Skip to content

Chemical nature of the surface and subsurface region in nitrogen plasma terminated polycrystalline diamond studied by electron spectroscopy

FieldValue
JournalThin Solid Films
Published2026-10-01
AuthorsMohan Kumar Kuntumalla, Arsène Chemin, Maria M. Brzhezinskaya, A. Hoffman
DOI10.1016/j.tsf.2026.141041

Microwave N₂ plasma produced the least near-surface disorder on hydrogen-terminated polycrystalline diamond. A simple depth model gave an sp² layer of 1.0 ± 0.5 Å after microwave treatment, compared with 11.0 ± 0.5 Å after non-damaging RF plasma and 18.0 ± 0.5 Å after damaging RF plasma. The study also found treatment-dependent nitrogen bonding, diamond crystallinity, and oxygen uptake.

The study used polycrystalline diamond films deposited on 1 × 1 cm² p-type Si (100) by hot-filament chemical vapor deposition. A 99/1 vol% H₂/CH₄ mixture was used for 100 min, and the resulting films were 2 to 3 μm thick. Before treatment, the samples were annealed in situ at 600 °C to remove oxygen adsorbed during transfer into a UHV system.

RouteNitridation or implantation conditionsBonding evidenceSurface result
MW(N₂) plasma1.47 kPa, 600 W, 15 min, estimated 100 to 150 °COne N K-edge pre-edge peak near 399.4 eVLowest sp² disorder among plasma routes
RF(N₂), nondamage9.3 Pa, 14 sccm, 40 kHz, 36 W, 30 min at room temperatureC-N and C=N componentsDiamond overtone retained
RF(N₂), damage4.0 Pa, 11 sccm, 40 kHz, 36 W, 30 min at room temperatureBroad N(1s) peak with C-N and C=N statesDiamond overtone absent
N₂⁺ implantation200 eV, 3 to 4 × 10¹⁴ N₂⁺/cm², room temperatureC-N and C=N states with no detected N-HHighest near-surface crystallinity

The 9.3 Pa RF route was designated nondamage, while the 4.0 Pa route was designated damage. H-Di(100) and O-Di(100) served as reference surfaces for assigning carbon and oxygen states.

Conventional XPS used Mg Kα photons at 1253.6 eV. Its C(1s) and N(1s) mean free path of 15 to 20 Å averaged surface and subsurface signals. HREELS used 8.4 eV primary electrons at a 55° incidence angle.

At BESSY II, variable photon energies changed the depth sampled by each core level. C(1s) photoelectron kinetic energies of 50, 150, 250, and 415 eV corresponded to diamond inelastic mean free paths of 7.7, 8.9, 11.2, and 15.1 Å. C(1s) at 335 eV and N(1s) at 450 eV both gave about 50 eV photoelectrons and an approximately 7 Å mean free path. C K-edge and N K-edge NEXAFS used total electron yield with a -30 V retarding potential.

The reported relative C(sp3)/C(sp2) spectral-component ratio was 6.59 after MW plasma. It was 0.26 after non-damaging RF plasma and 0.09 after damaging RF plasma at 335 eV. At 700 eV, the corresponding ratios rose to about 17, 3.5, and 0.4. The increase with sampling depth identifies sp³ carbon as the bulk phase and sp² carbon as a surface phase.

HREELS detected the diamond optical phonon overtone near 300 meV after LEII, MW plasma, and non-damaging RF plasma. It was absent after damaging RF plasma. The authors ranked near-surface diamond crystallinity as LEII, then MW plasma, then non-damaging RF plasma, then damaging RF plasma. The C(1s) peak width increased from about 1.2 eV on H-PCD to about 1.3 to 2.5 eV after nitrogen treatment.

HREELS found an N-H stretching peak near 420 meV on every plasma-treated surface. The implanted sample had no 420 meV peak and showed only a weak feature near 200 meV. The authors therefore placed implanted nitrogen in subsurface layers without N-H bonding.

N K-edge NEXAFS resolved separate C-N and C=N components near 399 and 400 eV after RF plasma and LEII. The 400 eV component was stronger for damaging RF plasma and LEII. MW plasma produced one pre-edge peak near 399.4 eV, associated with C-NH(ads) and/or C-N(ads). Synchrotron N(1s) widths increased from 1.7 eV for LEII to 1.9 eV for MW, 2.0 eV for non-damaging RF, and 2.4 eV for damaging RF.

The 700 eV XPS surveys gave 5.6 at.% nitrogen and 3.6 at.% oxygen after MW plasma. LEII gave 1.5 at.% nitrogen and 5.8 at.% oxygen. Non-damaging RF gave 4.8 at.% nitrogen and 6.2 at.% oxygen. Table 1 lists 9.4 at.% nitrogen and 10.6 at.% oxygen for damaging RF, although the nearby text gives 10.2 at.% oxygen. The H-Di(100) reference measured 99.5 at.% carbon and less than 0.5 at.% oxygen. O-Di(100) measured 6.6 at.% oxygen.

