Fabrication of freestanding Pt nanowires for use as thermal anemometry probes in turbulence measurements

Fully resolved measurements of velocity fluctuations in highly turbulent flows are challenging because of the broad range of active spatial and temporal scales, which grows with Reynolds number. For fixed facility size, reaching high Reynolds numbers requires resolving micrometer-scale structures with turnover times that demand bandwidths up to O(100 kHz) in some flows. The study aims to overcome resolution limits of conventional hot-wire anemometry by fabricating robust, freestanding platinum nanowires with substantially reduced cross section, thereby enabling shorter effective sensing lengths without excessive end-conduction losses and improving thermal inertia and bandwidth. The presented approach combines electron-beam lithography for the nanoscale wire with standard photolithography and a sequence of wet and dry etches to realize freestanding Pt nanowires between SiO2 beams on Si cantilevers, and demonstrates their suitability for turbulence measurements in high-density flows.

Hot-wire anemometry (HWA) offers the best spatial resolution and bandwidth for turbulent velocity measurements by exploiting velocity-dependent convective cooling of a heated wire. Sensor size is set by wire length l, but reducing l increases end-conduction losses unless the wire diameter d is reduced correspondingly, motivating high aspect ratio l/d designs. Traditional Wollaston wires (Pt core clad in Ag, Ag etched away) achieve minimum Pt diameters of ~1 µm. Attempts to push further faced performance issues: Willmarth and Sharma produced 0.5 µm-diameter, 50 µm-long wires with low aspect ratio and high end-conduction; Ligrani and Bradshaw achieved d ≈ 0.625 µm with l ≈ 125 µm (aspect ratio ~200), but length remained large. MEMS-based probes provided finer features but suffered from end-conduction or geometric constraints: Löfdahl et al. had large sensing areas; Jiang et al. fabricated a polysilicon thermal anemometer with good spatial resolution but significant end-conduction losses; Chen et al. developed multicomponent hot-wire probes (50 µm × 6 µm × 2.7 µm) fixed to a wall, unsuitable for conventional turbulence measurements. The NSTAP introduced nanoscale Pt wires (~100 nm thick, ~2 µm wide due to photolithography limits) via combined dry and wet etches and remains state-of-the-art for size. A q-NSTAP using e-beam lithography reduced width to 600–800 nm but with l ≈ 10 µm for humidity sensing, not anemometry, and experienced structural integrity issues from wet SiO2 etching during release. Despite progress, resolution for very high-Re turbulence remains limited, motivating the present freestanding Pt nanowire approach with smaller cross section (300 nm × 100 nm) and sufficient length (70 µm).

Substrate preparation: Conventional 4-inch (100) Si wafers (385 µm thick) were wet thermally oxidized to form a SiO2 layer. Electron-beam lithography (EBL) of Pt nanowires: A ~13 nm Ti adhesion layer was sputtered, followed by Pt sputtering. Pt nanowires were patterned by EBL at 100 kV (Raith EBPG 5150). The wire design targeted ~300 nm width, ~70 µm length, and ~100 nm thickness, with slightly expanded tips to facilitate electrical connection. Choice of Ti facilitates its removal alongside SiO2 in buffered HF (BHF), yielding freestanding pure Pt after release. Formation of Pt connections: Electrical connections to the nanowires were fabricated by standard photolithography and lift-off to define Pt micropatterns that overlap the expanded wire tips. Accurate overlay/alignment was critical to ensure electrical continuity. Backside processing and device base: The wafer frontside (containing Pt features) was protected with photoresist (PR). The backside SiO2 was removed in BHF, leaving the protected frontside SiO2 intact. A PR mask defining the device base was patterned on the backside, hard-baked at 120 °C for 10 min, and Si was etched by ICP deep reactive ion etching (DRIE; Bosch process) to form the base with a negatively tapered profile; this taper informed subsequent holding-bridge design for self-release. Frontside protection and cantilever formation: A PR structure was aligned and patterned on the frontside to fully cover the Pt features; a narrow PR line covered the Pt nanowire to shield it during subsequent etches. The wafer was immersed in BHF (~30 min) to underetch and remove SiO2 beneath the PR line, producing a freestanding PR line with the Pt nanowire adhered. Spin-drying removed trapped liquid without damaging the freestanding PR-supported wire due to hydrophobicity of PR and Si. Frontside DRIE of Si formed the support cantilevers; etching was stopped with ~10 µm Si membrane remaining to avoid backside cooling-gas leak, which could terminate etching and risk overheating the Pt wire. PR removal and final release: The PR covering the Pt nanowire was gently removed via low-power O2 plasma to avoid burning and breaking the wire. PR removal was performed before final device release to prevent brittleness-induced damage. Final isotropic Si etch with XeF2 enabled self-release of the device, leaving a freestanding Pt nanowire suspended between two SiO2 beams on Si cantilevers. Design features: A dedicated holding bridge facilitated gentle device release without wire damage. Process flow minimized EBL usage to the nanowire only; all other structures employed standard photolithography to reduce fabrication time. Imaging and inspection: HR-SEM and optical microscopy verified feature fidelity (wire dimensions, alignment of Pt connections), backside/base etch profiles, and cantilever geometry.

