AVS2009 Session PS1-ThM: Applications of Plasma-Surface Interactions

Recent experimental and simulation studies have revealed that plasma-organic surface chemistry can be strongly affected by synergistic interactions among ions, vacuum ultraviolet (VUV) photons and electrons at surfaces. In this talk, I summarize our recent studies of these synergies, focusing on various polymer and SiCOH low-k dielectric materials. When plasma and vacuum beam measurements, coupled with molecular dynamics (MD) simulations, are compared for various materials and various exposure conditions, certain patterns emerge. MD shows that (~ 100 eV) ion effects are restricted to several nm near the surface, but that their effects can depend strongly on the type of polymer and other species present. The synergistic effects of plasma-generated ions, photons and electrons can be understood in terms of a competition between bond-breaking scissioning reactions and bond-forming cross-linking and other reactions. The complexity of the results is due in part to the fact that these species have different depths of penetration, and that their bond breaking and bond forming reactions depend on the structure of the material. However, even greater complexity results from the fluence- or dose-dependent nature of electrons and ions: low dose result in scissioning and higher doses result in cross-linking. The effects of simultaneous exposure depend on both position relative to the surface and time. I illustrate these ideas with examples taken from PMMA-based 193 nm photoresists; poly-alpha methyl styrene (PaMS) and poly-four methyl styrene (P4MS); and ultra low k, nanoporous SiCOH. MD simulations and models of VUV photon penetration into polymers are used to interpret both plasma and vacuum beam experimental results.

Polymer photoresists are commonly used for pattern transferring in plasma etching of sub-0.1 micron features in microelectronics fabrication. Degradation and erosion of the photoresist (PR) is a major issue in controlling feature profiles, especially for high aspect ratio (HAR) features where etch times are long. For example, during fluorine based plasma etching of SiO₂, erosion of the photoresist leads to bowing of the top of the profile as the etch proceeds. Multilayer masking is one approach to minimize these effects. By depositing a hard mask layer under the PR, the pattern is transferred to the hard mask layer before the PR is eroded away. Another promising strategy is to deposit a hard mask layer onto the PR surface during the process. This can be achieved in-situ by low energy ion bombardment of the PR surface to both promote cross-linking and produce dangling bonds, coincident to there being a flux of Si radicals. The resulting Si-C bonding provides a hard-mask like surface. Another is to promote fluorocarbon deposition on the PR mask to slow its erosion. These opportunities may be afforded by dc-augmented capacitively coupled plasmas (CCPs) in which the silicon covered dc electrode is sputtered. In this talk, we discuss scaling laws for profile and PR control derived from a computational investigation of a dc augmented single/dual frequency CCP reactor to generate an Ar/C₄F₈/O₂ plasma and fluxes of Si radicals by sputtering the dc electrode. Fluxes (energy and angle resolved) of ions, radicals and electrons are obtained from the Hybrid Plasma Equipment Model (HPEM) as a function of dc voltages, rf frequencies and rf bias powers. Profiles of features are then simulated by the Monte Carlo Feature Profile Model (MCFPM). Both multilayer mask and Si deposition strategies will be discussed. Etching selectivity between SiO₂ and mask material and feature profiles will be discussed as functions of Si fluxes, initial patterns and thickness of the photoresist.

*Work supported by Tokyo Electron Ltd., and the Semiconductor Research Corp.

Experiments were run in a remote expanding thermal plasma (ETP) reactor, in Ar and Ar/H₂ gas mixture compositions. The substrate holder was negatively biased (up to -100V) by means of a home designed pulsed power supply operating with a frequency up to 200 kHz and a variable duty cycle. Ion energy distributions have been measured by means of a planar gridded retarding field energy analyzer.

Two pulsed biasing approaches will be presented (asymmetric rectangular pulses and modulated pulses with a linear voltage slope during the pulse) and their applicability is discussed on the basis of an intrinsic capacitance of the processed substrate-layer system. The substrate voltage and current waveforms were measured and mutual relations with the obtained ion energy distributions will be shown for both aforementioned cases. To demonstrate the IED control achieved, the effective carrier lifetime of n-type c-Si wafers, passivated by an a-Si:H thin film, as a function of the flux and energy of bombarding argon ions was determined. The ion energy and ion flux was independently varied and threshold ion bombardment characteristics leading to degradation of the effective lifetime will be presented.

