General

  • No, the resolution of a spectrum depends only on the spectrometer’s resolution (slit width and diffraction grating used) and the sample’s alignment to the CL system. The beam current and accelerating voltage do not change the emitted spectral profile but do influence the number of photons emitted so that it will have an impact on the spectrum’s signal-to-noise ratio.

    Was this helpful?
  • The exact working distance (WD) depends on the type of detector you have. Many advanced CL detectors include a collection optic in the form of an off-axis parabolic mirror. In this setup, it is critical to locate the specimen at the precise focal point of the parabola. Most systems have this set to be 10– 13 mm, but there are now versions available that enable CL imaging at <4 mm WD that also allow the use of in-lens detectors.

    Was this helpful?
    • Dedicated color CL imaging system: The ChromaCL2™ detector allows the simultaneous capture of color CL, BSED, and SE images in a single pass of the electron beam.

    • Spectrometer-based system: The Monarc™ detector forms color images by sequentially scanning the specimen and filtering the emission by red, green, and blue wavelengths at each image acquisition step. As an added advantage, this detector performs spectral analysis and collects hyperspectral images.

    Was this helpful?
  • The spatial resolution is dependent upon the beam spot size and interaction volume of the electron beam. The interaction volume is related to the beam voltage, spot size, and material investigated. Furthermore, it is not binary, meaning the contribution to the CL signal comes from wherever electrons can scatter inside the sample, the population of which decays as a function of depth and distance from the point of injection. For a 2 keV beam with spot size of about 2 nm on GaN, the CL generation is effectively from a range (radius of a few nm with a depth of about 40 nm), with the bulk of the CL coming from less than half of this range (about 14 nm), and lateral extent about the size of the beam spot.

    Was this helpful?
  • Charging of the sample can be a big issue when analyzing many of the wide bandgap semiconductor and insulator materials in the SEM. However, the impact is typically only associated with one’s ability to position the electron beam precisely. The CL results do not degrade further than the secondary electron images.

    Was this helpful?
  • The image resolution depends on many things, including the SEM settings, sample luminosity, and sensitivity of the detection hardware. In the SEM, CL images can achieve resolutions down to ~10 nm. However, for many practical geological specimens, the best achievable spatial resolution is of the order of 50– 100 nm.

    Was this helpful?
    1. Use the panchromatic path and low magnification (~100x) to place the specimen at the sweet spot.
    2. Adjust the sample stage Z so that you can observe a ~100 µm bright spot in the CL image. The bright spot will be independent of the sample region. This is ~1 mm below the bottom surface of the mirror.
    3. Use the SEM's Image Shift controls to bring the hot spot to the center of the image. If this is not possible, you will need to adjust the position of the CL system in X, Y (refer to the manual).
    4. For proper alignment, repeat this process in the monochromatic light path with small slits as this is the condition where the system has the smallest field of view and, therefore, the tightest tolerance on alignment. On a sample that you don’t know the wavelength of emission, you can use the zeroth-order light (where the diffraction grating acts like a weak mirror) – enter 0 nm in the software.
    Was this helpful?
  • The interaction volume is related to the beam voltage, spot size, and material investigated. Furthermore, it is not binary, meaning the contribution to the CL signal comes from wherever electrons can scatter inside the sample, the population of which decays as a function of depth and distance from the point of injection. For a 5 keV beam with spot size of about 5 nm on GaN, the CL generation is effectively from a radius of a few nm with a depth of about 200 nm, known as range. There are numerous publications that estimate the interaction volume of the electron beam for the generation of CL (e.g., Kurniawan, O. et al. Scanning. 29, p. 280-286 (2007)).

    Was this helpful?
  • Cathodoluminescence detectors are a very effective microanalysis tool for understanding our geological history. They are useful to determine mineral provenance for geochronology and metamorphic alteration studies as well as thermobarometry for petrographic applications.

    Many CL detectors only allow the collection of spatial information in the form of unfiltered (black and white) or filtered (colored) images. These techniques provide very limited spectral information that prevent the identification of trace elements, their valence and structural position, or quantitative analysis. In contrast, the Monarc™ detector provides hyperspectral imaging (spectrum imaging) analysis modes to correlate results with spectral information and enable identification of the above attributes.

    Was this helpful?
  •   SEM TEM
    Variety of samples Wide Limited to efficient emitters
    Sample preparation complexity Low High
    Sample thickness Thin-bulk Electron transparent
    Field of view Large Small
    Observe microstructure No Yes
    Spatial resolution (sample dependent) 10 nm 1 nm
    Polarization analysis Yes No
    Angle-resolved analysis mode Yes No
    Was this helpful?
  •   Photoluminescence Cathodoluminescence
    Spatial resolution (nm) 250 – 2,000 1 – 50
    Excitation energy (eV) 1 – 10 100 – 30,000
    Excitation location Volume Surface
    Vacuum required No Yes
    Cryogenic stages available Yes Yes
    Electron microscope required No Yes

    Was this helpful?
  • Since CL is an emission spectroscopy technique, the background signal is inherently low in most materials and conditions.

