TEM · Beam effects

Electron beam contamination: why a dark square appears where you scanned

The first picture shows a 10 µm field scanned for 1 min: an even dark square, as measured scanned squares look at first. Scan longer to see a frame grow. Then plan a session and test the fixes. It applies to SEM, TEM and STEM.

1 Why the dark square forms 2 Where your dose goes 3 What stops it Takeaways Questions References

Why does a dark square appear where I scanned in the SEM?

Hydrocarbon molecules (carbon and hydrogen) move along the surface into the beam, where electrons break them and bond the pieces into a carbon-rich film.

Top view: 2.4 times wider than the scanned field; darker means a thicker carbon film. Graph: film along the middle line. Model values fitted by Müller et al. (2024) to a parked round beam: 20 keV, carbon film.
Beam
A 21 µm square (or 3 × the field) flooded with the beam beforehand.
Film at the centre
Film at the edge (outer 5%)
Dose (electrons per Ų)
Molecules left in the middle

Try it: drag Scan time from 1 min to 30 min. The edge film grows from 0.44 nm to 4.5 nm and the centre from 0.40 nm to 2.3 nm, so the graph shows a frame. Measured scanned squares were also even at first and framed after long scans [2].

Why does the dark square appear exactly where I focused?

Dose is electrons per area, D = I t/(e A). At ten times the magnification, the same current falls on a hundredth of the area.

Picture: the area after the session; dashed blue: your image; garnet: where you focused. Bars: dose of each step. Film: the model of section 1, run for each step and added.
Field you focus at (step 3)
Focus on
Watch the session step by step
Dose where you focused
Dose at your image centre
Focus step ÷ one overview frame
Film at the edge of the darkest square (model)

Try it: press Play, then pick A spot next door. The dose at your image centre drops from 378 to 3.93 e/Ų, and the dark square leaves the dashed outline.

How do I reduce hydrocarbon contamination in the SEM and TEM?

Each box changes the model for a 1 µm field scanned for 20 minutes; grey text says what was measured.

Graph: average film in the 1 µm square. Grey dashed: no measures; garnet: yours. Squares: the field after 20 min. The model frames more strongly than measured squares [2], so compare the reductions.
Measured: about 70% less in the first 6 min, 60% after 20 min [2]. Model set to match.
Measured: clearly less, but less than plasma cleaning [2].
Measured: almost 100% less for 6 min on one SEM, little on another [2]; very effective in a STEM [5].
Measured: liquid nitrogen cooling stopped it in a STEM [5, as summarised in 2].
Measured: the beam also removes carbon; at a suitable flux the film stopped growing [8].
Not in the model: wear gloves and keep solvents away; handling is one source, with the pumps and chamber parts [7]. Heating graphene to 800 °C in the microscope prevented deposition [9].
No measures, after 20 min
With your measures
Less at 6 min
Less at 20 min

Try it: tick Plasma clean: 70% less at 6 min and 60% at 20 min, as measured. Then tick only Beam shower: 46% at 6 min falls to 24% at 20 min, because molecules drift back.

What to take away

Usually a carbon film, not your sampleThe beam fixes hydrocarbons into carbon where it scanned. In one SEM study, a 1 µm square scanned for 10 min (20 keV, 50 pA) gained 13 nm of carbon one day and 40 nm another.
Dose goes as one over the areaAt the same current and time, a 1 µm field gets 10 000 times the dose of a 100 µm field. So the square sits where you focused.
Clean, shower, focus next doorPlasma cleaning cut contamination by about 70% in one SEM study. Focus next door and your image keeps 3.93 e/Ų instead of 378.
More detail: the equations, the numbers and the limits

The model. n is the surface density of mobile hydrocarbon molecules, as a fraction of N0, the density they reach with the chamber gas. It follows ∂n/∂t = D∇²n − σn + η(1 − n): surface diffusion, use by the beam, and supply from the gas (Müller et al. 2024, Eq. 2). Utke et al. (2008) and Toth et al. (2015) review the same kind of rate equation for beam-induced deposition. The beam term is σ = σc · 2δ · I/(eA). Here δ is the secondary electron yield, and 2 counts both faces of a thin film. Each molecule used adds one carbon atom of a 1.6 g/cm³ film.

