AVS2012 Session PS-ThP: Plasma Science and Technology Poster Session

Due to their extraordinary range of properties, polymeric materials play an essential and ubiquitous role in everyday life. Plasma-modifications of polymeric surfaces further extend the range of their applications, especially with fluoropolymers due to their widespread in biocompatibility and self-cleaning coatings applications.

The aim of our work is to compare the treatments carried out with two distinct plasma sources on a selection of fluoropolymers: PVF, PVDF, TrFE, PTFE, Nafion©, FEP and PFA. The first treatment is performed with a post-discharge generated by an RF plasma torch, supplied with a helium-oxygen mixture and operating at atmospheric pressure. The second treatment is achieved in the plasma phase of a DBD and operates at atmospheric pressure, also in the same gases. In both cases, we discuss the influence of the treatment time and of the O₂ flow rate on the surface properties. A comprehensive study of the plasma/polymer interface has also been performed by mass spectrometry and optical emission spectroscopy to identify the species responsible of the surface texturization, i.e. the species breaking the atomic bindings on the surface top atomic layers.

The surface properties are characterized by determining the advancing and receding WCA so as to show if the treated polymers can be classified into the Wenzel or Cassie-Baxter models. The super-hydrophobicity of PTFE has already been achieved. Moreover, two liquid probes have been used: milliQ water (H₂O) and diiodomethane (CH₂I₂) to discuss the validity of the Fowkes theory.

We report results for a new plasma source which uses a low pressure (~10^-4 Torr) discharge with applied electric and magnetic fields. A dc voltage of 20-100 V is applied between the RF plasma cathode and the anode-chamber. The magnetic field is varied between 50-500 G. Under such conditions this cross-field discharge is shown to sustain an efficient ionization of xenon and argon gases (n_e^max ~ 10¹¹-10¹² cm^-3, T_e~ 1-10 eV). Probe measurements revealed that the magnetized plasma of the DC-RF discharge has a non-Maxwellian EEDF with a depleted high-energy tail. It is also shown that spatial variations of the EEDF are governed by a non-local electron heating and anomalous electron transport across the magnetic field. An important implication of the above results is that the anomalous electron transport may degrade the magnetic filter effect, which supposes to separate “hot” and “cold” groups of plasma electrons.

Here we present a practical estimation on deposition rate for a sputter system. The sputter system consists of four sputter sources in a high vacuum chamber, with an independent power supply for each sputter source. Quick estimation of deposition rate could be helpful for efficient experimental planning and reducing the number of experiments and time.

First, theory: Our early JAP paper* disclosed an energy balance model and derived a sputter rate = k (P –P_o). The physical meaning is that the sputter rate has a linear relationship with plasma source power P, with a nearly constant offset P_o, and constant slope k. Here we present several different materials results which illustrate the good fit between deposition rate and power within the working range of 100W to 550W.

Second, P_o estimation: The physical meaning of P_o is the power consumed to sustain plasma, which is dependent on the plasma boundary condition (chamber/plasma source geometry), ion diffusion co-efficient, electron temperature etc. In this situation, since the plasma ion is primarily Agon ion with fixed chamber geometry, P_ois nearly constant within the range of 25 to 52 W. With a simple estimation of 39W, an error of only 13W could be introduced. For a typical 200~400W experimental condition, the error from this estimation is small (2~5% error).

Third, slope k estimation: The different materials trends with slope k are calculated M/D *Y* V and plotted. (M: is the atomic weight, D: is the density, Y: is the sputter yield at 300V for Argon ion, and V is the DC voltage of the sputter source). The model assumes that the slope k is proportional to the sputter yield at the DC voltage, and converts the mass to volume for thin film thickness based on the deposition rate calculation. In our case, DC voltage V is close to 300V, so that sputter yields at 300V were used. However, further investigation showed that the simple sputter yield might not be accurate enough, and the correction for the sputter DC voltage could provide a better fit. The slope k trend agreed well with model estimation M/D*Y *V. Thus, results here implied that the slope k could be estimated for different materials.

