and give every part of the imaged surface a continuous electrical path to the stub. A coating that stops at the edge of the sample is electrically useless, because the charge has nowhere to go.
Coating adds a conductive metal layer a few nanometres thick. The choice of metal is a trade between signal and grain size. Gold gives an excellent secondary electron yield and is the default for routine work, but it forms an island structure that becomes visible at high magnification and can be mistaken for sample texture. Gold with palladium gives a finer grain, platinum and iridium finer still, and chromium is used where the finest surface detail matters. For elemental analysis, coat with carbon instead, because metal coatings put lines into the region of the spectrum you probably want to read.
Thickness is where preparation goes wrong most often. The coat needs to be thick enough to conduct and thin enough not to hide what you came to see. A heavy coat cracks under the beam into a pattern that looks like dried mud and is regularly presented in draft papers as a material feature. If your smallest feature of interest is at the scale of tens of nanometres, an over enthusiastic coat has already removed it.
Water cannot go into the chamber, and taking it out is where soft specimens are destroyed. Air drying pulls the structure apart through surface tension as the liquid front retreats, and the collapsed, shrunken result is easy to spot and impossible to fix afterwards.
The established sequence is chemical fixation, buffer washing, then dehydration through a graded series of ethanol, then removal of the ethanol without ever crossing a liquid to gas boundary. Critical point drying with carbon dioxide is the reference method. Chemical drying with hexamethyldisilazane is a practical alternative that many laboratories in Bangladesh adopt because it needs no pressure vessel, and for sturdy specimens such as leaves, seeds and insect parts it gives perfectly publishable results. Freeze drying is the third route. Whichever you use, the specimen must be genuinely dry before it goes in, and beam sensitive material should be imaged at reduced voltage and current with fast scans.
A cross section is the only honest way to measure a layer thickness or to look inside a composite, and it is the preparation most likely to introduce features that were never in the material. Mount in resin, using a cold curing resin for anything heat sensitive, then grind through a decreasing sequence of abrasives and polish. Clean ultrasonically between every step, because a single coarse particle carried forward will scratch the surface you spent an hour producing.
For four months of the year the ambient conditions work against you. A hygroscopic powder left on the bench while the coater warms up will take up water in minutes. Paper, textiles, soil, cement and most biological material behave the same way. The consequence is a long pump down, a chamber that will not reach its normal pressure, contamination deposited on the sample under the beam, and a queue of irritated users who assume the microscope is faulty.
Three habits fix nearly all of it. Store prepared stubs in a desiccator with silica gel that is regenerated on a schedule, rather than a jar with grey gel that has been saturated since last year. Vent the chamber with dry nitrogen where the instrument supports it, so the chamber is not filled with monsoon air at every sample change. And keep the coater, the desiccator and the microscope in the same continuously conditioned room, so a dried sample does not cross a humid corridor on its way to the column.
| What you see | Most likely cause | What to do |
|---|---|---|
| Bright flaring patches that move or fade | Charging: the conduction path is incomplete | Recoat, add a painted bridge to the stub, reduce accelerating voltage, try variable pressure mode |
| Scan lines displaced sideways in bursts | Charging, or an external magnetic field if the pattern is regular and follows the mains | Change scan speed: charging changes with it, a mains field does not |
| Steady drift in one direction | Specimen not held firmly, or the stage still settling after exchange | Remount rigidly, allow settling time, avoid tall unsupported mounts |
| A dark rectangle appearing where you were just imaging | Hydrocarbon contamination polymerised by the beam | Handle with clean tweezers and gloves, keep samples covered, plasma clean if available |
| A fine cracked pattern over the whole surface | Coating too thick, or damaged by beam current | Recoat thinner, reduce current, image at lower voltage |
| Surface bubbling, shrinking or perforating | Beam damage on polymer or biological material | Lower voltage and current, coat, use fast scans with frame averaging |
| Pump down far longer than usual | Wet sample, uncured paint, or a plastic that outgasses | Dry in a desiccator, cure paint before loading, avoid non vacuum plastics |
| Grainy texture at high magnification that looks the same on every sample | The coating, not the specimen | Change to a finer grained coating metal or reduce thickness |
Kept as a laminated sheet beside the instrument, that table settles most of the arguments that otherwise become a service call. The preparation bench, the coater, the desiccator and a stock of stubs, tabs and paint cost a small fraction of the microscope, and they decide what fraction of the microscope you actually get to use.