University of Houston • University of Houston-Clear Lake • ISSO Annual Report Y2002—pp. 80-84

High Intensity, Large Area, Energetic (10-100s of eV) Neutral Beams

Demetre J. Economou (UH)

Abstract
A novel neutral beam source can generate a high flux (~ 10 mA/cm² equivalent) of directional neutrals with controlled energy. Ions are extracted from a plasma through a grounded metal grid with high aspect ratio holes. Ions suffer grazing angle collisions with the inside walls of the holes turning into neutrals. Molding of the plasma sheath over the holes controls the resulting neutral beam properties. Modeling of plasma molding was performed in an effort to optimize the flux, energy and directionality of the outcoming neutrals. A thin sheath (compared to the hole diameter) favored a larger energetic neutral flux at the expense of neutral energy and directionality. A relatively thick sheath produced neutrals of higher directionality at the expense of neutral flux. Neutral energy and directionality both increased by increasing the plasma (sheath) potential. Patterned polymer films etched with an oxygen neutral beam using the source exhibited anisotropic profiles, implying that the neutral beam can, indeed, be made highly directional.

ENERGETIC NEUTRAL BEAMS (10-100S OF EV) ARE USEFUL IN many applications ranging from the manufacture of charge-free microelectronics to tests of materials of interest to NASA such as International Space Station (ISS) materials exposed to atomic oxygen. Sources of energetic neutral beams developed^1,2 suffer from one or more of the following problems: low intensity (flux), complicated apparatus, substrate contamination, and small area coverage.

We have developed a novel neutral beam apparatus (Fig. 1), which can generate a high intensity (~10 mA/cm² equivalent current) directional neutral beam of controlled energy (10-100s eV) over relatively large areas.¹ Positive ions created in a high density inductively coupled plasma are extracted through a grounded metal grid with high aspect ratio holes. Ions suffer grazing angle collisions with the inside walls of the holes and turn into neutrals. Experiments have shown that we can realize greater than 99 percent neutralization. Even more recently, Japanese researchers³ have used the same source concept also to generate an energetic neutral beam. Although this novel neutral beam concept has been demonstrated for microelectronics applications requiring beam energies of over 100 eV, testing materials for space applications requires beam energies of ~ 10 eV.

Plasma Molding
A sheath forms over any surface in contact with plasma, including the surface of the grid used to extract ions from the plasma (Fig. 2). Inside, the sheath ions are accelerated to the desired energy controlled by the sheath potential (the difference between the plasma potential and the grounded grid potential). Ideally, ions should be accelerated in the sheath so that they have a small angular dispersion when they enter the holes. Ions would then suffer a grazing angle collision with the inside wall of the holes and neutralize.

Since little energy is lost in grazing angle collisions, resulting neutrals would retain most of the energy and directionality of the parent ions. The motion of ions inside the sheath depends on the electric field distribution, which, in turn, depends on how the plasma sheath "molds" over the holes.

The important length scales that control system behavior are the plasma sheath thickness, L_sh, and the size of surface features. As an example, Fig. 1 provides a schematic of plasma molding over a trench of width D. In case (a), L_sh << D, the sheath thickness is much smaller than the trench width. The plasma-sheath interface (meniscus) conforms to the shape of the surface topography. At the other extreme (case c), L_sh >> D, the plasma-sheath interface is essentially planar, as if the trench were non-existent. The plasma simply does not feel the presence of the surface topography. In the intermediate case (b), L_sh ~ D, the plasma-sheath meniscus "bends" gently over the trench mouth becoming planar away from the feature. The depth of the trench H (or aspect ratio (H/D)) is another important parameter which affects ion flow inside the trench. Cases (a)-(c) would result in drastically different ion flow inside the trench. Specifically, the ion flux striking the sidewalls of the trench reflecting as a neutral beam will depend critically on plasma sheath molding over the surface feature.

Problem Formulation
Investigators utilized a combined fluid/Monte Carlo simulation. The self-consistent fluid simulation predicted the two-dimensional profiles of ion density, ion fluid velocity, and electric field in the sheath over the surface feature. Electric field profiles were then used in the Monte Carlo simulation to predict the ion (and energetic neutral) energy and angular distributions along the contour of the feature. At low operating pressure, charge exchange in the gas phase is not likely to occur. Thus, energetic neutrals resulted mainly from ion neutralization on the sidewalls of the trench. Details of the simulation have been provided in a previous publication.⁴