    Was this helpful?
  • Similar to EDS analysis in the SEM, the spatial resolution of CL images is typically not limited by the ability of the user to focus the electron probe. Instead, the lateral spread of energetic electrons as they interact with the sample limits the spatial resolution, which is highly dependent on the SEM accelerating voltage used. Using typical analysis conditions for many geological materials (10 or 15 kV), there is little or no difference between a FEG or tungsten SEM since the CL spatial resolution is ~1 µm. However, there are cases where you can observe sub-100 nm features in CL images when low voltage electrons are used (<5 kV). In tungsten SEMs, it may not be possible to focus the electron beam well at voltages below 5 kV and in this case, FEG-SEMs are superior.

    Was this helpful?
  • In an SEM, this radiation tends to be negligible when you compare it to the luminescence signal.

    However, in a TEM, it is important to consider both of these radiation mechanisms. Most CL experiments in the TEM are performed in the 60 – 80 kV range, where the scattering cross-section is higher (produce more photons) and the Cherenkov and transition radiation processes are less efficient.

    Was this helpful?
  • Silicon is an excellent choice as a substrate. As an indirect bandgap semiconductor, the background signal contribution from the substrate is very low.

    Was this helpful?

Data Analysis

  • In many insulator materials, including ceramics, rocks, minerals, and gems, CL can reveal the presence of many different trace elements. This detection capability is compatible with certain use cases and not recommended as a generic fingerprinting tool.

    CL can map many of the rare earth elements at concentrations (sub-parts per million) that are far lower than are possible with EDS or WDS in the electron microprobe and detect transition metal cations (and their charge state). More recently, quantitative analysis of the titanium cation in quartz is useful for thermobarometry purposes.

    Was this helpful?
  • It is essential to understand the effect of surface topography on the CL data. To detect light from the sample, it must escape from the sample into the vacuum. The probability of a photon escaping depends on the refractive index of the material, the wavelength of a photon, and the angle of incidence; one must consider all of these effects during data interpretation. Fortunately, many of the samples we investigate are thin films or polished samples with flat surfaces where the impact of surface topography is small. However, in materials like polycrystalline, it can be beneficial to use argon beam milling to create a flat surface to aid quantitative data interpretation in things like grain boundary contrast.

    Was this helpful?
  • The elemental mapping comes from the decomposition of a spectrum by non-linear least squares (NLLS) fitting, which in the case of the GaN/InGaN core-shell pillars shows three distinct nearly Gaussian peaks. The central wavelengths of these peaks relate to the energy bandgap in the material. Using published reference data, one can map the local bandgaps (central wavelengths) back to the indium concentration required to achieve them in the InxGa1-xN compound. Using this CL technique, you can estimate the local indium concentration across a map of the spectrum image. If you require an exact concentration, we recommend using EDS data.

    Was this helpful?

Materials

  • We are not aware of studies that use CL for the identification of clay. The closest application is by Juergen Schieber (Indiana University), who uses CL to determine the provenance of silt grain populations in shales and mudstones.

    Was this helpful?
  • Yes, it is possible to investigate a 20 nm luminescent nanocrystal. At 20 nm, the overall intensity of the CL signal is less due to the sample’s small volume. Previous results from metal plasmonic nanostars (50 nm in diameter) demonstrate this CL capability. In the Monarc, you would likely be able to image this type of sample since the bandgap luminescence should be stronger.

    Was this helpful?
  • The sensitivity for trace elements is material- and case-dependent. For example, you can detect Ti in SiO2 at a parts-per-million level. However, CL may be insensitive to Fe in the same material.

    Was this helpful?
  • The thickness of a specimen cannot be thought of in absolute terms, instead, we must think about its thickness relative to the size of the interaction volume of the electrons incident on the sample (which is itself dependent on their energy). SEMs typically operate at 1 – 30 kV and under these, conditions, electrons typically penetrate less than ~10 microns below the surface, thus, any sample thicker than 10 microns can be thought of as semi-infinite.
    As samples become thinner, the sample surface makes a much larger contribution to the data collected. This could be as simple as fewer photons being emitted (surfaces tend to have more defects/traps that lowers the photon generation rate) or give rise to new luminescence signatures associated with the surface.
    When the sample becomes sufficiently thin – or the electron penetration depth sufficiently large (like in a TEM) – the sample becomes effectively electron transparent and the electrons have insufficient chance to spread within the sample creating a probe with very small lateral spread and therefore potentially increasing spatial resolution. However, this is specimen dependent as charge carrier drift and diffusion must also be considered.

    Was this helpful?
  • Yes, similar applications exist in geology. In many oxides, particularly SiO2, there are a large number of crystal defects (e.g., point defects or clusters) that produce luminescence. These are studied less frequently than trace elements since the number of basic science experiments are much less common in the geological sciences. Many researchers use CL to gain insight into a geological process that they have a basic understanding of through EDS analysis. The materials where these are studied, tend to fall at the crossover between material and Earth science, e.g., diamond and silicon dioxide.

