Matthew Goeckner - Academia.edu (original) (raw)
Uploads
Papers by Matthew Goeckner
International Conference on Plasma Sciences (ICOPS), 1993
ABSTRACT
Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 2014
ABSTRACT The etch rate of deep features in silicon, such as trenches and vias, can vary significa... more ABSTRACT The etch rate of deep features in silicon, such as trenches and vias, can vary significantly with the feature aspect ratio (AR). Small AR features generally etch faster than large AR features. The reasons for this AR dependence include a slowing of the etch rate with increasing AR due to the necessary transport of molecules into and out of the features as well as ion flux reductions at feature bottom due to the angular spread of the ion flux and ion deflection caused by differential charging of the microstructures. Finding ways to reduce, eliminate, or reverse this AR dependence is both an active subject of research and difficult. In this work, instead of focusing on methods to reduce or prevent AR dependence in an etch process, the authors focus on methods to correct it after the fact. The authors show that an inhibitor film deposition step can be used under some circumstances to allow feature depth disparities to be corrected. This process can be used to correct feature depth disparities whenever the AR dependence of the inhibitor film deposition step is worse (larger) than the AR dependence of the following inhibitor etch step. To test the theory, a plasma process through SF6/C4F8/Ar mixtures was used to both produce trenches of various ARs having significant depth disparities and correct those disparities. The etch depth of small AR features can be held essentially constant while that of larger AR features is increased to match or even exceed.
ABSTRACT There are numerous computational models that can be used to study plasmas, each with adv... more ABSTRACT There are numerous computational models that can be used to study plasmas, each with advantages and disadvantages. Two of the most commonly used are the ambipolar model and the classical fluid model. The ambipolar model takes advantage of the ambipolar approximation, while the classical model includes the solving of the Poisson's equation. Both models were used to simulate the operation of the ICP electron source of the Electron Beam Exciter. Process conditions and geometrical variations were performed to optimize the electron density and electron to ion ratio in the beam extraction region. Results of this optimization along with deviations between the two models will be presented.
A systematic experimental study of the power, length, and gas pressure dependence of plasma densi... more A systematic experimental study of the power, length, and gas pressure dependence of plasma density and temperature profiles in thermally collapsed magnetized helium plasmas has been performed on the Princeton divertor simulator apparatus. The recombination edge and recombination radiation from highly excited levels (to n = 12) have been identified. The importance of radiation and neutral particles to heat transport has been demonstrated. A region of stable plasma detachment has been found and is suggested as being suitable for divertor operational space.
Hyperthermal-neutral beams have been proposed as a charge-free alternative to plasmas in select p... more Hyperthermal-neutral beams have been proposed as a charge-free alternative to plasmas in select processing steps. Existing prototype sources include three generic types: gas-dynamic, ion-neutral-charge-exchange and ion-surface-neutralization beam sources. The authors find that in surface-neutralization type sources, which have the highest flux, residual current can still flow to the substrate. However, this charged particle flux is several orders of magnitude smaller than the hyperthermal-neutral flux. Here they discuss the source of this residual current and methods for further reductions. This is important to the semiconductor industry.
Using laser-induced fluorescence (LIF) the ion density near the edge of an expanding plasma sheat... more Using laser-induced fluorescence (LIF) the ion density near the edge of an expanding plasma sheath has been measured. These measurements utilized a transition of Nzf [the P12 component of the X "X,$ (v=O) 3 3 '2: (Y=O) band] in a Ns plasma. The strength of the laser-induced fluorescence was used as a measure of the temporally and spatially varying ion density. The expanding sheath was produced by applying a -5 kV pulse to a polished planar electrode in the plasma source ion implantation device [J. R. Conrad et aL, J, Vat. Sci. Technol. A 8, 3146 ( 1990)]. The laser beam was aligned normal to the surface and was reflected off the center of the electrode. The LIF diagnostic used here is nonperturbing whereas previous researchers have used Langmuir probes, which perturb the plasma, to make their measurements. As such, the data reported here represent a benchmark measurement of pulsed sheaths and allow a better comparison between experimental measurements and theoretical predictions. It has been found that the sheath edge moves approximately 16 times faster than the ion-acoustic velocity during the early part of the pulse, t< 1 ,us, and then slows to approximately the ion-acoustic velocity after 6 ps. In addition to the LIF measurements, a biased probe was used far from the cathode to determine the sheath edge location. Good agreement is found when the LIF and probe data are compared. The LIF data also are compared to the predictions of a simulation that is based on a time-varying two-fluid model of the sheath [G. A. Emmert and M. A. Henry, J. Appl. Phys. 71, 113 ( 1992)]. While the predictions of the model show moderate agreement with the data, substantial discrepancies are observed. These discrepancies are attributed to a number of physical phenomena that are not included in the present model.