These atomic percentages have comparative meaning because nitrogen and oxygen were localized near the surface. Oxygen uptake ranked from highest to lowest as damaging RF, non-damaging RF, LEII, and MW plasma. No high-binding-energy N(1s) component indicated N-O bonds, and NEXAFS found no detectable NOx species within the measurement sensitivity. The oxygen was therefore assigned mainly to COx groups on surface and defect sites.

C K-edge NEXAFS also showed broad structure from 286.5 to 287.0 eV, assigned to C=N, and a shoulder near 288.5 eV, assigned to C-N. Similar oxygenated features appeared in the O-Di(100) reference, so the authors could not exclude some oxygen contribution to these peaks.

Agreement with earlier work and measurement limits

Section titled “Agreement with earlier work and measurement limits”

The depth comparison agrees with the authors’ earlier MW nitrogen-plasma study. That work found nearly full nitrogen saturation, two dominant bonding configurations, and about 5% coverage of graphene-like islands. In situ annealing to 700 °C caused no detected nitrogen diffusion into the first atomic layers and preserved their crystallinity. The present work extends that analysis to RF plasma and low-energy implantation. The two RF conditions also agree with earlier XPS results, where higher nitrogen pressure produced less damage despite lower nitrogen concentration.

All measurements were ex situ. Samples were stored in airtight boxes in a low-pressure desiccator for up to two weeks. Direct exposure to ambient air during beamline loading lasted up to 10 min.

The sp² thickness estimates used an exponential model for a flat, homogeneous surface and a 7 Å mean free path. Polycrystalline facets produce a broad mean free path distribution and different local plasma interactions at grain boundaries. The reported thicknesses therefore represent beam-scale averages, not local disorder around individual crystallites.

The stronger C=N contribution after LEII remains unresolved. The authors call for further study of whether implanted nitrogen bonds to defects or changes local carbon bonding. The paper also did not fabricate or measure NV centers or quantum sensor performance.

MW nitrogen plasma is the most favorable route in this dataset when low sp² disorder and one dominant nitrogen bonding state are the selection criteria. LEII retained the strongest diamond character and introduced subsurface nitrogen without detectable N-H, but its mixed bonding requires further study. RF plasma cannot be treated as one process, because its pressure and flow settings changed the carbon phase, nitrogen uptake, and oxygen adsorption.

Surface qualification should use the final terminated and air-exposed state. Variable-energy XPS tracks the carbon phase with depth, while NEXAFS can distinguish nitrogen bonding more clearly than the N(1s) peak alone. For near-surface NV work, this combination can screen termination chemistry before device fabrication.

For near-surface NV quantum sensing, the material requirement is hydrogen-terminated polycrystalline diamond that can accept nitrogen plasma termination or 200 eV N₂⁺ implantation while limiting sp² disorder. 6CCVD’s diamond materials catalog includes MPCVD-grown polycrystalline diamond plates and wafers from 0.1 to 500 micrometers thick, with substrates up to 10 mm. Engineering support can help define starting-material and finishing requirements for a termination evaluation, but the paper used HFCVD films and does not directly characterize MPCVD stock.

View Original Abstract

Nitrogen-modified diamond surfaces are increasingly investigated for near-surface NV quantum sensing. Plasma termination and low-energy implantation followed by ambient exposure lead to heterogeneous nitrogen-bonding configurations, a mixed near-surface C(sp³)/C(sp²) region, and co-adsorption of hydrogen and oxygen species. In the context of the increasing diversity of nitrogen incorporation techniques, this study compares surface defects, bonding configurations and electronic states produced by three distinct approaches on polycrystalline diamond (PCD) surfaces: exposure to microwave (MW) and radio-frequency (RF) nitrogen plasmas, and low-energy nitrogen ion implantation (LEII) followed by ambient exposure. High-resolution electron energy loss spectroscopy is used to probe variations in C-H and N-H vibrational modes following nitridation and subsequent exposure to ambient conditions. Depth-resolved high-resolution X-ray photoelectron spectroscopy, performed at multiple photon energies, reveals that MW(N 2 ) plasma treatment induces the lowest degree of surface graphitization. From a simple model which assumes flat and homogeneous PCD surfaces, the thicknesses of the disorder (sp 2 hybridized carbon) layers are 1.0 ± 0.5, 11.0 ± 0.5, and 18.0 ± 0.5 Å for the MW(N 2 ), RF(N 2 ) nondamage and RF(N 2 ) damage nitrogen plasma treated surfaces, respectively. Near-edge X-ray absorption fine structure spectroscopy resolves the nature of CN bonding, indicating predominantly C-N species upon MW(N 2 ) plasma exposure, whereas RF(N 2 ) plasma and LEII treatments result mainly in C-N and C=N bonding configurations. Adventitious oxygen mostly exists in CO x configuration on diamond surface/defect sites and no detectable NO x species are observed within the measurement sensitivity.