Thus, a quick deposition rate estimation method was derived in comparison with experiments, which is expressed as:dep rate= M/D*Y*V * k’ *(P-P_o), and K’ and P_o are chamber specific that need to be experimentally determined. Once calibrated from one experiment, the deposition rates of other materials can be estimated from this formula.

In addition, this study could be helpful in applying to other chambers and sputter sources for a quick deposition rate estimation, even for alloy target or reactive sputtering.

EUV lithography provides much bigger dry etch challenges than 193-nm lithography did. Its depth of focus (DOF) is

so small that the thickness of EUV resist is much thinner than that of 193-nm resist. Although EUV resist requires

higher etch selectivity than 193-nm resist does, UV/VUV cure, which has been used in 193-nm resist, is not an

effective technique to enhance etch selectivity and physical strength. This is because UV/VUV light is essentially

transparent in EUV resist. Etch selectivity normally attributes to polymer which prevents material from being etched.

In the case of EUV resist, line width roughness (LWR) is easily enhanced by polymer because its physical strength

is so low, and the resist width is so small that the resist cannot tolerate the stress of polymer. This is quite contrary

to the ideal reactive ion etch (RIE) which diminishes LWR. Therefore, an alternative curing technique for EUV resist

We introduced the e-beam curing technique for 193-nm resist utilizing a direct current superimposed (DCS) capacitively-coupled plasma (CCP) at the 55^th AVS in 2008. Negative high DC voltage is applied in DCS CCP,

and positive ions collide with the upper electrode. Thus, secondary electrons are emitted from the upper electrode.

The applied negative high DC voltage also accelerates the emitted secondary electrons, which turn into ballistic

electrons. The curing mechanism of 193-nm resist is scission and cross-linking of polymer by e-beam which is a

consequence of ballistic electrons. Considering its mechanism, the e-beam curing should be available to any

polymer. Thus, we applied this technique to EUV resist, and investigated the effect of e-beam in EUV resist. In order

to identify the cured thickness and chemical structure change, the surface was analyzed with cross-sectional SEM

and ToF-SIMS, respectively. As a result, the curing effect was confirmed. In addition to the curing technique, we

also invented a coating technique with a silicon compound material by sputtering the upper electrode utilizing the

DCS technology. This coating technique increases the etch selectivity to EUV resist. In this technique, silane type

gases are not required, making it easily applicable to manufacturing.

Incident species from plasma needs to be strictly controlled to fabricate advanced devices. We found that simultaneous injection of VUV/UV radiation and radicals caused the marked etch rate enhancement of SiN_x:H films [1]. This indicates that not only ions, but also VUV/UV radiation affects the surface reaction of radicals with SiN_x:H. In this study, we investigated the effect of transmitted VUV/UV radiation on underlying interface traps of SiN_x:H/Si substrate by using the pallet for plasma evaluation (PAPE) [2].

A dual frequency (60/2 MHz) CCP reactor was used. SiN_x:H films were deposited on Si substrates by PECVD and those were exposed to CF₄/O₂ plasma. The thicknesses of SiN_x:H films were 200 nm. To investigate the effect of radiation, MgF₂ (> 115 nm), quartz (> 170 nm), and borosilicate crown glass BK7 (> 300 nm) windows were put directly on the SiN_x:H film. The capacitance-voltage (C–V) characteristics of SiN_x:H on Si substrate were analyzed to study the interface-trap density (D_it).

The penetration depths of photons strongly depend on the wavelength of the VUV/UV radiation and the absorption coefficient of materials. The wavelengths of photons were classified into three major ranges by their interaction with the material:

1) Photons are absorbed in the material and damage the film and/or enhance the surface reaction [1].

2) Photons are transmitted through the material and absorbed in the underlying interfaces, resulting in the increase in D_it.

3) Photons are transmitted through both the material and interface, and there is no impact on the damage generation.

