We never see an exciton directly. We see its effects on light: the colors a material absorbs, the colors it emits, how those colors shift with time, field, and temperature. Spectroscopy is the art of turning those effects into knowledge about binding energies, lifetimes, coherence, and interactions. This article surveys the main tools — photoluminescence, absorption and reflection, time-resolved methods, and ultrafast pump-probe — and explains what each one reveals and what it cannot.
Photoluminescence: the workhorse
Photoluminescence, or PL, is the simplest and most widely used exciton measurement. Excite the material with light above the band gap, and collect the light it emits as excitons recombine. The emission spectrum — intensity versus photon energy — shows peaks at the exciton energies. The peak positions give energies, the widths give a measure of disorder and dephasing, and the intensities give a sense of population and radiative efficiency.
PL is sensitive and easy to implement, which is why it is often the first measurement on a new sample. It is also indirect in important ways. PL measures only excitons that recombine radiatively. Dark excitons, nonradiative losses, and excitons that dissociate before recombining do not appear. The PL intensity is not a direct measure of exciton density, because it depends on the radiative rate, which varies between species and with conditions. And the PL peak can be shifted from the true exciton energy by reabsorption, by Stokes shifts from relaxation, or by the presence of multiple overlapping species.
Despite those caveats, PL is remarkably informative when combined with knobs. Temperature-dependent PL shows which peaks survive to room temperature and which quench, giving clues about binding energies. Power-dependent PL distinguishes excitons, which grow roughly linearly with excitation, from biexcitons, which grow superlinearly. Gate-dependent PL in a field-effect device shows how peaks evolve with doping, identifying trions and charged species. Polarization-resolved PL measures valley polarization and coherence, by comparing emission in different circular or linear polarizations. Time-resolved PL, discussed below, adds lifetimes.
PL excitation spectroscopy, or PLE, is a variant that scans the excitation energy while monitoring emission at a fixed energy. It reveals which absorption resonances feed a given emission, and it can resolve excited exciton states — 2s, 3s — that are weak or absent in direct emission.
Absorption and reflection
Absorption spectroscopy measures what the material takes in, rather than what it gives out. In transmission, the absorbed fraction is inferred from what is not transmitted. In reflection, the exciton appears as a dispersive feature — a derivative-like shape — because the refractive index changes rapidly near a resonance. Differential reflection, in which the reflection with and without excitation is compared, is a sensitive method for thin layers on reflective substrates, where transmission is impractical.
Absorption has an advantage over PL: it sees both bright and dark states that have oscillator strength, and it is not biased by relaxation or by which states happen to emit. The exciton Rydberg series — 1s, 2s, 3s — is often clearer in absorption or reflection than in PL, because absorption probes the joint density of states directly. The separation between the 1s peak and the onset of free-carrier absorption gives an estimate of the binding energy, though extracting a precise value requires modeling the lineshape and the continuum.
Absorption also enables measurements that PL cannot. Electroabsorption — absorption under an electric field — shows the Stark shift and broadening of excitons. Magnetoabsorption shows Zeeman splitting and diamagnetic shifts. Both are standard tools for characterizing exciton size, dipole, and magnetic moment.
The challenge for absorption in atomically thin materials is signal. A monolayer absorbs only a small fraction of incident light — on the order of a few percent per layer for strong resonances in some materials — so the measurement requires sensitive detection, clean substrates, and careful normalization. Encapsulation and interference effects from the substrate can enhance or suppress the apparent absorption, and quantitative analysis must account for the multilayer stack.
Time-resolved photoluminescence
Steady-state PL tells you what energies are present. Time-resolved PL tells you how long they live. The material is excited with a short pulse, and the emission is recorded as a function of delay after the pulse. The decay of the emission — often a single exponential, often multi-exponential in real samples — gives lifetimes.
Exciton lifetimes vary enormously: picoseconds for bright intralayer excitons in some monolayers, nanoseconds for interlayer excitons, microseconds for momentum-dark or spin-dark states. The measured lifetime is the total lifetime, combining radiative and nonradiative rates. A short lifetime does not necessarily mean fast radiative recombination; it can mean fast nonradiative loss at defects. Separating radiative from nonradiative contributions requires additional measurements, such as quantum yield — the ratio of emitted photons to absorbed photons — or comparison across samples with different defect densities.
Time-resolved PL also reveals dynamics that steady-state PL hides. A rise time after excitation indicates relaxation or transfer — for example, the formation of interlayer excitons from intralayer excitons, or the capture of carriers into trions. A change in decay shape with density indicates exciton-exciton interactions or Auger processes. A long-lived tail can indicate a reservoir of dark excitons that slowly feed the bright state.
The instrumentation — time-correlated single-photon counting, streak cameras, or time-gated detectors — sets the time resolution, which ranges from tens of picoseconds for standard systems to sub-picosecond for streak cameras. Choosing the right tool depends on the timescale of interest.
Pump-probe: watching excitons move and interact
Pump-probe spectroscopy is the ultrafast method. A pump pulse creates excitons. A probe pulse, delayed by a controlled amount, measures the change in absorption, reflection, or transmission caused by those excitons. By scanning the delay, the experiment watches the exciton population evolve in real time, with femtosecond resolution.
Pump-probe can answer questions that PL cannot. It can watch the formation of excitons from free carriers — the timescale on which a hot electron-hole plasma cools and binds into excitons. It can watch charge transfer in heterostructures — the femtosecond-to-picosecond transfer that converts intralayer to interlayer excitons. It can watch cooling, scattering, and diffusion. And it can measure coherence directly, through photon echoes and related methods, as discussed below.
