November_EDFA_Digital

edfas.org 29 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 22 NO. 4 mirrors are used, whereas in reflective objec- tive systems, specificallydesignedaberration- corrected objectives are used. The main distinction between these two approaches is that the mirrors in the add-on approach have a smaller field of viewwhen focused on the sample, meaning that for CL maps larger than a few microns, constant adjustment of the mirror position during acquisition is necessary to retain maximum collection effi- ciency. The aberration correction approach circumvents this issue by using a fixedmirror collinear with the electron beam, thus ensur- ing constant collection efficiency on a field of views as large as hundreds of microns, thus improving the reproducibility of cathodolu- minescence andmaking it quantitative. [2] This enables optical and electronic alignments that are both faster and easier; an efficient automation and a significant reduction in the training needed to safely operate the tool. Figure 2 shows a typical SEM-CL collection apparatus, highlighting themain components used in light collection and analysis. Historically, two types of CL images have been acquired: a panchromatic one, generally using a point/single channel detector and integrationof the signal over a certainwavelength range, and a hyperspectral one (multi-channel detection), where the measured signal is resolvedwith respect towavelength. In the first case, each pixel simply contains the registered intensity, whereas in the second, a full wavelength spectrum is contained in each individual pixel, greatly increasing the wealth of information as well as the size of the dataset. An example of these acquisitionmodes can be found in Fig. 3. While in some applications, a panchromatic image is sufficient, the additional information brought by the hyperspectral map can be invaluable; in this case, it allows for a distinction between the various layers of the stack. A combination of both the panchromatic and hyperspectral data is usually applied to solve a given issue. The next section reviews a selection of applications focusing specifically on com- pound semiconductors. MAIN APPLICATIONS While CL spectroscopy has been extensively used in fundamental and proof-of-concept research in the past 30 years, its potential for use in a fab environment is only developing now as clear use cases are emerging. Presented here are several applications of CL to fab- related issues and challenges. These applications can be separated into three categories following the lifecycle of Fig. 1 Schematic of the various signals that are emitted as an electron beam strikes the surface of a material. Typical cathodoluminescence spectroscopy implementations have at least a secondary electron detector aswell as a cathodoluminescence collectionmirror toperform correlations. Fig. 2 Example of a light collection apparatus in a SEM dedicated to cathodoluminescence spectroscopy. The electronbeam(inblue) excites the sample,which in turn emits light (in yellow) that is collected using dedicated collection optics (red arrow), the light is then fed into a monochromator to split it into its color components usingagratingand finallydetected either using a CCD array for hyperspectral imaging or a point detector for panchromatic imaging. Hyperspectral imaging can also be performed using a point detector by rotating the grating and putting a slit in front of it, though this approach is significantly slower than the CCD. Courtesy of Attolight AG. a semiconductor device: development, production, and failure analysis. PROCESS DEVELOPMENT Alloy composition determination. One very basic use of hyperspectral CL acquisition is to determine local