Plasma treatment helps reveal how molecular coatings dope silicon

Semiconductors

EPFL Shows O2 Plasma Ashing Enables Tunable Molecular Doping of Silicon

EPFL researchers show O2 plasma ashing preserves phosphorus while stripping carbon from mixed monolayer doping of silicon, enabling tunable near-surface doping for nanoscale devices.

By
Tom Whitfield
Filed
Channel
Semiconductors
Read
2 min read

Researchers at Ecole Polytechnique Federale de Lausanne (EPFL) have demonstrated that oxygen plasma ashing can strip residual carbon from molecularly doped silicon surfaces while largely preserving the phosphorus dopant — a result that strengthens mixed monolayer doping (MMLD) as a controllable route to ultrashallow junctions for nanoscale and quantum devices.

The team, led by Pol Torres-Vila, grafted mixed monolayers of the phosphorus-containing molecule allyldiphenylphosphine (ADP) onto silicon, diluting it with the nondopant molecule 1-undecene to tune the dopant concentration available at the surface before thermal diffusion. They then applied O₂ plasma ashing after grafting to remove carbon residues. X-ray photoelectron spectroscopy (XPS) confirmed the treatment cut carbon content while retaining most of the phosphorus.

The findings appear in the journal Small Methods (DOI: 10.1002/smtd.70970), with the study dated 2026.

Work Function Tracks Dopant Density

Kelvin probe force microscopy (KPFM) revealed a systematic decrease in the silicon work function as ADP concentration increased — direct evidence that surface chemistry translates into electronic changes. But the measured surface response depended heavily on how the SiO₂ capping layer was deposited. Evaporated SiO₂ combined with O₂ plasma treatment produced the clearest evolution toward n-type behavior, whereas sputtered SiO₂ caused strong work-function pinning that masked the doping signal.

The capping-layer result carries practical weight. Any fab or lab evaluating monolayer doping by surface-sensitive metrology could misjudge the process simply by choosing the wrong dielectric deposition method, the comparison suggests.

Surface Versus Bulk

Four-point probe and Hall-effect measurements confirmed rising conductivity and carrier concentration with increasing ADP content. However, these bulk-integrated electrical measurements showed no significant difference between plasma-treated and untreated samples. The plasma step matters at the near-surface, where ultrashallow junctions actually form, but its effect disappears once current integrates through the wafer thickness.

That distinction is the study's central methodological lesson: evaluating monolayer-doped silicon requires combining surface-sensitive techniques such as XPS and KPFM with bulk electrical characterization, because each probes a different part of the doping profile.

Why It Matters

Controlling dopant concentration near the silicon surface is increasingly important as junctions shrink. Conventional ion implantation struggles with the abruptness and dose control needed at the nanoscale; MMLD instead uses self-assembled molecular layers to set the dose chemically before a thermal diffusion step drives the dopant in.

The EPFL work shows that small molecules, combined with plasma ashing, make this approach tunable — adjust the ADP-to-1-undecene ratio and the surface dopant budget follows. The researchers point to nanoscale and emerging quantum-device applications as the near-term beneficiaries, where near-surface doping control is most demanding.

The study positions plasma-assisted MMLD with small molecules as a promising, tunable approach to silicon doping, and future work will likely need to reconcile the plasma step's surface-level benefits with diffusion and activation results if the technique is to move from lab characterization toward device integration.

Source: Phys.org

Share this article:

More from Tom Whitfield

Tom Whitfield

Show full bio

Staff writer covering consumer brands and retail at Chip Dispatch.

73 articles

Related articles

« Previous articleNext article »