The numbers come from the fit of Müller et al. (20 keV, amorphous carbon film). D = 2 × 10−15 m²/s, σc = 2 × 10−20 m², δ = 0.08, η = 0.002 s−1 and N0 = 4.5 × 1019 m−2. So one full layer of molecules makes 0.56 nm of carbon, and about 3 e/Ų uses up about 60% of the molecules under the beam, if no new ones arrive. Hugenschmidt et al. (2023) estimated D = 1.5 × 10−14 m²/s, a slider stop in section 1.

Two limits [4]. Electron-limited: σ is small next to the supply, and the film grows evenly with current. Molecule-limited: the middle is used up, growth follows the supply, and molecules from outside are caught at the edge (a frame or ring). The readout uses above 60% and below 25% molecules left in the middle.

Dose. D = I t/(e A), e = 1.602 × 10−19 C; 1 e/Ų = 100 e/nm² = 1.602 mC/cm².

How it is solved. Finite volumes on a grid over a quarter of the beam area, backward Euler in time (stable for any step). A strip-shaped beam matches the exact steady state to 0.1%.

Limits. The model spreads the scan evenly in time, and its values were fitted to a parked round beam. It runs out of molecules sooner than scanned squares did. Measured 1 µm squares at 20 keV stayed nearly even for 10 min at 50 pA, framed only after long scans, and thickened with current up to about 190 pA [2]. The authors explain the even squares by molecules refilling each spot between passes of the scan. In the model, a 1 µm square at 100 pA frames within a minute. From 10 to 200 pA, its mean film after 10 min only rises from 5.3 to 6.3 nm. Measured squares gained 13 to 40 nm in 10 min on different days [2]; the model gives 6 nm. One energy (contamination grows as the energy falls to about 1 keV [2]); no charging or heating. Plasma in section 3 is set to match [2]. The shower leaves 30% of the molecules, as Müller et al. estimated for their own shower [1]. Bake, cooling and water are illustrative. Section 2 runs each step on a fresh surface.

Questions people ask

Why is there a dark rectangle in my SEM image?

Usually a thin carbon film, laid down where the beam scanned before, often where you focused. The electrons break hydrocarbon molecules on the surface and bond the pieces into a carbon-rich film [2]. In one SEM, a 1 µm square scanned for 10 min at 20 keV and 50 pA gained 13 to 40 nm, depending on the day [2].

Is SEM contamination carbon?

Mostly. The beam breaks hydrocarbons and cross-links them into a carbon-rich film [2]. Postek (1996) names three sources: the pumps, chamber parts such as stage lubricants and O-rings, and the sample with its preparation and handling.

How do I stop carbon contamination in the SEM?

Keep hydrocarbons off the sample and the dose off your area: gloves, plasma cleaning of sample and holder, a beam shower, and focusing next door. In one SEM, plasma cleaning cut contamination by about 70% in the first 6 minutes [2]. Cleaned TEM specimens stayed clean for several hours [6].

Does plasma cleaning damage my sample?

It can. The plasma cleans by removing carbon, so it also attacks carbon support films and organic material. Mitchell (2015) says plasma cleaning can be used on carbon support films, but with care. Hugenschmidt et al. call it effective if the sample can withstand it.

What is a beam shower?

You first flood a much larger area around your region with the beam, for example for 1 to 10 min. The beam fixes the mobile hydrocarbons there, so fewer can drift in. Hugenschmidt et al. scanned 21 × 21 µm for 10 min at 20 keV; the effect fades as molecules diffuse back.

Why does contamination form a ring in STEM?

A small or parked beam uses up molecules faster than they arrive. New ones are fixed at the beam edge before they reach the middle. Müller et al. (2024) saw ring-like contamination under a parked round beam about 1.2 µm wide at 112 pA and 20 keV. Pick Parked round beam in section 1.

What is electron dose in e/Ų?

