Measuring the absorption coefficient of a semiconductor thin film becomes difficult when absorption is weak, particularly below the band gap. Photoluminescence (PL) offers a way to probe this region with high sensitivity. When combined with a model of light propagation through the complete sample stack, PL can be used to determine the photoluminescence absorption coefficient in multilayer structures—not just in thick, self-supporting wafers.
This matters for materials characterization because sub-band-gap absorption reflects defects and disorder. These can affect carrier transport and the performance of semiconductor devices, including thin-film solar cells. PL-based measurements can help characterize band tails and quantify their energy dependence, provided the optical model and emission process are handled appropriately.
Why measure absorption below the band gap?
The absorption coefficient, α, describes how strongly a material absorbs light at a given photon energy. Near and below a semiconductor’s band gap, absorption may be associated with extended states, defects, or disorder-related states. In many materials, the absorption edge includes an exponential region known as Urbach behavior. The slope of this region is used to determine the Urbach energy, a measure of the tail’s energy scale.
These weak absorption signals are challenging to measure in thin films. Conventional spectrophotometry can be affected by internal light trapping, while measurements such as photothermal deflection spectroscopy (PDS) can face limits from weak signals or absorption in the substrate. PL has a different advantage: its intensity is strongly linked to the material’s absorptivity, making it sensitive to very low absorption. The reported method resolves band-tail absorption across several orders of magnitude and, in the studied samples, agrees with PDS where both techniques provide reliable data.
Extending PL analysis to thin-film stacks
For a thick wafer, PL analysis can use an expression that accounts for emission and absorption within the semiconductor. Applying that expression directly to a film on a substrate is not generally sufficient. Light can reflect at the front surface, at the film-substrate interface, and from the back of the substrate. Optional front layers add further optical interfaces. These effects influence how much internally generated light escapes and reaches the detector.
A transfer-matrix method provides a way to include those interfaces in the calculation. In the approach described for thin films, an optical matrix represents the front layers and interfaces, the semiconductor layer is modeled with its thickness and absorption, and a rear matrix accounts for the substrate and rear-side effects. The resulting optical correction factor describes the layer’s absorptivity and connects the measured PL flux to the light generated inside the semiconductor.
The method assumes conditions such as a homogeneous film with approximately constant absorption coefficient and quasi-Fermi-level splitting through its thickness. The analysis is restricted to systems where interference fringes are weak or absent; coherent propagation requires a different treatment. If carrier or optical properties vary with depth, the spatial dependence must instead be handled through integration or numerical analysis.
From a PL spectrum to an absorption coefficient
Extracting α involves more than recording an emission spectrum. The analysis first determines the quasi-Fermi-level splitting (QFLS), which describes the separation of electron and hole quasi-Fermi levels under excitation, along with the sample temperature. The PL spectrum is fitted in an energy range where the signal is usable and absorption is high enough for the optical properties to be measured reliably.
That fitting step depends on the optical correction factor. The refractive index is needed across the relevant spectral range, while the extinction coefficient—and therefore absorption—must be known in the higher-energy portion used to establish the fit. Spectrophotometry or spectroscopic ellipsometry can supply these optical data where their measurements are reliable. Once the QFLS and temperature have been established, the low-energy PL signal can be used to retrieve absorption in the weak-absorption region.
Nonradiative recombination reduces the PL yield and affects the QFLS, but it does not necessarily invalidate the extracted absorption coefficient: the method determines the QFLS before evaluating α. Under the conditions discussed in the study, the absorption calculation depends on a ratio involving PL and the corresponding emission term, so absolute PL calibration is not required for α. Accurate measurement of the spectrum’s shape remains important.
Why the multilayer correction matters
A simpler wafer-based optical approximation may be adequate for estimating QFLS when the film absorbs nearly all relevant light in the fitting range. That does not guarantee an accurate absorption spectrum at lower energies. Simulations in the study show that approximate optical factors can shift the recovered absorption or distort its magnitude, particularly when films are thinner or when the substrate reflects strongly.
For a reliable thin-film absorption measurement, the practical implication is clear: use the multilayer optical model when deriving α, rather than assuming the film behaves like a wafer. The sample thickness and substrate influence the result, and the model should represent the actual optical stack as closely as the available measurements allow.
Interpreting defect-related emission
The method is most straightforward when one transition process dominates the PL signal and can be described by a single QFLS. Some materials emit through both band-to-band transitions and transitions involving deep defect states. These processes can respond differently to excitation and may require separate quasi-Fermi levels.
In that situation, applying a single-QFLS model to the full spectrum can produce misleading absorption values in the defect-emission region. The reported work demonstrates that band-tail absorption can still be resolved when the band-to-band PL signal is distinct from the defect-related signal. However, extracting defect-related absorption itself is more difficult because its QFLS and the relevant optical factor must be determined reliably at very low absorptivity.
Practical implications for materials characterization
PL-based absorption analysis provides a sensitive route to studying thin-film semiconductor band tails, especially where conventional optical methods lose confidence at weak absorption. Its value depends on treating the complete layer stack, establishing QFLS and temperature in a suitable spectral range, and distinguishing emission pathways when defects contribute to the PL.
For engineers and technical teams evaluating semiconductor materials, the method is best viewed as a complementary characterization tool rather than a replacement for every optical measurement. Used with appropriate optical constants and a multilayer transfer-matrix model, PL can extend absorption analysis into regions that are otherwise difficult to access and support more detailed assessment of disorder-related absorption and Urbach behavior.