Light-emitting diodes (LEDs) are more efficient, faster, and have a much smaller footprint than filament-based light sources. LEDs are composed of semiconductor materials whose electronic density of states includes a forbidden band known as the bandgap. When electrons and holes are excited and allowed to recombine across the bandgap in a semiconductor, the emitted photon (observed as light) carries the energy of the bandgap. The advent of the blue LED in the 1990s, and the subsequently explored tunability of the III-nitride materials paved the way for LED-based lighting and display technologies.

For LED applications, GaN is usually grown in a heterostructure form with a number of In_xGa_1-xN quantum wells (QWs). These generate efficient light emission in the visible part of the electromagnetic spectrum rather than at the native (GaN) bandgap energy of 3.4 eV (365 nm). They are useful to enhance gain and optical confinement by increasing the active region volume but importantly, also allow tunability of the emission color (wavelength).

The cathodoluminescence (CL) technique is a fast and highly relevant method in the process development of GaN LEDs. The energetic electron beam can effectively excite the wide bandgap of nitride semiconductors (even AlN) and provide spatial resolution down to 10 and 1 nm for CL measurements performed in the scanning electron microscope (SEM) and transmission electron microscope (TEM) instruments, respectively. One can directly visualize the distribution of non-radiative defects, such as dislocations, as well as determine the identity of point defects. Furthermore, the spatial variation of strain, doping, growth direction, and carrier concentration can all be determined as well as the different constituents of the stack (e.g., AlN, GaNs, quantum wells).

In addition, you can visualize the emission pattern using the angle-resolved or wavelength- and angle-resolved modes of operation. These modes offer another significant metric for light-emitting devices as they enable detailed characterization of the spectra using angle-resolved CL imaging. 

Experiment briefs and application notes

Cathodoluminescence as a technique for inspection, metrology, and failure analysis of microLED processing	Nano-cathodoluminescence enables the design of light-emitting diodes with higher efficiencies

3D metrology of an LED structure

While CL provides high-lateral spatial resolution to investigate semiconductor materials, it also enables depth-resolved measurements. Using the selected accelerating voltage of the SEM to determine the surface depth that the CL signal is generated from, it enables the analysis of layers buried within the bulk material. Furthermore, it allows investigation of three-dimensional structures by depth-resolved CL spectroscopy which compares the simulation of the primary electron scattering and carrier drift and/or diffusion. Such depth-resolved measurements enable:

• Mapping of the depth and lateral distribution of dislocations and basal plane stacking faults in fully processed devices
• Thickness measurements of sub-surface layers of different composition, including analysis of individual quantum wells within a multi-quantum well stack
• Location of point defects at specific interfaces within a device structure (with ~1 nm axial resolution)

Failure analysis

General lighting applications and car lights frequently use white and blue LEDs, but their use also penetrates more diverse purposes, including industrial environments. In such environments, the devices are subject to extreme environmental conditions such as mechanical vibrations, or enhanced or extremely low temperatures, humidity, or chemical substances. There is a growing realization that under these conditions, the LED chip itself rather than a peripheral element can lead to module failure. With such a diverse range of applications, laboratory testing procedures can’t replicate all use cases in real field applications. It is, therefore, necessary to use analytical techniques capable of observing the LED structure on a macroscopic and microscopic scale, characterizing both physical structure and the functional properties. To this end, electron microscopy, including CL, is an attractive and effective tool in the evaluation of failure causes in light-emitting devices.

MicroLED arrays

Light-emitting diodes (LEDs) are an attractive alternative display technology. The high emissivity and low power consumption of LEDs promise significant improvements over existing technologies in augmented reality and mobile device displays. In order to cover the full-color gamut for displays, LED pixels to contain red, blue, and green single LEDs arranged in an array. Hence, the resolution of displays depends on the size and pitch of single LEDs. In efforts to improve resolution by reducing the size of the emitters, the irradiance is consequently reduced. For smaller conventional LEDs to achieve comparable irradiance, they require a higher-input power that can lead to greater heat dissipation requirements.

