DOI: 10.1002/adma.200701994 Self-Sealing of Nanoporous Low Dielectric Constant Patterns Fabricated by Nanoimprint Lithography** By Hyun Wook Ro, Huagen Peng, Ken-ichi Niihara, Hae-Jeong Lee, Eric K. Lin, Alamgir Karim, David W. Gidley, Hiroshi Jinnai, Do Y. Yoon, and Christopher L. Soles* Highly porous nanoscale structures are critical for a range of emerging technologies. Prominent examples include semicon- ductor devices, [1–6] sensors, [7] separations, [8] supports for catalyst, [9] and photonic devices. [10,11] Recently, we demon- strated that it is feasible to directly pattern sub-100 nm features with high fidelity into porous spin-on organosilicate glass (SOG) materials using nanoimprint lithography (NIL). [12] This novel use of the NIL technology was motivated by the enormous cost savings that could be realized by directly patterning the interlayer dielectric (ILD) material suitable for semiconductor interconnect structures. Directly creating the pattern in the functional material eliminates the subsequent processing steps that are normally required to the transfer the sacrificial resist pattern into the functional material. [13,14] This has the potential to greatly reduce the Back-End-Of-The-Line (BEOL) manufacturing costs. Some of these eliminated steps, such as the photoresist etching and ashing, are especially harmful to porous dielectrics with interconnected pore networks because the etching byproducts can diffuse into and contaminate the interior regions of the porous material. However, a new concern with the direct imprinting of the ILD material is that the patterns are essentially used ‘‘as-imprinted’’. If the imprinting process affects the key properties of the functional material, the performance of the resulting device can be impacted. This manuscript explores how NIL patterning affects the characteristics of the highly porous SOG patterns that are critical to their performance for ILD applications. There are several technical and measurement challenges for highly porous SOG patterns. The porogen is typically an organic component that phase separates from the SOG during the spin coating process into nanoscale domains. The porogens volatilize at the elevated temperatures where the SOG is vitrified into a cross-linked network, templating nanoscale pores inside a hard, vitreous SOG matrix. The primary concern with direct patterning is that the process exposes this two phase material to large shear deformation flow fields, greatly increasing the surface or interfacial area. Then the two related issues are 1) pattern shrinkage/collapse and, 2) imprint-induced changes to the intended porosity upon vitrification and porogen volatilization. We previously showed that pattern shrinkage is of minimal concern in sub-100 nm features imprinted in a porous SOG material. [12] The focus here is to quantify how both NIL and porogen removal affect the pore characteristics of this material that are critical for determining the dielectric performance. The NIL patterning process is described elsewhere in greater detail. [12] Briefly, the thermal embossing form of NIL was used to create patterns onto as-cast films of a poly(methyl- silsesquioxane) (PMSQ)-based material blended with 20% by volume of a thermally degradable porogen (Tetronics 150R1, purchased from BASF [15] ). The PMSQ matrix material used in this study is prepared from methyltrimethoxysilane, dimethoxy- dimethylsilane and 1,2-bis(triethoxysilyl)ethane with a 7:1:2 molar ratio. [16] Fully vitrified films of this PMSQ-based resin (without porogen) exhibit an elastic modulus of 10.5 GPa and a dielectric constant of 2.84 at 1 MHz. [5] Specular X-ray reflectivity (SXR), critical dimension small angle X-ray scattering (CD-SAXS), and field emission scanning electron microscopy (FE-SEM) were used to quantify the pattern fidelity and vitrification induced shrinkage with sub-nm level precision. These earlier measurements show that the NIL can reproducibly create patterns with a high fidelity of pattern transfer, regardless of the existence of the porogen. Vitrifica- tion induced shrinkage was also precisely characterized, revealing a small amount of vertical shrinkage in the pattern height direction but negligible change of pattern width; the patterns maintain their fidelity/shape through the vitrification and the porogen burn-out. COMMUNICATION [*] Dr. C. L. Soles, Dr. H. W. Ro, Dr. H.-J. Lee, Dr. E. K. Lin, Dr. A. Karim NIST Polymers Division 100 Bureau Drive, stop 8541, Gaithersburg, MD 20899 (USA) E-mail: csoles@nist.gov Dr. H. Peng, Prof. D. W. Gidley Department of Physics, University of Michigan Ann Arbor, MI 48109 (USA) K.-i Niihara, Prof. H. Jinnai Department of Polymer Science and Engineering Kyoto Institute of Technology Kyoto 606-8585 (Japan) Prof. D. Y. Yoon Department of Chemistry, Seoul National University Seoul 151-747 (Korea) [**] Official contribution of the National Institute of Standards and Technology; not subject to copyright in the United States. This work is supported in part by NIST Office of Microelectronic Program, University of Michigan, the System IC 2010 Project of Korea, and the Chemistry and Molecular Engineering Program of Brain Korea 21 Project. We also acknowledge the Nanofabrication Laboratory of the Center for Nanoscale Science and Technology (CNST) in NIST for providing facilities for the NIL process. 1934 ß 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2008, 20, 1934–1939