Thermo Nicolet OMNIC

Status: Supported (scoped) · Vendor: Thermo Nicolet · Extensions: .spa, .spg, .srs (.srsx pending)

OMNIC is the native file format written by Thermo Nicolet’s OMNIC software for FT-IR / FT-NIR / Raman acquisitions. nirs4all-formats reads single-spectrum .spa files, grouped .spg files and the TGA/GC and rapid-scan .srs series layouts through a reverse-engineered key-table parser modelled on SpectroChemPy.

Instruments & software

Produced by Thermo Scientific OMNIC across the Nicolet FT-IR / FT-NIR and Raman ranges, including hyphenated TGA-GC-IR workflows. The committed corpus is sourced from SpectroChemPy documentation fixtures; rapid-scan series come from local-only SpectroChemPy fixtures.

The Thermo Antaris II FT-NIR analyzer (RESULT / TQ Analyst software on the OMNIC engine) writes the same .spa / .spg containers and is therefore read here; its RESULT .csv / .xlsx exports route to the generic table / Excel readers and .spc interchange to the Galactic SPC reader. A branded Antaris fixture is still to be sourced.

File structure

  • .spa / .spg — detected by the ASCII magic Spectral Data File. The layout is a fixed header followed by a key table (count at offset 294, entries from offset 304, 16 bytes each). Each entry carries a key byte, a payload offset and a payload length. Key 02 points to the spectral header, key 03 to the float32 intensity block and key 6B (107) to group spectrum titles and OMNIC timestamps. A .spg is recognised either by extension or by carrying more than one header key.

  • .srs — detected by the magic Spectral Exte File. The TGA/GC layout is located from three 02 00 00 00 18 00 00 00 00 00 anchors, which fix the data header, background header and spectral-matrix offsets; the y/time axis length and bounds come from the data header.

What nirs4all-formats extracts

  • Signals — .spa emits one SpectralRecord; .spg emits one record per sub-spectrum. The signal type is decoded from the header signal key: absorbance, transmittance (%), reflectance (%), log(1/R), Kubelka-Munk, interferogram detector signal (V), and labelled Raman / photoacoustic intensities. .srs emits one 2D record with dims = ["y", "x"].

  • Axis — values are generated from the header first_x / last_x bounds and point count. The axis kind follows the header axis key: wavenumber (cm-1), wavelength (nm / um) or index. Wavenumber axes are emitted in their native descending order.

  • Metadata — OMNIC title and timestamp, scan counts, zero-path difference, reference frequency, optical velocity, key-table offsets and (for series) the series_variant, series_name and series_y_* fields. The series y/time axis is preserved both as a first-class ["y", "x"] coordinate (min, AxisKind::Time) and as series_y_len / series_y_first_min / series_y_last_min / series_y_step_min metadata.

  • Provenance & warnings — every record carries a reverse-engineering warning (nicolet_omnic_reverse_engineered_key_table or the matching series warning) plus source file and SHA-256.

Variants & support status

Variant

Status

Notes

.spa single spectrum

Supported

One record; semantic signal type from the header key.

.spg grouped spectra

Supported

One record per sub-spectrum, with per-spectrum titles.

.srs TGA/GC series

Supported

One 2D ["y", "x"] record; minute/time y-axis.

.srs rapid-scan (raw / reprocessed)

Supported

Raw interferograms use a generated index axis until the model grows an interferogram-domain axis.

Other .srs series anchors

Detected / refused

Series magic without exactly three TGA/GC anchors is refused as an unsupported variant.

.srsx

Planned

No redistributable fixture yet.

Limitations & known gaps

  • .srs support is intentionally limited to the layout fixed by the three TGA/GC anchors; other anchor patterns are refused explicitly rather than guessed.

  • .srsx and additional high-speed / rapid-scan variants remain pending until a real fixture and reference export are available.

  • Raw rapid-scan interferograms fall back to a generated index x-axis because the shared data model does not yet carry a richer interferogram-domain axis.

Reference readers

The implementation follows the public reverse-engineering model used by SpectroChemPy and spa-on-python; SpectroChemPy is the practical cross-check for the .spa, .spg and .srs paths.

Samples & validation

SPA/SPG/SRS fixtures live under samples/nicolet_omnic/ and are golden-backed with direct semantic tests, including the 2D matrices, offsets and series_y_* metadata. Control fixtures include 2-BaSO4_0.SPA (absorbance, cm-1, 11098 points), wodger.spg (2 records, 5549 points), GC_Demo.srs (1738 x-points x 788 y rows, transmittance) and TGAIR.srs (1868 x-points x 335 y rows). Three local-only SpectroChemPy .srs files cover the tg_gc, rapid_scan_raw and rapid_scan_reprocessed variants. The probe reports nicolet-omnic at Confidence::Definite for standard extensions, and the series probe at Confidence::Possible so dispatch can route to the read-time layout detector.