Cathodoluminescence as a technique for inspection, metrology, and failure analysis of microLED processing

Hexagonal core-shell pillar

To combat the aforementioned issues that use conventional flat emitters, researchers have turned attention to three-dimensional microLEDs (μLEDs). Such emitters have the obvious advantage of increased surface area, where further the increased dimensionality allows for the formation of quantum wells on the sides of the emitter. The increased dimensionality also provides access to various crystal faces. In the case of GaN, it includes the m-plane, which has a minimal native electric field that reduces the influence of the quantum confined stark effect (QCSE), and thereby improves the recombination efficiency of excited carriers.

Compositional variation and defects

A GaN/InGaN core-shell pillar is a typical example of a three-dimensional emitter. Observing the cathodoluminescence (CL) from such a pillar allows for the extraction of spectral information as a function of position. Spectral information can be spatially mapped in bands of interest that are easy to display, e.g., the GaN, InGaN, and point defect bands. Further, in this analysis example, compositional spatial mapping of the InGaN quantum well active layer reveals the In-fraction lies between 19 – 26% and concentrated at the vertices. The native non-linear least squares (NLLS) spectra fitting tool in Gatan Microscopy Suite^® software is useful to estimate the compositional mapping and to determine the InGaN bandgap as a function of position. The NLLS fitting determined bandgap allows for the estimation of In concentration, x, and material composition is determined by observing the theoretical bandgap of In_xGa_1-xN as a function of x. Thus far, we have information on spectral CL emissions, but without information concerning the angles of emissions, such shaped specimens can emit anisotropically. To observe such anisotropy, the angle-resolved CL mode of operation can be used.

Angle-resolved cathodoluminescence

When light is emitted from a specimen at the focal point of the parabolic mirror, it reflects down the axis of the mirror. Mathematically, the paraboloidal shape is 1-to-1. Therefore, the position of the light’s reflection on the collimating mirror is unique to the angles of emission from the sample. Hence, acquisition of the angular emission information is done by simply imaging the backplane of the paraboloidal collection mirror. With a known mirror shape, you can transform the data into a polar representation for easier interpretation.

Emission selectivity and shape effects

In the case of the single out-of-array microLED pillar, the ARCL revealed an emission selectivity of about 2.4x stronger at ~70˚ as compared to the normal direction when integrating over the entire pillar. As one might expect, the direction of emission depends on the position of excitation. In fact, the emission is most intense in the direction opposite from the side of the pillar being excited and includes an interference pattern. The intensity directionality is likely due to the effect of total internal reflection from the index contrast at the pillar surface with vacuum, while the interference pattern is due to reflections from the sample surface.

microLED hexagonal pillar array

MicroLEDs are most frequently arranged in an array configuration. In this case, the array configuration is hexagonal with (edge-to-edge) pillar width ~800 nm and a (nearest neighbor) period of ~2 μm. Such an array using GaN opens the possibility of photonic influences on optical behavior. Therefore, it is important to observe them in such an array to determine the impact on light emission.

Effective compositional variation and defects

An in-array single pillar was observed using the same data acquisition as the previously displayed out-of-array pillar. Again, we observe in this spectrum image, a spectrum for every point scanned, which allows for emission band mapping of the GaN, InGaN, and point defect bands, as well as an estimation of In content by position across the pillar surface. Here we see an estimate of indium content in In_xGa_1-xN content of between 11 and 19%.

Several interesting observations can be made in this comparison of in- vs. out-of-array microLED pillars. In the out-of-array device, the overall intensity is coming from the edge, as one might expect since the quantum well is being excited directly. While for the in-array device, the intensity is more evenly distributed across the surface. Observing spectra from various positions further reveals an enhancement in the overall intensity of the in-array device, especially in the 500 – 600 nm range where multiple smaller peaks are present. The differences are perhaps even more obviated by the color band images of the GaN, InGaN, and point defect bands. In these images, the array influence produces very different emission map for each excitation position.

