In this post I present some information about multi-view 3D SEM which is a technique used in scanning electron microscopy (SEM) for recovering the three-dimensional shape and surface texture of a sample and with high-resolution. The methodoly is based on algorithms that have been developed for decades. It is gaining impulse thanks to the interest in developing advanced 3D computer vision algorithms, moreover when these can be processed nowadays by powerful mobile phones, tablets and PCs together with the more and more refined integrated digital cameras.
Several different techniques are available for the three-dimensional (3D) characterization of micro- and nano-scale structures using electrons. In the transmission electron microscope (TEM), electron tomography and single particle reconstruction can be applied to studies of thin samples. In the scanning electron microscope, scanning TEM (STEM) tomography can be used to study very thin samples while for sample dimensions of between nm and mm an ensemble of techniques based on the destructive slice and view approach has been developed.
SEM in contrast to TEM probes only the surface of the sample under examination, therefore, and stricly speaking 3DSEM can be used to reconstruct the three-dimensional surface of your sample. The idea is based in the acquisition of images of the object of interest (your sample inside the SEM) from different viewpoints. For example, stereophotogrammetry can be used to obtain three-dimensional topographic reconstructions, typically from two images recorded in the form of a single stereo pair, i.e., one image at zero tilt and another one at a tilt of about 10 or 15 degrees. The technique is used extensively in metrological applications.
The stereopair technique can be seen as a simplified version of the multi-view 3DSEM. In multi-view 3DSEM you record many different images around your sample combining many tilts and rotations that form like multiple stereo pairs. At the end, you want to get a full map of how your sample looks from all possible perspectives. Of course, because the sample will not stand in vaccuum and it is on a support (the pin mount and may be the carbon, copper or aluminium tape), images of the sample from beneath it cannot be recorded. In consequence, the information in the 3D model from below will be missing, or in other words the 3D model of the sample will be always an open surface. Nonetheless, you can still obtain very nice 3D models that you can rotate, manipulate and measure opening new possibilities for the characterisation and metrology at the micro- and nanoscale in three dimensions.
In the paper I give a detailed description of the steps required for recovering the 3D surface shape and texture of your sample using an SEM. The core of the reconstruction of the 3D model uses 123DCatch an amazing freeware application (at least until now) from the company Autodesk. The software is intended for the use with optical images acquired with tablets or mobile phones. It is incredibly prowerful because it can generate 3D models from a series of uncalibrated images, i.e., without any prior information that would simplify certainly the reconstruction task, such as the magnification or the angles that define the viewpoint of each image.
The paper goes beyond the acquisition and reconstruction process and describes also how to take advantage of other freeware for manipulation of the images (ImageJ), the 3D datasets (Open3Mod, Meshlab) or for preparing a file suitable for 3D printing either in basic plastic or even in color (Netfabb, Slice3, Cura). The next flowchart summarises different steps that you can can follow for studying a sample using 3DSEM and freeware.
Flowchart for 3DSEM
If you prefer a video than reading a paper, I have made this 7 minutes movie. Or enjoy the music at least, “Il barbiere di Sevilla” of Rossini.