München, Techn. Univ., Diss., 2003. / Computerdatei im Fernzugriff.

München, Techn. University, Diss., 2003.

Carbon nanotubes (CNTs) are a highly promising material for future interconnects. It is expected that the decoration of CNTs with Cu particles or also the filling of the interspaces between the CNTs with Cu instead of the currently used SiO2 can enhance the performance of CNT-based interconnects. Due to the high aspect ratio of CNTs an appropriate deposition technique has to be applied which is able to coat such structures uniformly. The current work is therefore considered with thermal atomic layer deposition (ALD) of CuxO from the liquid Cu (I) β-diketonate precursor [(nBu3P)2Cu(acac)] and wet oxygen at 135°C on variously pretreated multi-walled CNTs. The different in-situ pre-treatments of the CNTs with oxygen, water vapor and wet oxygen in a temperature range from 100 to 300°C at a pressure of 1.33 mbar have been carried out prior to the ALD to enable uniform nucleation on the otherwise chemical inert CNT surface. The reduction of the CuxO as well as the filling of the space between the CNTs is not part of this work. Variations of the oxidation temperature as well as the oxidation agents resulted in different growth modes of the CuxO. An oxidation with wet oxygen at 300°C yielded in a partially layer like growth of the CuxO. It is expected that this growth mode is connected to a partial destruction of the outer CNT shell due to the oxidation. However, the damage introduced to the CNTs was not high enough to be detected by Raman spectroscopy. For all other investigated pretreatments, the formation of nanoparticles (NPs) was observed by electron microscopy. This formation of CuxO NPs can be explained by the metal-tube-interaction. Furthermore, the NPs probably decorate defect sites of the CNTs due to their higher reactivity. Additionally, analysis of energy-dispersive X-ray spectroscopy and spectroscopic ellipsometry measurements suggests that the used precursor [(nBu3P)2Cu(acac)] requires reactive oxygen surface groups for initiating the ALD growth. The observation of layer-like growth of CuxO on CNTs pretreated with wet oxygen at 300°C appears promising for deposition processes of Cu seed layers on CNTs. However, more aggressive pretreatments at higher temperatures or with more aggressive oxidation agents could be required to enable layer like growth on the entire CNTs.

Copper has become the material of choice for metallization of high-performance ultra-large scale integrated circuits. As the feature size is continuously decreasing, metal-organic chemical vapor deposition (MOCVD) appears promising for depositing the Cu seed layer required for electroplating, as well as for filling entire interconnect structures. In this work, four novel organophosphane and organophosphite Cu(I) trifluoroacetates were studied as precursors for Cu MOCVD. Details are reported on CVD results obtained with Tris(tri-n-butylphosphane)copper(I)trifluoroacetate, (ⁿBu₃P)₃CuO₂CCF₃. Solutions of this precursor with acetonitrile and isopropanol were used for deposition experiments on 100 mm Si wafers sputter-coated with Cu, Cu/TiN, and Al(2 % Si)/W. Experiments were carried out in a cold-wall reactor at a pressure of 0.7 mbar, using a liquid delivery approach for precursor dosage. On Cu seed layers, continuous films were obtained at low deposition rates (0.5 to 1 nm/min). At temperatures above 320°C, hole formation in the Cu films was observed. Deposition on TiN led to the formation of single copper particles and etching of the TiN, whereas isolating aluminum oxyfluoride was formed after deposition on Al(Si)/W. It is concluded that the formation of CF₃ radicals during decarboxylation has a negative effect on the deposition results. Furthermore, the precursor chemistry needs to be improved for a higher volatility of the complex.

