Plasma diagnostics and ITO film deposition by RF-assisted closed-field dual magnetron system

Deposition of indium tin oxide (ITO) among various transparent conductive materials on flexible organic substrates has been intensively investigated among academics and industrials for a whole new array of imaginative optoelectronic products. One critical challenge coming with the organic materials is their poor thermal endurances, considering that the process currently used to produce industry-standard ITO usually involves relatively high substrate temperature in excess of 200°C and post-annealing. A lower processing temperature is thus demanded, among other desires of high deposition rate, large substrate area, good uniformity, and high quality of the deposited materials. For this purpose, we developed an RF-assisted closed-field dual magnetron sputtering system. The “prototype” system consists of a 3-inch unbalanced dual magnetron operated at a closed-field configuration. An RF coil was fabricated and placed between the two magnetron cathodes to initiate a secondary plasma. The concept is to increase the ionization faction with the RF enhancement and utilize the ion energy instead of thermal energy to facilitate the ITO film growth. The closed-field unbalanced magnetrons create a plasma in the intervening region rather than confine it near the target, thus achieving a large-area processing capability. An RF-compensated Langmuir probe was used to characterize and compare the plasmas in mirrored balanced and closed-field unbalanced magnetron configurations. The spatial distributions of the electron density ne and electron temperature Te were measured. The density profiles reflect the shapes of the plasma. Rather than intensively concentrated to the targets/cathodes in the balanced magnetrons, the plasma is more dispersive in the closed-field mode with a twice higher electron density in the substrate region. The RF assistance significantly enhances ne by one or two orders of magnitude higher. The effect of various other parameters, such as pressure, on the plasma was also studied. The ionization fractions of the sputtered atoms were measured using a gridded energy analyzer (GEA) combined with a quartz crystal microbalance (QCM). The presence of the RF plasma effectively increases the ITO ionization fraction to around 80% in both the balanced and closed-field unbalanced configurations. The ionization fraction also varies with pressure, maximizing at 5-10 mTorr. The study of the ionization not only facilitates understanding the plasma behaviors in the RF-assisted magnetron sputtering, but also provides a criterion for optimizing the film deposition process. ITO films were deposited on both glass and plastic (PET) substrates in the 3-inch RF-assisted closed-field magnetrons. The electrical resistivity and optical transmission transparency of the ITO films were measured. Appropriate RF assistance was shown to dramatically reduce the electrical resistivity. An ITO film with a resistivity of 1.2×10-3 Ω-cm and a visible light transmittance of 91% was obtained with a 225 W RF enhancement, while the substrate temperature was monitored as below 110°C. X-ray photoelectron spectroscopy (XPS) was employed to confirm the ITO film stoichiometry. The surface morphology of the ITO films and its effect on the film properties were studied using atomic force microscopy (AFM). The prototype of RF-assisted closed-field magnetron was further extended to a larger rectangular shaped dual magnetron in a flat panel display manufacturing system. Similar improvement of the ITO film conductivities by the auxiliary RF was observed on the large-area PET substrates. Meanwhile, significant deposition rates of 25-42 nm/min were achieved.

Datasets, features, learning, and models in visual recognition

Visual recognition is a fundamental research topic in computer vision. This dissertation explores datasets, features, learning, and models used for visual recognition. In order to train visual models and evaluate different recognition algorithms, this dissertation develops an approach to collect object image datasets on web pages using an analysis of text around the image and of image appearance. This method exploits established online knowledge resources (Wikipedia pages for text; Flickr and Caltech data sets for images). The resources provide rich text and object appearance information. This dissertation describes results on two datasets. The first is Berg’s collection of 10 animal categories; on this dataset, we significantly outperform previous approaches. On an additional set of 5 categories, experimental results show the effectiveness of the method. Images are represented as features for visual recognition. This dissertation introduces a text-based image feature and demonstrates that it consistently improves performance on hard object classification problems. The feature is built using an auxiliary dataset of images annotated with tags, downloaded from the Internet. Image tags are noisy. The method obtains the text features of an unannotated image from the tags of its k-nearest neighbors in this auxiliary collection. A visual classifier presented with an object viewed under novel circumstances (say, a new viewing direction) must rely on its visual examples. This text feature may not change, because the auxiliary dataset likely contains a similar picture. While the tags associated with images are noisy, they are more stable when appearance changes. The performance of this feature is tested using PASCAL VOC 2006 and 2007 datasets. This feature performs well; it consistently improves the performance of visual object classifiers, and is particularly effective when the training dataset is small. With more and more collected training data, computational cost becomes a bottleneck, especially when training sophisticated classifiers such as kernelized SVM. This dissertation proposes a fast training algorithm called Stochastic Intersection Kernel Machine (SIKMA). This proposed training method will be useful for many vision problems, as it can produce a kernel classifier that is more accurate than a linear classifier, and can be trained on tens of thousands of examples in two minutes. It processes training examples one by one in a sequence, so memory cost is no longer the bottleneck to process large scale datasets. This dissertation applies this approach to train classifiers of Flickr groups with many group training examples. The resulting Flickr group prediction scores can be used to measure image similarity between two images. Experimental results on the Corel dataset and a PASCAL VOC dataset show the learned Flickr features perform better on image matching, retrieval, and classification than conventional visual features. Visual models are usually trained to best separate positive and negative training examples. However, when recognizing a large number of object categories, there may not be enough training examples for most objects, due to the intrinsic long-tailed distribution of objects in the real world. This dissertation proposes an approach to use comparative object similarity. The key insight is that, given a set of object categories which are similar and a set of categories which are dissimilar, a good object model should respond more strongly to examples from similar categories than to examples from dissimilar categories. This dissertation develops a regularized kernel machine algorithm to use this category dependent similarity regularization. Experiments on hundreds of categories show that our method can make significant improvement for categories with few or even no positive examples.