Breath carries the molecular story of your health.
OSCAii reads it — no blood, no needles, just breath.
Dual-comb spectroscopy measures the precise optical fingerprint of each molecule, allowing OSCAii to identify hundreds of species simultaneously in real time — from trace biomarkers at parts-per-billion to major gases at percent-level concentrations.
Molecules absorb light at specific frequencies determined by their chemical structure. A frequency comb acts as a precision optical ruler — at 200 MHz spacing, OSCAii places hundreds of thousands of evenly spaced laser frequencies across the molecular fingerprint region.
When this comb light passes through breath, each molecule absorbs its own characteristic set of frequencies, leaving a distinct pattern in the transmitted light. Reading that pattern reveals which molecules are present and how strongly they absorb — giving both broad spectral coverage and high precision in a single measurement.
Peer-reviewed validation: Xing et al., CLEO 2025, SS122.6 — First compact single-cycle comb empowered dual-band IP-DFG mid-IR combs for breath analysis.
Dual-comb spectroscopy works by combining two frequency combs with slightly different spacings. After one comb passes through the breath sample and both meet on a detector, the optical molecular fingerprints are converted into radio-frequency signals that standard electronics can read in real time.
OSCAii's advantage starts at the light source. Single-cycle pulses generate mid-infrared combs that span key molecular fingerprint windows — covering not just inorganic trace gases, but also the spectral regions where volatile organic compounds and functional-group vibrations leave their strongest signatures. The result is a massively parallel molecular readout: all species measured at once, rather than scanning one molecule or one wavelength at a time.
Peer-reviewed validation: Zhang and Xing, CLEO 2026, STH2H.7 — High-dynamic-range multi-species DCS with real-time coherent averaging, resolving CO, CO₂, NO, and H₂O in mixed gas samples and measuring CO from trace levels to 0.1% at 1 atm.
| Parameter | Value |
|---|---|
| Spectral range (current) | 2–12 µm (Mid-IR) |
| Spectral range (Phase 2) | 2–25 µm |
| Dynamic range (current) | ppb to % — 4+ orders |
| Dynamic range (Phase 2) | ppt to % — 7+ orders |
| Resolution | 200 MHz real-time + sub-Hz with interleaving |
| Acquisition speed | Seconds-scale spectra, application dependent |
| Coherent averaging | 11,000 real-time coherent averages in <10 s |
| Gas interaction length | 32 m (current) |
| Moving parts | No mechanical delay lines |
| Frequency traceability | RF-referenced · TAI-traceable |
| Pulse duration | <7 fs single-cycle |
| Frequency stability | 1×10⁻¹⁸ @ 1 s · 1×10⁻²⁰ @ 1,000 s |
OSCAii is building toward a future where breath analysis offers a fast, continuous, non-interruptive and non-invasive window into the body's molecular health.
Every claim is grounded in physical experiment. These are snapshots from our active development environment — not renders, not simulations.
Signal Detection · Time & Frequency Domain
Signal Detection
Active Detection · Frequency Locked
Frequency-Locked Detection in Progress
The OSCAii DCS platform enables a growing range of molecular detection applications — across clinical diagnostics, pharmaceutical research, and environmental monitoring.
A single frequency-comb acquisition captures many molecular signatures at once, from trace-level biomarkers to major breath gases. Instead of tuning to one target molecule at a time, OSCAii reads a broad spectral window in parallel and separates species by their unique absorption fingerprints.
OSCAii resolves isotopologues — molecules with the same chemical formula but different isotopes — even when rare species appear alongside much stronger dominant absorption lines. By tracking isotope enrichment or dilution relative to natural abundance, it enables isotope-ratio measurements for metabolic tracer studies, drug metabolism monitoring, and non-invasive biochemical analysis.
Disease leaves molecular traces in breath long before clinical symptoms emerge. OSCAii's broadband spectral fingerprinting identifies distinct VOC profiles across biological states — opening a path toward earlier, non-invasive detection across metabolic and respiratory conditions.
OSCAii is built as a platform, not a single-purpose instrument. It begins with dual-comb spectroscopy for real-time molecular fingerprints, extends toward electro-optic sampling for broader mid-infrared coverage, and is designed to incorporate squeezed-light sensitivity and physics-informed machine learning — scaling across diseases, research environments, and clinical deployment as breath analysis matures.
Every existing method forces a tradeoff between depth and accessibility. Breath eliminates it.
Three applications validated in our laboratory. The following results demonstrate the platform's core capabilities using real experimental data from our in-house DCS system.
CO detection from 50 ppb to 1,000 ppm in a single acquisition under 10 seconds — four orders of magnitude, no instrument reconfiguration. This establishes Requirement 1 as a standalone proof before the full five-capability test in PoC II.
A useful breath spectrometer must see background molecules and trace molecules simultaneously — without sacrificing sensitivity for one to measure the other.
CO₂ in ambient air contains four isotopologues spanning 98.42% to 0.07% natural abundance. The rarest species sit directly within the dominant absorber's spectral band. Resolving them requires five instrumental capabilities working simultaneously.
Why it matters: Within the same absorption band, OSCAii must distinguish signals that differ by more than three orders of magnitude — all at ambient pressure. Detecting the rare isotopologue lines beside the dominant absorber demonstrates the core capability needed for isotope-ratio analysis, metabolic tracer studies, and non-invasive biochemical monitoring.
¹²C¹⁷O¹⁶O at 0.07% must be detected alongside ¹²C¹⁶O₂ at 98.42% within the same absorption band — a 3+ order intensity contrast at a single spectral region.
A single interferogram contains too much noise to resolve a 0.07% signal. 1,000 coherent averages suppress the noise floor — completed in under 10 seconds. Speed matters: slow averaging introduces drift that destroys coherence.
After averaging, the absorption feature of ¹²C¹⁷O¹⁶O must rise clearly above the residual noise floor. SNR is the ceiling on how rare a species can be reliably identified.
Each isotopologue absorbs at precisely known frequencies. If the optical grid drifts between measurements, lines shift and molecular identity assignment becomes ambiguous. RF-referenced, TAI-traceable comb eliminates this error.
CO₂ isotopologue lines are closely spaced. Without sub-MHz spectral resolution, adjacent lines from different species blend together and cannot be individually assigned.
All four CO₂ isotopologues resolved simultaneously from ambient lab air at 0.1 ATM — ¹²C¹⁶O₂ (98.42%), ¹³C¹⁶O₂ (1.1%), ¹²C¹⁸O¹⁶O (0.4%), ¹²C¹⁷O¹⁶O (0.07%). Compact gas cell: 137 × 110 × 36 mm. No gas concentrating or dehydration required.
DCS spectra collected from mango and banana across Pre-Peak, Peak, and Post-Peak stages. Each stage produces a distinct molecular fingerprint — spectrally separable with high accuracy across biological subjects, without pre-labeling individual compounds.
Fruit VOC emission profiles change continuously across lifecycle stages — a directly analogous challenge to tracking metabolic state changes in complex biological breath over time.
Have questions or want to explore a partnership? We'd love to hear from you.