Raman Mapping of Monolayer Graphene on Silicon Micro‑Ring Resonators: Unveiling Doping and Strain Effects

Abstract

We present a detailed Raman mapping study of monolayer graphene’s G and 2D bands on silicon strip‑waveguide micro‑ring resonators (MRRs). By correlating peak positions, intensities, and widths with spatial location, we demonstrate that graphene remains electrically intrinsic when suspended, but becomes moderately p‑doped (≈3×10¹² cm⁻²) where it contacts the silicon/SiO₂ waveguide. This pinning of the Fermi level down by ~0.2 eV is attributed to trapped interfacial charges, not to geometry, as confirmed by identical results for 10‑µm and 20‑µm radius devices. The G‑peak asymmetry observed over the waveguide is explained by phonon stiffening from doping and by reduced degeneracy due to out‑of‑plane wrinkling of suspended graphene.

Background

Integrating graphene with silicon photonics promises breakthroughs in photodetection, modulation, and biosensing, thanks to CMOS‑compatible, low‑cost fabrication. However, the transfer process can introduce defects and unintentional doping, altering the material’s band structure and performance. Raman spectroscopy, sensitive to strain, doping, and defect density, is the go‑to technique for characterizing these changes. Its three principal peaks—D (≈1350 cm⁻¹, disorder), G (≈1580 cm⁻¹, in‑plane phonon), and 2D (≈2700 cm⁻¹, second‑order of D)—provide clear fingerprints of layer number, strain, and carrier concentration.

Graphene’s two‑dimensional electronic structure allows electrostatic gating to tune its optical absorption via Pauli blocking, enabling ultrafast (GHz) modulation. Previous work on graphene‑on‑silicon waveguides reported widely varying absorption coefficients, suggesting that transfer chemistry and interfacial quality critically influence device behavior. Our study addresses this gap by spatially mapping Raman signatures across graphene‑coated MRRs.

Methods/Experimental

The MRRs were fabricated in a 220‑nm SOI wafer with 2‑µm BOX, using 335‑nm wide strip waveguides to maximize overlap with graphene. Two racetrack resonators—10‑µm and 20‑µm radius—were studied. Devices were cleaned (acetone, IPA, DI water, NMP), oxygen‑plasma etched for 40 s, and then transferred with a polymer‑mediated wet CVD graphene process. Patterning and oxygen plasma etch defined graphene coverage; a 270 °C anneal in reducing atmosphere removed residual resist.

Raman mapping was performed at room temperature using a Horiba LabRAM HR Evolution spectrometer (600 g/mm grating, 633‑nm He–Ne laser, ×50 objective, NA = 0.75). The laser power density was <2 mW to avoid heating. A 120 × 120 point grid (0.25 µm step) produced high‑resolution maps of G and 2D peak positions, intensities, and widths. Lorentzian fits extracted peak parameters; the instrument resolution was ~4.6 cm⁻¹ (from Si peak).

Results and Discussion

Initial single‑point spectra (514‑nm excitation) confirmed monolayer graphene: weak D peak, strong symmetric 2D peak, G at ~1587 cm⁻¹ (≈ 1581.6 + 11/(1+n)^1.6), matching literature for n = 1. Figures 2c–d show G and 2D peak maps: both exhibit up‑shifts of ~11 and ~8 cm⁻¹ where graphene rests on the waveguide, versus suspended regions.

Because strain would shift the 2D peak ~6× the G peak, the comparable shifts of G and 2D indicate that strain is negligible. Instead, the direction and magnitude of the shifts match p‑type doping: G rises for both electron and hole doping, while 2D rises only for moderate hole doping (~15 cm⁻¹ at 3×10¹³ cm⁻²). Empirical relations (1) and (2) yield a Fermi level shift of ≈−0.2 eV (hole concentration ≈3×10¹² cm⁻²) on the waveguide, while suspended graphene remains intrinsic.

Equation (1): \[\left|E_{F}\right|\times 41.5=\Delta\omega_{G}=\omega(G)-\omega_{0}(G)\] Equation (2): \[\left|E_{F}\right|\times 31.5=\Delta\omega_{2D}=\omega(2D)-\omega_{0}(2D)\]

G‑peak asymmetry, modeled with a double Lorentzian (G⁺ and G⁻), reveals a ~25 % narrowing of G⁺ (10→7.5 cm⁻¹) and ~35 % narrowing of G⁻ (20→13 cm⁻¹) when supported, consistent with phonon stiffening from doping and reduced degeneracy due to localized wrinkling of suspended graphene.

The intensity ratio I_2D/I_G drops from ~3 (suspended) to ~2.5 (supported), indicating increased carrier‑phonon scattering, though the change is modest compared to the G‑peak shift. Integrated intensity ratio A_G/A_2D further corroborates higher hole concentration on the waveguide, allowing extraction of E_F via Eq. (3): \[\sqrt{A_{G}/A_{2D}}=C\left[\gamma_{e-ph}+|E_{F}|f\left(e^{2}/\varepsilon v_{f}\right)\right]\]

Spatial line scans (Fig. 4) show a Gaussian‑shaped Fermi level peak (~−0.2 eV) aligned with the waveguide, identical for both resonator radii, confirming substrate‑induced doping rather than geometric effects.

Vector decomposition of G vs. 2D peak positions (Fig. 5) places suspended data near the origin (intrinsic) and supported data along the p‑doping vector, with negligible strain contribution. This analysis yields an average carrier density of (2–3)×10¹² cm⁻², in line with Eq. (4): \[n=(E_{F}/\hbar\nu_{F})^{2}/\pi\]

Conclusions

Monolayer CVD graphene integrated on silicon micro‑ring resonators exhibits intrinsic behavior when suspended but undergoes moderate p‑doping (~3×10¹² cm⁻²) where it contacts the silicon/SiO₂ waveguide, shifting the Fermi level down by ~0.2 eV. Raman G‑peak asymmetry reflects both doping‑induced phonon stiffening and curvature‑induced degeneracy lifting. These substrate‑induced effects must be considered when designing graphene‑based photonic devices, particularly modulators and sensors, to ensure accurate interpretation of graphene’s optoelectronic properties.

Abbreviations

CCD

Charge‑coupled device

CEA‑LETI

Commissariat à l’énergie et aux énergies alternatives – laboratoire d’électronique des technologies de l’information

CMOS

Complementary metal‑oxide‑semiconductor

CVD

Chemical vapour deposition

DR

Doubly resonant

FWHM

Full width at half maximum

MRR

Micro‑ring resonator

NMP

N‑Methyl‑2‑pyrrolidone

Si

Silicon

SiO₂

Silicon dioxide

SWCNT

Single‑walled carbon nanotube