The paper's model: the intrinsic CH2 vibration of cristae lipids sits at 87 THz. Couple it to light inside a functional mitochondrion and the level splits — Rabi-style — into 71 THz and 103 THz. Drag the coupling to zero and the split collapses. The 71 THz line is the paper's headline: it appears only in living cells and tissues, and only when mitochondrial structure is intact.
What the spectrometer sees, per the paper. Living: a sharp anomalous peak at 71 THz that cannot be assigned to any specific molecule; the 103 THz partner hides inside the broad 90–110 THz water/biomolecular band. Kill the cell or disrupt the cristae and the 71 THz line vanishes — the ordinary 87 THz CH2 mode returns. The mode is a property of intact structure, not of any molecule.
The paper's boldest claim: the superposition state acts as a channel that modulates ATP production — tune the state, tune the cell's energy output. This scene is the model, drawn schematically: the gold mode in the cristae folds, feeding the ATP flux. Whether the channel is real is exactly what replication has to decide.