Spectroscopic ellipsometry is emerging as a routine tool for in-situ and ex-situ thin-film characterization in semiconductor manufacturing. For interconnects in ULSI circuits, diffusion barriers of below 10&nbsp;nm thickness are required and precise thickness control of the deposited layers is indispensable. In this work, we studied single films of tantalum and two stoichiometries of tantalum nitride as well as TaN/Ta film stacks both on bare and oxidized silicon. Spectroscopic ellipsometry from the UV to the NIR was applied to determine the optical properties of the films for subsequent modeling by a Lorentz-Drude approach. These models were successfully applied to TaN/Ta thin-film stacks where the values of the film thickness could be determined exactly. Moreover, it is shown that considerable differences in the optical properties arise from both film thickness and substrate. <br> <br> ©2006 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE. <br>

Copper has become the material of choice for metallization of high-performance ultra-large scale integrated circuits. As the feature size is continuously decreasing, metal-organic chemical vapor deposition (MOCVD) appears promising for depositing the Cu seed layer required for electroplating, as well as for filling entire interconnect structures. In this work, four novel organophosphane and organophosphite Cu(I) trifluoroacetates were studied as precursors for Cu MOCVD. Details are reported on CVD results obtained with Tris(tri-n-butylphosphane)copper(I)trifluoroacetate, (ⁿBu₃P)₃CuO₂CCF₃. Solutions of this precursor with acetonitrile and isopropanol were used for deposition experiments on 100&nbsp;mm Si wafers sputter-coated with Cu, Cu/TiN, and Al(2&nbsp;%&nbsp;Si)/W. Experiments were carried out in a cold-wall reactor at a pressure of 0.7&nbsp;mbar, using a liquid delivery approach for precursor dosage. On Cu seed layers, continuous films were obtained at low deposition rates (0.5 to 1&nbsp;nm/min). At temperatures above 320°C, hole formation in the Cu films was observed. Deposition on TiN led to the formation of single copper particles and etching of the TiN, whereas isolating aluminum oxyfluoride was formed after deposition on Al(Si)/W. It is concluded that the formation of CF₃ radicals during decarboxylation has a negative effect on the deposition results. Furthermore, the precursor chemistry needs to be improved for a higher volatility of the complex.

In multi-level Co/low-k based interconnect systems of ultralarge-scale integrated electronic devices the removal of plasma etch residues by common plasma cleaning processes has been shown to alter material properties like k-value and leakage current of the low-k dielectric. Besides the development of less damaging plasma processes their substitution by wet cleaning steps is in the focus of research and development. With further decreasing feature dimensions the development of wet cleaning processes has to include wetting issues like the non-wetting of small features due to the surface energy of the liquid or pattern collapse effects of low-k dielectric trenches due to high capillary forces This work at first focuses on the determination of the surface energetic character of common cleaning solutions for PERR and differently etched or ashed low-k dielectric surfaces by contact angle analysis, to outline which combinations of solid and liquid will be critical regarding their wetting behavior. Besides the determination of the wetting behavior the contact angle analysis turned out to be a fast and sensible analytic tool to understand the surface modifications introduced by different plasma processes and can help to understand the mechanisms of plasma damage of low-k dielectric surfaces. The analysis showed that especially polymerizing plasma etch processes produce a low-energetic low-k dielectric surface with a negligible polar energy contributions, which inhibits their wetting by high energetic water based cleaning solutions, which actually are favored by semiconductor manufacturers. The strategy to overcome these wetting issues followed in the present work is the reduction of the surface energy of the cleaning liquids by the application of surfactants. Several types of surfactants have been applied to the cleaning liquids and the compatibility of the surfactant solutions to BEOL materials like low-k dielectrics, copper and diffusion barriers as well as their dynamic behavior has been studied. The analysis showed that choosing the appropriate rinsing solution after the cleaning process is essential to ensure its compatibility to porous low-k dielectrics. Optical, electrical and structural data indicated that DIW rinse in most of the cases was not able to remove residual surfactant species within the material, while for an IPA rinse most of the residual surfactants have been removed. Considering the data received for compatibility to low-k materials, copper and barriers, the dynamic behavior of the surfactant solutions as well as influences of increased bath temperature and long term stability a general advice about surfactant selection and processing of surfactant aided solutions within BEOL is given. / In mehrlagigen Kupfer/low-k basierten