The goal of this study was to elucidate structural and compositional modifications of an organic photoconductor after an extensive exposure to plasma discharge by a charging element within an electrophotographic system. An organic photoconductor, commonly found in a variety of applications ranging from simple copiers to advanced high-speed digital presses, is the key element of the modern electrophotographic printing system. It facilitates formation of the latent image resulting from the area-selective light discharge of uniformly distributed charges deposited on its surface by the plasma. Its undesirable modifications may adversely impact the print quality.

Modifications of the photoconductor's surface layer were investigated with the help of Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy (ATR) and X-ray Photoelectron Spectroscopy (XPS). The experimental set-up was designed to simulate interactions between the plasma discharge and photoconductor occurring in a typical electrophotographic printing environment. Experiments were performed over an extended period of time corresponding to printing multiple pages. UV radiation and energetic particle bombardment originating from the plasma are the two major processes responsible for the photoconductor modifications. Therefore, further elucidation of the photoconductor modification phenomenon was obtained by employing the UV-only and the particle bombardment-only experimental conditions.

A long time exposure of the photoconductor to energetic neutral and charged species, and UV photons caused massive oxidation, substantial chemical bond breakage, and reformation of bonding configuration within a thin layer below the surface. This layer can be divided into two regions corresponding to the aforementioned photoconductor modification processes. The top region, having thickness of approximately 20nm - 30 nm, is formed primarily by the particle bombardment. It is heavily oxidized and its thickness is limited by the bombardment induced sputtering. Thicker region below is relatively free of oxygen and its molecular composition is distinctly different from the original photoconductor. It is primarily formed by UV-induced cracking of the benzene rings followed by reformation of the excited radicals into new molecular species. Thickness of this region corresponds to the UV penetration depth. The possibility of preventing the formation of a parasitic surface layer will be further discussed.

Selective etching of high dielectric constant (high-k) films over the underlying Si (and/or SiO₂) is indispensable in the fabrication of high-k gate stacks. In practice, the selectivity is usually not so high, owing to highly volatile halogen compounds of Si, and also to strong metal-oxygen bonds of high-k dielectrics and less volatile metal-halogen compounds. Profile control is also indispensable during etching of high-k: anisotropic profiles are required for high-k, and also profiles of gate electrodes on high-k are required to remain unchanged. This paper presents the control of selectivity and profile for high-k HfO₂ etching under low ion-energy conditions in BCl₃-containing plasmas, with emphasis being placed on a better understanding of the etching mechanisms concerned. Experiments were performed in both electron cyclotron resonance (ECR) and inductively coupled (ICP) plasmas, by varying pressure, additive concentration of O₂, Cl₂, and Ar, rf bias power, and also substrate temperature. Samples for etching were blanket HfO₂ and TaN films, and separate Si and SiO₂ substrates were also employed for reference. Samples of TaN/HfO₂ stack as well as separate HfO₂ and TaN masked with line-and-space patterns were also employed to examine the etched profile. We examined substrate surfaces by x-ray photoelectron spectroscopy, and investigated reactant and product species in the plasma during etching by optical emission spectroscopy and quadrupole mass spectrometry. A transition from deposition to etching regimes was found to be caused on all substrate surfaces, by varying pressure, by using additives such as O₂ and Cl₂, and by increasing rf bias power. In practice, surface inhibitor deposition was less significant for HfO₂ than for Si, SiO₂, and TaN; and the threshold bias power or ion energy for HfO₂ etching was in the range 10-20 eV, while the threshold was more than 20 eV for the other. A difference in pressure, additive concentration, and bias power for the transition between HfO₂ and Si (and/or SiO₂) gave rise to high or infinite selectivity of high-k over Si (and/or SiO₂), together with vertical high-k profiles. The difference for the transition between HfO₂ and TaN also gave no significant distortion of TaN profiles during HfO₂ etching, owing to passivation layers deposited on TaN sidewalls. Plasma and surface diagnostics indicated that inhibitor species for deposition are primarily boron-chloride polymers produced in the plasma, whose concentration largely depends on pressure, additive concentration, and plasma reactor (ECR and ICP), which in turn leads to a marked difference in etching characteristics.