How Raman spectroscopy works
An interactive walkthrough: photon scattering, why the shift is a chemical fingerprint, and what peak positions, widths, and the anti-Stokes ratio tell you.
Shine a laser at a material and almost every photon bounces off unchanged. But about one in ten million photons trades a tiny, precisely quantized amount of energy with the material's vibrations, and that tiny exchange encodes a chemical fingerprint. This post walks through the physics with three interactive models.
1. A photon meets a molecule
Three things can happen when a photon scatters off a molecule. In Rayleigh scattering (the overwhelming majority of events) the photon leaves with exactly the energy it arrived with. In Stokes Raman scattering, the photon deposits one quantum of vibrational energy in the molecule and leaves red-shifted. In anti-Stokes Raman scattering, a molecule that was already vibrating hands its quantum to the photon, which leaves blue-shifted.
Pick a process below and watch both the photon and the molecule. The energy-level diagram alongside shows the same event as a transition through a short-lived "virtual" state: the photon lifts the molecule up, and where the molecule lands when the photon is re-emitted determines whether energy was exchanged.
A photon meets a molecule
Pick a scattering process. Photon color tracks its energy: unchanged, red-shifted (lost energy), or blue-shifted (gained energy).
Energy-level view
The photon lifts the molecule to a short-lived “virtual” state. Where the molecule lands when the photon is re-emitted determines whether energy was exchanged.
Stokes Raman
~1 in 10 million photons
The photon gives up one quantum of vibrational energy to the molecule and leaves red-shifted (lower energy, longer wavelength). The molecule is left vibrating. This is the signal most Raman instruments measure.
Rayleigh scattering carries no chemical information, so the spectrometer's notch filter throws it away. Everything a Raman instrument measures lives in that one-in-ten-million inelastic fraction, which is why Raman signals are weak, why lasers (not lamps) are required, and why fluorescence can so easily swamp the measurement.
2. Why the shift is a fingerprint
The energy the photon loses equals one quantum of a molecular vibration, ℏω. Model a bond as two masses connected by a spring and the vibrational frequency depends on exactly two things: the spring's stiffness (bond strength) and the masses on its ends (the atoms). Both are fixed properties of the material.
The bond as a spring: ω = √(k / μ)
Change the bond stiffness and atomic masses and watch the vibrational frequency (and the Raman shift) respond.
~509 cm⁻¹
Resulting Raman shift
Stiffer bonds vibrate faster (bigger shift); heavier atoms vibrate slower (smaller shift). This is why each material has its own fingerprint peaks.
Raman shifts are reported in wavenumbers (cm⁻¹) relative to the laser line, so the fingerprint is the same no matter which excitation wavelength you use. Light, stiff bonds sit high: C–H stretches near 2,900 cm⁻¹. Heavy atoms on softer bonds sit low: the Mo–S modes of MoS₂ land near 400 cm⁻¹, and the W–Se modes of WSe₂ lower still. No two materials share the same full set of vibrational frequencies, which is what makes a Raman spectrum an identification tool.
3. Reading a real spectrum
A Raman spectrum plots scattered intensity against Raman shift. Three separate pieces of information are encoded in it:
- Peak positions identify the material (and its phase, strain state, and doping).
- Peak widths reveal disorder: defects interrupt the lattice and broaden the vibrational lines.
- The Stokes/anti-Stokes intensity ratio measures temperature, because the anti-Stokes signal requires molecules that are already thermally vibrating, and their population follows Boltzmann statistics.
The simulated MoS₂ spectrum below puts the last two under your control. Heat the sample and watch the anti-Stokes trace grow toward the Stokes one; broaden the linewidth and watch the sharp fingerprint smear out the way it does in a defective film.
Simulated MoS₂ Raman spectrum: Stokes vs anti-Stokes
Synthetic Lorentzian peaks at the known E₂g (383 cm⁻¹) and A₁g (408 cm⁻¹) positions. Anti-Stokes intensity follows the Boltzmann factor at your chosen temperature.
0.141
Anti-Stokes / Stokes ratio at 408 cm⁻¹
Heat the sample and the anti-Stokes peaks grow: more molecules start out vibrating. Broaden the peaks and you are seeing what defects do to a real spectrum, and this linewidth is exactly what defect-density estimators invert.
4. What you can learn from it
Everything above compounds into a remarkably information-dense measurement for something that is non-destructive, needs no sample preparation, and works through glass:
- Identity. Peak positions are a chemical fingerprint: MoS₂, graphene, silicon, and even polymorphs of the same compound each have distinct peak sets.
- Quality and defects. Defects broaden peaks and activate new ones (like graphene's D band). Mapped across a sample, linewidth becomes a quantitative defect-density map. The calibrated inversion behind that is its own story, told in How a Raman linewidth becomes a defect density.
- Strain, doping, and temperature. Peaks shift when bonds are stretched, when carriers stiffen or soften phonons, and the anti-Stokes ratio reads out local temperature.
Point a Raman microscope at a grid of positions instead of a single spot and each of these becomes a spatial map: a hyperspectral picture of where a film is clean, strained, doped, or damaged. That mapping workflow (parsing the raw hyperspectral files, segmenting clean regions, and inverting linewidths into defect densities) is exactly what Matter42 automates. For how those pieces fit together in one project, see AI-native characterization for 2D materials.