    Was this helpful?

Methodology

  • Gatan's SEM-based system does not use fibers. This allows the Monarc detector to preserve the angular information in the image projected from the collection mirror.

    Was this helpful?
  • The geometry of the collection mirror limits the light collection efficiency and calculations estimate it to be ~80% of a Lambertian source. Lambertian sources obey the cosine emission law and have a radiation pattern that is approximately a sphere.

    Was this helpful?
  • A version of the Monarc includes all of the components (optics, detector, and software) to capture and analyze ARCL emission patterns and spectrum images (4D data).

    Was this helpful?
  • Several exciting applications may take advantage of angular-resolved CL. 3D light-emitting devices, particularly those with micro- or nano-structure, are interesting candidates since the sample morphology directly impacts the pattern of light emission (as well as the emitted wavelength in a given direction). The same is true for light-absorbing materials because the processes of light emission and absorption are mostly reversible.

    But many technologies are being developed to manipulate how light and matter interact such as nanophotonics, metamaterials and photonic crystals. During these investigations, angular (energy) and wavelength-angular (energy-momentum) studies can reveal much about the fundamental processes. For conventional thin-film semiconductors or oxides, there are fewer compelling reasons to use this analysis since traditional optical models accounting for absorption and total internal reflection can easily simulate results. Early tests of this analysis mode measure the interference pattern generated from light emitted from thin 2D films.

    Was this helpful?
  • The entire optical design is different from previous-generation CL systems and is virtually aberration-free. This design allows the Monarc to significantly increase the light throughput since it rejects fewer photons at the entrance slit while it maintains a high spectral resolution.

    Was this helpful?
  • CL Analysis in the SEM and (S)TEM have very different benefits and choosing the right technique is dependent on your sample.

    In the SEM, the sample requires little or no preparation, and many different types of samples can be analyzed trivially. The SEM is more suited to analyzing weak CL signals due to the greater flexibility in excitation conditions (large probe currents). Also, the achievable spatial resolution is sufficient for many applications but is often limited to no better than ~10 nm (depending on the sample).

    In the (S) TEM, you can combine CL with other techniques (e.g., EELS) to directly correlate the luminescence with the sample’s microstructure and composition. This correlation potentially provides better spatial resolution (e.g., 1 nm) due to the low electron scattering in thin samples with high energy electrons. However, the low scattering means that signal levels are (very) low and only a limited range of specimens are suitable for analysis in the TEM.

    Was this helpful?
  • Most CL detectors allow the user to collect an image or map that reveals the spatial distribution of the light emission properties of a sample. Most detectors only provide limited information about the emitted light e.g. total intensity (MiniCL) or color (ChromaCL). Other detectors also allow spectroscopic (and spatial) analysis, such as the MonoCL or Monarc. The most recent detector (Monarc) also allows the polarization state and the emission pattern (angular distribution) of the light to be measured. Furthermore, the Monarc provides easier operation and better correlation with other signals in the electron microscope.

    Was this helpful?
  • SEM cathodoluminescence offers much higher spatial resolution and sensitivity levels.
    SE CL uses an electron beam focused to (sub-)nanometer diameter scanned across a sample to collect and analyze the emitted light pixel-by-pixel. The spatial resolution is limited only by the ability to focus the electron beam to a small point and its interaction(s) with the sample. In optical cathodoluminescence, the detection hardware limits the spatial resolution to several microns typically.

    Was this helpful?
  • Yes, you can use CL to analyze samples in (some) low vacuum conditions. In the same way as other SEM imaging modes, the resolution of images is degraded. Under most conditions, there is less impact to CL than other signals – the CL image resolution is less affected by the “skirting effect” as the spatial resolution of the CL image is predominantly dependent on the lateral spread of the generation volume that is typically much larger than the focal spot of the primary beam. However, some gases (notably water) CL significantly increases the background signal and prevents the formation of the CL images. This effect is not present with many other gases, e.g., nitrogen.

    Was this helpful?
  • Many specimens will exhibit drift during prolonged measurements. To deal with this, we have built drift correction algorithms into our DigitalMicrograph control software. This feature allows the user to specify a region of the specimen to repeatedly image during the data acquisition and determine and correct for drift.

    Was this helpful?
  • The Monarc™ detector uses a mirror to collect the light, and, when the sample is located at the precise focal point of the collection mirror, the projected image contains angular information – each emission direction corresponds to a different location in the image. The Monarc records this image using a 2D sensor. You can then use the mirror geometry to calculate the emission angle.

    Was this helpful?
  • In most cases, the SNR improves with decreasing temperature but this is mainly driven by the fact that the signal increases due to luminescence processes becoming more probable at lower temperatures.

    Was this helpful?