Library

Notes: Ionized physical vapor deposition of Cu in a mixture of three rare gases (He–Ar–Xe) is explored in this article. Results indicate that total Cu flux to the wafer, ionization fraction of Cu at the wafer, and ratio of total effective ion flux to total Cu flux increase with increasing Xe concentration in the gas mixture. This is because of enhancement of electron density and Xe+ ions having a larger sputter yield on Cu than other ions. Increase in He concentration decreases the ionization fraction due to a lower electron density. However, Cu flux to the wafer increases because He is less effective in thermalizing the hot sputtered neutrals. One major consequence of these trends is that one can independently control total Cu flux to the wafer (corresponding to deposition rate) and ionization fraction (a major factor controlling the deposition profile) over a wide range by means of the buffer gas composition. © 2001 American Institute of Physics.

Notes: A two-dimensional model has been used to understand the physics and process engineering issues associated with a conceptual 300 mm Cu internal-coil ionized physical vapor deposition reactor. It has been found that inductive coupling from the coil is the primary source of plasma production. Since the coil is in direct contact with the plasma, a significant fraction of the coil power is deposited in the gas capacitively as well. This results in sputtering of the Cu coil, which tends to improve Cu flux uniformity at the outer edges of the wafer. Since the Cu ionization threshold is much lower than Ar, Cu+ density is comparable to Ar+ density even though ground state Cu density is much smaller than Ar. Significant fraction of the neutral Cu flux to the wafer is in the metastable or athermal state. The effects of several actuators, reactor dimensions, and buffer gas on important plasma and process quantities have also been investigated. Electron density in the reactor and Cu ionization fraction increases with increasing total coil power because of enhanced ionization. Total coil power however does not affect the Cu density appreciably, except near the coil where enhanced coil sputtering increases the Cu density. Decrease in dc target voltage with increasing coil power decreases Cu+ loss to the target and results in an increase in total Cu flux to the wafer. Electron and Cu density in the reactor increase with increasing dc target power. This is due to enhancement in target sputtering and consequent ionization of the sputtered Cu. While this increases the total Cu flux to the wafer, ionization fraction is not affected much. It is demonstrated that uniformity of Cu flux to the wafer and ionization fraction can be controlled by means of the terminating capacitor at the coil. Decreasing the terminating capacitance increases the coil voltage, enhances coil sputtering and enhances Cu flux toward the outer edges of the wafer. This, however, decreases the amount of power that is transferred to the plasma inductively, reducing the ionization efficiency. Increasing the coil–wafer distance results in fewer sputtered Cu atoms being ionized as the target–coil distance becomes smaller than the mean free path for thermalization of hot sputtered Cu atoms. Also, one can control the ionization fraction of Cu flux to the wafer by replacing Ar by Ne or Xe, without significantly impacting the total Cu flux. © 2001 American Institute of Physics.

Notes: Hydrodynamic phenomena from KrF excimer laser ablation (10−3–20 J/cm2) of polyimide, polyethyleneterephthalate, and aluminum are diagnosed by schlieren photography, shadowgraphy, and dye laser resonance absorption photography (DLRAP). Experiments were performed both in vacuum and gaseous environments (10−5–760 Torr air, nitrogen, and argon). In vacuum, ablation plumes are observed to expand like a reflected rarefaction wave. As the background gas pressure is increased, shock waves and reduced-density ablation plumes become visible. Below 10 Torr, the ablation plume follows closely behind the shock wave. Between 20 and 100 Torr, the plume recedes behind the shock wave. Below 10 Torr and above about 200 Torr, both the plume and the shock expand with the same temporal power law dependence. Agreement is found between these power law dependences and those predicted by ideal blast wave theory. The DLRAP diagnostic clearly shows that the ablated material (CN molecule from polyimide and ground state neutral aluminum atoms from laser-ablated aluminum) resides in the ablation plume. CN molecules are detected in both argon and air environments proving that CN is generated as an ablation product and not by reaction with the background gas. As the background gas pressure and the time after ablation is increased, the film darkening due to the laser-ablated material begins to fade leaving only the nonresonant shadowgraphy component of the plume. The plume dynamics observed by DLRAP are discussed in terms of gas dynamics, plume chemical kinetics, material diffusion in the plume, and cluster/particulate formation.

Notes: A new method of gathering statistics for Monte Carlo methods, Legendre polynomial weighted sampling (LPWS), is presented. LPWS requires only a minimum of particles to extract higher-order derivative information about a particle's distribution function. In this technique, when calculating a particle's distribution function, higher-order derivative information about the Monte Carlo particles is recorded along with just counting the number of particles in a bin. The distribution function is then constructed from this information. Specifically, in this paper, second-order Legendre polynomial weighted sampling is employed. Legendre polynomial weighted sampling is demonstrated by calculating the electron energy distribution functions in an inductively coupled plasma reactor.

