Field-Tunable Meissner-Levitated Ferromagnetic Microsphere Sensor for Cryogenic Casimir and Short-Range Gravity Tests

Abstract

Near-field force measurements at submicron separations can probe Casimir effects and hypothetical short-range interactions, but require cryogenic operation and stable, \textit{in situ} control of separation-dependent backgrounds. We propose a self-calibrating quantum force-gradient sensor in which a ferromagnetic microsphere is Meissner-levitated above a type-I superconducting plane, while a bias magnetic field reproducibly tunes the equilibrium gap for in situ separation scans without mechanical approach. The force gradient is encoded as a resonance-frequency shift tracked by a phase-locked loop, and the motion is read out with a SQUID-coupled, flux-tunable microwave resonator that provides adjustable measurement strength without optical heating. Using the input--output formalism, we derive the conditions for reaching the standard quantum limit (SQL) and identify a counterintuitive scaling law: because displacement-to-flux transduction increases with microsphere size, larger microspheres require fewer photons to reach the SQL, enabling a pathway to macroscopic quantum metrology. We quantify the trade-off between suppression of electrostatic patch potentials (via Au coating) and eddy-current dissipation, project force sensitivities of $\sim 10^{-19}\,\rm{N\,Hz^{-1/2}}$ at millikelvin temperatures, and outline protocols to extract Casimir pressure and constrain Yukawa-type deviations from Newtonian gravity over $0.1$--$10\,μ\mathrm{m}$.

Figures (4)

• Figure 1: Overview of the proposed force-gradient sensing platform based on a ferromagnetic Meissner-levitated oscillator (FMLO) integrated with a SQUID--flux-tunable microwave-resonator readout and a phase-locked loop (PLL). (a) Mechanical subsystem: a ferromagnetic microsphere (radius $R$) with a conductive Au coating (thickness $d$) is magnetized by a bias field $B_{\mathrm{ext}}$ and Meissner-levitated above a type-I superconducting Pb plane; the equilibrium gap $h_0$ is tuned in situ via $B_{\mathrm{ext}}$, and small vertical motion is denoted by $x$ so that $h=h_0+x$. A piezoelectric actuator (PZT) beneath the plane provides inertial excitation for coherent oscillation. (b) Readout subsystem: the displacement-dependent pickup flux $\Phi_p(x)$ in a SQUID loop is transferred as $\Phi=\lambda_{\Phi}\Phi_p$ to the readout circuit, tuning the Josephson inductance $L_J(\Phi)$ and hence the resonance frequency $\omega_c(\Phi)$ of the flux-tunable microwave resonator (FTMR); the resulting microwave quadrature/phase modulation is detected at the output. (c) PLL: a phase detector (PD) and loop filter (LF) generate an error signal that steers a voltage-controlled oscillator (VCO) to track the mechanical resonance frequency, enabling frequency-shift readout of the force gradient.
• Figure 2: Radius scaling of the force-noise budget and SQL measurement-strength requirement for near-field FMLO operation (representative parameters: $h_0=1\,\mu\mathrm{m}$, Au thickness $d=100\,\mathrm{nm}$, bath temperature $T=1\,\mathrm{mK}$). (a) Equivalent force-noise amplitude spectral density $\sqrt{S_{FF}}$ evaluated at the mechanical resonance: mechanical noise (solid blue; thermal noise dominated by Au eddy-current damping plus the PZT drive-voltage contribution), quantum readout noise $\bar{n}\le 400$ (solid orange; achievable with an intracavity photon constraint $\bar{n}_{\mathrm{SQL}}\le 400$, include imprecision+backaction noises from the SQUID+FTMR readout), the standard quantum limit (SQL, green dashed), and the total noise (solid red; quadratic sum). (b) Bare coupling $G_0$ (blue) and the optimal coupling $G^{\ast}$ required for SQL operation (orange) versus sphere radius. (c) Intracavity photon number $\bar{n}_{\mathrm{SQL}}=(G^{\ast}/G_0)^2$ required to reach the SQL; under the photon-number cap the readout approaches the SQL only for $R\gtrsim 500\,\mu\mathrm{m}$, whereas for $R\lesssim 30\,\mu\mathrm{m}$ the readout noise exceeds the mechanical noise.
• Figure 3: Projected Casimir-pressure extraction and power-law fitting with the FMLO force-gradient protocol. (a) Casimir pressure $P_C(h_0)$ inferred from the force-gradient spectrum after calibrating the slowly varying magnetic background with large-$h_0$ data and subtracting it; symbols show a representative data set generated using the noise budget (integration time $100\,\mathrm{s}$ per separation), and the solid curve shows the nominal Casimir model used for the illustration. The shaded band indicates the propagated $1\sigma$ uncertainty, including the contribution from background calibration/subtraction; for visibility, the error bars and band are multiplied by $10^{9}$ (data values unchanged). Inset: extracted $P_C$ at larger separations ($1$--$10\,\mu\mathrm{m}$), where the Casimir signal is small and the calibration residual is expected to be consistent with zero within uncertainty. (b) Reduced chi-square $\chi_\nu^2$ versus the fitted exponent $S$ in the model $P_C(h_0)=K h_0^{-S}+B$; the best-fit $S_0$ occurs at the minimum of $\chi_\nu^2$, and $\Delta S$ is obtained from the $\Delta\chi^2=1$ criterion. (c) Scaling of the fitted-exponent uncertainty (standard deviation of $\Delta S$) with the per-point signal-to-noise ratio $\mathrm{SNR}\equiv P_C/\Delta P_C$, highlighting that the precision of the exponent test is set primarily by relative (fractional) pressure uncertainty.
• Figure 4: Projected constraints on Yukawa-type deviations from Newtonian gravity in the sphere--plane geometry, expressed as an upper bound on the coupling $|\alpha|$ versus interaction range $\lambda$ [Eq. (20)]. The red dashed curve ("This work") shows the projected $1\sigma$ sensitivity obtained by fitting the background-subtracted residual force-gradient spectrum to the Yukawa force-gradient model [Eq. (21)] using the FMLO noise budget. Colored curves show representative existing laboratory limits from Casimir-force, isoelectronic, Cavendish-type, and optomechanical experiments (labels indicate references and years); parameter values above a given curve are excluded. Over the interaction-length window shown (tick labels correspond to $\lambda\sim 10^{-7}$--$10^{-5}\,\mathrm{m}$, i.e., $0.1$--$10\,\mu\mathrm{m}$), the FMLO projection suggests potential improvements of order $10^{3}$--$10^{6}$ in $|\alpha|$, assuming near-field backgrounds can be modeled and subtracted at the quoted noise level.