Iso-dense Bias Research Articles

Quasi-atomic layer etching of silicon nitride

Atomic layer etching (ALE) is a promising technique that can solve the challenges associated with continuous or pulsed plasma processes—trade-offs between selectivity, profile, and aspect ratio dependent etching. Compared to silicon, oxide, and other materials, atomic layer etching of silicon nitride has not been extensively reported. In this paper, the authors demonstrate the self-limited etching of silicon nitride in a commercial plasma etch chamber. The process discussed in this paper consists of two sequential steps—surface modification in hydrogen plasma followed by the removal of modified layers in fluorinated plasma. In addition to the ALE characteristics, the authors also demonstrate that the process is anisotropic and the selectivity to oxide is >100. Although the saturated etch rate of one monolayer per cycle could not be attained, self-limited etching of silicon nitride still enables us to incorporate the benefits of atomic layer etching such as an absence of isodense bias and an extremely high selectivity to oxide into practical etch applications.

Implementation of atomic layer etching of silicon: Scaling parameters, feasibility, and profile control

Atomic or layer by layer etching of silicon exploits temporally segregated self-limiting adsorption and material removal steps to mitigate the problems associated with continuous or quasicontinuous (pulsed) plasma processes: selectivity loss, damage, and profile control. Successful implementation of atomic layer etching requires careful choice of the plasma parameters for adsorption and desorption steps. This paper illustrates how process parameters can be arrived at through basic scaling exercises, modeling and simulation, and fundamental experimental tests of their predictions. Using chlorine and argon plasma in a radial line slot antenna plasma source as a platform, the authors illustrate how cycle time, ion energy, and radical to ion ratio can be manipulated to manage the deviation from ideality when cycle times are shortened or purges are incomplete. Cell based Monte Carlo feature scale modeling is used to illustrate profile outcomes. Experimental results of atomic layer etching processes are illustrated on silicon line and space structures such that iso-dense bias and aspect ratio dependent free profiles are produced. Experimental results also illustrate the profile control margin as processes move from atomic layer to multilayer by layer etching. The consequence of not controlling contamination (e.g., oxygen) is shown to result in deposition and roughness generation.

Rigorous electromagnetic field simulation of the impact of photomask line-edge and line-width roughness on lithographic processes

The impact of edge profile roughness of the absorber lines on an optical photomask has been studied by means of rigorous electro-magnetic field (EMF) simulation for the mask diffraction spectrum and subsequent imaging. Roughness has been modeled using two different approaches, a sinusoidal description and an algorithm known from literature-based on Fourier transformation. The latter can arbitrarily create rough profiles and surfaces based on the three morphological parameters standard deviation σ, roughness exponent α, and correlation length ξ. In this study, the standard deviation has been kept fixed while varying the remaining two morphological parameters, still showing impact on the lithographic process. A software interface for use of the generated profiles with the Waveguide EMF solver of the Dr.LiTHO lithography simulation suite has been implemented. It was shown by means of image analysis and study of the resulting process windows that mask roughness is partially transferred to the aerial image. Isolated and dense features behave differently, leading inter alia to an iso-dense bias different to that of ideal lines. Due to the roughness, process windows now depend on the position along the line. Determing process windows at multiple positions for statistical analysis implies a reduction of the effective process window. Correlation length ξ has shown to be an important parameter and, thus, morphology should not be ignored in the modeling of rough lines. Tapered sidewalls can add to the shift of the process windows in the same order of magnitude.

