AVS2013 Session PS-TuP: Plasma Science and Technology Poster Session

Fluorocarbon films deposited using c-C₄F₈/N₂/Ar capacitively coupled plasmas are studied with the goal of understanding the effects of N₂-addition on surface chemistry. The plasma system used is a parallel-plate discharge sustained by a 13.56 MHz power source with the electrode diameter 25.4 cm and a gap distance 6 cm . Quartz crystal microbalance (QCM), Fourier transform infrared spectroscopy (FTIR), x-ray photoelectron spectroscopy (XPS), and contact angle measurement are used to characterize the film properties. QCM measurements show a monotonically increase in the deposition rate with N₂ addition, suggesting that CN_xF_y deposition rate surpassed the rate of forming volatile species. FTIR spectra show an additional adsorption near 1350 cm^-1 for conditions with N₂ addition. This can be caused by either the incorporation of nitrogen into the polymer chain or by creation of additional disordered sp² carbon in the film. The XPS C_1s spectrum of the film shows four major peaks, assigned to CF_x (x= 1, 2, 3) and C-CFbonds. With N₂ admixture, CFN bonds appear in the C_1s spectrum, while the amount of CF₃ and CF₂ bonds only decrease slightly. In support of the XPS data, the water contact angle on FC films remains 100~110°and does not change noticeably with N₂-addition. At the end of the presentation, the process significance in N₂ addition into c-C₄F₈/Ar/N₂ plasmas will be discussed.

Gallium nitride and related alloys have been expanding their applications for next-generation optoelectronics such as HEMTs or white LEDs. Plasma etching processes are necessary especially for ternary compounds such as AlGaN and InGaN. To control precisely profiles at nano-scale and to reduce plasma-induced damages, which influencing the device performance,^1-3 the mechanism of reaction of ions and radicals in plasma with surfaces should be clarified. In this research, we investigated the etching mechanism of AlGaN in chlorine plasma with focusing on surface stoichiometric composition.

Samples were Al_0.32GaN, Al_0.47GaN and GaN epitaxial films grown on Si substrate. Photolithographic patterning of UV photoresist was conducted. After chemical cleaning, an inductive coupled plasma etcher was used; the typical condition is antenna power of 200 W, bias RF power of 0 - 120 W, Cl₂/Ar mixture gas flow rate of 30 sccm at 1 Pa. The samples were chucked electrostatically on the stage kept at room temperature. X-ray photoelectron spectroscopy (XPS) was used for stoichiometric composition analysis.

When changing the ratio of Cl₂/(Cl₂+Ar) from 0 to 100%, the etch rate of GaN and AlGaN was increasing with the Cl₂ ratio. The etch rate in all cases saturated as increased the Cl₂ flow ratio above 40%. Order of the etch rate among the films were Al_0.47GaN < Al_0.32GaN < GaN, which could be explained by their sputtering threshold energies for Al-N of 11.52 eV and Ga-N of 8.92 eV.⁴ Since the etching products such as AlCl_x have high volatility compared with GaCl_y, a layer consisting of low Al-to-Ga ratio was possibly formed on the AlGaN surface. We will discuss the effects of plasma parameters on both etching profiles and surface stoichiometry in order to elucidate comprehensively understanding of the etching mechanism.

This work was supported by the Knowledge Cluster Initiative (Second Stage), MEXT, Japan. We would like to thank Taiyo Nippon Sanso Corp. for sample preparations.

