Sliding Speed Effects on Tribochemical Wear of Oxide‑Free Silicon

Abstract

Understanding how sliding speed governs the tribochemical wear of oxide‑free single‑crystalline silicon is essential for optimizing ultra‑precision surface finishing. We report the speed‑dependent nanowear of freshly etched silicon against SiO₂ microspheres in humid air (60 % RH) and deionized water. When the contact pressure is insufficient to trigger silicon yield, the presence of water molecules facilitates a tribochemical reaction that progressively reduces wear volume logarithmically with increasing speed, eventually plateauing. Transmission electron microscopy and Raman spectroscopy reveal that the dynamic rupture and reformation of interfacial Si_substrate–O–Si_tip bonds underpin this behavior.

Background

Material wear can be mechanical—fracture, plastic deformation, or viscous flow—or tribochemical, driven by stress‑assisted bond breaking and sometimes chemical corrosion. Single‑crystal silicon is a cornerstone of semiconductor manufacturing, and chemical mechanical polishing (CMP) remains the most effective method for achieving atomically smooth substrates. In CMP, material removal before silicon yields is predominantly tribochemical.

Because CMP involves a complex mix of pad, slurry, load, and speed, many studies simplify the system by examining silicon against a single SiO₂ microsphere. Earlier work has identified a tribochemical pathway in which interfacial bonding bridges transfer mechanical energy to the silicon substrate, promoting atom removal. However, these studies largely used silicon with a native oxide layer, which markedly alters wear behavior. Few investigations have addressed truly oxide‑free silicon, the state relevant to CMP after oxide removal.

Our goal was to clarify how sliding speed influences tribochemical wear of oxide‑free silicon in humid air and deionized water, providing insights that could enhance CMP efficiency.

Methods

We employed p‑type Si(100) wafers, removing the native oxide with 40 % HF for 2–3 min followed by ultrasonic cleaning in methanol, ethanol, and DI water. The resulting surface, terminated with Si–H, had an RMS roughness of 0.12 ± 0.02 nm over a 500 × 500 nm area and a static water contact angle of 82 ± 2°, indicating moderate hydrophobicity.

Using an AFM (SPI3800N, Seiko), we slid a 1.25 µm radius SiO₂ microsphere mounted on a cantilever (k = 10.5–13.8 N/m) against the silicon surface at room temperature, 2 µN load, 200 nm scratch amplitude, and 100 cycles per test. Sliding speeds ranged from 0.08 to 50 µm/s, under 60 % RH or in DI water.

Post‑test, wear scars were imaged with a sharp Si₃N₄ tip (R ≈ 10 nm) in vacuum (<10⁻³ torr). Cross‑sectional TEM (Tecnai G2, FEI) and Raman (RM2000, Renishaw) analyses quantified subsurface damage and chemical changes, respectively.

Results and Discussion

Sliding Speed‑Dependent Nanowear in Aqueous Environments

AFM images (Fig. 1a,b) show clear material removal after 100 cycles in both humid air and DI water. Wear volume decreased logarithmically with speed and plateaued at ~2 × 10⁴ nm³ in air and ~5 × 10⁴ nm³ in water beyond ~8 µm/s.

AFM images and cross‑sectional profiles of wear scars at 0.08–50 µm/s in humid air (a) and in water (b). Volume versus speed (c). Load = 2 µN, amplitude = 200 nm, cycles = 100.

In air, the Si surface, terminated with Si–H, can adsorb water molecules, facilitating the formation of interfacial Si_substrate–O–Si_tip bonds that enable tribochemical removal. In DI water, the higher water concentration lowers the energy barrier for hydrolysis, yielding more Si–OH groups and consequently higher wear volumes.

Sliding Speed‑Dependent Nanowear in Dry Air

Under the same 2 µN load, the contact pressure (~1 GPa) is far below silicon’s yield stress (7 GPa). Instead of material removal, hillocks formed (Fig. 2a), originating from stress‑induced amorphization. Hillock volume decreased with speed, indicating incomplete crystalline‑to‑amorphous transformation at high speeds. This behavior differs from the tribochemical wear observed in humid air and water, where removal dominates.

Hillock formation in dry air (a). Hillock volume versus speed (b). Load = 2 µN, amplitude = 200 nm. Inset shows a typical hillock cross‑section.

TEM Observation of Wear Tracks

High‑resolution TEM of wear scars at 0.08 µm/s and 50 µm/s (Fig. 3) revealed well‑ordered silicon lattices beneath the worn surface, with no amorphization or dislocations. Scar depths were ~11 nm and ~2.3 nm, respectively, supporting a tribochemical mechanism where Si_substrate–O–Si_tip bonds mediate atom removal.

High‑resolution TEM of wear scars at 0.08 µm/s (a) and 50 µm/s (b) in humid air. Insets show scar depths of ~11 nm and ~2.3 nm.

Raman Analysis of Dehydration and Hydrolysis

Raman spectra (Fig. 5) of wear debris displayed pronounced O–Si–O and Si–OH peaks relative to the pristine surface, indicating that dehydration and hydrolysis reactions contribute to tribochemical wear. The relative intensities of these bonds increased with wear, confirming their role in bond formation and breakage during sliding.

Raman spectra of (a) original Si and (b) wear debris. Normal load = 1 N, cycles = 2000.

Mechanism of Sliding Speed‑Dependent Tribochemical Wear

We propose that interfacial Si_substrate–O–Si_tip bridges form under mechanical stress and water presence. At low speeds, sufficient contact time allows bridge reformation via dehydration, strengthening the bond and enabling material removal. As speed increases, reduced contact time limits bridge reformation, diminishing the tribochemical reaction. Beyond ~8 µm/s, a dynamic equilibrium between rupture and reformation yields a constant wear volume.

Schematic of interfacial states at increasing sliding speed.

Conclusions

We demonstrated that tribochemical wear of oxide‑free silicon against SiO₂ microspheres decreases logarithmically with sliding speed and stabilizes in both humid air and deionized water. TEM confirms subsurface integrity, while Raman shows concurrent dehydration and hydrolysis. The observed behavior can be modeled by the kinetics of interfacial Si_substrate–O–Si_tip bond formation and rupture. These insights advance the fundamental understanding of silicon CMP and suggest that controlling hydrolysis (e.g., via slurry pH) can optimize polishing efficiency.

Abbreviations

AFM

Atomic force microscope

CMP

Chemical mechanical polishing

DI water

Deionized water

RMS

Root‑mean‑square

TEM

Transmission electron microscopy

ToF‑SIMS

Time‑of‑flight secondary ion mass spectrometry