PEB bake optimization for process window improvement of mixed iso-dense pattern

We have shown that process effects induced by extending the post-exposure bake temperature in the process flow of chemically amplified photoresists can lead to significant improvements in depth-of-focus (DOF) and exposure latitude (EL) and small geometry printing capability. Due to improved acid dose contrasts and a balanced optimization of acid diffusion in the presence of quencher, PEB temperature increase has enabled the printing of iso and semi-dense space of 0.2 µm and below with a large DOF, using binary masks and 248 nm lithography without expensing the iso-dense bias. The results and findings of a full patterning process in a device flow, with different PEB temperatures as a process enhancement, are presented. The main objective of this study is to demonstrate how, using KrF lithography with binary masks and no optical proximity correction (OPC) nor other reticle enhancement technique (RET), the process latitude can be improved. Lithographic process latitudes, intra-field critical dimension (CD) uniformity and resist profiles of different PEB processes are evaluated. Then, the after-etch profiles are also investigated to ensure the feasibility of this technique.

Verification of point-spread-function-based modeling of an extreme-ultraviolet photoresist.

A crucial component of lithographic modeling is the resist. Resists typically used at extreme-ultraviolet (EUV) wavelengths are derivatives of deep-ultraviolet chemically amplified resists. Models that describe these resists are often very complicated and are dependent on a large number of free parameters. Point-spread-function-based resist modeling serves as a simple alternative. I show this type of modeling to be a viable technique at EUV wavelengths by directly comparing modeling results with a variety of printing metrics, including process windows and isodense bias.

Optimization of 193 nm Contact Hole Resists for 100 nm Node.

This paper discusses the 193nm lithography of contact holes with various pitches for 100nm node. We have studied 193nm contact hole resists in view of resist components, process conditions and optical settings. Sidewall roughness was improved by optimizing photoacid generators. Side lobes were eliminated by applying higher post exposure bake temperature or modification of polymers. The influence of optical settings, types of masks and mask bias was discussed with simulation and lithographic results and guidelines for better resolution and iso-dense bias were proposed. The optimized formulation, AZ® AXTM1O50P has a high resolution combined with a large depth of focus and an iso-dense overlap window (130nm (NA=0.63) DOF 0.38μm @ Exposure latitude 10%).

Application of VEMA type ArF Resist to Sub-100nm Lithography.

It is expected that ArF lithography will be introduced for device manufacturing for sub-100nm nodes, as high NA ArF step and scan systems (NA=O.75) become available. We previously reported on a platform, based on a vinyl ether-maleic anhydride (VEMA) alternating polymer system. This platform demonstrated both good resolution and high dry etch resistance in comparison to other platforms based on acrylate and cyclic-olefin-maleic anhydride (COMA) polymer systems. The VEMA platform has been continuously improved to meet the increasing requirements, such as resolution, depth of focus (DOF), iso-dense bias, and post-etch roughness for real device manufacturing. This VEMA system is being implemented for sub-100nm device with high NA (NAO.75)=ArF exposure systems. In this paper, recent experimental results are reviewed.

Impact of exposure induced refractive index changes of photoresists on the photolithographic process

In many commercial and noncommercial photoresists the real and the imaginary parts of the refractive index are changed during exposure. Using a finite-difference beam-propagation algorithm, we analyze the impact of these nonlinear optical effects on the photolithographic process. Changes of the real part of the refractive index have a considerable impact on dose latitudes, sidewalls, swing curves, iso-dense bias and other process parameters. These effects become more dominant as the thickness of the resist layer increases.

A Novel Platform for Production-worthy ArF Resist

ArF lithography, in combination with chemically amplified resists, has been investigated as one of the most promising technologies for producing patterns below 100nm. In considering the polymer matrix for 193nm photoresist applications, factors such as sensitivity, transparency to 193nm radiation, adhesion to substrate, dry etch resistance, ease of synthesis, and availability of monomers are very critical. In these respects, remarkable progress has been made in development of ArF resist material. Polymers of acrylic and methacrylic esters show good imaging performance at 193nm, but have insufficient dry-etch resistance under oxide or nitride etch condition. On the other hand, cyclic olefin-maleic anhydride (COMA) alternating copolymers exhibit good dry etch resistance, but have poor resolution capability. We previously reported a new platform, based on a vinyl ether-malefic anhydride (VEMA) alternating polymer system, that demonstrated both good resolution and high dry etch resistance.In this paper, VEMA systems with improved lithographic performance are presented. The new platform (VEMA) showed good performance in resolution, depth of focus (DOF), isodense bias, and post-etch roughness. With conventional illumination (NA=0.6, sigma=0.7), 120nm dense line/space patterns with 0.4μM DOF were resolved. And 90nm L/S patterns 0.6μM DOF were resolved with off axis illumination (NA=0.63). Another important factor to be considered for the dry-etch process is post-etch roughness. In the case of VEMA system a clean surface was observed after etch under oxide, nitride, and poly conditions. The VEMA resist system is regarded as one of the most production-worthy material for real device manufacture.