1. S. Chen, et al., AIP adv. 2, 022149 (2012).

2. S. Chen, et al., J. Appl. Phys. 112, 053513 (2012).

3. S. Chen, et al., Jpn. J. Appl. Phys. 51, 111002 (2012).

4. J. Ladroue, et al., J. Vac. Sci. Technol. A 28, 5 (2010).

Extreme UltraViolet Lithography (EUVL) requires reflective mask blanks, manufactured by ion beam sputtering a multilayer stack of thin films, primarily Mo and Si, onto a mask substrate. At least 40 bilayers of Mo and Si are necessary to produce a surface which has sufficient EUV light reflectivity for use in high volume manufacturing exposure tools. When contaminant particles deposit between these layers, the EUV light is absorbed or scatters irregularly, rendering the mask blank unusable. One possible source of such particles is bombardment of shields in the deposition chamber by energetic particles scattered from the ion beam and target and “overspill” of the tails of the ion beam off the edge of the target under oblique target angle of incidence. Shields are used to cover targets that are not in use and prevent deposition or sputtering nearby surfaces and equipment. These shields must be able to accept many successive layers of deposition without flaking and forming particles of deposited material. They must also be able to withstand ion beam overspill bombardment, while forming a minimal amount of particles.

In order to evaluate improved shield materials and surface finishes, shield samples were placed under a broad angle ion beam and particles were collected on a witness plate. Shields of various materials and surface finishes were compared to determine the lowest level of particle formation. The total number of particles on the witness plates was quantified using laser scattering particle detection. Particle sizes and shapes were quantified with high resolution SEM imaging of the witness plate, and their composition was determined using backscattered electron imaging. The shield samples themselves were also analyzed using SEM to check for qualitative features, such as plateau formation, that may indicate the mechanism of particle formation.

Time-dependent dielectric breakdown (TDDB) is a major concern for low-k dielectrics. The TDDB properties of porous organosilicate glass (OSG) and photopatternable low-k dielectric (PPLK) films after plasma and/or VUV exposure are evaluated using contstant voltage time-to-breakdown measurements. To examine the effect of plasma exposure on TDDB degradation, dielectric films were exposed to Ar plasma in an ECR reactor. To separate the effect of vacuum ultraviolet (VUV) radiation on TDDB from charged particle bombardment, synchrotron radiation was used. This also has the advantage of being able to vary the wavelength of the radiation continuously. Dielectric films were exposed under different photon energies with the same fluence. After plasma or VUV exposures constant-voltage time-to-breakdown measurements were made for each sample. Weibull statistics were used. Both the results from PPLK and OSG showed that when the samples were exposed to plasma, significant degradation in breakdown time was observed. The VUV radiation under different photon energies shows less effect on the TDDB of both PPLK and OSG for the same fluence as the plasma. Charge-to-breakdown measurements also show similar results.

This work has been supported by the Semiconductor Research Corporation under Contract No. 2012-KJ-2359 and by the National Science Foundation under Grant CBET-1066231.

Plasma process control applications require acquisition of diagnostic data at a rate faster than the characteristic time scale of perturbations to the plasma. Diagnostics based on optical emission spectroscopy (OES) of intense emission lines permit rapid noninvasive measurements with low-resolution (~1nm), fiber-coupled spectrographs, which are included on many plasma process tools for semiconductor processing. The use of OES is an established practice to determine when a process is completed, i.e., the process “endpoint,” by detecting changes in intensity in optical emissions of key gas-phase chemical species, and OES can also be used to detect the presence of impurities and monitor changing reactor wall conditions. More detailed real-time information about the plasma state is increasingly desirable for process monitoring, however, due to progression in the semiconductor industry toward plasma processes with both tighter tolerances and multiple steps, i.e., where operating parameters are varied over the course of the process. In this work,* we examine the utility of plasma optical emissions from argon measured with a low-resolution spectrograph (Verity 1024 SH) as a real-time monitor of plasma parameters during the course of a plasma process, based on a rapid method to monitor and analyze the intensities of a select group of Ar emission lines to dynamically determine the following plasma parameters. Electron temperature and density are relevant parameters for characterization of the dynamic behavior of processing plasmas, because gas phase reactions are driven by collisions involving energetic plasma electrons. Metastable and resonance level concentrations are also relevant as these species play significant roles in plasma processing, through energy released when they de-excite upon reaching substrate surfaces, and through the emission of VUV photons which enhance surface reactions. These parameters are unambiguous indicators of the instantaneous plasma state and as such may play a valuable role as monitors for closed- loop process control. Results will be presented for argon and argon/mixed-gas (Ar/N₂, Ar/O₂, Ar/H₂) inductively coupled plasmas. Accuracy of the results (which are compared to measurements under static conditions made by Langmuir probe and white-light absorption spectroscopy) are typically better than ±15%. The system time resolution is ~0.1 s, which is more than sufficient to capture the transient behavior of many processes, limited only by the time response of the spectrograph used.

Support by NSF grant PHY-1068670 and Applied Materials Inc. is gratefully acknowledged.

* J. Vac. Sci. Technol. A 31(2), 021303 (2013).

Coupling oscillation is a newly discovered plasma oscillation mode due to the coupling between discharge circuit and magnetic circuit[1],[2],[3]. As we know, Hall thrusters always depend on the coils to produce the indispensable magnetic field. Line and line, layer and layer between grading coils exist stray capacitance. The inevitable stray capacitance, the coils and the equivalent resistances of magnetic circuit consists of a high level resistance-inductance-capacitance resonance network with natural frequency. With the excitation of the time-varying Hall current, the time-varying induced electromotive force is induced in the magnetic coils, which leads to the oscillation of the coil current. The resonance of the resistance-inductance-capacitance network gives rise to the oscillation of the coil current with a large oscillating component at the natural frequency. This oscillation affects the discharge oscillation. As a result, a self-cosistent equilibrium state is formed and the discharge current has the same oscillation frequency. Roughly speaking, oscillation frequency spectrum ranges from several kilohertzs to megahertzs. A parameter - coupling coefficient was applied to represent the intensity of coupling between the magnetic and discharge circuit[4] . According to the previous studies, the coupling coefficient is related to the materials and the cross-sectional area of the magnetic coils and the magnetic circuit of the Hall thruster. However, in our recent study on coupling oscillation, it was found that hall current equivalent position has important effects on coupling intensity between discharge circuit and magnetic circuit. This cause the different of coupling coefficient on different operation condition of Hall thrusters. Through non-intrusive measuring methods of hall current equivalent radius and axial position, it indicates that with the increase of discharge voltage and magnetic field intensity the hall current equivalent radius increases and its axial position move to the exit plane. Thus, the coupling coefficient grows larger and the coupling intensity between discharge circuit and magnetic circuit increases. Therefore, the suppressing methods of this instability should be involved in the future.

[1] L.Q.Wei, C.S.Wang,et al Contrib. Plasma Phys. 52, 761(2012)

[2] D.R. Yu, L.Q. Wei et al., Plasma Sources Sci. Technol. 16, 757 (2007).

[3] H. Pfister, K. Gaff et al., “50th Annual Meeting of the Division of Plasma Physics” 53 (2008).

[4] L.Q. Wei, B.H. Jiang et al., Plasma Sources Sci. Technol. 18, 045020 (2009).

Capacitively coupled SiH₄/H₂ plasmas (CCP) have been widely employed for depostion of Si film for applications such as thin film transistors and thin film solar cells. For Si thin film solar cells, requirements for yielding higher conversion efficiency at a lower cost calls for a high rate and high uniformity plasma porcess for deposition of microcrystalline silicon (m-Si) film. A better understanding of the plasma discharge is needed. In this study, a 80 MHz SiH₄/H₂ CCP discharge has been investigated by numerical simulation based on fluid model, with 27 species and 47 reactions. Simulation results show that plasma density, as well as the number densities and fluxes of H, SiH₂, and SiH₃, important reactive species for the growth of m-Si film, as well as the ratio of H/SiH₃ flux, a key indicator for crystal fraction of m-Si film, increase with rf power, as expected. Compared with the case in 27.12 MHz, the results also show that plasma density for the 80 MHz discharge are higher, due to frequency effect in the dominate electron heating mechanism, and, consequently, the densities of the key radical species also drop. More importantly, the ratio of of H/SiH₃ flux is also higher for the 80 MHz cases. Further more, the plasma potential, which determines the energy of ions incident on the grounded substrate electrode, decreases as operation frequency decreases. Consequently, a higher crystal fraction can be obtained by increasing the frequency of the rf power of plasma reactors, due to less damage on films causes energy ions bombardment. Simulations have also been carried out for different gap spacing between two electrodes and calculation results show that plasma density increases as the gap spacing decreases, as a result of lower electric field strength.

*Work supported by the National Science Council of ROC.

Neutral Loop Discharge (NLD) plasmas are being investigated as a new plasma source to produce good directionality and high fluxes of ion bombardment at substrate surface¹⁾ for semiconductor device fabrication. In an NLD plasma reactor, a Neutral Loop (NL) can be produced (zero-field path) that is generated by cusp-shaped magnetic field from direct current magnets. It has been shown theoretically that magnetic reconnection occurs in the neutral loop region when radio-frequency fields are applied²⁾. During the reconnection process, the DC magnetic field and the r.f. magnetic field cause disconnection and reconnection of the magnetic-field lines over an r.f. period. The reconnection results in the potential to heat electrons stochastically on the NL. Changing the position and the diameter of the NL can modify processing uniformity over a large area and thus improves plasma processing without need to move or rotate the workpiece. However, to confirm this, the details of plasma parameters as a function of position of the NL are produced by numerical simulations of electron motions near the NL in the presence of an r.f. The relationship between the r.f. magnetic field and its corresponding electric field and the average electron energy is also investigated.

This Work was supported by the Semiconductor Research Corporation under Contact No. 2012-KJ-2359 and the National Science Foundation under Grant CBET-1066231.

[1] T. Uchida and S. Hamaguchi., J. Phys. D: Appl. Phys. 41 (2008) 083001

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

Hydrocarbon conversion plays an important role in our energy economy. For example, the conversion of natural resources such as methane into hydrogen, carbon monoxide, ethanol, and other chemical fuels is essential to current and future energy needs. Typically, hydrocarbon conversion (specifically, methane reforming) is carried out by high-temperature (>500 °C), catalytic processes. To lower the temperature requirements and improve the conversion and selectivity, non-equilibrium plasmas have been explored for the conversion of hydrocarbon feedstocks¹. However, the power requirements and stability at high pressures have been obstacles to achieving significant improvement. In this study, we explore the application of a novel class of atmospheric-pressure plasmas, microplasmas², for the conversion of hydrocarbon gases including CH₄ and CO₂. As carbon dioxide emissions increase globally, reactions consuming CO₂ may become a necessity. The reaction of CO₂ with CH₄, known as dry methane reforming, is endothermic and normally requires high temperature and pressure and a catalyst; however, a non-equilibrium microplasma could potentially carry it out at room conditions. Additionally, when coupled with a catalyst, plasmas in general have been shown to have a synergistic effect³ that improves conversion beyond that of just the plasma alone or catalyst alone.

In this study, CO₂ and CH₄ were introduced into a microplasma in varying feed ratios and flow rates. The effluent was characterized by simultaneous gas chromatography and mass spectrometry to identify and quantify products. Reactive intermediate species were monitored by optical emission spectroscopy and soot formation was detected by aerosol ion mobility measurements. Results for CO₂ and CH₄ conversion as well as selectivity to specific products such as CO, H₂, and higher order hydrocarbons will be presented, as well as the formation of soot.

References

[1] Fridman A, Gutsol A, and Rabinovich A. "Combustion-assisted Plasma in Fuel Conversion." J. Phys. D: Appl. Phys. 44 (2011).

[2] Mariotti D, and Sankaran R M. "Microplasmas for Nanomaterials Synthesis." J. Phys. D: Appl. Phys. 43 (2010).

[3] Chang M B, et al. "Review of Plasma Catalysis on Hydrocarbon Reforming for Hydrogen Production—Interaction, Integration, and Prospects." Applied Catalysis B: Environmental 85 (2008).

Methane is of particular interest due to its use in diamond deposition processes, and presence in fusion and combustion plasmas. Methane is known to be a difficult molecule to simulate, combined with the difficulty of obtaining electronic excitation and dissociation for both theorists and experimentalists alike[1], this presents a relative lack of data for this important molecule.

