Caleb T. Nelson

Temperature dependence of the infrared absorption cross-sections of neutral species commonly found in fluorocarbon plasmas

This article serves as a reference for the analysis of Fourier transform infrared spectroscopy data from processing plasmas. Until now, there has been a lack of accurate reference data for addressing the problems of species identification and density measurements in cases of increasing gas temperatures. Our results show that, while the integrated absorption cross-sections do not change significantly as temperature increases, the temperature of the absorbing species can be estimated from the rotational band maximum in most cases. Integrated absorption cross-sections for c-C3F6, C4F8, C3F8, C2F6, C2F4, and CF4 are presented for all fundamental bands in the 650 cm−1 to 2000 cm−1 region. In addition, the binary combination bands up to 4000 cm−1 are presented for all species. The temperature of each species has been varied to correspond to neutral temperatures commonly found in processing plasmas.

Relationship between gas-phase chemistries and surface processes in fluorocarbon etch plasmas: A process rate model

In a typical plasma tool, both etch and deposition occur simultaneously. Extensive experimental measurements are used to help develop a general model of etch and deposition processes. This model employs reaction probabilities, or surface averaged cross sections, to link the measurable surface processes, etch and deposition, to the flux of various species to the surfaces. Because the cross sections are quantum mechanical in nature, this surface rate model should be applicable to many low temperature plasma processing systems. Further, the parameters that might be important in reaction cross sections are known from quantum mechanics, e.g., species, energy, temperature, and impact angle. Such parameters might vary from system to system, causing the wide processing variability observed in plasma tools. Finally the model is used to compare measurements of ion flux, ion energy, and fluorocarbon radical flux to the measured process rates. It is found that the model appears to be consistent with calculations of gain/loss rates for the various radicals present in the discharge as well as measured etch and deposition rates.

Kinetics of the deposition step in time multiplexed deep silicon etches

The time multiplexed deep silicon etch (TMDSE) process is the etch process of choice to make MEMS devices and through wafer vias. It has been used to produce deep trenches and vias at reasonable throughputs. