A diffraction-limited laser spot can illuminate one quantum dot, one grain boundary, or one point on a composition-graded nanowire, yet the detected light may originate well beyond that spot. Absorbed carriers diffuse, drift, transfer between layers, become trapped, and recombine before photons traverse a wavelength-dependent confocal path. Micro-photoluminescence adds spatial selection to PL spectroscopy, but a nominal objective magnification or spot diameter is not the measurement resolution. Excitation profile, carrier transport, collection point-spread function, stage calibration, focus, temperature, power, spectrum, and sample geometry must be resolved together.

Micro-PL combines localized optical excitation with spatially filtered emission spectroscopy. A microscope objective focuses continuous or pulsed light and usually collects luminescence through the same optical path. A pinhole, single-mode fiber, spectrometer slit, or camera defines confocal detection. Rastering the sample, beam, or image produces spectra or selected-band maps. Cryogenic operation can reduce thermal broadening and reveal excitons, charge states, fine structure, phonon replicas, and localized emitters, but room-temperature micro-PL remains valuable for epitaxy, devices, two-dimensional materials, perovskites, wide-bandgap defects, and process uniformity.

For free-space wavelength $\lambda$, objective numerical aperture $\mathrm{NA}$, and refractive index $n_m$ in the object space, a conventional lateral diffraction scale is

This is an optical scale, not a guaranteed micro-PL resolution. Excitation and collection wavelengths differ, the confocal pinhole changes the combined point-spread function, aberration and refractive-index mismatch distort focus, and carriers can move before emission. A claimed 0.5–1 µm spot must be measured under the actual wavelength, objective, cryostat window, sample depth, and alignment.

The detected spatial response is the convolution of optics and carrier motion. A schematic variance budget is

when Gaussian and independence assumptions are reasonable. The material term may be strongly non-Gaussian: diffusion from a point source, drift in a field, transfer into a quantum well, photon recycling, and guided modes create tails. Pixel pitch or stage step smaller than the point-spread function oversamples the image but does not improve resolution.

Measure excitation focus with a suitable knife edge, bead, reflective feature, or nonlinear response, and measure collection response separately when the experiment needs quantitative localization. Cryostat windows, covers, substrate thickness, immersion mismatch, and objective correction collar affect aberration. Confocal pinhole reduction can reject background but lowers counts and changes alignment sensitivity. Report its projected sample-plane size, not only a fiber core diameter.

Stage coordinates require calibration for scale, orthogonality, rotation, distortion, backlash, hysteresis, creep, and temperature contraction. Cryogenic drift during a long spectral map can displace features by more than the nominal spot. Interleaved reflectance or fiducial images, bidirectional scans, repeated anchor points, and image registration reveal the error. A fitted centroid can be localized more precisely than the optical resolution when signal and calibration support it; localization precision must not be mislabeled resolution.

Define whether the question concerns transition energy, uniformity, transport, or a single emitter
  -> Choose excitation wavelength, power, spot, polarization, repetition, and temperature
  -> Select objective, confocal aperture, spectral range, grating, detector, and scan geometry
  -> Calibrate spatial scale, distortion, focus, PSF, wavelength, response, power, and dark counts
  -> Record reflectance or fiducials and a low-dose overview without selecting only bright sites
  -> Acquire local spectra with nearby background and substrate controls
  -> Run excitation-power, temperature, polarization, or magnetic/electric-field series as needed
  -> Monitor peak shift, linewidth, intensity, blinking, drift, focus, and sample change
  -> Build hyperspectral maps with raw counts, uncertainty, registration, and failure masks
  -> Model optical PSF, generation depth, carrier transport, reabsorption, and collection
  -> Test transition assignments against alternate peaks and correlated observables
  -> For single-emitter claims, measure lifetime and background-aware photon correlation
  -> Correlate coordinates with AFM, Raman, CL, EBSD, TEM, chemistry, or device response
  -> Propagate calibration, fitting, background, drift, power, and model uncertainty
  -> Archive raw spectra, images, timing events, metadata, corrections, and provenance

Excitation power changes the population and can change the specimen. Incident power is not excitation density without spot profile, absorption, reflectance, pulse structure, and focus. A small spot creates high irradiance, so local heating, trap filling, screening, exciton-exciton annihilation, biexcitons, Auger loss, photodoping, ion migration, oxidation, bleaching, or damage can occur at powers that look modest on a meter. Measure power at the sample plane or traceably correct the optical path and state whether it is average or per pulse.

For a candidate transition, a descriptive power law is often fitted:

The exponent $m$ can help compare regimes but does not uniquely label excitons, biexcitons, or defects. Saturation, heating, state filling, competing capture, and detector nonlinearity change it. Fit over a justified range with uncertainty and inspect spectral shape, linewidth, and peak energy simultaneously. A power series should be acquired in both directions or followed by a low-power check to identify irreversible change or hysteresis.