Quantum-mechanical calculations can be the answer. The calculations presented use the R-Matrix method: where the configuration space is divided into two regions. There is an inner region, where the physics is complicated by exchange and correlation effects, and an outer region, where greatly simplified equations can be solved. The programs used to carry out these calculations are provided by the UKRmol suite. See Tennyson [2][3] for a detailed review of electron-molecule collisions using the R-matrix method. Quantemol-N provides an expert system for running these codes as well as adding extra functionality tailored to provide data for plasma models.

Using Quantemol-N, several different cross sections and properties were calculated for methane, including electron impact elastic, excitation, rotational excitation, differential, momentum transfer, ionisation, and dissociative attachment cross sections. This data can be coverted in the reaction rates and used for modeling of plasma processes.

[1]: M. Ziolkowski et al., J. Chem. Phys., 137, 22A510 (2012)

[2]: J. Tennyson, Phys. Rep. 491, 2-3 (2010)

[3]: J. Tennyson et al, J. Phys. Conf. Series, 86, 012001 (2007)

The magnetic neutral-loop discharge (NLD) was developed in 1994.[1] In this work we designed an NLD reactor using a stainless-steel chamber, instead of the commonly used quartz chambers in previous work because of the need for such a system in microelectronic processing. The vacuum chamber lies in the middle of three sets of magnet coils. With DC currents flowing in the opposite direction in the middle set of coils, a circle on which magnetic field is zero, i.e. a neutral loop(NL), can be produced in the middle of the chamber. In order to generate plasma, 13.56 MHz RF is inductively coupled into the chamber with a spiral antenna, through a quartz window located on one end of the chamber. The reactor can be operated in two modes, (1) an NLD mode when there are oppositly directed DC currents in the magnet coils, or (2) an ICP mode when there are either no DC currents or same direction DC currents in the magnet coils. In the NLD mode, the plasma was observed to be brighter near the location of the NL than in the center. This difference was further confirmed with measurement of the optical spectrum using an OceanOptics spectrometer, which shows the relative plasma glow brightness at the NL is as twice high as from the center of the chamber, and about 10% higher than the ICP plasma mode.

By adjusting the ratio of the DC currents running in the magnet coils, the radius of the NL can be changed. Both experiment and simulation show that the glow follows the change of the NL radius, especially at low pressure measured with a monochrometer and photomultiplier and compared with that observed from the ICP mode as well as other reactors. Although the location of the argon peaks are the same, the relative heights of the peaks and their widths are strong functions of the operating pressure and r.f. power for both modes.

1. H Tsuboi, M. Itoh, M. Tanabe, T. Hayasi and T. Uchida, Jpn. J. Appl. Phys 34 2476 (1995)

In the paper, we show the size effect of metal particles on PEPVD of high-k dielectric thin film for MOSFETs.

In the case of a fewer deposition of Hf, high density hafnium metal nano particles with 2-6 nm diameter are formed on SiO₂/Si surface and subsequent nitrogen plasma exposure (~10 min.) leads to the growth of HfSiON high-k dielectric film with a SiN interface layer. The exposure of atomic nitrogen and ions from the non-equiliblium plasma enables the introduction of N into the film and increases the interfacial reaction rate of Hf and SiO. Within the first 1 min., the Hf nano particles are oxynitrided with the N atoms from the plasma and the O atoms supplied from the lower interface judging from the XPS analysis. The nitridation rate is quite high and the quantitative measure shows the sticking of the N atom to the Hf nano particles reaches 1.0, while it is an order lower for thicker Hf layers more than 50 nm . A contact AFM survey of the Hf nano particles indicates the nano particles are liquidous due to Melting-point depresssion [2]. The following plasma exposure (~ 5 min.) enables the diffusion of Si atoms into the higk -k film from the underlying SiO layer. Following N₂ ICP exposure continuously increases the N atom fraction in the film which is relatively slow compared to the initial stage. The XPS spectrum shows the sea incorporated is mostly nitrides in the film. The spectrum also indicates the interfacial SiO layer is nitrided and this leads to the minimized EOT of the high-k stack structure.