The D_it increase is the most serious issue for advanced device fabrication since the D_it directly degrades electrical performance. Thus, the damage of SiN_x:H/Si substrate interface was investigated by C–V measurement. The D_it of SiN_x:H/Si substrate was unaffected by the VUV radiation (< 170 nm) since all the high-energy photons were absorbed in the SiN_x:H film (case 1). When the photons in the UV region (> 170 nm) were irradiated, the D_it increased and a negative charge was generated in the interface (case 2). This indicates that the VUV/UV radiation transmitting through the upper dielectrics causes the electrical characteristics of underlying devices to fluctuate. The UV radiation (> 300 nm) had almost no effect on the increase in D_it (case 3) due to the lack of absorption in both the material and interface.

Thus, the wavelength dependence of increases in D_it needs to be investigated for possible interfaces in advanced devices.

[1] M. Fukasawa et al., Jpn. J. Appl. Phys., 51 (2012) 026201.

In setups for plasma research there are in most cases several sensors available to monitor the plasma conditions. However there are also numerous plasma systems for non-research applications that have limited or no plasma diagnostics available at all.

To solve this problem we have designed, a so called plasma diagnostics box. The plasma diagnostics box is made out of two units, a sensor head and the data logger box containing the data logger, batteries and electronics, connected by a flexible tube.

The sensor head contains multiple stand-alone sensors. The prototype device contains five basic plasma monitoring sensors, a PT100 temperature sensor, a heat flux sensor, silicon photodiode, a double Langmuir probe and a Faraday cup for measuring several key plasma parameters

The two unique features of the plasma diagnostics box is that it is a complete standalone unit and that it can be placed within the vacuum/ plasma system. This means that there is no need for a feedthrough flange on the plasma/ vacuum system. However, if a feedthrough is available, the data logger box can be placed outside the vacuum, and the sensor head alone can be inserted in the system.

Because of the flexible connection the sensor head can be placed anywhere within the plasma. This makes it possible to characterise the plasma at different positions.

In this contribution we will discuss the design and manufacturing of the plasma diagnostics box. Besides the design we also will present the data we have measured with the plasma diagnostics box in one of our research plasma setups that we have available at TNO. The results obtained with these measurements will be benchmarked against the commercially available diagnostics on our research plasma setup.

Non-neutral plasma, or plasma made up of particles of a single sign of charge can be isolated and confined for long periods of time, permitting detailed examination. Non-neutral plasma possesses characteristics that are unique (such as a dynamic equilibrium state), while exhibiting some phenomena that are similar to neutral plasma (such as supporting the propagation of electrostatic waves). Previous work on non-neutral plasma has been conducted largely in cylindrical traps, but our toroidal trap offers the opportunity to test theoretical predictions that are not observable in cylindrical geometry. One such effect is the transport and mode damping due to a phenomenon called magnetic pumping. We report on experiments in which we excite toroidal analogs of two different so-called ‘diocotron’ (or flute-like) modes in the plasma in order to diagnose its characteristics and behavior. The frequency of the m=1 diocotron mode is proportional to the total trapped charge in the plasma. The damping of this mode has a strong dependence on magnetic field that is not presently understood, though theory suggests there is a dependence on, as yet unmeasured, plasma temperature. The frequency of the m=2 diocotron mode provides information about the average density of the plasma and its time variation, as well as its dependence on magnetic field, and other experimental control parameters. The combination of total charge and average plasma density provides a measure of plasma transport. This work is supported by National Science Foundation Grant No. PHY-0812893.

Magnetic random access memory (MRAM) has made a prominent progress in memory performance and has brought a bright prospect for the next generation nonvolatile memory technologies due to its several advantages. Dry etching process of magnetic thin film is one of the important issues for the magnetic devices such as magnetic tunneling junctions (MTJs) based MRAM. MTJs which are the basic elements of MRAM can be used as bits for information storage. CoFeB is a well-known soft ferromagnetic material, of particular interest for magnetic tunnel junctions (MTJs) and other devices based on tunneling magneto-resistance (TMR), such as spintransfer-torque MRAM. Recently, transferring the pattern by using an Ar+ ion milling is a commonly used, although the redeposition of sputter etch products on the sidewalls and the low etch rate are main disadvantages of this method. Other method, which reported the etch rates higher than 50 Å/s for magnetic multilayer structures using Cl₂/Ar plasmas, is also proposed. However, the chlorinated etch residues on the sidewalls of the etched features tend to severely corrode the magnetic material. To remove this problem, the etching of MTJ layer by using organic-based gases such as CO/NH₃, CH₃OH, etc. are actively investigated currently.

Economical Si etching is required to fabricate through silicon via ( TSV ) integration architecture for multi-layered packaging, micro-electro-mechanical system (MEMS), and surface patterning of solar panels. Chemical dry etching techniques by highly reactive ClF₃ gas [Ibbotson et al. J. Appl. Phys. 56 (1984) 2939] has been reported but the contamination by Cl is problematic. In this study, we have been investigating a new and economical Si chemical dry etching technique using nitric oxide (NO) and molecular fluorine (F₂) gas mixture. Atomic fluorine (F) is generated by mixing these gases at room temperature by the reaction of F₂ + NO à FNO + F. Kinetic energy of F at 0.8 eV and the change in post-reaction bonding energies of Si, SiF_x (x = 1~4), Si-NO, Si-NOF, and HF were calculated by B3LYP/6-311+G(d) in Gaussian 09. Si etch rate, etch directionality, etch selectivity, surface chemical bonding states, and surface roughness were evaluated by scanning electron microscopy (SEM), X-ray photoelectron spectroscopy, and Fourier transform infrared spectroscopy, and atomic force microscopy to elucidate the Si etch dynamics in molecular level.

A prototype etching reactor was fabricated using the quartz tube with the inner diameter of 7.5 mm and the length of 150 mm. Flow rate of gases introduced into this chamber was Ar/5%F₂ ~ 38.5 sccm, F₂ ~ 1.9 sccm and NO ~ 1.5-5 sccm while maintaining the constant pressure at 620 Pa. The corresponding flow rate ratio, NO/NO + F₂, was 0.44 ~ 0.72. Etching was performed for 15 ~ 600 s. Single crystal and poly-Si samples were prepared to determine the etch rate, which is calculated from the cross-sectional SEM images. Preliminary results show that the etch rate was increased from 0.2 ~ 1.8 m m/min at NO/NO + F₂ = 0.44 ~ 0.72. This measured etch rate was more than 100 times faster than the estimated value by F atom reaction with H terminated Si. The etch rate sharply increased with NO up to NO/NO + F₂ ~0.57 and became almost constant when NO/NO + F₂ > 0.57. The etched surface after the exposure to the atmosphere mainly consisted of SiO₂, indicating that the reaction between Si-F_x, Si-NO, Si-NOF and H₂O in the air may occur rapidly. The detail analysis of the change in surface chemical bonding state during and after etching is in progress. The etched surface became rough and scalloped Si was formed when NO flow rate was increased. The surface roughening may be due to the presence of Si-NO and/or Si-NOF bond prior to the formation of Si-F_x that would eventually form SiF₄. From these results, Si etching is initiated not only the presence of F but also the existence of F₂, NO, and FNO.

Fuel cells have been recognized as a potential clean energy-converting device due to their high efficiency and low emissions. However, two major technical gaps limit their commercializations: cost and reliability. Currently, platinum (Pt)-based catalysts and their corresponding cathode catalyst layers are among the major causes which limit their performance and cost for proton exchange membrane (PEM) fuel cells, although these are the most promising and practical fuel cell catalysts. Some approaches have been studied to reduce the cost and to improve the performance over twenty years, but there has been no real breakthrough yet. Recently, most researchers focused on the use of carbon catalyst and non-rare metal catalyst for replacement of Pt-based catalysts. The former groups are divided in organic dope methods and inorganic dope methods on carbon material such as nanotubes or graphene sheets. The major drawback of organic and inorganic dope method are, respectively, low thermal stability and limited controlling of the process. Meanwhile, non-rare metal catalysts show low catalytic activities.

In this study, we synthesized nickel/carbon nano-particles produced by solution plasma process.

