Liquids relax and unify strain in graphene

The study addresses how supporting environments influence strain and charge doping in graphene and whether liquids can serve as superior, minimally perturbing supports compared to conventional solids. Solid substrates introduce strain/doping heterogeneity that degrades graphene’s intrinsic electronic/optical properties and mobility, especially in CVD-grown films that exhibit spot-to-spot variability due to growth and transfer inhomogeneities. Prior work suggested enhanced mobility for graphene caged between liquids but lacked direct, spatially resolved characterization of strain/doping and had no reports for CVD graphene at liquid–liquid interfaces. The purpose is to determine if liquid supports (water/air and water/oil interfaces) relax strain and reduce doping uniformly, and to test whether such configurations can cleanly reveal chemical modification effects (e.g., hydrogenation) without substrate artifacts.

Background work shows substrates strongly affect graphene via strain and doping, impacting Raman signatures and mobility (numerous refs). CVD graphene on copper exhibits large strain variability due to lattice mismatch and surface inhomogeneity. Suspended graphene can present localized low-strain regions but typically exhibits strain fields from transfer and geometry. Graphene on SiO2 can partially relax strain; graphene on h-BN yields high mobilities and improved conformity. Raman spectroscopy is a key tool to separate strain vs doping via correlated G and 2D peak analysis and to quantify defects via D band intensity. Earlier liquid-related studies reported mobility improvements and ensemble Raman of exfoliated flakes in water but lacked single-flake, spatially resolved assessment of strain/doping on liquid supports, especially for liquid–liquid interfaces.

• Materials and configurations: CVD monolayer graphene grown on 25 μm copper foils (after annealing at 1035 °C). Tested in multiple configurations: as-grown on Cu; transferred to Si/SiO2 wafers; free-standing on quantifoil TEM grids; floating at water/air; and at water/1-octanol and water/cyclohexane interfaces. Some samples underwent hydrogenation via H2 plasma (1 mbar, 10 W) for 10 s or 60 s before characterization.
• Liquid-interface preparation: For water/air, graphene was floated after copper etching in 0.1 M APS and replacement with ultrapure water; immobilization (frame or thin water layer) minimized movement. For water/oil, cyclohexane or 1-octanol was layered over APS during Cu etching to position graphene at the interface; after etching, APS was replaced with ultrapure water.
• Raman spectroscopy: Confocal Raman (WITEC), ≤2 mW laser power; excitation at 457 nm and 532 nm. Objectives: ×100 (Cu, Si/SiO2, free-standing) and ×70 immersion (liquid supports). For each support, 3–10 samples; per sample 10–20 spectra at different spots. Confocal setup enabled resolving graphene bands against strong liquid signals. G (~1585 cm^-1) and 2D (~2700–2730 cm^-1 depending on laser) bands analyzed for frequency and FWHM; D (~1350 cm^-1) used to assess defects/hydrogenation.
• Strain/doping analysis: Correlative G vs 2D frequency plots (correlation maps) constructed using non-orthogonal axes for hydrostatic strain and doping (with 2D dispersion of 100 cm^-1 eV^-1) to separate and quantify strain (%) and carrier density (cm^-2) from peak shifts.
• Stability tests: Monitored Raman signatures of graphene-on-water over eight days to assess temporal stability.
• Molecular dynamics (MD) simulations: LAMMPS, NVT at 300 K with Nosé–Hoover thermostat; systems included graphene with water and/or hydrocarbons (1-octanol or cyclohexane), and graphene on Cu. Force fields: AIREBO for graphene, OPLS-based models for solvents; TIP3P for water; Lennard-Jones for graphene–solvent interactions (parameters including C–O epsilon 0.0033869 eV, r0 3.19 Å; no C–H LJ for water). Reactive COMB3 for Cu interactions. Simulations probed how liquids vs metals affect graphene strain and interfacial structure (e.g., capillary layering under graphene).