Once the array's influence on light generation is understood, it is common to explore the emission anisotropy as a function of wavelength to determine whether a microLED is feasible for implementation in a display device. To do this, we employ the wavelength- and angle-resolved CL (WARCL) technique that includes wavelength information along with angular. This reveals any preferential direction of emission and allows for the observation of photonic bandgaps.

Here is a side-by-side comparison of WARCL frames at several wavelengths. The weaker emission in the direction normal to the pillar is present in both the out- and in-array LEDs. However, an interesting behavior is present for the in-array demonstrating the usefulness of the WARCL technique for microLED arrays.  

Further information on the photonic crystal behavior of micro-pillar arrays can be found here.

Application Note

Cathodoluminescence as a technique for inspection, metrology, and failure analysis of μLED processing

Aging studies and failure analysis of laser diodes

Semiconductor lasers are important for many applications such as material processing, medical surgery, pumping of solid-state lasers and fiber lasers, and space applications. Many of these applications require high reliability and predictability of the lifetime due to the expense and/or difficulties in accessing their installation locations in environments like undersea fiber cables or satellite laser terminals.

Cathodoluminescence (CL) analysis is useful to reveal defects that nucleate during the useable life of a laser diode, how these evolve under further operation, plus the type and location of defects that can lead to the device failure.

As an example, CL images of three different broad area laser diodes (50 µm stripe width and 1.5 mm resonator length) at different stages of lifetime testing are shown below. The laser diodes consist of a GaAsP/AlGaAs single quantum well embedded in a 2 µm thick AlGaAs waveguide structure. The images show example images of laser diodes that exhibit (a) gradual degradation, (b) freak (early) failure, and (c) sudden failure after gradual degradation. During extended lifetime testing, an increasing concentration of dark spots can be observed from a CL image (e.g., images a and c). The dark spot contrast originates from newly nucleated dislocation loops, and the dislocation loop concentration correlates well with a gradual degradation in device performance, e.g., need for higher drive currents to maintain a constant output power. However, while laser diodes exhibit only these dark spots, catastrophic failure is avoided. In devices that exhibited catastrophic failures (b and c) you can observe dark lines through the laser stripe. These dark lines are extended dislocations formed by glide and climb processes of dislocation loops. The suddenly failed device (c) indicates dark lines running in [110] along the laser stripe, and in [100] directions, latter inclined 45° towards the laser stripe.

Photovoltaic (PV) materials and devices play a crucial role in our fight to reduce human-made greenhouse gas emissions. However, fossil fuel-derived electricity is still cheaper to produce (typically) than electricity from renewable sources like PV. To make (or in some cases to keep) renewable electricity cost-competitive with fossil fuel electricity, we must continue to improve PV and other renewable technologies.

Cathodoluminescence (CL) is useful to reveal the microscopic mechanisms that determine the macroscopic performance of PV devices and modules. This allows researchers to understand and analyze fundamental physical and chemical processes in such materials as thin-film, perovskite, silicon, and other semiconductors; industrial applications include quality control and failure analysis.

Thin-film          Multi-crystalline silicon

Perovskites

Despite only recently becoming an area of intense research focus, halide perovskite solar cells achieve exceptional efficiencies that even surpass thin-film PV materials like CdTe and CuInGaSe₂. They exhibit high carrier mobility, long diffusion lengths, and tunable spectral absorption windows. As a result, many researchers claim halide perovskites have the potential to be the most extraordinary and market competitive optoelectronic material. With that in mind, there remain significant hurdles to mainstream adoption, and researchers will require characterization techniques capable of exposing the interplay among structure, chemistry, and optoelectronic properties to inform processing strategies to increase device efficiencies and long‐term stability.

