Cathodoluminescence workflow for STEM

Specimen Prep     Holders & Transfer     Optical Setup     Data Acquisition     Analysis

Step 1: Specimen preparation

To reveal structural information samples need to be electron transparent. Many nanoparticles already are, but for bulk sample preparation mechanical grinding, polishing, and argon ion milling is often used to achieve electron transparency. You may prepare site-specific samples by focused ion beam (FIB) method, however, we recommended that these are put through a low energy argon ion polishing routine to remove the damage caused by the FIB gallium beam.

Step 2: Holders and transfer to microscope

In the scanning transmission electron microscope (STEM), cooling semiconductor samples to <150 K is often critical because of the low light levels and to reduce the point defect creation caused by the high energy electron beam. You usually mount specimens on 3 mm grids, then load into the cooling stage for insertion into the transmission electron microscope (TEM). Some special holders are available that integrate the cryogenic stage and collection of the cathodoluminescence (CL) signal, such as the 465.N holder of the Vulcan system.

Step 3: Optical setup

It is critical that you locate the specimen’s region of interest in the correct position for the cathodoluminescence collection optics to work efficiently. With the Vulcan holder, a thin sample loaded into the holder automatically aligns to the correct focal plane so specimen adjustment in the X,Y plane is all that is necessary; specialist tools are provided to make this straightforward. In systems where the collection optics and sample are independent, it is necessary to go through a complex setup that optimizes the sample position relative the electron beam axis (X, Y and Z), collection mirror (Z) and the mirror relative to the electron beam axis (X and Y). This can be particularly difficult with samples that are non-uniform or weakly emit cathodoluminescence.

Step 4: Data acquisition

The light emitted from the sample can vary from position to position in intensity, wavelength (energy), angular distribution, and lifetime (the time it takes for the light to be emitted after a sample is excited by the electron beam). There are many different methods you can use to capture this information:

Imaging: Measures the variation in emitted light intensities as the electron beam is scanned in an X,Y pattern across a sample. Panchromatic or polychromatic imaging is the simplest method to quantify the total light emitted at each pixel, and is often used to measure the dislocation density in semiconductor samples or reveal zonation in zircons for geochronology studies. Alternatively, you can use monochromatic or bandpass imaging to spectrally filter the light to select only the luminescence associated with a particular spectral feature. This method can reveal the distribution of specific minerals, and crystal defects such as stacking faults or impurities.

In earth sciences, color cathodoluminescence imaging is a favorable method to reveal and interpret reservoir quartz texture as well as shale provenance. Here the collected light is dispersed (split) into red, green and blue (RGB) wavelengths then sensed separately before being recombined in software. Color images may be created through multiple passes of the electron beam with sequential measurement of the RGB signals or, preferentially in a single pass of the electron beam using simultaneous detection of the colors.

Spectroscopy: Analyzes the spectral distribution of the light emitted. The emitted light wavelength can be analyzed to determine the electronic process taking place, thus reveal the nature of the luminescence center (e.g., a particular (trace) impurity, material band gap or crystal defect).

Spectrum imaging: In an advanced acquisition mode, as the electron beam is scanned in an X,Y pattern, you can acquire a full spectrum at every pixel to form a data cube with X and Y spatial dimensions as well as a wavelength dimension. Therefore you can acquire all spatial and spectral information in a single scan of the electron beam.

Angular-resolved emission pattern: The angle that a photon is emitted varies and will depend on the sample structure and mechanism that leads to the photon creation. In some applications, this is useful to measure the angular emission pattern to characterize the viewing angles of LEDs or displays, as well as to understand the interaction of light and matter.

Step 5: Analysis

In semiconductors, cathodoluminescence is usually an intrinsic property whose luminescent features are associated with the material band gap, doping, and crystal defects. However, the competitive nature of the cathodoluminescence generation process can complicate data interpretation. Specifically, recombination of electron-hole pairs and non-radiative recombination pathways (often via phonon-assisted processes) can perturb the measured intensity of spatially resolved cathodoluminescence profiles. Researchers frequently measure dislocation density in semiconductors where threading dislocations are centers of high non-radiative recombination rate and are revealed as dark spots in a cathodoluminescence image. The ability to correlate luminescence from individual quantum wells, discs and dots in the STEM opens the opportunity for complete characterization of novel device structures.

The study of the optical properties of metal nanoparticles has attracted wide interest recently as novel optical device designs are proposed. Cathodoluminescence and low-loss electron energy loss spectroscopy (EELS) are proven to be key analytical techniques in understanding the basic interactions of light and matter. Cathodoluminescence in the TEM enables the local density of optical states to be revealed from individual nanoparticles at spatial resolutions far below the optical diffraction limit.

Gaussian or Lorentzian curves are typically fit the acquired spectral peaks (in eV) during spectral analysis. This enables determination of the intensity, central wavelength, and full width half maximum (FWHM). In three-dimensional (spectrum image) datasets, you can apply curve fitting to every pixel within the dataset to extract distribution maps of particular luminescence centers (even if there are spectral overlaps with other features) and to reveal subtle shifts in the central wavelength or FWHM. This allows you to quantify compositional variation or stress.