Junhao Cheng

Lab researchAt the bench

Moving liquids with sound

Huang Acoustofluidics Laboratory, Duke University · Summers 2025 and 2026

Eight projects between the bench and the market

Bench experiments, devices, measurementsMarket prices, sourcing, portfolios

  1. Acoustofluidic devicesDuke · 2025–26
  2. Respiratory diagnosticsSEFH · 2025–26
  3. Dual-ion batteriesUT Dallas · 2026
  4. Waste-to-hydrogenSEFH · 2023–24
  5. Host-city modelHiMCM · 2025
  6. Battery price covarianceIndependent · 2026–
  7. Anode sourcingCATL · internship
  8. Asgard Assets portfolioWharton · 2025–26

Acoustofluidics uses sound waves to move liquids and particles inside chips whose channels are measured in micrometers. The Huang lab builds sensing and diagnostic tools around it: detecting molecules, reading a colorectal-cancer liquid biopsy through DNA methylation, and barcoding single cells for sequencing.

I spent two summers there, working across three of those projects.

A hexagonal microfluidic chip on a pale background with the tip of a pen pointing at its center
A fabricated PDMS chip.
Two stacks of clear plastic trays holding batches of microfluidic devices on a cutting mat
Batches of devices ready for testing.

The lab’s three projects

  • Acoustic reaction circuits for molecular sensing: molecular sensing on a chip, with sound used to speed up transport to the sensing surface.
  • A digital acoustofluidic platform for colorectal-cancer liquid biopsy, paired with machine learning on DNA-methylation data.
  • Droplet acoustofluidics for barcoding single cells ahead of high-throughput sequencing.
  • The lab reports a readout in 15–20 minutes and a sensitivity of 0.3 copies/µL for its sensing platform. These are laboratory figures, not a clinical validation.

What I did

  • Fabricated the PDMS devices: more than 500 chips and over 10 wafer molds.
  • Tested them under the optical microscope, simulated acoustic streaming in COMSOL, and analyzed reaction-efficiency and detection data in Origin.
  • For the liquid-biopsy project, built a methylation matrix of roughly 5,000 × 20,000 and implemented the machine-learning analysis on it.
  • Analyzed the sensing platform’s diagnostic performance for my own science-fair poster on respiratory diagnostics, which draws on a subset of this work.
  • Took part in the single-cell sequencing project.