434 research outputs found
Modeling the reactive magnetron sputtering process
Get PDFReactive magnetron sputtering is a versatile plasma technique to deposit thin layers of compound material on all kind of objects. The purpose of these thin films is to add or enhance interesting properties to the object. The term ``thin'' should here be interpreted as ranging from a few nanometers up till several microns. The technique is well appreciated in industry and found its application in numerous technological products ranging from big building glass windows, over car parts and drill chucks to the touch panel of smart phones. The basics of the technique are conceptually simple. A mostly metallic target is bombarded by the ions from a low-pressure noble gas plasma. These ions are accelerated over the applied negative voltage difference between the chamber and the target. The particles that are sputtered due to this bombardment, condense on surfaces within the vacuum chamber and ideally on the object to be coated. To modify this metallic layer into a compound layer, one or more reactive gases are added to the process. This reactive gas chemically reacts with the deposited material to form the desired compound. The purpose of the magnetron setup is to enhance this method by the creation of a magnetic field in the close neighborhood of the target surface. This magnetic field locally intensifies the plasma density and the ion production to optimize the deposition flux as function of the electrical power provided to the process.
In the first chapter the technique is qualitatively explained and the most important aspects are introduced. Over the decades that the technique has been used, several magnetron setups came to existence. These designs can differ in the target geometry, the shape of the magnetic field and the electrical operation characteristics. The more general known design decisions are touched on. Characteristic for reactive sputtering as deposition technique is the occurrence of hysteresis phenomena in the process curves. These hystereses cause that the operation conditions are not uniquely determined by their instantaneous operation parameters, but will be history dependent. These hystereses can be investigated by so called direct controlled or feedback controlled hysteresis experiments. They differ in the way how an operation point is established. For direct control, the operation parameters are manually set irrespective on how the system will behave. For feedback control these parameters are automatically adapted by system monitoring in order to realize a certain system state. The technological relevance of these hysteresis phenomena is the impact on the process efficiency through the deposition rate and on the operation stability. Modeling of these hystereses is then interesting to understand their origin, their dependencies and their impact. Two modeling approaches are here recognized to do this. The first ``atomistic'' approach individually models in great detail and with high quantitative power the many physical and chemical subprocesses which are involved in the technique. Combining them would result in a strong predictive tool to cut away unnecessary trial-and-error experiments. The goal of this approach is very ambitious but is hampered by the big difference in temporal and spatial scales which have to be combined across the subprocesses. The second ``holistic'' approach starts to model the technique as a whole but in a way that strongly (over)simplifies the reality. Its primary goal is the basic qualitative understanding and modeling of the whole process and only in second order, the exact quantification of it. It is the second approach, also coined the top-bottom approach, which will be followed in this work. This contrasts with the former ``atomistic'' or bottom-up approach.
The second chapter sets off with a historical overview of simple reactive sputtering models which prelude the original Berg model. The Berg model can be viewed as the ancestor of the Reactive Sputtering Deposition (RSD) models which follow. The development of the Berg model fits the ``holistic'' approach as it models the essence of the hysteresis during reactive sputtering deposition with a minimum in model complexity and parameters. The dependency of the hysteresis in the reactive gas pressure (system observable), as function of the introduced reactive flow (operation parameter), is studied with respect to the material and the operation parameters within the Berg model. In the thesis, new solution strategies for the original Berg model are proposed which enables the simulation of the hysteresis as function of the pumping speed, for example. Starting from the original Berg model, two kinds of extensions can be formulated. The first kind embodies extensions which add physical mechanisms to the model or differently describe included mechanisms. The second kind of extensions add spatial or temporal resolution to the model without touching the essence of the model. For the original Berg model, the inclusion of a deposition profile (spatial), an ion current profile (spatial) and the time dynamics (temporal) are examples of second kind extensions. In this work, the modeled time dynamics are shown to qualitatively correspond with the experimental behavior.