• Liquids relax strain and reduce doping uniformly:
  • Graphene at water/air and water/oil interfaces exhibits strain values clustered around zero with deviations within ~0.1%.
  • Doping variations on liquid supports are low: typically within 2–3 × 10^12 cm^-2, versus up to ~10 × 10^12 cm^-2 on Cu and free-standing, and up to ~8 × 10^12 cm^-2 on SiO2.
• Solid vs liquid supports (from correlation maps):
  • On Cu (as-grown): strain ranges from about −0.8% to +0.7% (broad spot-to-spot variability).
  • Free-standing: strain varies from −0.1% to +1% (predominantly tensile; strong spatial dependence).
  • On Si/SiO2: narrower strain range, approx. −0.3% to +0.2%.
  • On liquids (water/air, water/cyclohexane, water/1-octanol): tightly clustered near zero strain and low, uniform doping.
• Raman peak positions and widths:
  • 2D band positions are narrowly distributed for liquids with averages ~2727 cm^-1 (457 nm) and ~2690 cm^-1 (532 nm); G band less sensitive to strain but also shows narrower distributions on liquids.
  • FWHM(2D) on liquids appears similar to solids, likely due to higher spectral noise from graphene motion on liquid surfaces rather than intrinsic strain inhomogeneity; peak maxima positions remain reliable.
  • FWHM(G) on liquids comparable to Cu and larger than on SiO2 and free-standing, indicating intrinsic charge density is not significantly affected by Cu or liquid supports.
• Stability: Raman signatures and statistical strain/doping distributions for graphene-on-water remained unchanged after 8 days floating.
• Different liquids: Water/air, water/cyclohexane, water/1-octanol, and D2O/air all yield small deviations from zero strain (within 0.1–0.2%) and low doping (2–3 × 10^12 cm^-2). Minor differences (e.g., slightly more uniform on water/1-octanol vs water/cyclohexane) correlate with interfacial tension and association but overall effects are similar and small.
• MD simulations: Fluids, lacking shear modulus, enable strain-free conditions in graphene; metals (e.g., Cu) with large shear modulus promote lattice matching and strain. Hydrocarbons can form molecular layers under graphene, possibly inducing residual corrugation and minor residual strain.
• Hydrogenation case study:
  • For 10 s H2 plasma: h-G on water shows much narrower D peak position distribution (ωD ≈ 1368 ± 1 cm^-1) compared to h-G on Cu (≈ 1376 ± 5 cm^-1). ID/IG ratio is less variable on water, reflecting uniform hydrogenation; on Cu, broader distributions indicate substrate-induced strain/doping contributions to D peak.
  • Correlation maps: h-G/water exhibits smaller doping variability (within ~1 × 10^13 cm^-2) than h-G/Cu (~1.5 × 10^13 cm^-2); strain on water ~ −0.2% to 0%, on Cu ~ −0.6% to +0.1%.
  • Hydrogenation increases doping relative to pristine graphene on water: G/water ~0–5 × 10^12 cm^-2 vs h-G/water ~2–12 × 10^12 cm^-2 (10 s) and ~5–5 × 10^12 cm^-2 (60 s; text indicates elevated range). Strain remains near zero for both G/water and h-G/water.
  • For 60 s H2 plasma: trends persist; h-G/water D peak and ID/IG indicate higher and less uniform hydrogenation vs 10 s; Cu-supported data remain inconclusive due to large strain/doping variability.
• Overall: Liquid supports minimize substrate-induced artifacts, enabling accurate assessment of chemical modifications (e.g., hydrogenation) via Raman, isolating defect-related signals from strain/doping effects.

The findings demonstrate that liquid interfaces provide a structurally adaptive, contamination-minimizing, and mechanically soft support that inherently relaxes strain and limits doping in CVD graphene. Correlated Raman analysis shows that both the magnitude and spatial variability of strain and carrier density are dramatically reduced on liquids compared to solid supports or free-standing configurations, thereby revealing graphene’s intrinsic spectral signatures. This uniform low-strain, low-doping environment allows clean separation of defect-related Raman features from substrate-induced effects, as evidenced by the hydrogenation study where the D band on water reflects true chemical functionalization rather than strain/doping artifacts. MD simulations rationalize the observations: liquids’ vanishing shear modulus prevents imposition of interfacial strain, while metal substrates with finite shear modulus promote strain through lattice interactions. These results are relevant to accurate characterization of graphene chemistry and physics and suggest that liquid supports can improve reliability in monitoring structural modifications and may support higher-performance device studies where substrate perturbations must be minimized.

The study establishes liquids as effective, minimally perturbing supports for graphene. Confocal Raman spectroscopy of CVD graphene at water/air and water/liquid interfaces reveals near-zero strain with minimal spatial variation and low, uniform doping, independent of liquid chemistry. The 2D and G bands retain stable, narrow distributions on liquids, and the approach enables accurate assessment of chemical modifications (e.g., hydrogenation), where the D band reflects true defect chemistry without substrate-induced artifacts. Practical advantages include simple, contamination-free handling, high reproducibility, and minimal sample-to-sample variability. Future work could extend liquid-supported Raman methodologies to other functionalizations, liquid chemistries, and two-dimensional materials, and integrate liquid supports into in situ electrochemical or spectroscopic platforms to further decouple intrinsic material responses from substrate effects.

• Raman measurements at liquid interfaces are technically challenging; increased spectral noise from graphene motion on liquid surfaces can broaden apparent 2D linewidths, potentially reducing precision in width analysis (though peak positions remain reliable).
• Conventional (non-confocal) Raman is generally insufficient to resolve monolayer graphene on liquids due to strong liquid signals; specialized confocal setups are required.
• Free-standing graphene and CVD samples exhibit inherent variability due to growth and transfer history; comparisons to exfoliated graphene may yield different absolute strain/doping ranges.
• Comparison to h-BN is limited to CVD graphene on CVD h-BN; extrapolation to mechanically exfoliated graphene on h-BN may show even lower strain/doping.
• Minor residual strain may arise from hydrocarbon layering/corrugation under graphene, as suggested by MD, which could depend on liquid composition and experimental conditions.