The group-II element beryllium when introduced into silicon forms more than one acceptor center. ... more The group-II element beryllium when introduced into silicon forms more than one acceptor center. Their infrared absorption spectra have been investigated using a high-resolution FTIR spectrometer.
The first direct measurement of a collisional Bohm presheath from plasma potential measurements i... more The first direct measurement of a collisional Bohm presheath from plasma potential measurements is given. By measuring the presheath thickness in front of a grounded wafer stage, a determination of the collision mean free path for ions in an electron cyclotron resonance etching tool has been made. Presheaths were measured in N2 and CF4 plasma using an emissive probe. The presheath thickness in N2 was found to be linearly dependent on the mean free path. Measurements of CF4 plasmas, for which the collision cross sections are unknown, have shown results similar to those found for nitrogen. This result has enabled an extrapolation to be made of the effective cross section for collisions in plasmas created from CF4.
In this letter, we describe a unique method of producing hyperthermal neutrals for material proce... more In this letter, we describe a unique method of producing hyperthermal neutrals for material processing. The hyperthermal neutrals are produced by accelerating ions across a sheath from a plasma onto a surface. On impact, the ions are neutralized and reflected with ϳ50% of their incident energy. These neutrals then bounce off of additional surfaces prior to impacting the target. This unique multiple bounce system was developed for the following reasons: to reduce contamination from sputtered surface material, improve beam uniformity, and reduce UV radiation in the beam path. As a test of this method, we built a prototype beam source and used it to ash photoresist at rates up to 0.022 m/min. These rates are consistent with a predicted neutral beam flux, 2 ϫ10 14 cm Ϫ2 s Ϫ1 . In addition, a simple model is used to indicate that this method is capable of producing economically acceptable ash rates. Comparisons with other neutral-beam production methods are made.
International Conference on Plasma Sciences (ICOPS), 1993
ABSTRACT
Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 2014
ABSTRACT The etch rate of deep features in silicon, such as trenches and vias, can vary significa... more ABSTRACT The etch rate of deep features in silicon, such as trenches and vias, can vary significantly with the feature aspect ratio (AR). Small AR features generally etch faster than large AR features. The reasons for this AR dependence include a slowing of the etch rate with increasing AR due to the necessary transport of molecules into and out of the features as well as ion flux reductions at feature bottom due to the angular spread of the ion flux and ion deflection caused by differential charging of the microstructures. Finding ways to reduce, eliminate, or reverse this AR dependence is both an active subject of research and difficult. In this work, instead of focusing on methods to reduce or prevent AR dependence in an etch process, the authors focus on methods to correct it after the fact. The authors show that an inhibitor film deposition step can be used under some circumstances to allow feature depth disparities to be corrected. This process can be used to correct feature depth disparities whenever the AR dependence of the inhibitor film deposition step is worse (larger) than the AR dependence of the following inhibitor etch step. To test the theory, a plasma process through SF6/C4F8/Ar mixtures was used to both produce trenches of various ARs having significant depth disparities and correct those disparities. The etch depth of small AR features can be held essentially constant while that of larger AR features is increased to match or even exceed.
ABSTRACT There are numerous computational models that can be used to study plasmas, each with adv... more ABSTRACT There are numerous computational models that can be used to study plasmas, each with advantages and disadvantages. Two of the most commonly used are the ambipolar model and the classical fluid model. The ambipolar model takes advantage of the ambipolar approximation, while the classical model includes the solving of the Poisson's equation. Both models were used to simulate the operation of the ICP electron source of the Electron Beam Exciter. Process conditions and geometrical variations were performed to optimize the electron density and electron to ion ratio in the beam extraction region. Results of this optimization along with deviations between the two models will be presented.