Local temperature differs from cryostat sensor temperature under focused excitation. Bandgap shifts and linewidth changes can act as temperature indicators only after calibration because strain, carrier density, fields, and composition also shift peaks. Varying spot size or chopping excitation can separate average heating from carrier-density effects. Low temperature sharpens features but can alter charge trapping, diffusion, phase, strain, condensation, and surface adsorbates.

Spectral assignment requires more than a narrow peak at one coordinate. Photon energy is

but measured wavelength depends on spectrometer calibration, slit, grating, detector pixels, optical throughput, and instrument line shape. A reported linewidth near resolution must be deconvolved or bounded rather than quoted as intrinsic. Cosmic rays, etalons, Raman lines, laser leakage, substrate luminescence, and detector defects can imitate narrow emission.

Transition assignment strengthens through correlated behavior: power dependence, polarization, temperature, lifetime, electric or magnetic field, spatial coincidence, excitation spectrum, and known band structure. Quantum-dot neutral exciton, charged exciton, biexciton, and fine-structure lines can move or exchange intensity with charge environment. Defect zero-phonon lines can accompany phonon sidebands. Peak fitting should preserve alternate decompositions and residuals; adding Gaussians until residuals vanish is not physical identification.

Spectral maps create thousands of correlated fits. Fixed peak counts can fail where bands merge or vanish, while unconstrained fits can swap labels between pixels. Global or continuity constraints help only when physically justified and should not erase abrupt boundaries. Report raw representative spectra, calibration, fit uncertainty, covariance, model-selection rule, detection limit, and failed pixels beside energy, width, and intensity maps.

Polarization measurements require calibration through objective, windows, beamsplitters, fiber, grating, and detector. Rotating only an analyzer can confuse source polarization with instrument diattenuation. High-NA collection mixes polarization components, and sample orientation matters. Stokes or Jones/Mueller treatment should match whether emission is coherent and whether depolarization occurs.

Single-emitter claims require photon statistics and background-aware controls. An isolated diffraction-limited spot or spectrally narrow line may still contain multiple emitters. A Hanbury Brown–Twiss setup estimates the normalized second-order correlation

Antibunching near zero delay supports sub-Poissonian emission, but the measured depth depends on background, detector timing jitter, dead time, afterpulsing, beamsplitter balance, count-rate drift, blinking, repetition period, binning, and fit model. The familiar ideal threshold should be applied to a clearly defined raw or background-corrected value with uncertainty, not to a cosmetically normalized curve.

For pulsed excitation, correlation peaks repeat at the laser period and the zero-delay peak area is compared with side peaks after accounting for blinking and memory. For continuous excitation, the antibunching width combines pumping and decay rates. A lifetime measurement, saturation curve, spectrum, stability trace, polarization, and correlation together describe emitter performance. Brightness at the detector is not source extraction efficiency without calibrated optical loss and collection geometry.

Emitter localization can exceed diffraction-limited resolution statistically, but traceability matters when aligning a quantum dot to a cavity or fabricating around it. Magnification distortion, chromatic registration between excitation, emission, reflection, and alignment-mark channels, stage error, fabrication overlay, and cryogenic contraction contribute. Calibrated standards and a full uncertainty chain are needed before quoting tens-of-nanometers placement accuracy.

Correlative micro-PL separates optical consequence from structural origin. Raman maps constrain strain, composition, temperature, and phase; reflectance or transmission constrains optical resonances and layer thickness; AFM reveals topography; CL and EBIC connect electron-excited emission and collection; EBSD or diffraction constrains orientation; TEM reveals interfaces and defects; electrical maps test device impact. Registration uncertainty and different interaction volumes must be included before declaring coincidence.

In nanowires, the substrate and neighboring wires can contribute, waveguiding redirects emission, and excitation or collection along the axis differs from transverse geometry. In two-dimensional materials, wrinkles, bubbles, edges, dielectric screening, contamination, strain, and doping all shift PL. In perovskites, illumination can move ions or phases. In wide-bandgap materials, a bright defect may not be the electrically important defect. The specimen-specific alternatives belong in the model.

A defensible deliverable records excitation wavelength and bandwidth, continuous or pulsed mode, repetition and pulse width, power at sample, spot and PSF, polarization, objective and NA, confocal aperture, optical path, sample orientation and temperature, cryostat window, scan step and dwell, stage calibration and drift, wavelength and instrument-line calibration, spectral response, detector gain and linearity, dark and cosmic-ray treatment, raw spectra and maps, fit model and failures, power history, and correlation/timing calibration where used.

The conclusion should distinguish focus size from material spatial resolution, pixel size from resolution, centroid precision from resolution, a narrow line from a single emitter, brightness from quantum efficiency, peak shift from a unique cause, and spatial coincidence from mechanism. Micro-PL is most decisive when calibrated microscopy, controlled excitation, spectral dynamics, carrier transport, photon statistics, and structural correlation support the same interpretation. Read micro-photoluminescence through the excitation-PSF-transport-collection-spectrum-dose-localization-and-correlation lens.