Notes: Models capturing the periodic steady-state behavior of rf capacitively coupled discharges are now commonplace. New plasma sources have been motivated by selectivity, charge-damage mitigation, and general process control needs in plasma processing of electronic materials. These new sources require models that can accurately capture the transient behavior of the plasma source. Such models are not commonplace because the behavior of transport parameters in transients is still not well understood and because the problem is inherently stiff, i.e., widely disparate time scales are important. In this paper, we present the results of an investigation of the simplest type of transient, known as a step disturbance, in a 2 cm gap parallel-plate argon discharge at 1 Torr. As examples, two classes of step transients are considered: step increases in the peak-to-peak (pp) applied voltage (300 to up to 450 V pp) and step decreases (300 to as low as 150 V pp). The resulting transients are interpreted in terms of time scales representative of electron and ion motion in the sheath, ionization dynamics, and neutral transport processes. The possibility of using these transients as a means of process identification is discussed. © 1997 American Institute of Physics.

Notes: Quantitative measurements of ablated material from the surface of polyethyleneterephthalate (PET) by 248-nm excimer laser fluences up to 10 J/cm2 are performed by HeNe laser-beam deflection in vacuum and by photoacoustic depth profiling in air. HeNe laser-beam deflection measures the density of gas phase material present in the ablation plume. Photoacoustic depth profiling is a nonintrusive diagnostic that directly measures the etch depths from laser ablation. A hydrodynamic model consisting of a centered rarefaction wave that reflects off the PET surface is shown to describe the laser deflection signals. From these measurements an estimate of the initial temperature of the ablated species is found.

Notes: Pulsed schlieren photography and fast helium-neon laser deflection are used to study the hydrodynamics of laser ablation of polyethyleneterephthalate and polymethylmethacrylate by pulsed KrF (248 nm) radiation in atmospheric air, Ar and N2. Schlieren measurements show the evolution of shock waves, sound waves, and reduced-density, hot gas plumes. A transition from sound to shock at the ablation threshold for both polymers is observed. The shock velocity of PET tends to approach agreement with blast wave theory at fluences higher than 1 J/cm2. Plumes in air are consistently larger than those produced in Ar and N2 (at fluences below 5 J/cm2) suggesting that combustion may occur. Laser deflection measurements for PET at 150 mJ/cm2 indicate a plume density of 0.6 kg/m3 (50% atmospheric density).

Notes: Photoacoustic and photothermal laser-beam deflection were applied as diagnostics of the pulsed ultraviolet (UV) laser ablation of a polymer polyethyleneterephthalate. Here, a continuous-wave (cw) laser beam is passed parallel to the sample, but displaced from it by a few hundred micrometers. A density gradient caused by the pulsed UV laser heating or ablation of the sample deflects the cw laser beam. This deflection is measured directly using a position-sensitive detector. A quantitative model of the photothermal deflection at low fluence was developed which fits the data very well. This enabled a new method of measuring the thermal diffusivity of the fluid in contact with the sample. Distortion of the photothermal and photoacoustic signal as the excimer fluence is raised through the ablation threshold allowed the determination of the threshold. Also, the velocity of the ablation products was measured through a time-of-flight analysis and found to be dependent on the laser fluence used, the nature of the gas above the sample, and the distance above the sample at which the velocity is measured. The beam deflection in a vacuum is used to measure the ablation product velocity.

Notes: The transient response of electronegative radio-frequency glow discharges is important for process control, charge free etching, and highly selective etch applications. The step response of typical electronegative process gases (silane at 1 Torr and chlorine at 100 mTorr) is studied using a drift-diffusion model for silane and a three-moment model for chlorine. The silane simulations include a blocking capacitor whereas the chlorine results do not. For the silane results with a blocking capacitor in series with the plasma, it is found that there are three types of transients. Depending on the final steady-state value of the source rf voltage, the step response can be characterized either by smooth transitions in the number densities of species in the discharge from one steady state to the next, temporary extinction of the discharge or a discharge mode characterized by temporary extinction and reignition of the discharge. In the case of silane definite thresholds separate the phenomena. The step response of the chlorine discharges is always characterized by a smooth transition from one steady state to the next. Smooth transitions from one steady state to the next in the case of step decreases in the source voltage are possible since decrease of the negative ion density in the bulk is controlled by ion–ion recombination. It appears that the temporary extinction of the discharge and natural pulsed steady state is the consequence of how the voltage is divided between the gap and the blocking capacitor during the transient and the fact that the attachment coefficient becomes larger than the ionization coefficient at low values of reduced electric field. © 1998 American Institute of Physics.

Notes: Charge damage considerations are prompting the development of neutral beam sources for etching applications. Anisotropic etching with hyperthermal Cl2 and SF6 beams has been demonstrated. We describe a two-dimensional plasma chemistry fluid model of laser ablation of frozen Cl2 in vacuum as a neutral beam source. In this scheme an externally applied electric field would be used to enhance the dissociation rate of Cl2 potentially providing an enhanced Cl content in the beam for a greater etch rate. Laser ablation generated neutral beams also may contain a desirable and controllable ion content which may be used to further enhance the etch rate. Limitations of the concept are discussed. © 1996 American Institute of Physics.