Casimir forces on suspended thin dielectric membranes covered with graphene

AI summary

Overview

Research area: Quantum physics — specifically the Casimir effect — combined with graphene optoelectronics and nanomechanical force sensing.

Technical level: Advanced. The paper pairs a Lifshitz-formalism calculation of finite-temperature Casimir pressure with a torsion-free cantilever-based force-gradient measurement, and assumes familiarity with Matsubara frequencies, Kramers-Kronig analysis, and the proximity force approximation.

Scope: A combined theoretical and experimental study of how replacing a bulk substrate beneath graphene with a thin suspended dielectric membrane changes the size and temperature dependence of the Casimir force against a gold surface.

What This Paper Is About

Graphene's electrons move roughly 300 times slower than light, which pushes thermally driven Casimir effects to much shorter separations than in ordinary metals and dielectrics — short enough that the force is strong and measurable. But graphene normally sits on a solid substrate, and that substrate suppresses the thermal enhancement. The authors ask whether swapping the substrate for a thin suspended membrane of silicon oxide or silicon nitride restores the enhancement, and they test the idea by measuring the force gradient between a gold-coated sphere and graphene on a suspended silicon nitride membrane.

Key Contributions

1. A quantitative theoretical prediction that thin suspended membranes restore thermal sensitivity. For a gold plate facing graphene on a suspended 15 nm silicon oxide membrane, the Casimir pressure changes by up to about 54% between 10 K and 294 K at a separation near 500 nm, versus only 7% at 479 nm for the same graphene on 285 nm of silicon oxide on a silicon wafer.

2. The same prediction for a fabrication-friendly material. For graphene on a suspended 13 nm silicon nitride membrane, the predicted change reaches 40%, compared with 5% at 495 nm for graphene on solid silicon nitride — the authors handle incomplete optical data for SiN by shifting the Kramers-Kronig-derived permittivity up by a constant 3.562 to recover the known static value of about 8.

3. An experimental realization and measurement. The team fabricated 100 μm by 100 μm suspended non-stoichiometric silicon nitride membranes (13 nm measured by atomic force microscopy) covered with transferred monolayer graphene, and measured the Casimir force gradient against a gold-coated sphere.

4. A record-low force gradient and a route to reduced stiction. After scaling by sphere radius, the measured force gradient is, to the authors' knowledge, the smallest reported in the sphere-plane configuration — about a factor of 5 smaller than earlier experiments using gold-coated surfaces and about 40% smaller than prior measurements on solid dielectric layers covered with graphene.

Main Findings

• Substrates mask the thermal effect. For graphene with chemical potential 165 meV on 285 nm of silicon oxide on silicon, the calculated pressure at 294 K exceeds that at 10 K by only 7% at z = 479 nm; the thick oxide and silicon largely hide the thermal contribution.

• Suspending the membrane restores it. Replacing that stack with a suspended 15 nm silicon oxide layer raises the predicted 10 K to 294 K change to roughly 54% at around 500 nm. The analogous jump for silicon nitride is from 5% (solid) to 40% (13 nm suspended).

• Silicon oxide membranes proved impractical. The material is under compressive stress, the suspended films are fragile and tend to crumble, and graphene transfer onto them often caused breakage; the authors could not produce a sample that was simultaneously intact, clean, and large enough for measurement.

• Silicon nitride was used instead. It is durable, stays flat under tensile stress, and deforms less than 0.3 nm under a 200 pN load spread over a 30 μm radius circle, justifying the flat-membrane assumption.

• Graphene quality was verified. Raman spectroscopy at the membrane center (5 μm spot radius) showed the symmetric 2D peak at 2679 cm⁻¹, confirming monolayer graphene, with a small D (1341 cm⁻¹) to G (1585 cm⁻¹) intensity ratio indicating the layer is largely defect-free.

• Chemical potential was estimated indirectly. From the conductance minimum versus backgate voltage on a separate graphene-on-285 nm-oxide sample, the chemical potential was estimated at 0.165 eV; no equivalent gated measurement was possible on SiNₓ because of weak optical contrast.

• The measured force gradient agrees with theory. Data for gold sphere 1 (radius 38.7 μm) on graphene over 285 nm silicon oxide and for gold sphere 2 (radius 43.4 μm) on graphene over suspended SiNₓ both fall within theoretical expectations at 294 K.

• Thermal effects were not resolved. Measurement uncertainty, driven mainly by incomplete knowledge of the silicon nitride optical properties, thickness non-uniformity (vendor-specified variations up to 1 nm within a membrane, with a ±3 nm band across samples), and possible organic residue from graphene transfer, prevented distinguishing the 294 K prediction from the 10 K one.

• Suspension lowers the force. The force gradient on the suspended nitride membrane is smaller than on thick silicon oxide, as expected, and the reduction is attributed to thinning one of the interacting bodies to roughly 10 nm rather than to any change of sign.

Methodology in Plain English

Theory side. The authors model the system as a graphene-covered dielectric body facing a gold half-space separated by distance z. They compute the Casimir pressure with the Lifshitz scattering formula, summing over discrete Matsubara frequencies ξ_n = 2πnk_BT/ħ and integrating over parallel wavevector, using reflection coefficients for each body. Graphene enters through a frequency-dependent surface conductivity split into intraband and interband terms, with a relaxation rate Γ = 10¹³ rad/s. Gold is treated with a Drude model using ħω_p = 9 eV and ħγ = 35 meV. Silicon oxide's dielectric response at imaginary frequencies comes from a Kramers-Kronig transform of tabulated measured data; silicon nitride's comes from measured data spanning 10¹² to 10¹⁸ rad/s, but because the measured window misses at least one resonance, the transformed permittivity falls short of the known static value, so the authors apply a constant upward shift of 3.562 and restrict the calculation to the frequency region where that shift is valid, terminating before 10¹⁶ rad/s. Sphere-plane forces are converted to parallel-plate pressures using the proximity force approximation, F′ = 2πRP.

