Spectroscopic techniques play a critical role in identifying and characterizing defects in semiconductors, providing insights into their electronic and optical properties. Defects such as vacancies, interstitials, and complexes significantly influence material performance, making their detection and analysis essential for optimizing semiconductor devices. Among the most effective methods for defect identification are deep-level optical spectroscopy (DLOS) and photoinduced transient spectroscopy, which reveal defect signatures through their interaction with light.

Deep-level optical spectroscopy (DLOS) is a powerful tool for probing defect states within the bandgap of semiconductors. DLOS measures the optical absorption associated with electronic transitions between defect levels and the conduction or valence bands. By analyzing the absorption spectrum, researchers can identify the energy levels of defects and their concentrations. For instance, in silicon, the neutral monovacancy (V0) exhibits a characteristic absorption peak near 1.8 eV, while the divacancy (V2) shows distinct features at 1.0 eV and 1.5 eV. In compound semiconductors like GaAs, the arsenic vacancy (VAs) produces a signature peak around 0.8 eV, whereas gallium interstitials (Gai) contribute to absorption near 1.2 eV. These peaks arise due to transitions involving localized defect states, and their positions are highly sensitive to the defect's atomic configuration and charge state.

Photoinduced transient spectroscopy (PITS) complements DLOS by examining the transient optical response of defects after photoexcitation. When a semiconductor is illuminated with light of sufficient energy, charge carriers are excited and subsequently trapped by defect states. The relaxation of these carriers back to equilibrium emits photons or phonons, producing transient signals that are characteristic of specific defects. PITS detects these signals as a function of time and temperature, allowing for the extraction of defect parameters such as activation energy and capture cross-section. For example, in zinc oxide (ZnO), oxygen vacancies (VO) generate a transient peak at 2.3 eV, while zinc interstitials (Zni) produce a feature at 2.0 eV. The kinetics of these transients provide additional information about defect stability and interactions with other lattice imperfections.

Defect complexes, which consist of multiple point defects bound together, often exhibit more complex spectroscopic signatures than isolated defects. In silicon carbide (SiC), the carbon vacancy-silicon antisite pair (VC-SiSi) gives rise to multiple absorption peaks between 1.5 eV and 2.5 eV, reflecting the hybridized electronic states of the complex. Similarly, in gallium nitride (GaN), the nitrogen vacancy-gallium interstitial complex (VN-Gai) shows a broad emission band centered at 2.8 eV due to overlapping transitions involving both defects. The presence of such complexes can be confirmed by comparing experimental spectra with theoretical predictions based on density functional theory (DFT) calculations.

The temperature dependence of defect-related optical transitions further aids in their identification. Many defects exhibit thermally activated behavior, where the intensity or position of their spectral features shifts with temperature. For instance, the emission peak of the silicon divacancy in 4H-SiC shifts from 1.35 eV at 10 K to 1.28 eV at room temperature due to electron-phonon coupling. Similarly, the absorption linewidth of defects often broadens at higher temperatures as a result of increased lattice vibrations. These thermal effects must be accounted for when interpreting spectroscopic data to avoid misidentification of defect states.

Phonon coupling is another critical factor in defect spectroscopy, as lattice vibrations modulate the energy levels of defects and contribute to sidebands in optical spectra. In diamond, the nitrogen-vacancy (NV) center exhibits a zero-phonon line at 1.945 eV accompanied by a series of phonon replicas at lower energies. The relative intensities of these replicas provide information about the local vibrational modes associated with the defect. Similarly, in transition metal dichalcogenides like MoS2, sulfur vacancies (VS) introduce phonon-assisted luminescence peaks that are redshifted from the main excitonic emission.

The choice of excitation wavelength in defect spectroscopy is crucial for selectively probing specific defects. Below-bandgap excitation minimizes bulk carrier generation, enhancing the visibility of defect-related signals. For example, using 532 nm light to study defects in silicon avoids exciting band-to-band transitions, allowing clearer observation of sub-bandgap features. Conversely, above-bandgap excitation can populate defect states via carrier trapping, useful for studying defects with low concentrations. In CdTe, excitation at 800 nm preferentially probes tellurium vacancies (VTe) due to their higher absorption cross-section at this wavelength compared to other defects.

Time-resolved spectroscopy adds a dynamic dimension to defect analysis by revealing the lifetimes of defect-related emissions. In perovskites, lead vacancies (VPb) exhibit a fast decay component (~10 ns) attributed to direct recombination, while iodine interstitials (Ii) show a slower decay (~100 ns) due to trapping-detrapping processes. Such measurements help distinguish between different defect types and assess their impact on carrier dynamics. Additionally, polarization-dependent spectroscopy can uncover the anisotropic nature of defects, as seen in wurtzite crystals where defects aligned along the c-axis exhibit polarization-sensitive absorption.

The combination of multiple spectroscopic techniques enhances the accuracy of defect identification. Cross-referencing DLOS data with photoluminescence (PL) spectra helps confirm defect assignments by matching absorption and emission energies. For instance, in Ga2O3, the gallium vacancy (VGa) absorbs at 3.5 eV and emits at 3.2 eV, forming a Stokes-shifted pair that distinguishes it from other defects. Similarly, comparing steady-state and transient spectra can reveal metastable defects that only appear under certain conditions, such as light soaking or electric field application.

Advanced spectroscopic methods, such as Fourier-transform infrared (FTIR) spectroscopy, extend defect analysis to vibrational modes. Hydrogen-related defects in silicon produce FTIR peaks at 1990 cm−1 (Si-H stretching) and 810 cm−1 (Si-H bending), providing evidence of passivation. In oxides like TiO2, oxygen vacancies introduce mid-infrared absorption bands due to localized vibrational modes of the defect-distorted lattice. These vibrational signatures are often complementary to electronic transitions, offering a more comprehensive picture of defect properties.

The interpretation of defect spectra requires careful consideration of the semiconductor's band structure and the defect's charge state. A defect may exhibit different optical signatures depending on whether it is neutral, positively, or negatively charged. In GaAs, the EL2 defect shows distinct absorption features for its neutral (1.0 eV) and singly ionized (1.3 eV) states, necessitating control of Fermi level position during measurements. Similarly, the charge state of silicon vacancies in SiC influences their luminescence polarization, with neutral vacancies emitting unpolarized light and charged vacancies showing polarization anisotropy.

Quantitative analysis of defect concentrations from spectroscopic data relies on calibration against known standards or complementary techniques like secondary ion mass spectrometry (SIMS). The absorption coefficient of a defect peak can be related to its concentration using the Lambert-Beer law, provided the oscillator strength of the transition is known. For example, the 0.97 eV peak in irradiated silicon corresponds to the divacancy with an absorption cross-section of 2×10−16 cm2, enabling quantification at concentrations as low as 1012 cm−3.

Emerging spectroscopic techniques, such as hyperspectral imaging, enable spatially resolved defect mapping across semiconductor wafers. Variations in defect density or type can be visualized by scanning the sample while recording full spectra at each pixel. This approach is particularly useful for identifying defect clusters or gradients in epitaxial layers, where localized defects may degrade device performance. In GaN-based devices, hyperspectral imaging has revealed inhomogeneous distributions of carbon-related defects that correlate with regions of increased leakage current.

The continuous development of spectroscopic methods promises even greater precision in defect identification. Ultrafast spectroscopy with femtosecond resolution can capture defect dynamics on the timescale of carrier trapping, while cryogenic spectroscopy at millikelvin temperatures reduces thermal broadening for sharper spectral features. These advancements, combined with theoretical modeling, will further refine our understanding of defects and their role in semiconductor materials.