Deep learning models have been proven to automate metrology tasks. It provides accurate, robust and fast results if it is trained with proper data. Nonetheless, obtaining training data remains tedious. It requires an expert user to delimitate objects boundaries in several images representing tens to hundreds of objects. Instead of drawing precise boundaries, we propose a tool relying on a rectangular bounding box to detect and segment objects. For complex applications with non-homogeneous background, the user must draw one box per object to segment them. For more homogeneous objects such as contacts, one box on the whole image can successfully segment all objects at once. To further improve the capabilities of the tool, we provide the possibility to segment the different material regions inside the found objects. The process's robustness is demonstrated through benchmarking in two contexts. Firstly, we trained two Mask R-CNN models, one with manual segmentations and the other with segmentations obtained using our tool. We compared the two models to the manual reference and found that the tool is consistent with human annotations while reducing annotation time by a factor of 30. Additionally, the tool greatly reduces user bias as the selected segmentation features are more stable. Furthermore, we suggest extending the tool to identify objects within the already found objects.
KEYWORDS: Image segmentation, Data modeling, Defect detection, Defect inspection, Photomicroscopy, Object detection, Deep learning, Scanning electron microscopy, Transmission electron microscopy
Defect inspection is an important part in the semiconductor manufacturing. This task is tedious and time consuming if done manually. Therefore, reliably automating this task is a major challenge for many semiconductor manufacturers. In the recent years, deep-learning methods for object detection have demonstrated ever better performances. However, most of the publicly available models are trained on natural images and objects. Hence, most of them needs a long and data greedy training step to be used on industrial Transmission Electron Microscopes or Scanning Electron Microscope images. In this context, we propose a deep-learning based model to detect and segment defects in electron micrographs. Using SmartDef3 from Pollen Metrology, we annotated defects on images from several industrial applications. We split them in a training and validation dataset, with which an Instance segmentation model with state-of-the-art backbone is trained. The model is then evaluated on different use cases. Competitive performances on new data in terms of detection rates and segmentation quality are demonstrated and discussed. Furthermore, the model showed a relevant defect detection rate even on images that are not in the semiconductor domain, providing an interesting tool for defects detection on a new use case without new training data. This shows how deep learning strategies can help save time and costs by automation of defects inspections. Furthermore, advanced metrological analysis of the defect can be simultaneously obtained that help optimizing the manufacturing processes and reduce defect production rate.
Critical Dimension measurement in the final slider-level fabrication is essential in the development and manufacture of magnetic read/write heads for hard disk drives (HDD). It validates the device level geometry that plays a dominant role in the magnetic performance of the writer and provides critical feedback to the wafer-level head fabrication process control. Measurement at the slider-level affords the true Air Bearing Surface (ABS) view of the real device that can only be approached by the destructional cross-section at wafer level [1,2]. While the large set of CDSEM images of writer ABS at slider level enables an excellent statistical view of wafer uniformity, it also poses special challenges to the metrology due to a substantial number of variations from the upstream wafer process. The large structure variations observed at the sliderlevel is particularly prevalent in the initial development phase where large DOE (Design of Experiment) are designed to produce intended structure variations, and low process maturity yields large unintended variations among the devices. A traditional metrology used in such a variant data set requires extensive tuning or even a set of separate solutions with each solution in the set only applicable to a small subset of the variations. However, this approach is inefficient and demands high engineer resources. In this work, we use a machine learning based metrology approach to process the large set of magnetic writer device images at the slider level. For the current study, we use a model-based solution that was trained with deep learning (DL) using a dataset from 4 different head designs. The model aims to retrieve precise boundaries of the head to perform accurate measurements. We demonstrate the progressive robustness of the model-based solution by expanding the training set to measure the CD of writer poles with different designs and large process variations due to the intrinsic wafer level structure variation and the image distortions from the slider fabrication process. In addition, we will demonstrate the efficiency of the Deep Learning (DL) based solution in comparison with traditional metrology and manual measurements on the same set of data.
Germanium is a very good candidate to host a versatile spintronics platform thanks to its unique spin and optical properties. Recently we focused on two approaches in order to tune the spin-orbit interaction (SOI) in this Ge-based platform. The first one relies on growing high quality epitaxial topological insulators (TIs) on a Ge (111) substrate, we developed an original method to probe the spin-to-charge conversion at the TI/Ge(111) interface by taking advantage of the Ge optical properties. The latter approach is to exploit the intrinsic SOI of Ge (111). By investigating the electrical properties of a thin Ge(111), we found a large unidirectional Rashba magnetoresistance, which we ascribe to the interplay between the externally applied magnetic field and the current-induced pseudo-magnetic field applied in the spin-splitted subsurface states of Ge (111). Both studies open a door towards spin manipulation with electric fields in an all-semiconductor technology platform.
Fe films grown on Ag(001) as well as MgO films on Fe(001) have been studied by spin-polarized electron
reflection experiments. The three central observations are: 1) Oscillations with monolayer periodicity of the electron-spin
motion angles ε and Φ are observed as a function of the Fe thickness. They are attributed to the oscillatory behavior of
the surface-lattice strain that is relaxed at island edges of the incompletely filled top Fe layer. 2) For strongly relaxed
thick Fe films a giant spin precession angle of 180o, which is accompanied by a pronounced minimum in the reflected
electron intensity, is observed for an electron energy of 7.3 eV. 3) For the interface system MgO/Fe(001) a very strong
sensitivity of the spin motion angles on the MgO coverage is observed for certain energy ranges.
Ab-initio band structure and spin-dependent electron reflection calculations reveal that lattice relaxations during
growth of Fe on Ag(001) as well as MgO on Fe(001) are responsible for the strong changes of the electron-spin motion
angles.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.