Skip to main content

Zero damage Ultra-Low-K etch using the precursor CONDensation technique

Periodic Reporting for period 1 - ULKCOND (Zero damage Ultra-Low-K etch using the precursor CONDensation technique)

Reporting period: 2016-04-01 to 2018-03-31

Since the beginning of the 21st century, the scaling down in micro-electronics is impacted by the RC delay: the average time of transport of the electrons between two transistors is too long. A possible solution to this problem is to reduce the capacitance (C) value of the circuit connecting various active components (the interconnects). This is done by reducing the permittivity (k) of the dielectric present in between the metal wires. The emergent material family to replace the (dense) SiO2 is the porous organo-silicate glass (p-OSG). These compounds are porous, SiOCH-based and with methyl (Si-CH3) terminations. They show low-k value (k < 2.3 vs 4.2 for SiO2) and good mechanical properties (Young modulus ≈ 6 GPa). However, the integrations steps required to fabricate the interconnects are detrimental for both electrical and mechanical properties of the p-OSG. The ULKCOND project (for zero damage Ultra-Low-K etch using the precursor CONDensation technique) aimed at protecting the dielectric from damage during the plasma etching step (in-situ). The goal is to use micro-capillary condensation to condensate a gas into the porous low-k, without condensing on the surface. The condensation precursor are used together with other gases to etch the low-k. Because the pores are filled with a liquid/solid phase during the etch process, the methyl groups are expected to be protected from the plasma’s reactive radicals and VUV emission.
The low-k protection during critical steps (like etching) is of main importance for the scaling of future CMOS devices. ULKCOND has demonstrated that the cryo-etching is protecting very well the low-k. First of all, it avoid detrimental reaction with radicals. The best reagent also absorb detrimental VUV photons. Etching process can then be implemented on low-k, ensuring good electrical and mechanical properties. The direct application of this work will enable the fabrication of a new generation of more efficient microchips.
WP1 focused on sample preparation. A key objective of this task was to prepare and characterize thin porous low-k film with various porosity on 300 mm Si substrate. These films have been measured by ellipsometry porosimetry to measure the level of open porosity. Mechanical characterization (Hardness and Young modulus) and electrical measurements allowing k-value extraction have also been done using the methods developed in IMEC. Electrical devices (nano-interconnected structures, composed of a series of Cu wires separated by low-k dielectric) have been also prepared and characterized.
WP2 focused on the study of the condensation process into the porous material by micro-capillary condensation. State-of-the-art literature does not give satisfying explanations for this phenomenon. The fellow proposed a new paradigm to describe micro-capillary condensation into hydrophobic microporous materials. The Kelvin equation used up till now does not give an accurate representation of the micro-capillary condensation. A new general equation has been proposed by the fellow, which will be submitted as a Letter very soon. Regarding the experimental selection of molecules for micro-capillary condensation, five different organic candidates were compared. The condensation behavior was found to be driven not only by the pore radius but also by the wettability of the precursor with the surface. A better wettability allows to lower the condensation pressure in comparison with the vapor pressure of the gas at a given temperature (this is predicted by our new model but not by the Kelvin equation).
WP3 was focused on the mechanism of plasma etching in presence of a solid/liquid phase condensed in the pore network. The most relevant results were provided by studying the material after etching in different condition. The best candidate to protect the material was found to be the nerima (commercial name of this gas, which use has been patended by Air Liquide in the course of ULKCOND). Without plasma activation, nerima is inert with respect to the p-OSG material. The mechanisms of plasma damage propagation were determined in a SF6/nerima plasma discharge. VUV photons, emitted by the plasma, were found to be the single origin of damage in the low-k. With a different discharge, NF3/nerima, more reactions with the by-products/radicals are found. After processing in a nerima-containing plasma discharge, the mechanical and electrical properties are maintained at a temperature of around -10°C (or lower) in the conditions of the experiment. As expected, when condensed, nerima mitigate damage by avoiding detrimental reactions between the plasma by-products and the low-k; in addition nerima absorb the VUV between 100 to 130nm allowing even better protection of Si-CH3 bonds. This range of wavelength include the emission of SF6 at 106 nm for example.
WP4 was focused on pattern transfer optimization. A DOI was performed with NF3/nerima plasma in a dedicated 300mm tool. Good pattern transfer was demonstrated on a 45nm ½ pitch vehicle.
The use of micro-capillary condensation was proposed in 2013 for the first time. The first gas used for this was C4F8, a commodity gas used routinely for plasma etch, allowing condensation and good protection at -110 °C. In the present work, some new reagents provided by Air Liquide were screened and allowed some condensation up to -20 °C for a partial pressure of 2.5 mTorr. This allows to work above -50°C, enabling cost-effective industry-relevant plasma etch tool concepts. Moreover, the phenomenon of micro-capillary condensation is now described with a new paradigm, which is relevant for the present application but also in many other fields as MEMS technology or geology (oil extraction). Also, the damage propagation has been explained allowing improvement of this cryo-etching process in the future.
The work performed in the framework of the ULKCOND project allowed to develop, understand and improve the plasma etching of ultra-porous low-k’s. The new concept demonstrated within ULKCOND might be applied in the coming few years for various technologies, advanced CMOS and more-than-Moore (sensors, DNA detection).