Atomistic Insights into Oxidation Mechanisms in Chemical–Mechanical Polishing of Diamond
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Micromachines |
| Published | 2026-09-22 |
| Authors | Kechong Wang, Yunkai Wang, Meng Li, Xiang Li, Chongjun Wu |
| DOI | 10.3390/mi17101107 |
H2O2-assisted CMP produced a CVD single-crystal diamond surface with Sa = 0.256 nm over a 5 µm × 5 µm area after 1 h. The accompanying ReaxFF molecular dynamics model attributes the process to controlled interfacial oxidation through stepwise H2O2 decomposition. Pressure mainly accelerated early activation and reactant consumption, while oxidant availability governed later bond evolution.
ReaxFF model and CMP conditions
Section titled “ReaxFF model and CMP conditions”The reactive model used ideal single-crystal diamond (100) beneath a planar SiO2 patch. An aqueous gap separated the abrasive from the substrate.
The cell measured 28.4 × 28.4 × 45.4 ų and contained 1536 carbon atoms in diamond. It also contained 256 Si atoms and 448 O atoms in the abrasive.
The OH slurry held 100 H2O molecules and 100 hydroxyl radicals. The H2O2 slurry held 100 H2O molecules and 100 H2O2 molecules, giving 2740 and 2940 total atoms, respectively.
Simulated pressures were 10, 50, 100, and 150 MPa at 300 K. The authors state that these nanoscale loads do not correspond directly to nominal macroscopic CMP pressure.
After energy minimization, the system used NVT equilibration with a Langevin thermostat. NVE sliding then followed a 10 ps loading stage, 50 ps equilibration, and 150 ps sliding at 10 m/s.
The time step was 0.1 fs, and the velocity-Verlet algorithm integrated motion. QEq updated atomic charges at every step. A bond-order cutoff of 0.30 defined bonds, and output was sampled every 1000 steps.
Validation used a 10 mm × 10 mm × 1 mm CVD single-crystal diamond (100) sample. The slurry combined 50 nm SiO2 with 30 wt% H2O2 solution at a 1:3 mass ratio.
Polishing ran for 1 h at 25 °C under a 30 N load. Slurry flow was 15 mL/min, while disk and holder speeds were 80 and 60 r/min. The holder also reciprocated radially at three cycles per minute.
Direct oxidation in hydroxyl radical slurry
Section titled “Direct oxidation in hydroxyl radical slurry”The OH system oxidized carbon more directly because radicals attacked mechanically activated surface sites. At 10 MPa, contact remained localized, so C-O and C-OH groups appeared only at scattered surface positions.
Minor C-H formation passivated dangling bonds during equilibration. Sliding caused limited bond reconstruction because shear was insufficient at low pressure.
At 100 MPa, contact area, charge redistribution, and reactive-species consumption all increased. Radical and species populations fluctuated more strongly than at 10 MPa.
The bond analysis compared 10 and 150 MPa. At 150 MPa, C-O bonds formed rapidly during loading and remained high through equilibration and sliding. At 10 MPa, the C-O population stayed low.
Low-pressure sliding increased C-H bonds, but their large fluctuations indicated repeated formation and rupture. Under higher pressure, oxidation occupied carbon sites that otherwise supported hydrogen termination.
Decomposition-limited oxidation in H2O2 slurry
Section titled “Decomposition-limited oxidation in H2O2 slurry”H2O2 oxidation required stepwise decomposition into reactive intermediates, including OH and HO2 radicals. This supply step made the pathway more gradual and spatially selective than direct OH attack.
Under 10 MPa, H2O fell from 100 molecules to about 75 during loading. It then decreased to about 60 during equilibration before stabilizing with small fluctuations.
H2O2 fell below approximately 50 under the same condition. It approached full consumption during sliding, showing that low pressure restricted decomposition without stopping it.
Under 100 MPa, H2O2 was nearly depleted during the 50 ps equilibration stage. Higher pressure promoted decomposition and interfacial contact, accelerating the initial reaction.
Figure 11 shows C-O bonds near 30 after equilibration at 10 MPa. The population later reached a maximum near 75 during sliding.
At 100 MPa, C-O bonds stabilized around 30 to 40 through much of the run. The reported instantaneous count therefore did not scale directly with pressure.
Faster oxidation and oxidant consumption at 100 MPa preceded a plateau as reactive species became limited. At 10 MPa, C-H bonds grew as unoxidized sites remained available. At 100 MPa, C-H stayed extremely low.
The authors describe the H2O2 pathway as controlled surface modification. It caused limited lattice distortion and did not drive extensive subsurface oxidation.
XPS and AFM validation
Section titled “XPS and AFM validation”XPS found a pristine-surface C1s peak at 284.8 eV and no detectable oxygen-related signal. After CMP, an O1s peak emerged near 531.5 eV, supporting interfacial oxidation.
Deconvolution assigned O-H groups at 530.01 eV, C=O groups at 531.96 eV, and C-O groups at 535.18 eV. The assignments show hydroxyl, carbonyl, and single-bond oxygen groups on the polished surface.