The system employed in this work to simulate plasma molding over the holes of the extraction grid is shown in Fig. 3. A conductive substrate (Fe in this case) with a deep trench (width D and depth H) is located at the bottom of the domain. The potential was specified on the substrate (F_w) and at the upper boundary (F_o), while the sides were symmetry planes (Ñ_nF = 0). At the upper boundary, the argon ion density (n_o) and electron temperature were also specified. Note that the electron density was set equal to the ion density at the upper boundary, in accordance with the quasi-neutral plasma approximation. The plasma sheath, where charge neutrality breaks down, evolved self-consistently, provided that the upper boundary of the domain was several times thicker than the sheath. Larger electron (ion) densities and lower electron temperatures (smaller Debye lengths) result in thinner sheaths (for a given sheath potential). Actual neutral beam sources have bottomless holes instead of the geometry considered here. Nevertheless, it is believed that the results obtained in this study should be similar to bottomless hole geometry, especially for high aspect ratio features.

Results and Discussion
In neutral beam processing, energetic neutrals (instead of ions) promote surface reactions on target materials. In the neutral beam source described in Panda et al,⁵ ions are thought to neutralize by grazing angle collision with the inside surface of the high aspect ratio holes of the extraction grid. Neutralized ions exiting the bottom of the holes form an energetic neutral beam. Therefore, the flux, energy, and angular distributions of energetic neutrals at the trench bottom of Fig. 3 are of great interest.

The variation of (spatially) average neutral flux at the trench bottom as a function of the sheath thickness is shown in Fig. 4(a). For comparison, the ion flux at the trench mouth is also displayed. For the system pressure of 5 m Torr, the mean free path of Ar⁺ ions is ~1 cm. Thus, energetic neutrals are produced mainly by ion neutralization on the substrate surface. Therefore, the ion flux at the trench mouth should be equal to the total flux of ions and energetic neutrals at the bottom, assuming no surface trapping at the Fe substrate. When the sheath is thin and most of the ions entering the trench are diverted (to strike the sidewall and neutralize), the neutral flux at the bottom is almost equal to the ion flux at the mouth.

Figure 4. (a) Average (spatially) ion flux at the trench mouth and average (spatially) neutral flux at the trench bottom as a function of sheath thickness. (b) Average (spatially) energy and impact angle of fast neutrals at the trench bottom as a function of sheath thickness. The fluxes were normalized by the undisturbed value of ion flux on a flat wall. The trench was 500 m-wide and 2500 m-deep. The potential of the upper boundary was 100 V, and the wall was grounded.

For thick sheaths, the ion flux at the bottom increases and the neutral flux at the bottom tends to decrease. The (spatially) average impact energy and angle of neutrals at the trench bottom are shown in Fig. 4(b). When the sheath is thin, a strong horizontal component of the electric field is present, and ions are deflected to strike the sidewall at large angles (with respect to the y-axis). Ions then lose a large fraction of their kinetic energy upon collision with the wall, resulting in low energy neutrals.

For a thicker sheath, ions have more vertical momentum, producing fast neutrals with smaller angles and larger kinetic energies. For example, for L_sh = 113 m, only neutrals with kinetic energy of several eV are collected at the bottom. For L_sh = 1064 m, energetic neutrals with more than 80 percent of the sheath voltage (100 V in this case) impinge on the bottom surface with better directionality (impact angle is ~15 degrees off normal). Fast neutrals with high vertical momentum are desirable for neutral beam applications. This effect can be achieved by increasing the sheath thickness for a fixed trench width. However, a thicker sheath implies a smaller plasma density and, thus, a smaller ion (and energetic neutral) flux. One may increase sheath thickness without influencing plasma density by applying a larger bias voltage (Fig. 6). This can be achieved in systems in which plasma production is decoupled from substrate bias.

Flux, energy, and impact angle of fast neutrals at the trench bottom are considerably affected by the trench depth H. Figure 5 shows (spatially) average neutral distributions at the bottom of 500 m-wide trenches with different depths. The neutral flux increases with depth, since ions entering the holes have a better chance of striking the wall and neutralize for deeper holes. The average energy and angle of fast neutrals go through a shallow maximum as the depth is increased. For very thin sheaths, the produced neutrals keep colliding with the sidewall during their transit toward the bottom (multiple collisions); their kinetic energy is lower.

