Why does my SEM image charge and drift?

Charging and drift produce different artefacts and have different fixes, and treating them as one problem wastes weeks. How to tell them apart in two images, and what to change first.

An image looks acceptable at low magnification and falls apart above it: bright streaks that shift between scans, edges that glow, or a field that slides steadily in one direction over a slow scan. These are two separate faults with two separate cures, and groups routinely spend weeks applying the fix for one to the other.

They are different faults with different signatures

Charging is an electrical problem at the specimen. The beam injects electrons faster than the specimen can conduct them away, the surface potential rises, and it deflects the beam. The signatures are abrupt brightness jumps, glowing edges, dark bands, features that appear to move or smear between successive scans rather than progressively, and contrast that changes when you change magnification or scan speed.

Drift is a mechanical or thermal problem. The specimen and the field of view move smoothly and continuously relative to the beam. The signature is a steady shear: a slow scan image is skewed in the scan direction, features stretch, and the whole field has translated when you return to it.

The quickest way to separate them: capture a field, blank the beam or move it away, wait a few minutes, and capture the same field again. If the features have translated but look clean, it is drift. If they are in the same place but the contrast and brightness have changed, or the artefacts move each time you scan, it is charging. Charging can also cause apparent drift, so if both are present, fix the charging first.

Fixing charging

• Drop the accelerating voltage. There is a voltage for every material at which the emitted electrons balance the injected ones and the net charge is close to zero. Working down towards it costs some resolution and removes the artefact, and it is free to try.
• Reduce the beam current or the spot size, and use a faster scan with frame averaging rather than one slow scan. Line averaging and scan rotation both help.
• Use low vacuum or variable pressure mode if the instrument has it. Residual gas ionises and neutralises the surface, which is the fastest route to an image of an uncoated insulator.
• Coat the specimen. A thin conductive layer of gold, gold-palladium, platinum or carbon solves it outright. Use carbon if EDS of light elements matters, because gold and palladium peaks sit where they will interfere.
• Make a real ground path from the surface to the stub. This is the step that gets done badly. Carbon tape under a specimen does not ground its top surface. A stroke of silver paint from the surface down the side to the stub does, and on tall or awkward specimens a clamp does better than tape.
• Use the backscatter detector. Backscattered imaging is less disturbed by surface potential and often gives a usable image when the secondary signal will not.

Fixing drift

• Wait longer after loading. The specimen, the stub and the holder need to reach thermal equilibrium in the chamber. A freshly coated specimen is warm, and one taken from an air-conditioned prep room into a chamber at a different temperature will move for a while.
• Wait after a stage move. Stages settle. Moving a long distance and then imaging immediately at high magnification will show drift that is simply the stage relaxing.
• Check the mount. Double sided carbon tape creeps under load, particularly with a heavy or tall specimen, and it will produce a slow steady drift for as long as the session lasts. Silver paint, a clip or a mechanical clamp removes it.
• Look for outgassing. A specimen carrying moisture, solvent or an uncured adhesive will outgas, move and contaminate. Dry it, cure it properly, or pump it in the airlock for longer before imaging.
• Rule out charging before blaming the room. A slowly charging surface deflects the beam in one direction and looks exactly like mechanical drift.

The causes specific to a Bangladeshi building

• Stray alternating magnetic field. A transformer, a busbar, a lift motor or a UPS on the other side of a wall puts a mains frequency ripple into the image that gets worse at high magnification and slow scan. It looks like a regular waviness rather than random noise. Switching suspect equipment off one at a time during a scan identifies it quickly.
• Air conditioning cycling. A split unit that switches on and off changes the room temperature on a cycle, and the drift follows that cycle. If the drift has a period you can time with a watch, look at the air conditioner before the microscope.
• Generator transfers. A changeover during load shedding produces a step in the image and a disturbance that takes time to settle. Long acquisitions should be scheduled around the known load shedding pattern where there is one.
• Humidity on the specimen. Samples stored in an uncontrolled room during the monsoon carry a water film that outgasses in the chamber, giving both drift and contamination. A desiccator for specimen storage is one of the cheapest improvements available to a Bangladeshi laboratory.

Work in that order and most cases close quickly. Coat the specimen and lower the accelerating voltage, ground the surface properly, then leave it in the chamber long enough to settle. Only when a well-mounted, coated, thermally settled specimen still drifts is it worth measuring the room, and by then you will have a much better case for the money that involves.

Back to all Insights