AFM over 5 µm × 5 µm gave Sa = 0.256 nm. Directional Ra was 0.239 nm parallel to the polishing grooves and 0.407 nm perpendicular.
The XPS evidence remains qualitative. The authors state that survey-intensity changes cannot establish sp² carbon or amorphous carbon after CMP.
Pressure changes reaction onset more than final bond count
Section titled “Pressure changes reaction onset more than final bond count”Higher pressure increased the number of interfacial contact atoms and decreased mean interfacial separation. These changes promoted earlier charge transfer, C-O formation, and reactive-species consumption.
The effect was strongest during early activation. Oxidant-specific kinetics and reactive-species availability governed later C-O evolution.
Direct OH attack produced rapid, fluctuating oxidation. H2O2 required intermediate generation, which limited the reaction rate and kept structural change moderate.
At low pressure, hydrogenation competed with oxidation. At high pressure, C-H formation fell while oxidation-related bonds appeared earlier or increased.
Sliding renewed contact and exposed fresh carbon sites at both pressures. Shear did not overcome the H2O2 supply limit or cause extensive subsurface oxidation.
Comparison and unresolved issues
Section titled “Comparison and unresolved issues”Prior CMP work reported a 206.4 nm/h removal rate, Ra = 0.6 nm, and a 3.242 nm plastic damage layer with CeO2-containing abrasives. The present study reports neither removal rate nor subsurface damage, so it cannot establish better throughput or lower damage.
Another cited process combined 1168 nm/h removal with Sa = 0.079 nm and a 0.66 nm damage layer. The present Sa = 0.256 nm is higher than that cited value, although process conditions differ.
The simulation used an idealized, unpassivated diamond (100) surface without reconstructed steps, vacancies, or artificial defects. This differs from the chemically terminated CVD surface used for the experiment.
The 28.4 Å lateral cell approximated a flat patch of a 50 nm SiO2 particle. It resolves local interface chemistry rather than full particle contact mechanics.
The authors cite spatial and temporal scale limits. The 150 ps simulation also cannot reproduce the full 1 h polishing experiment.
Experimental validation covered the H2O2 system only. It did not directly test OH chemistry or an experimental pressure series.
Implications for diamond surface engineering
Section titled “Implications for diamond surface engineering”Pressure can shorten the onset of oxidation, but higher pressure can also consume the oxidant faster. The results imply that process control should track oxidant supply and reaction intermediates rather than pressure alone.
The authors suggest that oxidant-controlled surface modification can guide oxidation-assisted polishing. They do not establish an optimal slurry, pressure, or removal rate for a specific device.
Surface engineers can treat C-O growth, C-H suppression, and the XPS O1s signal as indicators of interfacial oxidation. These indicators should be paired with removal-rate and subsurface-damage measurements before process transfer.
Low-damage CMP remains relevant to optical windows, semiconductor devices, precision cutting tools, and quantum information technologies. The current experiment shows low roughness but does not measure subsurface damage.
A scale-up claim requires matched removal-rate and subsurface-damage measurements.
Materials and support
Section titled “Materials and support”The paper’s CMP specimen is 10 mm × 10 mm × 1 mm CVD single-crystal diamond with a (100) surface. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal and polycrystalline plates and wafers up to 125 mm, plus substrates up to 10 mm. For the required orientation and specimen geometry, review custom fabrication support on the engineering support page.
View Original Abstract
Understanding the interfacial oxidation behavior during chemical-mechanical polishing (CMP) of diamond is important for controlling surface chemical modification and subsequent material removal. In this study, reactive force field molecular dynamics (ReaxFF MD) simulations were employed to comparatively investigate the atomistic evolution of interfacial oxidation, charge transfer, chemical species, and bond structures of single-crystal diamond in •OH- and H2O2-containing environments under different pressure conditions. The results show that oxidant chemistry and mechanical loading jointly regulate the kinetics and pathway selectivity of interfacial reactions. Elevated pressure promotes early-stage interfacial contact, surface activation, and oxidant consumption; however, the subsequent evolution of C-O bonds is also governed by oxidant-specific reaction kinetics and the availability of reactive species. In the •OH system, highly reactive radicals interact directly with mechanically activated carbon sites, whereas oxidation in the H2O2 system is additionally constrained by the stepwise decomposition of H2O2 and the generation of reactive intermediates, resulting in a comparatively moderated and decomposition-limited oxidation pathway. The interfacial reactions involve the formation and dynamic evolution of C-O, C-OH, and C-H structures, reflecting the competition between oxidation and hydrogenation. XPS measurements after H2O2-based CMP provide qualitative experimental evidence for the formation of oxygen-containing functional groups on the diamond surface, supporting the occurrence of interfacial oxidation. These findings provide atomistic insight into the coupled effects of oxidant chemistry and mechanical loading during diamond CMP and may assist in the rational design of oxidation-assisted polishing processes.
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →