Cathodoluminescence (CL) microscopy can be performed in the scanning (SEM) or transmission electron microscope (TEM) and—as with other forms of microscopy—the requirements for sample preparation vary significantly. Experiments in the SEM allow analysis for a broader range of sample types, including bulk specimens, 6” wafers, packaged devices, thin geological sections, epoxy mounts, and nanoparticles dispersed on a supportive substrate. Whereas, the TEM requires specimens to be thinned to electron transparency to attain a potentially higher spatial resolution.

For TEM

SEM

CL analysis often requires little or no sample preparation beyond that needed for general SEM imaging and analysis. Samples should be macroscopically flat and free from sample preparation-induced defects, e.g., scratches from mechanical polishing or ion-induced defects from focused ion beam (FIB) sample preparation methods. For FIB-prepared specimens, we recommend a clean-up step using broad argon beam milling, such as the PECS™ II or Ilion™ II system. 

Many samples of interest are insulators or wide bandgap semiconductors. For these materials, it may be necessary to coat the sample's surface with a conductive layer to prevent charge accumulation. For this purpose, a thin 1 – 3 nm carbon layer is preferred over metal evaporation due to the low and uniform absorption of the light escaping from the specimen. Analysis in an environmental or low-vacuum SEM chamber may be possible. However, some gases (e.g., water vapor) provide a background luminescence signature that may obscure the luminescence signal.

CL in the SEM does allow the analysis of sub-surface features by providing the freedom to control the penetration depth of the incident electron beam. The depth below the surface is determined by the accelerating voltage of the SEM and depending on the sample extends the analysis range up to 5 – 10 μm. For this analysis, the interaction volume created by the electron beam must be able to access the region of interest. In some semiconductor devices, this may limit analysis to specific sample geometries or require a small degree of sample processing.

When using detection systems that use a light-collection mirror, it must be possible to locate the sample’s region of interest at the focal point of the collection mirror. The Monarc™ and ChromaCL™ systems allow for 1 mm clearance between the sample and mirror plus accommodate 6” and 3” diameter samples, respectively. Additionally, the Monarc supports sample tilts of 25° (in a single axis).

In many nanophotonic applications, surface plasmon polaritons are only supported at metal-dielectric interfaces (e.g., metal-vacuum surface); and any hydrocarbon contamination can lead to a dampening of the optical response and/or misleading results. For optimum results, we strongly recommend using a plasma cleaner (e.g., Solarus™ II) to prepare contamination-free surfaces.

TEM

CL requires little or no sample preparation beyond what is necessary for general TEM imaging and analysis. We recommend using an argon-milling system like the PIPS™ II to clean-up FIB-prepared specimens and remove gallium-induced defects. Plasma cleaning using a Solarus II plasma cleaner or similar system is recommended. In many specimens, the low intensity of the CL signal necessitates large probe currents, consequently leading to a higher rate of hydrocarbon contamination.

When preparing semiconductor devices and materials, note that as the sample becomes thinner, the effect of surface recombination becomes increasingly important. In cases where the sample is thinner than the minority carrier diffusion length, surface recombination may quench the emission entirely. For semiconductor samples that contain heterostructures and other nanoscale features, the effect may be less severe as the bandgap alignment is often designed to prevent surface recombination.

In many nanophotonic applications, surface plasmon polaritons are only supported at metal-dielectric interfaces (e.g., metal-vacuum surface); and any hydrocarbon contamination can lead to a dampening of the optical response and/or misleading results. For optimum results, we strongly recommend using a plasma cleaner (e.g., Solarus™ II) to prepare contamination-free surfaces.

Sample's contribution to image drift

Common environmental factors can cause your sample to drift during an experiment, including vibration, thermal variations, and stray electromagnetic fields. This drift may cause image blurring that impacts the resolution and quality of your data. Below are frequently encountered sources of drift related to the sample itself:

• Charging
  • Symptoms – Sample drift or jumping will increase as higher probe currents are applied
  • Solution – Coat your sample with carbon to make the sample conductive
• Improper sample mounting
  • Symptoms – Sample vibrates
  • Solution – Verify how securely fastened the sample is to the grid or holder

The practicalities of getting ready for a cathodoluminescence (CL) experiment vary between a scanning (SEM) and a transmission electron microscope (TEM). Please refer to the appropriate section below.

For TEM

SEM

To get ready to perform a CL experiment, please take the following steps:

1. Secure sample to a stub suitable for the SEM; for insulating specimens, ensure that the sample surface has a good electrical path to the SEM stage.

2. Insert your sample into the microscope chamber.

3. Set the electron microscope to the desired analysis conditions (refer to specimen-dependent recommendations).

4. Find the region of interest you want to investigate.

5. Insert collection mirror after ensuring that it is safe to do so, e.g., that the sample (or another detector) is not occupying the space that the CL mirror occupies.

6. Turn the electron beam on and perform a basic alignment of the electron column.

7. Perform alignment of the CL detector and sample; this step is critical to achieving the highest quality CL results (see the Alignment section).

8. Perform final alignment of the electron microscope to help maximize image brightness and minimize image distortions.

Selecting the best SEM settings

The way a user chooses to set up the microscope is particularly important when considering CL experiments. The selection of the optimum operating parameters such as accelerating voltage and probe current is highly dependent on the sample(s) for analysis.

The table below provides guidance for a range of specimens. To learn why and how to optimize these settings for your specimen(s), please refer to the optimization discussion in the Mapping, Spectroscopy, and Emission Pattern portions of the How To section.

Sample type	Strategy for sharper images	Possible limiting factors	Typical accelerating voltage (kV)	Typical beam currents (nA)
Insulators	Lower accelerating voltages	Signal-to-noise ratio Sample preparation	5 – 20	1 – 100
Semiconductors, direct bandgap	Lowest accelerating voltage	Signal-to-noise ratio Sample structure Minority carrier diffusion	0.5 – 10	0.001 – 20
Semiconductors, indirect bandgap	Lowest accelerating voltage	Signal-to-noise ratio Sample structure Minority carrier diffusion	5 – 30	1 – 100
Metals	Higher accelerating voltages	Signal-to-noise ratio Skin depth of material Propagation of surface plasmon polaritons	20 – 30	0.1 – 20

TEM

Before you start an experiment, the TEM must be well aligned. Perform the emitter, condenser, and optic axis adjustments typical for your TEM; this will help maximize image brightness and minimize lens aberrations.

It is important to note that the CL results that you can obtain will depend on your sample type and the accelerating voltage selected. The table below suggests TEM operating conditions for a range of specimens. 

Sample type	Possible limiting factors	Typical accelerating voltage (kV)
Insulators	Signal-to-noise ratio Sample preparation Cherenkov and transition radiation (at high accelerating voltages)	40 – 120
Semiconductors, direct bandgap	Signal-to-noise ratio Sample structure Minority carrier diffusion Cherenkov and transition radiation (at high accelerating voltages)	40 – 80
Semiconductors, indirect bandgap	Signal-to-noise ratio Sample structure Minority carrier diffusion Cherenkov and transition radiation (at high accelerating voltages)	40 – 80
Metals	Signal-to-noise ratio Skin depth of material Propagation of surface plasmon polaritons	200 – 300

To get ready to perform a CL experiment, please follow the below steps:

1. Set the TEM to the desired accelerating voltage for your specimen.

2. Load the sample into the Vulcan™ holder.

3. Set the sample so that the region of interest is within the lateral focal range of the optics of the Vulcan holder. This step is critical to achieving the highest quality CL results (see the Alignment section).

4. Plasma clean your sample using a Solarus™ II plasma cleaner, or similar system.

5. Insert your sample into the column. 

6. Find the region of interest you want to investigate.

7. Set the TEM to a eucentric focus value and bring the sample to the eucentric height.

8. Insert and align the hard x-ray aperture (some TEMs only).

9. Set the TEM to scanning (STEM) mode.

10. Perform final alignment of the electron microscope as the aperture in the collection mirrors may have disturbed the optimum alignment conditions.

Anyone who has used an electron microscope or looked down an optical microscope realizes the ability to align and focus the (electron) optical system is critical to achieving the sharpest images. During cathodoluminescence (CL) microscopy experiments, it uses both the electron optics of the scanning (transmission) electron microscope (SEM or STEM) and the light optics of the CL detector; and each must be aligned well. Even more importantly, the two optical systems and samples need to be co-aligned precisely to collect the sharpest CL maps, spectra, and emission patterns. As seen in the example spectrum (collected with the Monarc™ system), misalignment as few as tens of microns in a single axis leads to a significant reduction in performance. In other CL systems, a misalignment of just a few micrometers can prevent meaningful data from being collected at all. 

In the following section, learn why optical alignments are so important and the ways to adjust a system to capture the best results. For Monarc users, the system includes independent methods that automatically perform the alignment step without the need for user input. The following section is useful to develop understanding but, for the practicalities of collecting optimized results, proceed to the instructions for the Monarc system. 

For TEM

SEM

The electron beam generates light from within our sample, e.g., within the interaction volume. In many specimens, these photons radiate isotropically, e.g., propagate with equal intensity in all directions. It may help to visualize this process by considering the light emission as a point source located at the center of the globe that radiates such that there is equal illumination at all points on the surface of the globe:

In some cases, the sample may absorb light before it ever escapes the specimen before it reflects and refracts at the surface of the sample as well as absorbed by the electron microscope stage or holder. For these reasons, in bulk specimens, it is more typical to consider that photons are emitted from the sample surface into the upper hemisphere only:  

To have a detector with high sensitivity, it is of utmost importance to maximize the collection of photons. The preferred method is to use a high–NA (numerical aperture) collection mirror. A typical collection mirror is an off-axis paraboloid with an aperture to allow the electron beam to reach the specimen. When the sample is located at the focal point of the parabola, ~85% of the emitted photons are collected and reflected into a collimated beam for further processing by an optical system. The polar plot of the Northern Hemisphere below illustrates the solid angle covered by a typical collection mirror:

In these high sensitivity systems, optical alignment becomes critical to capturing sharp maps, high-resolution spectra, and emission patterns. See, for example, comparative CL maps and a collection of spectra captured with varying degrees of misalignment in (just) one axis below:

In these unfiltered CL maps of hexagonal boron nitride, a misalignment of 100 μm (in Z-axis) reduces the signal-to-noise ratio by a factor of 10, but with poor alignment in more than one axis, this could easily be a factor in excess of 1000.