Metallisierungssystemen hochintegrierter elektronischer Bauelemente kann die Entfernung von Residuen nach der Plasmastrukturierung des Dielektrikums mittels herkömmlicher Plasmareinigungsprozesse zur Schädigung der Isolatorschicht und damit zum Ansteigen der relativen Dielektrizitätszahl sowie der Leckströme führen. Neben der Entwicklung schädigungsarmer Plasmaprozesse stellt der Ersatz dieser Prozesse durch Nassreinigungsschritte zur Ätzresiduenentfernung eine vielversprechende Alternative dar. Mit stetig abnehmenden Strukturabmaßen ist bei der Entwicklung dieser Nassreinigungsprozesse neben der Materialkompatibilität auch das Benetzungsverhalten der Reinigungsflüssigkeit von entscheidender Bedeutung, da die Oberflächenenergie der Reinigungslösung das Eindringen dieser in kleinste Strukturen verhindern und es durch hohe Kapillarkräfte zum Kollaps von Grabenstrukturen im Dielektrikum kommen kann. In der vorliegenden Arbeit wurde zunächst mittels Kontaktwinkelanalyse die Oberflächenenergie verschieden prozessierter low-k Dielektrikaschichten sowie herkömmlicher Lösungen zur Entfernung von Ätzresiduen untersucht, um hinsichtlich ihres Benetzungsverhaltens besonders kritische Materialkombinationen aufzuzeigen. Neben der Bestimmung des Benetzungsverhaltens hat sich die Kontaktwinkelanalyse zur Oberflächenenergieberechnung als schnelle und empfindliche Methode zur Analyse der Auswirkung von Plasmaprozessen auf die Oberfläche von low-k Dielektrika erwiesen. Die Untersuchungen haben gezeigt, dass besonders polymerisierende Plasmaprozesse eine niederenergetische Oberfläche erzeugen, welche von den derzeit in der Halbleiterfertigung bevorzugten hochenergetischen wasserbasierten Reinigungslösungen nur schlecht benetzt wird. Um diesem Effekt entgegenzuwirken wurde in der vorliegenden Arbeit die Senkung der Oberflächenenergie der Reinigungslösungen durch Zugabe von Tensiden untersucht. Es wurden mehrere Tenside unterschiedlichen Typs den Reinigungsflüssigkeiten zugemischt und die Kompatibilität dieser Lösungen mit low-k Dielektrika, Kupferschichten und Diffusionsbarrieren untersucht sowie ihr dynamisches Verhalten analysiert. Dabei hat sich gezeigt, dass die Auswahl der geeigneten Spüllösung nach dem eigentlichen Reinigungsprozess von entscheidender Bedeutung ist. Optische, elektrische sowie strukturelle Daten deuten darauf hin, dass bei Verwendung einer Spülung mit deionisiertem Wasser in den meisten Fällen Tensidrückstände im porösen Dielektrikum verbleiben. Eine Spülung mit Isopropanol war hingegen in der Lage, einen Großteil dieser Tensidrückstände zu entfernen. Unter Einbeziehung der Daten zur Materialkompatibilität und dem dynamischen Verhalten der Tensidlösungen bei Raumtemperatur und erhöhter Badtemperatur sowie ihrer Langzeitstabilität konnte schließlich eine Prozessempfehlung für die Verwendung der benetzungsoptimierten Reinigungslösungen in der BEOL-Prozessierung gefunden werden.

Carbon nanotubes (CNTs) are a highly promising material for future interconnects. It is expected that the decoration of CNTs with Cu particles or also the filling of the interspaces between the CNTs with Cu instead of the currently used SiO2 can enhance the performance of CNT-based interconnects. Due to the high aspect ratio of CNTs an appropriate deposition technique has to be applied which is able to coat such structures uniformly. The current work is therefore considered with thermal atomic layer deposition (ALD) of CuxO from the liquid Cu (I) β-diketonate precursor [(nBu3P)2Cu(acac)] and wet oxygen at 135°C on variously pretreated multi-walled CNTs. The different in-situ pre-treatments of the CNTs with oxygen, water vapor and wet oxygen in a temperature range from 100 to 300°C at a pressure of 1.33 mbar have been carried out prior to the ALD to enable uniform nucleation on the otherwise chemical inert CNT surface. The reduction of the CuxO as well as the filling of the space between the CNTs is not part of this work. Variations of the oxidation temperature as well as the oxidation agents resulted in different growth modes of the CuxO. An oxidation with wet oxygen at 300°C yielded in a partially layer like growth of the CuxO. It is expected that this growth mode is connected to a partial destruction of the outer CNT shell due to the oxidation. However, the damage introduced to the CNTs was not high enough to be detected by Raman spectroscopy. For all other investigated pretreatments, the formation of nanoparticles (NPs) was observed by electron microscopy. This formation of CuxO NPs can be explained by the metal-tube-interaction. Furthermore, the NPs probably decorate defect sites of the CNTs due to their higher reactivity. Additionally, analysis of energy-dispersive X-ray spectroscopy and spectroscopic ellipsometry measurements suggests that the used precursor [(nBu3P)2Cu(acac)] requires reactive oxygen surface groups for initiating the ALD growth. The observation of layer-like growth of CuxO on CNTs pretreated with wet oxygen at 300°C appears promising for deposition processes of Cu seed layers on CNTs. However, more aggressive pretreatments at higher temperatures or with more aggressive oxidation agents could be required to enable layer like growth on the entire CNTs.

