The scheme below shows the main subprograms of the MASDET program and introduces an algorithm for mass determination. The following approach is used:
Display and selection of suitable image areas for mass determination. This part allows the display of the acquired microscope recording parameters and micrographs. Distance measurements, fast Fourier transform and image filtering are also possible.
Entry of selection box coordinates for the images and subsequent editing of the selected boxes (ROIs). Two different types of boxes are used optionally: square or circular. The box coordinates can be distributed to 9 different layers. All coordinates are saved to the XY-coordinate file.
Computation of the mass within selected boxes. Currently, two evaluation program-modes are available allowing the calculation of:
mass-per-area (MPA) of sheets and mass-per-box (MPB), i.e., mass of individual particles, using the program-mode AREA;
mass-per-length (MPL) of filaments using the program-mode FILAMENT.
The computation of the mass within the selected boxes depends on the type of sample; the main steps performed are integration of the scattered electrons per box, subtraction of the background counts arising from the very thin film supporting the sample, and correction by several factors (e.g., to account for the detector efficiency). The auxiliary parameters computed are also stored (e.g., a width parameter for filaments, the recording electron dose, the average background scattering). For "thick" samples the nonlinearity of the ADF-signal caused by multiple electron scattering should be taken into account using data from Monte Carlo calculations; usually this correction should be done for sample of thickness above 0.1L (L: mean free electron path). Before starting the mass calculations in individual boxes the box-size in the case of the program-mode AREA and some input parameters in the case of the program-mode FILAMENT can be optimized.
Computation of the mean mass and standard deviation by statistical analysis of the individual masses contained within the selected boxes. The statistics are calculated by fitting one to ten Gaussian curves to the mass histogram using the Levenberg-Marquardt method. All results can be displayed graphically as well as by text output.