Transient absorption is the most common pump-probe variant. The probe is broadband, so the full spectrum is recorded at each delay, showing bleaching at exciton resonances — reduced absorption because the states are already occupied — and induced absorption at other energies. The bleaching signal is roughly proportional to exciton density, though the relationship is not always linear, and careful modeling is needed to extract densities and rates.
Two-dimensional electronic spectroscopy, or 2DES, is an advanced variant that spreads the signal over two frequency axes — excitation and detection — revealing couplings between states, energy transfer pathways, and homogeneous versus inhomogeneous broadening. 2DES is powerful but technically demanding, and it is used when the question requires separating overlapping contributions that one-dimensional spectra cannot resolve.
Pump-probe also has limitations. The pump itself perturbs the system, creating high densities that can introduce many-body effects not present at low density. The signal is often small — a fractional change in reflection of parts per thousand or less — requiring lock-in detection and averaging. And interpreting transient spectra requires a model, because bleaching, shifting, and broadening can all produce similar changes.
Coherent spectroscopy: photon echoes and four-wave mixing
Excitons lose coherence through scattering and dephasing. Coherent spectroscopy measures that loss directly. In a photon echo experiment, two or three pulses create a coherence, let it dephase, and then reverse the dephasing, producing an echo whose amplitude decays with the coherence time. The echo separates homogeneous broadening — the intrinsic linewidth from dephasing — from inhomogeneous broadening — the spread of energies from disorder. That separation is crucial, because a broad PL peak can mean either fast dephasing or a wide distribution of exciton energies, and the two have very different implications.
Four-wave mixing is a related coherent method, in which three pulses interact to produce a fourth signal that carries information about coherence and interactions. It is sensitive to exciton-exciton interactions, to biexciton binding, and to dephasing mechanisms. Both photon echoes and four-wave mixing require phase-stable pulses and careful alignment, and they are used when the question is about coherence itself, not just population.
Coherence times in semiconductors are short — often picoseconds or less at room temperature, longer at low temperature — and they are a central concern for quantum applications, where coherence must survive long enough to perform operations. The companion article on coherence and dephasing discusses this in depth.
Spatial imaging: where excitons go
Spectroscopy is not only about energy and time; it is also about space. PL imaging maps where excitons emit, revealing diffusion, funneling, and localization. In a uniform sample, the PL spot expands beyond the excitation spot as excitons diffuse. In a strained or gated sample, the PL concentrates in regions of lower exciton energy, showing funneling. In a moiré sample, PL imaging can reveal the superlattice pattern, though the period is often below the diffraction limit and requires near-field methods.
Near-field optical microscopy beats the diffraction limit by using a tip to localize the excitation or collection, achieving spatial resolution on the order of tens of nanometers. It can map exciton emission from individual moiré sites, from edges, and from defects. Scanning probe methods that measure photocurrent or Kelvin probe signals provide complementary information about charge separation and potentials.
Diffusion measurements — tracking how far excitons travel before recombining — give the diffusion length and, combined with lifetime, the diffusion coefficient. Interlayer excitons, with their long lifetimes, can diffuse over micrometers, making transport observable. Intralayer excitons, with shorter lifetimes, diffuse over shorter distances. Diffusion is sensitive to disorder, temperature, and density, and it is a key parameter for excitonic devices that rely on transport.
Other probes: photocurrent, Raman, and beyond
Photocurrent spectroscopy measures the current produced when excitons dissociate, rather than the light they emit. It is the relevant probe for photodetectors and solar cells, and it can reveal states that are dark in PL but contribute to current. Gate-dependent photocurrent maps the band alignment and the efficiency of charge separation in heterostructures.
Raman spectroscopy measures vibrational modes, not excitons directly, but it is a valuable complement. Raman shifts reveal layer number, strain, doping, and stacking, all of which affect excitons. Resonant Raman, where the excitation is tuned to an exciton resonance, enhances specific modes and provides information about exciton-phonon coupling.
Angle-resolved photoemission spectroscopy, or ARPES, measures the band structure directly, including the valence bands and, with pump-probe methods, the transient conduction-band population. Time-resolved ARPES can watch excitons in momentum space, showing which valleys are populated and how they evolve. It is a powerful but specialized tool, requiring ultrahigh vacuum and careful sample preparation.
Putting it together: a measurement strategy
No single method tells the full story. A typical investigation combines several. PL identifies the peaks. Absorption confirms the resonances and the Rydberg series. Gate dependence identifies charged species. Power dependence identifies biexcitons. Time-resolved PL gives lifetimes. Pump-probe gives formation and transfer times. Coherent spectroscopy gives dephasing. Imaging gives diffusion. Together, they build a consistent picture.
The discipline is to check assignments against multiple knobs. A peak that appears in PL but not in absorption may be defect-related. A peak that grows superlinearly with power and appears only at high density is a candidate biexciton, but it must also show the expected temperature and gate dependence. A claim of a new species is strongest when it is supported by energy, power, gate, temperature, polarization, and time-resolved data together.
Sources and further reading
For photoluminescence, absorption, and reflection spectroscopy of excitons, see standard semiconductor optics textbooks such as those by Klingshirn, Yu and Cardona, and Fox. For time-resolved and ultrafast spectroscopy, see textbooks and reviews on ultrafast spectroscopy of semiconductors and on time-resolved photoluminescence. For pump-probe and two-dimensional electronic spectroscopy, see reviews on ultrafast spectroscopy of two-dimensional materials and on 2D electronic spectroscopy. For photon echoes, four-wave mixing, and coherence measurements, see reviews on coherent spectroscopy of semiconductors and on exciton dephasing. For near-field imaging and diffusion measurements, see reviews on nano-optics and exciton transport in two-dimensional materials. Where this article describes timescales as ranges, that reflects the variation across materials and conditions reported in that literature.