[1] H. Watanabe, Appl. Phys. Lett . 85, 449 (2004).

[2] M . J . Stowell, Proc . Roy . Soc . Lond . A., 318, 231 (1970).

What if a liquid is injected directly into low-pressure plasmas? The current material processing with low-pressure plasmas (< 100 Pa) requires the gas-phase precursors in many cases. The limitation is sometimes irritating. The technique we have proposed, the direct liquid injection into low-pressure plasmas, enables liquids for plasma processing without applying any heat. Also, this technique enables the injection of solids through a liquid. For example, one could control the number of nano-particles in a film that is produced with a plasma to change the mechanical property of the film. Or, one could leave the clusters of solids on a substrate by evaporating a liquid off to print a three-dimensional structure. It is also fine that one uses heat-sensitive materials such as proteins, or even bacterias because the temperature of plasma processing is generally low. This technique definitely expands the possibility of material processing with plasmas.

With our best knowledge, not so many people researched the combination of direct liquid injection and low-pressure plasmas. Coppins suggested these kinds of plasmas to call as misty plasmas[i] [#_edn1] because one can consider that the droplets are another material state of particles. In contrast to dusty plasmas, misty plasmas can regulate the increase of particle temperature due to the droplet evaporation. Ward patented the configuration to realize the plasma processing in 2005.[ii] [#_edn2] However, our experiences showed that the configuration should create unstable plasma so that the process should be difficult.

This presentation shows the some investigations of the technique from some theoretical calculations and from the experimental observations. Our calculations shows that the evaporation time of droplets becomes half even at a plasma with 10¹⁰ cm^-3 and 2 eV due to the contribution of three-body recombination on the droplet surface. On the other hand, our experimental results are far from the computational results because of our technical difficulties. Our time-resolved measurements of plasma density and optical emission intensity showed that a plasma dimmed due to the massive vapor at the first several milliseconds. This means that the plasma gives almost no contribution to the droplets’ evaporation because the transport time of vapor is much shorter than that of droplets. This result indicates that the control of initial vapor plays a role to realize this technique. In the poster, we will summarize these results shown above with recent progresses.

[i] [#_ednref] M. Coppins, Phys. Rev. Let. 104, 065003 (2010).

[ii] [#_ednref] Ward, US Patent No. 2005/0227018 A1.

Carbon dioxide is usually considered as an end-product in chemistry because of its high stability. Due to the high quantities of CO₂ produced, its conversion received more and more attention in the recent years [1]. The transformation of CO₂, with CH₄ as second reactant and using atmospheric plasma technology shows that in good conditions, both gases can be converted into valuable products [2-3].

The conversion of CO₂/CH₄ mixtures was performed using a dielectric barrier discharge atmospheric plasma using Argon as the main plasmagen gas. Gas Chromatography was used to determine the composition of the gas after plasma treatment and this treatment was carried out in a new type of reactor developed in the laboratory. We demonstrated that the synthesis of syngas (CO and H₂) and small organic molecules such as C₂H₆, C₂H₄ is totally possible in this type of discharge [Fig. 1].

The study is focused on the effect of the plasma parameters on the CO₂ and CH₄ conversion rate. The parameters which are evaluated in this work are the ratio of CO₂/CH₄ flow rates and the power supplied.

We demonstrated that the CO₂/CH₄ ratio in the mixture has an important impact on the conversion rate. However, there is no real interaction between active species of these two gases since oxygenated organic compounds have only been detected at trace amounts. Nevertheless, the absence of oxygen after CO₂/CH₄ plasma is always obtained, suggesting that atomic oxygen is consumed in a plasma which contains only a few quantity of CH₄.

On the other hand, the effect of the power was clearly visible, showing a linear increase for both the CO₂/CH₄ conversion rates and the production of syngas according to the supplied power [Fig. 2]. This suggests that the number of electrons circulating between the electrodes has a huge impact on the conversion rate.