CL microscopy is an ideal technique to provide this type of information due to its ability to capture high-spatial-resolution, plus robust optical and electronic information. CL can monitor degradation due to environmental factors, reveal phase segregation resulting from material processing, and other halide perovskite‐centric material issues.

It was once thought that the electron beam‐based analysis of halide perovskites was impractical due to the instability of material properties when exposed to high-energy electrons. The analysis below the damage threshold for many halide perovskites is now possible due to recent developments that stabilize formulations and refine optical designs for high sensitivity and repeatability.

Experimental brief

Revealing the spatial distribution of phases in perovskite solar cells for the development of high efficiency and stable devices

Thin-film

Thin-film solar cells based on CdTe or CIGS are leading PV technology that is widely commercialized. Further improvements to device and module efficiencies require increased minority-carrier lifetime within the CdTe absorber layer. The low minority-carrier lifetime in polycrystalline thin films is thought to be caused by non-radiative Shockley-Read-Hall (SRH) recombination at deep levels in the bandgap created by crystal defects—both point defects and extended defects are known to be prevalent in thin-film solar cells. However, there has been a failure to reach consensus on precisely which defect(s) is(are) most limiting with the ongoing debate surrounding the role of grain boundaries after thermal processing. To this end, cathodoluminescence (CL) is an important technique that enables researchers to analyze recombination processes at the defect-scale, allowing macroscale device performance to be understood from the microscopic behavior.

Many researchers use the nano-scale spatial resolution of CL to qualitatively and quantitatively compare carrier recombination and defects at grain boundaries and interiors before and after treatments designed to passivate defects. The contrast in a CL map of a polycrystalline material reveals regions of the specimen with strong(er) non-radiative SRH recombination, often grain boundaries. Statistical analysis and correlation using grain boundaries (e.g., EBSD analysis) provide a quantitative understanding of recombination at grain interiors and a variety of grain boundary types. Recently, CL measurements also allow researchers to determine the grain boundary recombination velocity. Furthermore, capturing a complete CL spectrum at every (spatial) pixel in a map (spectrum imaging or hyperspectral imaging) allows researchers to map compositional variation within the compound semiconductor grain interiors and to develop more accurate band model diagrams to model carrier behavior.   

Experimental brief 

Mapping the electronic bandgap of semiconductor compounds with milli-electron volt accuracy

Multi-crystalline silicon

Silicon has an indirect electronic band structure. Since this band-to-band radiative recombination has only a low probability in comparison to non-radiative transitions at room temperature, it makes CL largely impractical. However, CL is useful to measure the luminescence behavior of extended defects at cryogenic temperatures. These valuable measurements can simultaneously obtain spectrally- and spatially-resolved information to provide deep insights into the electronic structure of extended defects. Example defects include dislocations and grain boundaries, which have a controlling influence on the performance of many electronic and photovoltaic devices.

In many CL experiments, extended defects act as non-radiative recombination centers and, therefore, the luminescence from band-band-recombination is decreased in the vicinity of extended defects in comparison to the undisturbed semiconductor. In the case of Si, Ge, and some compound semiconductor materials, dislocations and grain boundaries can exhibit characteristic luminescence bands (with photon energies lower than that of the bandgap) that reveal the detailed electronic structure of defects. This allows researchers to develop accurate recombination models to improve device performance. Extensive CL investigations of bonded silicon wafers and of multi-crystalline silicon show that dislocations and low angle grain boundaries are responsible for the occurrence of shallow and deep levels in the bandgap—the radiative transitions associated with these defect states are denoted as D-lines (D­₁ – D₄). There remains ambiguity over the role of the strain field accompanying extended defects as well as oxygen and transition metal contamination. Correlation of the information revealed by CL with other techniques in the SEM, such as EBSD, shows strong promise as a means to resolve uncertainties and provide manufacturers guidance in optimizing materials and device processing.