Experiment side. Monolayer graphene grown on copper was transferred onto targets using a PMMA support layer, electrochemical delamination of the copper, and a wet transfer, followed by heating to 150 °C to melt the PMMA and acetone/isopropanol/water cleaning. A gold-coated sphere (150 nm of gold) on a gold-coated silicon cantilever (224 μm by 29 μm by 3.4 μm) acted as the force sensor. Two spheres were used: sphere 1 with radius 38.7 μm (resonance about 40963 Hz, quality factor 12053) and sphere 2 with radius 43.4 μm (about 28795 Hz, quality factor 9068). Cantilever vibrations were driven piezoelectrically and read out with a 1310 nm fiber interferometer; a second interferometer at 1550 nm monitored and removed long-term drift in separation. The whole apparatus sat in a bell jar below 10⁻⁶ torr. A phase-locked loop tracked resonance shifts, which are proportional to the force gradient through Δf = CF′(z) with C = −f₀/2k. Calibration of C and the initial separation z₀ came from fitting the electrostatic force at six applied voltages (±100 mV, ±140 mV, ±170 mV) relative to the residual voltage V₀. Fitted C values were (−3.17 ± 0.13)×10³ s kg⁻¹ and (−2.184 ± 0.027)×10³ s kg⁻¹ for the two measurements.

Why This Matters

Research impact. The work connects a theoretical prediction — that thin suspended dielectrics preserve graphene's unusual thermal Casimir behavior — to a real fabricated device, and it identifies precisely which material-characterization gaps (notably silicon nitride optical data and graphene chemical potential on nitride) currently block a definitive thermal measurement. It also supplies a benchmark: the smallest sphere-plane force gradient measured so far, in the authors' assessment.

Applications the authors point to:

• Stable operation of neutral nanomechanical elements held at small separations, where the Casimir force otherwise dominates.
• Alleviating stiction between neutral mechanical components by thinning one interacting body to about 10 nm.
• Design of graphene-on-membrane structures for optomechanical and electron-microscopy-style membrane platforms, building on silicon nitride's existing role in those areas.
• An alternative to the liquid-mediated approach that is otherwise required to obtain repulsive Casimir forces between planar surfaces in full thermal equilibrium.

Industry relevance. Any micro- or nanoelectromechanical device operating with sub-micrometer gaps must contend with Casimir attraction; a demonstrated recipe for substantially reducing it — while keeping the structure flat and robust — is directly relevant to MEMS/NEMS design and to metrology at the nanoscale. The result also matters to the graphene-transfer and membrane-manufacturing supply chain, since the limiting uncertainties were material properties and thickness control rather than the force measurement itself.

Future Directions

• Directly measure the optical properties of the actual silicon nitride used, over a wide frequency range, to remove the need for the constant-shift correction and the arbitrary termination of the Matsubara sum.
• Fabricate the membranes in house to reduce thickness uncertainty, which currently spans about ±3 nm across samples and up to 1 nm within a single membrane.
• Determine the chemical potential of graphene on silicon nitride, since Raman G-peak shift calibration exists for graphene on oxide but not on nitride.
• Substitute other thin dielectrics, such as few-layer hexagonal boron nitride, to further weaken the substrate's influence on the thermal dependence of the Casimir force.
• Reduce or eliminate organic residue from the transfer process, which may trap charge and shift the measured force.

Target Audience

Researchers in Casimir physics and quantum fluctuation phenomena; graphene and 2D-material device physicists; nanomechanics and optomechanics groups working with suspended dielectric membranes; and MEMS/NEMS engineers concerned with stiction and Casimir-induced instability at sub-micrometer gaps. The theory sections suit readers comfortable with Lifshitz/Matsubara formalisms, while the fabrication and measurement sections are accessible to experimentalists in thin-film transfer and cantilever-based force sensing.

Authors’ abstract

We investigate the Casimir force on thin suspended dielectric layers covered with graphene, both theoretically and experimentally. Thermal effects of the Casimir force on graphene occur at separations significantly smaller compared to ordinary metals and dielectrics. At such small separations, the Casimir force is sufficiently strong to be measured experimentally. However, the presence of substrates to support the graphene significantly reduces the enhancement. We show theoretically that replacing the solid substrate beneath graphene with a thin dielectric membrane partially restores this strong thermal dependence. One of the interacting objects is a solid gold plate. The other object is a suspended dielectric membrane made of either 15 nm of silicon oxide or 13 nm silicon nitride, covered with graphene. At a separation of $\sim$500 nm, theoretical predictions yield changes in the Casimir pressure of up to 54 % and 40 % for silicon oxide and silicon nitride, respectively, when the temperature is increased from 10 K to 294 K. Measurements were performed for graphene on a suspended silicon nitride membrane at room temperature, using a gold-coated sphere attached to a cantilever as the force sensor. The measured force gradient is consistent with the theoretical prediction, although the thermal dependence could not be resolved, in part because of uncertainties in the optical properties of silicon nitride. Replacing a bulk substrate with a thin dielectric membrane covered with graphene provides a substantial reduction of the Casimir force, which may facilitate the stable operation of neutral nanomechanical elements at small separations.