The detection of emitting species generated in the plasma (such as CO₂⁺ and CO₂ from the Fox, Duffendack and Barker’s system, different peaks of H, OH, O…) was also carried out via optical emission spectroscopy (OES).

Finally, we observed a different effect of two plasmagen gases (Argon and Helium) on the conversion of CO₂ and CH₄. Indeed, the conversion of CH₄ is better when He is used as carrier gas compared to the use of Ar, whereas we observed an opposite effect for the conversion of CO₂.

[1] T. Sakakura, J-C. Choi and H. Yasuda, Chem. Rev. 107 (2007) 2365−2387

[2] X. Tao, M. Bai, X. Li, H. Long, S. Shang, Y. Yin and Xiaoyan Dai, Progress in Energy and Combustion Science 37 (2011) 113-124

[3] A-J. Zhang, A-M. Zhu, J. Guo, Y. Xu and C. Shi, Chemical Engineering Journal 156 (2010) 601–606

The physical processes underlying low-pressure, low-temperature industrial plasma tools encompasses substantial ranges of space and time. This presents significant difficult for simulation. Large processing tools produce atomic scale features with process times often exceeding 1000's of seconds while applied frequencies are typically in the MHz or even GHz range. Tractable simulations, therefore, must typically make a number of rang limiting assumptions. An extreme simplification which is often employed when probing large parameters spaces is the “global-model” approximation, where volume-averaged chemistry balance equations are solved over the parameter space of interest. Though this may be regarded as an extreme approximation in view of the complexity of typical processing tool geometries, careful consideration of the transport and loss effects often yields surprising agreement with experiment. However, these models do have limited potential and cannot hope to account for spatially dependent phenomena, which may play a significant role in many industrial processes. Spatially resolved simulations, on the tool level, are much less commonly employed for state-space exploration and are often restricted to limited regions of parameter space, or significantly simplified chemistries. However, as the prevalence of multi-core desktop computers continues to grow rapidly, it is not unusual for anengineer to have easy access to a great many processing cores. Utilizing relatively coarse mesh tool simulations, in conjunction with such computer resources now makes spatially resolved state-space mapping tool simulations feasible for short term investigations. In this poster we will present the results of such a study with an SF6etching chemistry done using the new Quantemol - Virtual Tool (Q-VT) plasma simulation package.

A planar structure GaN-HEMT enables high-speed switching with a two-dimensional electron gas (2DEG), but it is a normally-on device while a normally-off device is preferable. To realize normally-off planar GaN-HEMTs, precise thickness control of the AlGaN layer at the gate is necessary. However, reproducible, stable thickness control of the AlGaN layer in the etching process is challenging when manufacturing normally-off GaN HEMTs. For this reason, we developed a slow etching process with low damage and high reproducibility, combined with in-situ thickness measurement using an optical interferometric film thickness monitor.

Manufacturing GaN-HEMTs with a triple channel (Camel) structure requires an etching process with high GaN to AlGaN selectivity on the p-GaN substrate. We developed an etching process with GaN to AlGaN selectivity of 60:1. We also applied interferometric film thickness monitoring for high etch process reproducibility. Another challenge in the Camel GaN-HEMT structure is to prevent current collapse, which prevents drain current, caused by electron traps on the interface of the AlGaN and insulation layers. We managed this issue by lowering the interface states of the AlGaN.

The substrate heating by the plasma on the magnetron sputtering process is a knowledge process that has not been extended studied such as involve complex effects like ion, excited and neutral surface bombardment and radiation heat from the plasma. To this end we have constructed a multi- point thermocouple to study the spatial variation of the substrate heating under a range of experimental conditions, plasma power and Ar gas pressure during DC magnetron sputtering of titanium. Additionally, we have studied the effect of allowing the thermocouples to be floating, connected to ground and reactive environment. The results showed that the substrate temperature can reach temperature more than 200ºC with a plasma power of 200W and that the highest temperature is localized in front at the race track due to a higher degree of particle bombardment.