It is the number of electrons that hit each square ångström: D = I t/(e A). For example, 1 nA for 1 s on a 1 µm square gives 62.4 e/Ų. One e/Ų equals 100 e/nm² or 1.6 mC/cm², so the 1.8 C/cm² used by Hugenschmidt et al. is about 1100 e/Ų.

Is it contamination or beam damage?

Contamination adds a carbon film that thickens with time where the beam was. Beam damage changes the sample itself: atoms move, bonds break, mass is lost (see the beam damage lab). A practical check (a suggestion, not a published test): plasma clean or beam-shower, then scan a fresh area the same way. If the new mark is much weaker, it was mostly contamination, because cleaning cuts contamination [2].

Related: Beam damage in the TEM · SEM imaging basics · Magnification vs resolution · EELS thickness calculator · FIB specimen prep artifacts · EBSD sample preparation · STEM detectors guide · EDS in the SEM

References

Show the 10 references
  1. E. Müller, K. Adrion, M. Hugenschmidt and D. Gerthsen, Reaction-diffusion study of electron-beam-induced contamination growth, Ultramicroscopy 264, 113995 (2024). doi:10.1016/j.ultramic.2024.113995. Model and default values.
  2. M. Hugenschmidt, K. Adrion, A. Marx, E. Müller and D. Gerthsen, Electron-beam-induced carbon contamination in STEM-in-SEM: quantification and mitigation, Microscopy and Microanalysis 29, 219 to 234 (2023). doi:10.1093/micmic/ozac003. Measured effects of cleaning, baking and showering.
  3. I. Utke, P. Hoffmann and J. Melngailis, Gas-assisted focused electron beam and ion beam processing and fabrication, Journal of Vacuum Science and Technology B 26, 1197 to 1276 (2008). doi:10.1116/1.2955728
  4. M. Toth, C. Lobo, V. Friedli, A. Szkudlarek and I. Utke, Continuum models of focused electron beam induced processing, Beilstein Journal of Nanotechnology 6, 1518 to 1540 (2015). doi:10.3762/bjnano.6.157. Growth regimes.
  5. D. R. G. Mitchell, Contamination mitigation strategies for scanning transmission electron microscopy, Micron 73, 36 to 46 (2015). doi:10.1016/j.micron.2015.03.013
  6. T. C. Isabell, P. E. Fischione, C. O'Keefe, M. U. Guruz and V. P. Dravid, Plasma cleaning and its applications for electron microscopy, Microscopy and Microanalysis 5, 126 to 135 (1999). doi:10.1017/S1431927699000094.
  7. M. T. Postek, An approach to the reduction of hydrocarbon contamination in the scanning electron microscope, Scanning 18, 269 to 274 (1996). doi:10.1002/sca.1996.4950180402
  8. M. Toth, C. J. Lobo, M. J. Lysaght, A. E. Vladár and M. T. Postek, Contamination-free imaging by electron induced carbon volatilization in environmental scanning electron microscopy, Journal of Applied Physics 106, 034306 (2009). doi:10.1063/1.3187926
  9. O. Dyck, S. Kim, S. V. Kalinin and S. Jesse, Mitigating e-beam-induced hydrocarbon deposition on graphene for atomic-scale scanning transmission electron microscopy studies, Journal of Vacuum Science and Technology B 36, 011801 (2018). doi:10.1116/1.5003034
  10. M. Ilett et al., Analysis of complex, beam-sensitive materials by transmission electron microscopy and associated techniques, Philosophical Transactions of the Royal Society A 378, 20190601 (2020). doi:10.1098/rsta.2019.0601.
Cite this page: Tripathy, Manisha. “Electron Beam Contamination Lab.” untethered atom, 2026, https://untetheredatom.com/tem/electron-beam-contamination.
BibTeX 
@misc{tripathy2026beamcontamination,
  author = {Tripathy, Manisha},
  title  = {Electron Beam Contamination Lab},
  year   = {2026},
  howpublished = {\url{https://untetheredatom.com/tem/electron-beam-contamination}},
  note   = {Interactive web tool}
}
Last updated 3 November 2026.