In the example of a CL spectrum shown at the top of the page, a 20 μm misalignment in the Z-axis reduces the collected signal by 50%, and at a misalignment of 60 μm, you can no longer resolve the double peak at 543 nm. In some analysis systems, a misalignment of just 5 μm may prevent data from being collected at all.

Basic alignment workflow

The basic steps to begin using a high performance, mirror-based CL system appear to be straightforward:

1. Insert the collection mirror.

2. Center the collection mirror.

3. Set the sample at the operating working distance.

4. Focus on the SEM image.

5. Start your experiment.

We have already seen that it is important to align the SEM, CL system, and sample to within only a few micrometers. This sounds challenging enough, even before considering that you cannot directly observe the focal point of the collection mirror. The collection mirror of a CL system is designed with an aperture to allow the electron beam to reach the sample. Thus, the CL mirror's focal point is approximately at the center of the aperture. However, the exact location depends not only on the accuracy of the mirror itself but also the alignment of the SEM’s electron beam. When the SEM’s accelerating voltage is changed, the electron beam axis can shift, causing an apparent translation to the mirror’s position and, if the optical axis of the SEM is not perpendicular, an apparent shift in the position of the focal point within the aperture.

With a specimen that emits light strongly and uniformly over many hundreds of microns, it is relatively straightforward to perform this alignment manually after good operator training. Most, if not all, samples of interest for CL measurements exhibit significant spatial variation that makes this challenging or impossible even for expert users on many specimens.

Monarc               MonoCL™/ChromaCL™                 Vulcan™

Schematic ray diagram

As we have seen, optical alignment is critical to collect optimal results (sharpest images or spectra or during the shortest acquisition time). In the following section, a simplified ray-trace diagram helps explain why alignment is so critical to many of the basic modes of operation of a CL system. One consequence of employing a collection mirror is that the alignment of the sample, the electron beam of the SEM, and the CL system become very important. Consider the simplified schematic of a CL detector shown below:

One can assume that light emerges from the sample at the location of the electron beam. If the sample and CL system are aligned well (blue lines), the light reflects from the collection mirror in a collimated ray bundle. For instance, the light rays travel in parallel to transfer optics (a generalized term for the lenses, mirrors or optical fibers used in the optical design). This transforms the ray bundle into a format suitable for the downstream analysis hardware. If the two systems are not aligned well around the CL system's focal point (red lines), the ray bundle no longer travels through the optical system in the intended manner, and few photons reach the downstream analysis hardware. This is even more critical when the light bundle must travel through an aperture, e.g., the entrance slit to a spectrograph prior to reaching the detector. This has some important consequences:

• High fluence only when the sample is located at the CL system's focal point 
• Low fluence when the beam is scanned out of the focal range of the optical system (vignetting) – A limited field of view in CL maps
• The field of view to capture CL spectra depends on the spectrograph's slit width unless you employ special transfer optics are employed (e.g., in the Monarc models 450 and 450.FS)
• Alignment must be sufficiently good not to destroy angular information in the image projected from the collecting mirror

TEM

The electron beam of a transmission electron microscope (TEM) causes light generation from within a sample. In many specimens, the photons radiate isotropically, e.g., propagate with equal intensity in all directions. The most efficient way to collect these emitted photons is to use large solid angle mirrors positioned above and below the specimen as in the holder of the Vulcan system. Light reflects from the mirrors and focuses on optical fibers that run out of the microscope to detect the hardware.

In mirror-based systems, proper optical alignment is critical. For example, a misalignment of just a few tens of micrometers can reduce the signal by 50% or more. With samples in the TEM typically emitting only weakly, precise optical alignment of a sample to the collection system is challenging and fraught with danger in the hands of inexperienced users unless, as with the Vulcan system, the system design guarantees optical alignment in most specimens.

During the planning phase of each experiment, it is important to establish what questions you want to answer. When combined with information you already know, such as your sample’s geometry or some approximate composition, it is easier to select an appropriate technique that delivers answers in the shortest time possible.

• Unfiltered images/maps – Reveal crystal texture in mineralogy, differentiate phases, and quantify extended defects based on their radiative efficiency in semiconductor materials and devices
• Individual spectra from a few well-chosen points – Will typically give the highest quality spectral data
• Wavelength-filtered imaging – Powerful technique to image one wavelength range (bandpass) to determine the spatial distribution of different phases or defects
• Wavelength spectrum imaging – Allows deconvolution of overlapping peaks and mapping of systematic changes in the emission spectrum; collects a complete spectrum at each point in an image
• Individual emission patterns – Determine the direction of photon emission from a few well-chosen points
• Spectroscopic analysis of emission pattern – Capture and analyze the emission spectrum emitted in a given direction or reconstruct emission patterns at a specific wavelength band
• Polarization – Ascertain the polarization state of the light emitted; can use in conjunction with other modes of analysis