SPP is a useful and simple method for the metal NPs synthesis because this non-equilibrium plasma can provide extremely rapid reactions due to the reactive chemical species, radicals and UV radiation produced in atmospheric pressure plasma operating in glow discharge limits and offering a suitable medium to control the chemical reactions inside the solutions.

The SPP was generated by the electrical discharge between opposite nickel electrodes. The glow discharge in 0.1 M KCl solution was produced by using bipolar pulsed power supply operated at 1~2 kV of voltage, 15 kHz of pulse frequency and 2 μs of pulse width. The diameter of the electrode was 0.6 mm and the interelectrode gap was 0.5 mm. In addition, 0.1 g of dispersion treated carbon (CNT, CNB) was inserted in 100 ml solution.

M/C NPs morphology was investigated by transmission electron microscopy, energy disperse X-ray microanalysis, X-ray Diffraction, Fourier transform infrared spectroscopy. Cyclic voltammetry was demonstrated for evaluating catalytic activities.

In conclusion, as applied voltage increased, diameter of synthesized NPs decreased. The smallest synthesized NPs with 3 nm showed the higher oxygen reduction rate in catalytic activities according to decrease diameter.

PTFE samples were treated by O₂/He plasma at various pressures, from low (6.67 Pa) to atmospheric pressure. Treatments were carried out using different mixtures of He and O₂. The first results highlight two completely different etching mechanisms depending on the gas mixture used and working pressure.

The etching was studied by mass loss measurements, weighing the sample before and after plasma exposure.

At low pressure the strongest etching occurs for pure oxygen plasma while no etching is detected for helium plasma. The opposite is observed at atmospheric pressure, the strongest etching takes place with pure helium.

X-ray photoelectron spectroscopy (XPS) analysis was used to study the chemistry of the PTFE surfaces as well as the etching products. High resolution F1s and C1s peak show different etching products as the pressure and/or the gas composition are changed. In all cases both carbon and fluorine are detected. Depending on the plasma parameters (pressure and gas composition) either CF₂ fragments (BE C1s = 292 eV) or CC (BE C1s = 285 eV) are detected.

Optical emission spectroscopy (OES) measurements were used to study the plasma phase and the etching products. A change in the He plasma emission was observed above a certain pressure (around 13.3 Pa). The plasma color changes from greenish to pink. This change could be related to the increase of the He metastable emission line at 389 nm. OES measurements also allowed us to detect etching products in the gas phase during the plasma exposure of the PTFE samples. Fluorine, CO and CO₂ lines were detected.

Those various results allow us to suggest two different mechanisms of PTFE etching.

In the case of low pressure O₂ plasma, the surface composition after the treatment and the etching products (detected by XPS and OES) suggest an etching mechanism where the C-F bonds are broken by charged particles (probably e^-). Oxygen then reacts with the carbon backbone to produce CO₂. The role of the charged particles was evidenced using different sample positioning and magnets.

In the case of atmospheric pressure treatments the etching occurs via the removal of CF₂ fragments. The main active specie responsible for the etching seems to be metastable He atoms.

In processing plasma sources for semiconductor device fabrication, both uniformity over a large area and controllability to obtain desirable plasma parameters are required. A magnetic neutral loop discharge (NLD) plasma¹⁾ has been proposed as a new plasma source which satisfies these requirements. The position and the diameter of the plasma can be easily controlled by changing the position and the diameter of a neutral loop (NL). It has been theoretically shown that the electron motion becomes nonlinear around the NL when the radio frequency (RF) electric field is applied along the NL perpendicular to the magnetic field lines²⁾. The electron makes meandering motions and acquires kinetic energy of several tens of electron-volts from the RF electric field where the meandering range contains the region between the NL and the electron cyclotron resonance (ECR) region, the length of which is designated by L^3,4). Through such a process, the electrons are heated efficiently without collisions. The electron behavior near the NL determines the uniqueness of this plasma production technique and therefore details of its motion are needed. In this work, in order to accurately understand the characteristics of an NLD plasma, a numerical analysis of electron behavior around the NL was performed based on a 2-dimensional model in which three-dimensional effects were taken into account. For obtaining the optimum conditions for plasma production, the relationship between the normalized electric field and the average electron energy were also investigated.