A systematic experimental study of the power, length, and gas pressure dependence of plasma densi... more A systematic experimental study of the power, length, and gas pressure dependence of plasma density and temperature profiles in thermally collapsed magnetized helium plasmas has been performed on the Princeton divertor simulator apparatus. The recombination edge and recombination radiation from highly excited levels (to n = 12) have been identified. The importance of radiation and neutral particles to heat transport has been demonstrated. A region of stable plasma detachment has been found and is suggested as being suitable for divertor operational space.
Hyperthermal-neutral beams have been proposed as a charge-free alternative to plasmas in select p... more Hyperthermal-neutral beams have been proposed as a charge-free alternative to plasmas in select processing steps. Existing prototype sources include three generic types: gas-dynamic, ion-neutral-charge-exchange and ion-surface-neutralization beam sources. The authors find that in surface-neutralization type sources, which have the highest flux, residual current can still flow to the substrate. However, this charged particle flux is several orders of magnitude smaller than the hyperthermal-neutral flux. Here they discuss the source of this residual current and methods for further reductions. This is important to the semiconductor industry.
Using laser-induced fluorescence (LIF) the ion density near the edge of an expanding plasma sheat... more Using laser-induced fluorescence (LIF) the ion density near the edge of an expanding plasma sheath has been measured. These measurements utilized a transition of Nzf [the P12 component of the X "X,$ (v=O) 3 3 '2: (Y=O) band] in a Ns plasma. The strength of the laser-induced fluorescence was used as a measure of the temporally and spatially varying ion density. The expanding sheath was produced by applying a -5 kV pulse to a polished planar electrode in the plasma source ion implantation device [J. R. Conrad et aL, J, Vat. Sci. Technol. A 8, 3146 ( 1990)]. The laser beam was aligned normal to the surface and was reflected off the center of the electrode. The LIF diagnostic used here is nonperturbing whereas previous researchers have used Langmuir probes, which perturb the plasma, to make their measurements. As such, the data reported here represent a benchmark measurement of pulsed sheaths and allow a better comparison between experimental measurements and theoretical predictions. It has been found that the sheath edge moves approximately 16 times faster than the ion-acoustic velocity during the early part of the pulse, t< 1 ,us, and then slows to approximately the ion-acoustic velocity after 6 ps. In addition to the LIF measurements, a biased probe was used far from the cathode to determine the sheath edge location. Good agreement is found when the LIF and probe data are compared. The LIF data also are compared to the predictions of a simulation that is based on a time-varying two-fluid model of the sheath [G. A. Emmert and M. A. Henry, J. Appl. Phys. 71, 113 ( 1992)]. While the predictions of the model show moderate agreement with the data, substantial discrepancies are observed. These discrepancies are attributed to a number of physical phenomena that are not included in the present model.
The group-II element beryllium when introduced into silicon forms more than one acceptor center. ... more The group-II element beryllium when introduced into silicon forms more than one acceptor center. Their infrared absorption spectra have been investigated using a high-resolution FTIR spectrometer.
The first direct measurement of a collisional Bohm presheath from plasma potential measurements i... more The first direct measurement of a collisional Bohm presheath from plasma potential measurements is given. By measuring the presheath thickness in front of a grounded wafer stage, a determination of the collision mean free path for ions in an electron cyclotron resonance etching tool has been made. Presheaths were measured in N2 and CF4 plasma using an emissive probe. The presheath thickness in N2 was found to be linearly dependent on the mean free path. Measurements of CF4 plasmas, for which the collision cross sections are unknown, have shown results similar to those found for nitrogen. This result has enabled an extrapolation to be made of the effective cross section for collisions in plasmas created from CF4.
In this letter, we describe a unique method of producing hyperthermal neutrals for material proce... more In this letter, we describe a unique method of producing hyperthermal neutrals for material processing. The hyperthermal neutrals are produced by accelerating ions across a sheath from a plasma onto a surface. On impact, the ions are neutralized and reflected with ϳ50% of their incident energy. These neutrals then bounce off of additional surfaces prior to impacting the target. This unique multiple bounce system was developed for the following reasons: to reduce contamination from sputtered surface material, improve beam uniformity, and reduce UV radiation in the beam path. As a test of this method, we built a prototype beam source and used it to ash photoresist at rates up to 0.022 m/min. These rates are consistent with a predicted neutral beam flux, 2 ϫ10 14 cm Ϫ2 s Ϫ1 . In addition, a simple model is used to indicate that this method is capable of producing economically acceptable ash rates. Comparisons with other neutral-beam production methods are made.