The two main peaks in the spectrum of graphene. Their shape, position and relative height reveal how many layers there are and whether the graphene is stretched or electrically doped.
Going deeper
The G and 2D bands are the two peaks every graphene sample is judged by. G is a single in-plane vibration; 2D is a two-phonon process whose shape counts layers, which is why one spectrum reports thickness, doping, strain and disorder at once.
Where the two peaks come from
The G band, near 1580 cm−1, is the straightforward one: the doubly degenerate in-plane stretching mode at the zone centre, allowed in first order, present in every sp2-bonded carbon. Its position and width respond to anything that changes bond stiffness or the electronic screening of the vibration.
The 2D band, near 2700 cm−1 for visible excitation, is not a simple overtone of a zone-centre mode. It is a second-order process in which an electron is scattered by one near the zone corner and back by another, so momentum is conserved without any help from defects. Because which electronic states participate depends on the energy, the pair of phonons selected changes with the laser – the peak moves by roughly a hundred wavenumbers for every electronvolt of excitation energy, which is why the excitation wavelength must always be stated. The D band near 1350 cm−1 is the same phonon scattered once, and needs a defect to supply the missing momentum, which is precisely why its presence measures .
Reading layers, doping and strain
Layer number is read from the 2D band’s shape rather than its height. In graphene the double-resonance condition is satisfied by a single set of transitions and the peak is a single, symmetric, relatively narrow Lorentzian. In bilayer graphene the bands split, several scattering paths contribute at slightly different frequencies, and the peak becomes visibly broader and asymmetric; in thicker samples it approaches the graphite lineshape.
and each leave their own fingerprints. Adding carriers of either sign stiffens the G mode and narrows it, because the vibration can no longer decay into pairs – a non-adiabatic effect that is much larger than the naive bond-stiffening argument suggests. Tension softens both peaks, and uniaxial strain splits the G band into two components as the degenerate mode is no longer degenerate.
Separating the effects, and the usual traps
The difficulty is that strain and doping both move the G peak, so one number cannot distinguish them. The standard trick is to plot the 2D position against the G position for many spots: strain moves a sample along one direction in that plane and doping along another, so the cloud of points separates into contributions that a single spectrum could not. It is now routine practice for mapping graphene, where both are present and spatially variable.
The remaining traps are experimental. Thin-film interference in the modulates absolute intensities, so intensity ratios are only comparable on identical substrates. A tightly focused laser heats the sample and shifts its peaks, which mimics strain. And since the 2D position depends on excitation energy, a peak position quoted without a laser wavelength cannot be compared with anything.
For specialists
Graphene’s first-order in-plane Raman mode and its second-order double-resonant overtone, which together report layer number, strain and doping.