In chapter three, a first kind extension of the original Berg model results in the RSD2007 model. The new mechanism that the original RSD2007 model introduces is reactive implantation and subsurface reaction. The derivation of this original RSD2007 model has here been investigated more thoroughly and lead to an analytical solution form of the subsurface implantation and reaction equations. For a uniform implantation profile an analytical closed form is achieved. In this work it is shown that these subsurface mechanisms can resolve at least three limitations of the original Berg model with experiments. The backside is the introduction of more model parameters which partially can be retrieved from experiments or other models. In this thesis, the remaining unknown parameters are estimated by fitting two RSD implementations to experimental hystereses of an aluminum and yttrium target sputtered in an argon/oxygen atmosphere. The first implemented RSD model which assumes an uniform current profile, is not able in reproducing the correct experimental oxide sputter yields. The second RSD implementation with a spatial resolved current profile however succeeds. The correlation between these sputter yields and the subsurface reaction rate is investigated in this work and the sputter yield ratio between the two metal systems showed to be constant and independent of the chosen model.
Chapter four proposes a new formulated RSD2013 model elaborated within this thesis. The complete model is given in all its details together with the implementation of the model in a public available and user-friendly software package, equally called RSD2013. The major extensions over the previous version RSD2009 is the remodeling of redeposition, the addition of a second subsurface layer, a saturation limit for implanted reactive species and a complete equivalent steady state model for all model options including a multi-cell description and/or redeposition. Furthermore the option to include or exclude certain aspects of the model is been realized. The latter is illustrated with a series of simulations where the detail of the modeled system is gradually incremented. But the major physical enhancement to the RSD model is the remodeling of redeposition, the deposition of sputtered material back on the sputtering target surface. The dependency of redeposition on the metal-gas combination, the target-sample distance, the angular sputtering distribution and the gas density is examined with a Monte Carlo transport code SiMTra. The relevance of redeposition during reactive sputtering deposition is put forward. In a case study, the influence of the redeposition fraction on the reactive sputtering system is investigated by RSD2013 modeling. More specific, the influence on the target condition, the effect on the hysteresis and the shape modification of the racetrack and the sputter profile is examined.
In the final chapter some future suggestions for next generation RSD models are put forward. Some of these suggestions form contemporary research topics of the research group DRAFT while others are disparately waiting to be tackled
Dual magnetron sputter deposition of yttria-stabilized zirconia thin films
Get PDFDriven by the global quest for alternative materials, several initiatives have been started to develop better materials to be applicable not only in research but also in industry. There are two possible ways to obtain these materials.
The first comprises the discovery of a new material, while the second involves the understanding of the behavior of an existing material in order to control its characteristics. Due to the strong driving force towards creation of new technologies, up to now more effort has been done in the development of new materials, leaving the understanding to a side.
This research focuses on Yttria-stabilized Zirconia (YSZ), which is commonly known as solid electrolytes. These types of electrolytes are electrically conductive materials due to the presence of mobile ions acting as charge carrier. The charge carrier can be a cation (such as H+, Li+, Ag+) or an anion (such as F− and O2−). A vast amount of literature data on these materials is mostly related to bulk materials and thick films. However, there is at present a limited amount of information on the fundamental aspects related to solid state electrolytes on academic level, specially in Flanders. Nevertheless, this fundamental research is needed to support the idea of understanding the material in order to control its properties.
In order to discern the relations happening in the solid state electrolyte, it is necessary to reduce its proportions to atomic level. Downscaling the electrolytes to thin films, for example, changes their fundamental aspects due to the occurrence of small scale or nano effects. Essentially, two fundamental questions arise when reducing the film thickness.
First, one can question if the results related to the influence of the composition on the ion conductivity as measured for bulk solid state electrolytes can be transferred to thin film materials. The pronounced scatter of the activation energies for the oxygen ion migration and electrical conductivity properties of solid state electrolytes as a function of their composition makes it difficult to answer this question.
The second question is related to microstructure and crystallinity. At present, it is not clear how the microstructure of solid state electrolytes affects the properties of the thin films. However, a good understanding of this relationship would open the possibility to manipulate the YSZ properties by controlling the nanocrystalline microstructure of the thin film.
Naturally, one must realize that this subdivision based on these two questions is somewhat artificial as the thin film morphology and texture can be related to the composition. Hence, only a multi-axis approach allows a full understanding due to the relationship between composition and microstructure (and texture). Therefore, it is necessary to investigate the morphological and textural characteristics of thin films as function of both composition and deposition conditions. In this way, the three principal axes are i) elemental composition, ii) microstructure and texture and, iii) properties
- …