In multi-level Co/low-k based interconnect systems of ultralarge-scale integrated electronic devices the removal of plasma etch residues by common plasma cleaning processes has been shown to alter material properties like k-value and leakage current of the low-k dielectric. Besides the development of less damaging plasma processes their substitution by wet cleaning steps is in the focus of research and development. With further decreasing feature dimensions the development of wet cleaning processes has to include wetting issues like the non-wetting of small features due to the surface energy of the liquid or pattern collapse effects of low-k dielectric trenches due to high capillary forces This work at first focuses on the determination of the surface energetic character of common cleaning solutions for PERR and differently etched or ashed low-k dielectric surfaces by contact angle analysis, to outline which combinations of solid and liquid will be critical regarding their wetting behavior. Besides the determination of the wetting behavior the contact angle analysis turned out to be a fast and sensible analytic tool to understand the surface modifications introduced by different plasma processes and can help to understand the mechanisms of plasma damage of low-k dielectric surfaces. The analysis showed that especially polymerizing plasma etch processes produce a low-energetic low-k dielectric surface with a negligible polar energy contributions, which inhibits their wetting by high energetic water based cleaning solutions, which actually are favored by semiconductor manufacturers. The strategy to overcome these wetting issues followed in the present work is the reduction of the surface energy of the cleaning liquids by the application of surfactants. Several types of surfactants have been applied to the cleaning liquids and the compatibility of the surfactant solutions to BEOL materials like low-k dielectrics, copper and diffusion barriers as well as their dynamic behavior has been studied. The analysis showed that choosing the appropriate rinsing solution after the cleaning process is essential to ensure its compatibility to porous low-k dielectrics. Optical, electrical and structural data indicated that DIW rinse in most of the cases was not able to remove residual surfactant species within the material, while for an IPA rinse most of the residual surfactants have been removed. Considering the data received for compatibility to low-k materials, copper and barriers, the dynamic behavior of the surfactant solutions as well as influences of increased bath temperature and long term stability a general advice about surfactant selection and processing of surfactant aided solutions within BEOL is given. / In mehrlagigen Kupfer/low-k basierten Metallisierungssystemen hochintegrierter elektronischer Bauelemente kann die Entfernung von Residuen nach der Plasmastrukturierung des Dielektrikums mittels herkömmlicher Plasmareinigungsprozesse zur Schädigung der Isolatorschicht und damit zum Ansteigen der relativen Dielektrizitätszahl sowie der Leckströme führen. Neben der Entwicklung schädigungsarmer Plasmaprozesse stellt der Ersatz dieser Prozesse durch Nassreinigungsschritte zur Ätzresiduenentfernung eine vielversprechende Alternative dar. Mit stetig abnehmenden Strukturabmaßen ist bei der Entwicklung dieser Nassreinigungsprozesse neben der Materialkompatibilität auch das Benetzungsverhalten der Reinigungsflüssigkeit von entscheidender Bedeutung, da die Oberflächenenergie der Reinigungslösung das Eindringen dieser in kleinste Strukturen verhindern und es durch hohe Kapillarkräfte zum Kollaps von Grabenstrukturen im Dielektrikum kommen kann. In der vorliegenden Arbeit wurde zunächst mittels Kontaktwinkelanalyse die Oberflächenenergie verschieden prozessierter low-k Dielektrikaschichten sowie herkömmlicher Lösungen zur Entfernung von Ätzresiduen untersucht, um hinsichtlich ihres Benetzungsverhaltens