On-Chip Magnetic-Levitation Architecture

• On-chip magnetic-levitation architectures are microfabricated platforms that generate tailored 3D magnetic field gradients to stably levitate microscopic particles for quantum control and precision sensing.
• These systems employ superconducting quadrupole traps, double-loop traps, and AC magnetic Paul traps, with tunable mode frequencies from tens of hertz to tens of kilohertz and varied operating temperatures.
• Finite-element modeling confirms excellent agreement with experimental data on trap frequencies, quality factors, and dissipation mitigation, ensuring reliable integration with chip-based readout electronics.

On-chip magnetic-levitation architectures use lithographically patterned microfabricated coils or electrodes to produce field landscapes capable of stably levitating single microscopic particles above solid substrates. These platforms provide three-dimensional field gradients—magnetostatic or time-dependent—tailored for ultra-high mechanical quality factors, motional ground-state cooling, and integration with chip-based readout and feedback electronics. The architectures discussed here include superconducting quadrupole traps for Meissner-state diamagnetic spheres, double-loop planar coil traps, and AC magnetic Paul traps for ferromagnetic particles. Designs accommodate picogram to microgram masses and operate at cryogenic to room temperatures, with mode frequencies spanning tens of hertz up to tens of kilohertz, supporting quantum control and precision inertial sensing (Latorre et al., 2021, Latorre et al., 2022, 2002.03868, Janse et al., 2024).

1. Geometric Configurations and Fabrication

Architectures are categorized into three principal geometries: anti-Helmholtz-coil (AHC) quadrupole traps, planar double-loop (DL) traps, and planar AC magnetic Paul traps (MPT).

• AHC Quadrupole Trap:
  • Implements two vertically displaced planar superconducting coils (typically Nb, thicknesses $\sim$300–1000 nm), patterned on double-side-polished Si chips, separated by 280 µm (Latorre et al., 2021, Latorre et al., 2022). Each coil features 10–20 windings of 30 µm $\times$ 1 µm cross-section, width 2 µm, radial footprint $\sim$0.4 mm with a 200–300 µm central loading aperture. Fabrication uses DC magnetron sputtering, lithography, reactive-ion etch, deep Si etch (Bosch ICP), manual flip-chip stacking, Nb wire-bonding, and precision micro-manipulation for particle loading.
• Double-Loop (DL) Trap:
  • Co-planar concentric Nb loops with opposing currents, diameters 12–18 µm, wire widths $\sim$1 µm, single-layer Nb (300 nm) on Si, electron-beam lithography plus ICP-RIE. A DL trap is fabrication-simplified, active for larger particles ($\sim$10 µm diameter) (2002.03868).
• Planar AC Magnetic Paul Trap (MPT):
  • Two planar loops on PCB or silicon, inner radius $R_1=0.7$ mm, outer radius $R_2=1.4$ mm, track width 0.3 mm, thickness 35 µm (PCB) or 1–5 µm (on-chip Cu/Au). Concentric AC currents $(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$ null homogeneous AC field components, a central aperture $\sim$1 mm for particle injection. Potential on-chip implementation uses lift-off, damascene, DRIE, and anodic bonding (Janse et al., 2024).

Materials are selected for high $j_c$ (Nb $\times$03–5$\times$1–$\times$2 A/m$\times$3), with trap geometry, current capacity, and local surface field bounded by Meissner-state limitations ($\times$4, $\times$5).

2. Physical Principles and Trap Modeling

Levitation exploits strong field gradients for diamagnetic or paramagnetic force transduction. Quadrupole field patterns—static or oscillatory—produce spatial minima trapping particles in 3D.

• Meissner-State Diamagnetism (Superconducting Sphere):
  • The potential $\times$6 ($\times$7 for ideal Meissner state) yields a restoring force $\times$8, with small oscillations governed by $\times$9 (Latorre et al., 2021, 2002.03868).
• AC Magnetic Paul Trap (Ferromagnetic Cube):
  • An oscillating quadrupole field $\sim$0 produces a time-averaged ponderomotive potential, reducing to Mathieu equations for secular and micromotion dynamics, with parameters $\sim$1, eigenfrequencies $\sim$2 (Janse et al., 2024).

Finite-element modeling (FEM, COMSOL Multiphysics) solves the Maxwell–London equations, incorporates finite $\sim$3, demagnetization, and flux quantization. For AHC and DL traps, FEM accounts for all coil, aperture, and particle effects, delivering quantitative predictions of $\sim$4, levitation force, and stiffness. Meshing and convergence studies yield sub-1% agreement in trap frequencies, with mesh sizes $\sim$5–$\sim$6 DOF for 3D (2002.03868).

3. Experimental Parameters and Mode Characterization

• Static Quadrupole Superconducting Trap:
  • Typical coil currents $\sim$7–$\sim$8 A, stably levitating spheres in the $\sim$9–$\sim$0 µm range. Trap gradients $\sim$1 T/m, trap depths $\sim$2 J ($\sim$3$\sim$4 K). Measured trap frequencies for 48–50 µm Pb/SnPb spheres at $\sim$5 A: $\sim$6–$\sim$7 Hz, $\sim$8 Hz, $\sim$9–$\sim$0 Hz (Latorre et al., 2022, Latorre et al., 2021).
  • Motional mode quality factors $\sim$1–$\sim$2 limited by eddy current damping. Trap frequencies are tunable $\sim$3 in the 30–160 Hz range. Duffing nonlinearities and inter-mode coupling are quantified using FEM, with quartic coefficients $\sim$4–$\sim$5 m$\sim$6s$\sim$7 (Latorre et al., 2022).
• AHC and DL Traps (Microscale):
  • AHC traps ($\sim$8 µm spheres, $\sim$9 A/m$R_1=0.7$0) produce $R_1=0.7$1 kHz, $R_1=0.7$2 kHz, $R_1=0.7$3 kHz. DL traps (10 µm sphere) yield $R_1=0.7$4 Hz (2002.03868).
  • Trap frequency scaling is captured by $R_1=0.7$5.
• AC Magnetic Paul Trap (Room-Temperature Ferromagnet):
  • 250 µm-edge cube, $R_1=0.7$6 A, $R_1=0.7$7 Hz, yields $R_1=0.7$8 Hz. Librational modes $R_1=0.7$9 Hz. Q-factors peak at $R_2=1.4$0, dominated by eddy currents; vibrational $R_2=1.4$1–$R_2=1.4$2 (Janse et al., 2024).