Analysis of Quencher Diffusion in Chemically Amplified Resists and Its Effect on Imaging Characteristics.

Diffusion of quencher contained in most chemically amplified resists today can cause proximity effects such as dependence of pattern feature size on pattern density or iso-dense bias. Diffusion coefficients were measured for several quencher compounds, and significantly large diffusion coefficients (e.g. 0.003μm2/sec at 100°C) were observed compared with those observed for acid. To evaluate the influence of quencher diffusion on macroscopic imaging behavior of resists, a new resist model for mutual diffusion of acid and quencher was introduced and was implemented to our in-house lithography simulator. Simulation results based on the model explained various proximity effects observed on real wafers, assuming quencher diffusion length as large as 0.2μm. This is significantly long compared with acid diffusion lengths reported and agrees with the results of direct measurement. Our results suggest that very small amount of quencher can have a strong influence on imaging characteristics of chemically amplified resists.

Recent advancements in 193 nm step and scan lithography.

With the introduction of full field step&scan systems, 193nm technology development is currently being accelerated and resists are used closer to their final area of application, i.e. under realistic conditions of lens aberrations, stray-light and wafer coverage. In this paper, the lithographic performance of advanced 193nm resist materials has been evaluated on a full field step&scan system. Single layer and hi-layer resist processes are compared in terms of performance and complexity. Very similar lithographic performance is observed for both the single layer and the bi-layer approaches at 130nm. Optimization of illumination conditions (NA, sigma) is investigated as a way to enlarge processing windows and to reduce iso-dense bias. The application of a PSM illustrates the extendibility of 193nm lithography for the 100nm technology node. In general, significant progress in resist performance has been made but further improvements are needed before 193nm resists will reach the maturity level of today's state-of-the-art 248nm resists.

Optimization of DUV Negative Resists for 0.15 .MU.m Lithography.

Recent performance improvements of a negative tone chemically amplified DUV resist consisting of a hydroxystyrene based copolymer, a melamine crosslinker, a photoacid generator (PAG) and certain amine additives are described. The general trends of formulation and process changes on the lithographic performance were investigated using a response surface method. PAG concentration variations and optimization of the prebake conditions allowed for the elimination of microbridging in the subquarter micron region. The pattern profiles were improved by the optimization of the polymer properties and the selection of a specific amine/ammonium hydroxide combination. The effects of certain polymer properties and of individual amine components are discussed in more detail and a correlation with their structures is given based on dissolution rate measurements. The optimized resist material has a resolution potential below 0.15μm (NA=0.55) combined with a large depth of focus (>1.0μm @ 0.15μm iso lines), acceptable iso-dense bias, and small sensitivity towards changes in the post exposure bake conditions.

Reducing Proximity Effects in Optical Lithography

Using simulation, the influence of stepper parameters on optical proximity effects is explored. In particular, numerical aperture and partial coherence will be examined for a variety of feature sizes and types. Both one-dimensional and two-dimensional mask features will be studied. The impact of resist contrast will also be explored. In addition to the iso-dense print bias as a metric of proximity effects, the depth of focus as an overlapping of two focus-exposure process windows, one for the isolated line and one for the dense line, will be used. The optimum NA and σ will give the maximum depth of focus calculated from the overlapped process window. Finally, the statistical CD distribution methodology will be used to find the stepper settings that minimize the linewidth distribution spread for a given process.