besonders kritische Materialkombinationen aufzuzeigen. Neben der Bestimmung des Benetzungsverhaltens hat sich die Kontaktwinkelanalyse zur Oberflächenenergieberechnung als schnelle und empfindliche Methode zur Analyse der Auswirkung von Plasmaprozessen auf die Oberfläche von low-k Dielektrika erwiesen. Die Untersuchungen haben gezeigt, dass besonders polymerisierende Plasmaprozesse eine niederenergetische Oberfläche erzeugen, welche von den derzeit in der Halbleiterfertigung bevorzugten hochenergetischen wasserbasierten Reinigungslösungen nur schlecht benetzt wird. Um diesem Effekt entgegenzuwirken wurde in der vorliegenden Arbeit die Senkung der Oberflächenenergie der Reinigungslösungen durch Zugabe von Tensiden untersucht. Es wurden mehrere Tenside unterschiedlichen Typs den Reinigungsflüssigkeiten zugemischt und die Kompatibilität dieser Lösungen mit low-k Dielektrika, Kupferschichten und Diffusionsbarrieren untersucht sowie ihr dynamisches Verhalten analysiert. Dabei hat sich gezeigt, dass die Auswahl der geeigneten Spüllösung nach dem eigentlichen Reinigungsprozess von entscheidender Bedeutung ist. Optische, elektrische sowie strukturelle Daten deuten darauf hin, dass bei Verwendung einer Spülung mit deionisiertem Wasser in den meisten Fällen Tensidrückstände im porösen Dielektrikum verbleiben. Eine Spülung mit Isopropanol war hingegen in der Lage, einen Großteil dieser Tensidrückstände zu entfernen. Unter Einbeziehung der Daten zur Materialkompatibilität und dem dynamischen Verhalten der Tensidlösungen bei Raumtemperatur und erhöhter Badtemperatur sowie ihrer Langzeitstabilität konnte schließlich eine Prozessempfehlung für die Verwendung der benetzungsoptimierten Reinigungslösungen in der BEOL-Prozessierung gefunden werden.

Since there is a continuous shrinking of feature sizes in ultra-large scale integrated (ULSI) circuits, requirements on materials and technology are going to rise dramatically in the near future. In particular, at the interconnect system this calls for new concepts and materials. Therefore, carbon nanotubes (CNTs) are considered as a promising material to replace partly or entirely metal interconnects in such devices. The present thesis aims to make a contribution to the CNT growth control with the thermal chemical vapor deposition (CVD) method and the integration of CNTs as vertical interconnects (vias) in ULSI circuits. Different support/catalyst systems are examined in processes for catalyst pretreatment and CNT growth. The investigations focus on the catalyst formation and the interactions at the interfaces. Those effects are related to the CNT growth. To get an insight into interactions at interfaces, film structure, composition, and CNT growth characteristics, samples are extensively characterized by techniques like AFM, SEM, TEM, XRD, XPS, and Raman spectroscopy. Screening studies on nanoparticle formation and CNT growth with the well known system SiO2/Ni are presented. This system is characterized by a weak support/catalyst interaction, which leads to undirected growth of multi-walled CNTs (MWCNTs). By contrast, at the Ta/Ni system a strong interaction causes a wetting of catalyst nanoparticles and vertically aligned MWCNT growth. At the system W/Ni a strong interaction at the interface is found as well, but there it induces Stranski-Krastanov catalyst film reformation upon pretreatment and complete CNT growth inhibition. Studies on the SiO2/Cr/Ni system reveal that Cr and Ni act as a bi-catalyst system, which leads to a novel nanostructure defined as interlayer CNT (ICNT) structure. The ICNT films are characterized by well crystallized vertically aligned MWCNTs, which grow out a Cr/Ni layer lifted off as a continuous and very smooth layer from the substrate with the growth. Besides, this nanostructure offers new possibilities for the integration of CNTs in different electronic applications. Based on the presented possibilities of manipulating