This Work was supported by Kyungsung University Research Program (2011) and by the Semiconductor Research Corporation under Contact No. 2008-KJ-1871 and the National Science Foundation under Grant CBET-1066231.

[1] Z. Yoshida and T. Uchida., Jpn. J. Appl. Phys. 34 (1995) 4213.

[2] Z. Yoshida et al., Phys. Rev. Lett. 81 (1998) 2458.

[3] Y. M. Sung et al., J. Vac. Sci. Technol. A18 (2000) 2149.

[4] Y. M. Sung et al., J. Vac. Sci. Technol. B20 (2002) 1457.

A laser-assisted plasma-coating technique at atmospheric pressure (LAPCAP) for depositing thin yttria-stabilized-zirconia (YSZ) films has been developed. This technique allows columnar-structured YSZ films with a thickness of 1~5 µm to be prepared on a Ni-based superalloy substrate at atmospheric pressure. The atmospheric pressure plasma is generated in a microwave-induced plasma torch system with a gas temperature T_g of more than 2,000 °C. Optical emission spectroscopy (OES) technique has been used to spatially analyze some critical characteristics of plasma, such as electron density (n_e > 10¹⁵ cm^-3), electron temperature (T_e ~ 1 eV), and plasma gas temperature (T_g ~ 800-1200 °C). The thermally grown oxide (TGO) layer is found to affect the film morphology significantly, and characteristics of TGO grown by pre-heating the substrate to 800-1200 °C are investigated. TGO in the form of α-Al₂O₃ with a thickness of ~ 1 µm is found to provide a means to deposit high quality, adhesive thin YSZ films on substrates with columnar microstructure, same as seen in films by high-vacuum electron-beam PVD method. The morphology and characteristics of the films have been compared at various deposition temperatures (100-1200 ºC) and laser energy density (1-10 J/cm²), using microanalysis techniques such as scanning electron microscope (SEM), focused ion beam (FIB), X-ray photoelectron spectroscopy (XPS), and X-ray diffraction (XRD).

Electron beam-generated plasmas are unique type of plasmas generated when high electron energy beam is injected into the gas. The beam preferentially ionizes the gas molecules, thereby producing daughter electrons with energies up to half of the initial beam energy. The daughter electrons, however, quickly lose energy as they collide with gas molecules. Thus, without an external electric field to heat them, the daughter electrons cool and they soon outnumber the beam electrons and the energetic daughter electrons responsible for ionization. The low-energy “plasma” electrons are not only far more populous, but also largely determine the plasma characteristics – i.e. the plasma density n_e, plasma potential f_p, and electron temperature T_e. Typically T_e is 1 eV or less in beam-produced plasmas and thus the kinetic energy of the ions bombarding the surface is below the polymer bond energies. In this talk we estimate the charged, excited and resonant species concentrations and the fluxes bombarding the polymer surface. We correlate the ion to radical ratios in different gas environments with observed morphological and chemical modifications in polystyrene, poly(methyl methacrylate) and polyethylene. This work was supported by Naval Research Laboratory Base Program.

We investigated the etch damage on the interface between the titanium nitride(TiN) and hafnium oxide(HfO2) during etching the carbon layer on the TiN. HfO2 is widely used as a high-k gate dielectric material to reduce the gate leakage current and improve the reliability in the semiconductor device. TiN and amorphous carbon layer(ACL) are deposited on the HfO2 as a gate electrode and mask material, respectably. We fabricated complimentary metal oxide semiconductor (CMOS) devices with patterning ACL on the TiN by dry etching and removing the TiN layer by wet chemical. We realized that an interlayer between the TiN and HfO2 layer was generated due to the plasma damage during the ACL etching and the interlayer was not removed clearly by the wet chemical, which resulted in device degradations. We studied the properties of the interlayer with changing the ACL etching recipe to find the key factor of generating the interlayer. The properties changed significantly with the ion energy and the gas composition of the ACL etch recipe.

*Work supported by the National Science Council of ROC.