4. Integrated Readout and Coupling Techniques

Chip-based platforms support integration of superconducting and magnetic readouts.

• DC-SQUID Magnetometry:
  • Superconducting pickup loops (L$R_2=1.4$3 nH), offset by $R_2=1.4$4 µm, coupled to commercial DC-SQUIDs (input $R_2=1.4$5 nH, mutual $R_2=1.4$6 nH). Flux transfer efficiency $R_2=1.4$7; noise floor $R_2=1.4$8 m$R_2=1.4$9/Hz$(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$0, displacement sensitivity $(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$1 nm/Hz$(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$2 (x), $(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$3 nm/Hz$(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$4 (y), $(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$5 nm/Hz$(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$6 (z). Improved matching and persistent-current traps will approach the intrinsic SQUID noise $(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$7 $(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$8/Hz$(+I_\text{trap}, -\xi I_\text{trap}, \xi=2.2-2.4)$9 (Latorre et al., 2022).
• Prospective Hybrid Readouts:
  • On-chip superconducting microwave cavities (sideband cooling, ground-state cooling $\sim$0), vector-resolved inertial sensing arrays, and coupling to NV centers for spin-mechanics (Janse et al., 2024).

5. Dissipation, Stability, and Noise Mitigation

Superconducting traps at cryogenic temperature ($\sim$1 mK–4 K) exhibit minimal intrinsic dissipation. Lifetimes exceed days in the absence of illumination.

• Primary Dissipation Channels:
  • Eddy currents in nearby normal metals, resistive heating of coil tracks, technical heating or flux trapping from optical illumination (trapped particles released after $\sim$2 s continuous exposure) (Latorre et al., 2021, Janse et al., 2024).
  • In AC Paul traps, eddy-current damping dominates over gas damping. Pressure dependence is weak ($\sim$3 barely increases from $\sim$4 bar to $\sim$5 mbar).
• Mitigation Strategies:
  • Use of superconducting microstrips, superconducting shielding (Nb/Cryoperm), mechanical isolation, removal of conductive coatings on particles, careful matching of coil apertures and separation, persistent-current operation to suppress current-noise-induced flicker (Latorre et al., 2022, Janse et al., 2024).

6. Functional Implications and Applications

Chip-based magnetic-levitation enables quantum optomechanics, inertial and force sensing, and fundamental macroscale quantum tests.

• Quantum Ground-State Cooling and Control:
  • FEM models predict chip-scale traps with $\sim$6 enabling MHz-scale cavity coupling; feedback/sideband protocols expected to reach motional ground state in $\sim$7 ms. Low noise and high $\sim$8 open access to macroscale quantum superpositions and tests of collapse models, decoherence, and gravity (Latorre et al., 2022, 2002.03868).
• Precision Sensing:
  • Sensitivity $\sim$9 with $j_c$0 kg, $j_c$1 Hz, $j_c$2 N/Hz$j_c$3 gives $j_c$4/Hz$j_c$5 (Latorre et al., 2022). Chip arrays allow vector-resolved inertial/gravitational sensing and tests of short-range forces.
• Quantum-Magnetic Coupling:
  • On-chip AC MPTs for micron-scale ferromagnetic grains offer MHz librational frequencies, direct magnetic coupling to spins (NV centers), and high phononic coupling strengths $j_c$6–$j_c$7 kHz (Janse et al., 2024).
• Scalability and Integration:
  • Scaling to sub-micron particles requires increasing $j_c$8 to $j_c$9 T/m$\times$00—superconducting tracks or pulsed operation for current density, cryogenic anchoring, and microfluidic or dielectrophoretic on-chip loading are necessary.

Table: Key Experimental Features of Representative Architectures

Architecture	Typical Particle Size	Trap Frequencies	Quality Factor
AHC Quadrupole Trap	$\times$01–$\times$02 µm (Pb/SnPb sphere)	$\times$03–$\times$04 Hz (exp), up to $\times$05 kHz (FEM)	$\times$06–$\times$07
DL Trap	$\times$08 µm (sphere)	$\times$09–$\times$10 Hz (FEM)	Similar to AHC
Planar AC MPT	$\times$11 µm cube (NdFeB)	$\times$12–$\times$13 Hz vib., $\times$14 kHz libr.	Up to $\times$15

Frequencies, materials, and Q-factors as reported in (Latorre et al., 2021, Latorre et al., 2022, 2002.03868, Janse et al., 2024).

---

On-chip magnetic levitation platforms advance precision control of massive quantum objects, offer scalable integration with superconducting electronics, and support multi-dimensional sensing schemes. Their design relies on careful coil geometry, low-dissipation refrigeration, and detailed electromagnetic modeling; levitation regimes, frequency scaling, and noise performance are in quantitative agreement with analytic and finite-element models. Persistent-current operation, microwave-cavity coupling, and hybrid quantum-mechanical protocols are under development, solidifying this architecture's role in next-generation quantum technology and inertial measurement (Latorre et al., 2021, Latorre et al., 2022, 2002.03868, Janse et al., 2024).