CNT growth, an integration technology was derived to fabricate CNT vias. The technology uses a surface mediated site-selective CVD for the growth of MWCNTs in via structures. Developments are demonstrated with the fabrication of via test vehicles and the site-selective growth of MWCNTs in vias on 4 inch wafers. Furthermore, the known resistance problem of CNT vias, caused by too low CNT density, is addressed by a new approach. A CNT/metal heterostructure is considered, where the metal is implemented through atomic layer deposition (ALD). The first results of the coating of CNTs with readily reducible copper oxide nanoparticles are presented and discussed. / Aufgrund der kontinuierlichen Verkleinerung von Strukturen in extrem hoch integrierten (engl. Ultra-Large Scale Integration − ULSI) Schaltkreisen werden die Anforderungen an die Materialien und die Technologie in naher Zukunft dramatisch ansteigen. Besonders im Leitbahnsystem sind neue Materialien und Konzepte gefragt. Kohlenstoffnanoröhren (engl. Carbon Nanotubes − CNT) stellen hierbei ein vielversprechendes Material dar, um teilweise oder sogar vollständig metallische Leitbahnen zu ersetzen. Die vorliegende Arbeit liefert einen Beitrag zur CNT-Wachstumskontrolle mit der thermischen Gasphasenabscheidung (engl. Chemical Vapor Deposition − CVD) sowie der Integration von CNTs als vertikale Leitungsverbindungen (Via) in ULSI-Schaltkreisen. Verschiedene Untergrund/Katalysator-Systeme werden in Prozessen zur Katalysatorvorbehandlung sowie zum CNT-Wachstum betrachtet. Die Untersuchungen richten sich insbesondere auf die Katalysatorformierung und die Wechselwirkungen an den Grenzflächen. Diese werden mit dem CNT-Wachstum in Verbindung gebracht. Für Untersuchungen von Grenzflächeninteraktionen, Schichtstruktur, Zusammensetzung sowie CNT-Wachstumscharakteristik werden Analysen mit AFM, REM, TEM, XRD, XPS und Raman-Spektroskopie genutzt. Zunächst werden Voruntersuchungen an dem gut bekannten System SiO2/Ni zur Nanopartikelformierung und CNTWachstum vorgestellt. Dieses System ist gekennzeichnet durch eine schwache Wechselwirkung zwischen Untergrund und Katalysator sowie ungerichtetem Wachstum von mehrwandigen CNTs (MWCNTs). Im Gegensatz dazu hat bei dem System Ta/Ni eine starke Interaktion an der Grenzfläche eine Katalysatornanopartikelbenetzung und vertikales MWCNT-Wachstum zur Folge. Für das W/Ni-System gelten ebenfalls starke Interaktionen an der Grenzfläche. Bei diesem System wird allerdings eine Stranski-Krastanov-Schichtformierung des Katalysators und eine vollständige Unterbindung von CNT-Wachstum erreicht. Bei dem System SiO2/Cr/Ni agieren Cr und Ni als Bi- Katalysatorsystem. Dies führt zu einer neuartigen Nanostruktur, die als Zwischenschicht-CNT (engl. Interlayer Carbon Nanotubes − ICNTs) Struktur definiert wird. Die Schichten sind durch eine gute Qualität von gerichteten MWCNTs charakterisiert, die aus einer geschlossenen, sehr glatten und von den CNTs getragenen Cr/Ni-Schicht herauswachsen. Darüber hinaus bietet die Struktur neue Möglichkeiten für die Integration von CNTs in verschiedene elektronische Anwendungen. Auf der Grundlage der vorgestellten Manipulationsmöglichkeiten von CNT-Wachstum wurde eine Integrationstechnologie für CNTs in Vias abgeleitet. Der Ansatz ist eine oberflächeninduzierte selektive CVD von vertikal gerichteten MWCNTs in Via-Strukturen. Diese Technologie wird mit der Herstellung von einem Via-Testvehikel und dem selektiven CNT-Wachstum in Vias auf 4 Zoll Wafern demonstriert. Um das Widerstandsproblem von CNT-Vias, verursacht durch eine zu niedrige CNT-Dichte, zu reduzieren, wird eine Technologieerweiterung vorgeschlagen. Der Ansatz geht von einer CNT/Metall-Heterostruktur aus, bei der das Metall mit Hilfe der Atomlagenabscheidung (engl. Atomic Layer Deposition − ALD) implementiert wird. Es werden erste Ergebnisse zur CNT-Beschichtung mit reduzierbaren Kupferoxidnanopartikeln vorgestellt und diskutiert.