Since the wall was grounded, the potential of the upper boundary F₀ was essentially equal to the sheath potential. According to Child’s law, the sheath thickness scales approximately as F₀^3/4n₀^-1/2.⁶ Thus, plasma molding depends on the sheath potential F₀. Figure 6 shows (spatially) average neutral distributions at the trench bottom as a function of potential of the top (upper) boundary of Fig. 3 (essentially the sheath potential). The variation of neutral flux at the bottom suggests that the ion trajectory becomes more anisotropic (vertical) at higher sheath potentials. Thus, fewer ions strike the sidewall, and a smaller flux of neutrals is observed when the sheath potential is increased. The enhanced ion directionality is also clearly seen in terms of the impact energy and angle of energetic neutrals. As the sheath potential is increased, higher energy neutrals impinge on the bottom at smaller angles off normal.

As seen in Figs. 4 and 6, there is a tradeoff between neutral beam energy and directionality. Specifically, in order to achieve a low energy (~10 eV) beam, of interest to materials testing for space applications, one must establish a small plasma (sheath) potential (Fig. 6) and/or a small sheath thickness (Fig. 4). The smallest plasma potential will be the floating potential (~15-20 V) under these conditions. Capacitive coupling from the coil must be suppressed completely for this purpose. A small sheath thickness can be obtained by increasing plasma power and, thus, plasma density (smaller Debye length). However, a small beam energy will be accompanied by a large divergence angle of the neutral beam.

Finally, an oxygen plasma was used to etch polymers with energetic O-atoms produced by the neutral beam source of Fig. 1. Researchers demonstrated high rate (6,000 A/min), microloading free, high aspect ratio (5:1) etching, with straight sidewalls of sub-250 nm features (Fig. 7). Importantly, there was no detectable (by XPS) contamination of the wafer by the grid material (aluminum).

Acknowledgments
This work was supported in part by the Institute for Space Systems Operations and the National Institute of Standards and Technology (NIST).

References
1K. Yokogawa, T. Yunigami and T. Mizutani. "Neutral-Beam-Assisted Etching System for Low-Damage SiO2 Etching of 8-inch Wafers," Japanese J. Appl. Phys. Part 1 35 (1996): 1901-05.
2M. J. Goeckner, T. K. Bennett and S. A. Cohen. "A Source of Hyperthermal Neutrals for Materials Processing," Appl. Phys. Lett. 71.7 (1997): 980.
3S. Samukawa, K. Sakamoto, and K. Ichiki. "Generating High-Efficiency Neutral Beams by Using Negative Ions in an Inductively Coupled Plasma Source," J. Vac. Sci. Technol. A 20.5 (2002): 1566-73.
4D. Kim and D. J. Economou. "Plasma Molding Over Surface Topography: Simulation of Ion Flow, and Energy and Angular Distributions Over Steps in RF High-Density Plasmas," IEEE Trans. Plasma Sci. 30.5 (2002): 2048-58.
5S. Panda, D. J. Economou, and L. Chen. "Anisotropic Etching of Polymer Films by High Energy (100s of eV) Oxygen Atom Neutral Beams," J. Vac. Sci. Technol. A 19.2 (2001): 398-404.
6M. Lieberman and A. Lichtenberg. Principles of Plasma Discharges and Materials Processing. NY: J. Wiley, 1994.

Publications
Kim, D. and D. J. Economou. "Plasma Molding over Deep Trenches and the Resulting Ion and Energetic Neutral Distributions," J. Vac. Sci. Technol. 21.4 (2003): 1248-53.
Kim, D. and D. J. Economou. "Simulation of Plasma Molding Over a Ring on a Flat Surface," J. Appl. Phys. (Accepted.)
Kim, D., D. J. Economou, J. R. Woodworth, P. A. Miller, R. J. Shul, I. C. Abraham, B. P. Aragon, and T.W. Hamilton. "Plasma Molding Over Surface Topography: Simulation and Measurement of Ion Fluxes, Energies and Angular Distributions Over Trenches in RF High Density Plasmas," IEEE Trans. Plasma Sci. (Accepted.)

Presentations
Economou, D. and D. Kim. "Plasma Molding Over Surface Topography and Resulting Ion/Fast-Neutral Distribution Functions," 55th Gaseous Electronics Conference, Minneapolis, MN, Oct. 15-18, 2002.
Economou, D. and D. Kim. "Plasma Molding Over Trenches and Resulting Ion/Fast-neutral Distribution Functions," 49th International Symposium of the American Vacuum Society, Denver, CO, Nov. 4-8, 2002.

Funding and proposals
Economou, D. J. "Sheath Structure, and Flux, Energy and Angular Distributions of Ions over Topographical Features in Plasma Processing." National Institute of Standards and Technology, Sept. 2002-Aug. 2003, $75,000.

PDF (K)
Table of Contents

Institute for Space Systems Operations - Y2002 Annual Report
Copyright © 2003