When Light Falls Silent: Understanding Coal in the Infrared
Coal is a dark material, hungry for light. It absorbs most of the incoming radiation—especially in the visible to near-infrared (VNIR) and shortwave infrared (SWIR) regions where hyperspectral scanners typically operate. As a result, the instrument receives only a faint reflected signal, often hovering near the noise threshold of its detectors. That is the first problem.
The spectral features that do appear in the VNIR/SWIR correspond to combinations and overtones of C–H and C=O bonds. These are inherently weak compared to the sharp, well-defined bands of minerals. They are, in effect, echoes of the fundamental vibrational modes that occur in the mid-infrared (MIR) region.
For example, the 2315 nm absorption band seen in the SWIR arises from overtones of aliphatic C–H stretching that has its fundamental near 3380 nm in the MIR. In low- to medium-rank coals, where hydrogen-rich aliphatic structures are abundant, this band is more pronounced. As coal rank increases and aromatization progresses, these absorptions fade—mirroring what FTIR shows: the loss of CH₂/CH₃ stretching near 3000–2800 cm⁻¹ (3330–3570 nm) in the MIR.
Because of this behaviour, the 2315 nm SWIR band cannot be used for rank estimation; it does not discriminate coals with vitrinite reflectance above ~1% (Rodrigues et al., 2021, 2023). However, in the MIR, the corresponding fundamental absorptions can be used effectively. Much like FTIR analysis, the ratio between the areas of the aromatic C–H band (centred near 3280 nm) and the aliphatic C–Hₓ bands (3100–3600 nm) shows an excellent correlation with vitrinite reflectance. This region of the infrared spectrum therefore offers genuine potential for rank determination.
Hyperspectral features of coal at different ranks from 450 nm to 14,500 nm. (For more detailed features see references below).
For core scanning, most commercial hyperspectral systems (e.g. Corescan, Specim) stop around 2500 nm. Beyond that, detector noise rises sharply, cooling becomes necessary, and environmental control must be much tighter. The exception is HyLogger 3, which already extended into the thermal infrared (TIR). The newer HyLogger 4 now integrates the full spectral range from VNIR to TIR (or LWIR), capturing the wavelengths where most minerals truly reveal themselves. The payoff is immediate for carbonates and organics, which display strong, distinct MIR and TIR features.
Yet one group of minerals remains largely silent: the sulphides. These include pyrite, chalcopyrite, galena, and sphalerite—opaque, electrically conductive minerals whose lattice vibrations (which would normally produce MIR absorption bands) are drowned by free-electron absorption, the same phenomenon that gives them metallic lustre.
In simple terms:
Incoming light is absorbed or reflected almost entirely at the surface.
There’s no transparent path for photons to interact with molecular vibrations.
Consequently, no diagnostic MIR or TIR bands appear—only a smooth, featureless continuum.
Sulphides do not “speak” in light; we must listen with other senses—chemistry.
Continuous XRF scanning could provide elemental fingerprints along the entire tray—mapping Fe, Cu, Zn, Pb, As, Ni, Co, and related elements that mark sulphide systems at core scale. Detecting sulphur itself remains difficult, though: its low-energy fluorescence requires a helium purge to improve sensitivity.
Perhaps this is the logical next step: a “HyLogger 5”, combining the laser profiler, RGB imaging, full spectral range, and elemental data. With that integration, we could finally resolve the spectral silence of sulphides through chemistry rather than light.
When we began scanning coal, we didn’t know if the spectra would make sense—or if coal’s darkness would simply swallow the light. It turned out that the answers were there, hidden in the mid-infrared, waiting for us to look differently. The same uncertainty now drives the next question—how to make sulphides visible. Progress in core imaging has always come from asking what seemed impossible to measure.
References:
Rodrigues, S., Fonteneau, L., Esterle, J., 2023. Characterisation of coal using hyperspectral core scanning systems. International Journal of Coal Geology 269, 104220. https://www.sciencedirect.com/science/article/pii/S0166516223000381
Rodrigues, S., Esterle, J., Fonteneau, L., Martini, B., 2021. Coal spectral libraries for scanning devices. ACARP Project C28045. 102 pp.
https://www.acarp.com.au/abstracts.aspx?repId=C28045