• Successfully fabricated freestanding Pt nanowires with dimensions ~300 nm width, ~70 µm length, and ~100 nm thickness, suspended between SiO2 beams on Si cantilevers. - Robust process integrating EBL (for the nanowire only) with standard photolithography and combined wet/dry etches; a holding-bridge design enabled gentle self-release without damaging the nanowire. - Established operational suitability in constant-current mode for turbulence measurements: produced a stable calibration between output voltage and flow velocity over 0.5–5 m s−1 in SF6 at 2 bar and 21 °C. - Demonstrated signal stability with negligible drift over several hours. - Verified mechanical robustness under high dynamic pressure: nanowires withstood air velocities up to 55 m s−1 at room temperature. - Process considerations: precise overlay of Pt connections to nanowires is critical; DRIE must be carefully terminated (~10 µm residual Si) to prevent overheating; low-power O2 plasma is required for safe PR removal.

The presented fabrication addresses the core limitation in turbulent flow measurements—insufficient spatial and temporal resolution—by reducing the nanowire cross section to 300 nm × 100 nm while maintaining a practical sensing length (70 µm). This reduction decreases end-conduction losses for a given length and lowers thermal inertia, which is expected to enhance frequency response and measurement bandwidth. Approaching unity aspect ratio between width and thickness mitigates angular sensitivity of the velocity signal. The hybrid lithographic approach (EBL limited to the nanoscale wire, photolithography for all other structures) streamlines fabrication without sacrificing nanoscale precision. Release strategies (PR undercut via BHF, careful DRIE control, gentle O2 plasma, and XeF2 isotropic etch) ensure structural integrity and yield freestanding wires with reliable electrical connections. Functionally, stable calibration in a dense SF6 environment (0.5–5 m s−1, 2 bar, 21 °C) and negligible drift over hours confirm suitability for turbulence studies, while survival at 55 m s−1 in air demonstrates mechanical robustness under high dynamic pressures. Collectively, these results validate the approach as a pathway to improved resolution and bandwidth for high-Re turbulence measurements.

This work introduces a robust, efficient process to fabricate freestanding Pt nanowire hot-wire probes with substantially reduced cross section (300 nm × 100 nm) and 70 µm length, achieved by combining EBL for the nanoscale wire with standard photolithography and carefully engineered wet/dry etches and release strategies. The devices provide stable operation in constant-current mode, exhibit reliable calibration in dense-gas flows over 0.5–5 m s−1, show negligible drift over hours, and withstand high-speed air flows up to 55 m s−1. These advances support higher spatial and temporal resolution measurements in turbulent flows. Potential future work includes: quantifying frequency response and bandwidth limits; extending calibration across wider velocity, pressure, and temperature ranges and different gases; optimizing wire geometry (length and cross section) for specific flow regimes; integrating multi-component arrays; and investigating long-term durability and fatigue under extended high-speed operation.

• Affiliations and broader experimental conditions are not detailed here; calibration was demonstrated over 0.5–5 m s−1 in SF6 at 2 bar and 21 °C, so generalizability to other fluids/conditions requires further validation. - Frequency response and bandwidth were not quantified in the provided text. - Fabrication demands precise overlay between nanowires and Pt connections; misalignment risks open circuits. - Process sensitivity: DRIE must be halted with ~10 µm residual Si to avoid overheating; PR becomes brittle post-etch, necessitating careful timing of PR removal; low-power O2 plasma required to prevent wire damage. - Long-term stability beyond several hours and repeated high-speed cycling were not reported.