Organische molecule gecontroleerde oppervlakte selectieve depositie als toepassing voor resist vrije lithografie

The semiconductor manufacturing industry faces enormous challenges due to the continuous downscaling of devices. These devices' miniaturization solely relies on top-down approaches using engineering patterning techniques like litho-etch-litho-etch (LELE) or self-aligned multiple patterning (SAMP). However, these methods encounter limitations at smaller dimensions, including pattern collapse at high aspect ratios, edge placement errors, and misalignment errors in multi-level dimension patterning. A more advanced solution to these challenges is the adoption of bottom-up approaches, such as area-selective deposition (ASD). ASD allows for the targeted deposition of materials exclusively in growth areas (GA) while preventing material deposition in non-growth areas (NGAs). It can be achieved through various schemes, including inherent selective pre-patterned surfaces, area deactivation, and area activation. These approaches create chemical contrasts between GA and NGA, enabling selective material deposition using surface-sensitive techniques like atomic layer deposition (ALD) or chemical vapor deposition (CVD). This thesis delves into the fundamental aspects of achieving ASD by two distinct approaches. First scheme involves a novel approach using photolithography defined pattern generation on organic inhibitors passivated photocatalytic surface. This route combines the effect of area activation and deactivation where NGA is passivated with organic layer and GA is generated by removal of this layer through lithography. The second scheme involves area deactivation scheme where NGA is passivated with organic layer to selectively deposit material by ALD on GA. The organic inhibitors used in this thesis to passivate the NGAs are self-assemble monolayers (SAMs) and small molecular inhibitor (SMI). The thesis is structured around four focused objectives, each addressing a critical aspect of achieving ASD. The first objective focuses on selecting an appropriate photocatalytic layer that is crystalline and active only under UV light, enabling the removal of the organic layer upon UV exposure to create GAs. The second objective involves a comprehensive study of various organic layers, analysing their molecular structure, bonding capability with the photocatalytic underlayer, UV sensitivity, and ALD inhibition capacity. The third objective centres on developing and optimizing the Ru ALD process using two different precursors, comparing their performance, and examining how ALD parameters impact selectivity, aiming to strike a balance between selective film growth and achieving an optimal growth per cycle (GPC). Finally, the fourth objective focuses on achieving Ru ASD through both described routes, including a detailed defect analysis in the NGAs, covering aspects such as particle size distribution (PSD), defect coverage, and defect density. The photocatalytic underlayer used in this ASD approach plays a vital role, as exposure to UV light with energy higher than its band gap generates electron-hole pairs that aid in the degradation of organic inhibitors. This mechanism is particularly important because it can potentially eliminate the pattern collapse issues that occur during the development step for photoresist (PR) removal in top-down processes. Selecting an appropriate underlayer requires considering several factors: a higher band gap, a crystalline structure to minimize charge recombination, and a suitable band gap position to facilitate the oxidation and reduction reactions necessary for removing organic molecules. With these criteria in mind, tin oxide (SnO₂) and titanium oxide (TiO₂) were selected, with TiO₂ being favored due to the availability of better deposition techniques. TiO₂ deposited using chlorine-based precursors tends to be amorphous unless subjected to high-temperature annealing. However, using an isopropoxide-based precursor proved more effective in achieving a crystalline surface without the need for an annealing step, although a minimum film thickness is required to ensure the surface remains crystalline. Various SAMs were thoroughly investigated with a focus on their molecular structure, aiming to balance their role in inhibiting ALD material deposition in NGAs with their ability to be removed during UV exposure to create GAs. Key parameters considered included the functional head groups for bonding with TiO₂, a halogenated backbone for enhanced UV absorption, the presence of unsaturated bonds in the backbone, and halogen-functional groups that facilitate easier bond breakage upon UV exposure, hydrophobic tail groups for ALD blocking. Among the candidates, octadecyl trimethoxy silane (TMODS), octyl trimethoxy silane (TMOS), and perfluoro decyl trimethoxy silane (FDTMS) emerged as the most effective organic inhibitors, exhibiting high hydrophobicity (>90° WCA) that indicates superior blocking capability. Moreover, a reduction in hydrophobicity upon UV exposure was observed, suggesting that these SAMs can promote material deposition achieving ASD on the exposed areas. Ru is a promising hard mask material in semiconductor device manufacturing, known for its high etch resistance, mechanical hardness, chemical inertness, and high density. While various Ru ALD precursors are available, this work focuses on exploring two specific precursors: ethylbenzyl 1-ethyl-1,4-cyclohexadienyl ruthenium (EBECHRu) and tri-carbonyl trimethylenemethane ruthenium (Ru(TMM)(CO)₃). The thesis discusses the demonstration of Ru ASD using both precursors, highlighting that Ru(TMM)(CO)₃ offers the distinct advantage of being deposited at a lower temperature of 220°C. This is particularly beneficial when using SAMs or SMIs for passivation, as many of these molecules exhibit low thermal stability. In contrast, EBECHRu requires an ALD temperature exceeding 250°C. Furthermore, the smaller size of the carbonyl precursor allows for a higher GPC than the bulkier EBECHRu precursor. Finally, our investigation examines the Ru blocking capabilities of organic inhibitors, including TMODS, TMOS, FDTMS, and DMA-TMS. These layers were deposited using either vapor phase or dip-coating methods on TiO₂ and SiO₂ surfaces. The defect growth mechanism was studied on SAM- and SMI-passivated surfaces during Ru ALD cycles, focusing on parameters such as defect density, PSD, and defect coverage. These metrics provide insights into the effects of chain length and the presence of alkyl or fluorine backbones on passivation quality. Among the tested inhibitors, vapor phase TMOS showed the best Ru-blocking performance for NGA while being removable by lithography for GA on TiO₂, whereas DMA-TMS was most effective on SiO₂. These learnings were extended for the ASD of Ru films in hard mask applications. This is divided into two parts. The first part involves achieving ASD on SiO₂/TiN 3D line patterns by deactivating the NGA, while the second part addresses GA creation through lithography and a photocatalytic TiO₂ layer. In the first section, DMA-TMS is utilized to passivate the SiO₂ NGA, allowing for selective deposition of Ru on the TiN underlayer. We emphasize the critical importance of the oxygen co-reactant pulse time in achieving selectivity. While a longer oxygen pulse time promotes a higher GPC of Ru, it is essential to maintain a process that does not negatively impact the NGAs. This balance is attained using the carbonyl precursor by adjusting the pulse time of the oxygen co-reactant, optimizing both GPC and selectivity. Additionally, selectivity is enhanced through corrective measures, such as re-passivation with DMA-TMS or employing an etch cycle. In the second part, ASD of Ru is achieved using EBECHRu and Ru(TMM)(CO)₃ precursors, with GA created by exposing the TMOS-passivated TiO₂ surface through UV lithography. However, while ASD is successful for larger patterns, the defects in the NGA increase significantly for smaller patterns, resulting in a diminished contrast between GA and NGA after Ru ALD. This may be due to the Gaussian distribution of UV energy, which degrades the organic inhibitor in the unexposed NGA, reducing selectivity—a challenge that becomes more significant at the nanometer scale. These findings highlight the importance of understanding the interactions between substrates, inhibitors, and precursors in ALD. Notably, our results demonstrate that the mechanisms of ASD can vary across different patterning scales. It is also crucial to recognize that not all ALD processes are inherently compatible with ASD schemes, and process modifications may be required to achieve optimal selectivity. These insights can contribute to the advancement of ASD processes for semiconductor manufacturing applications.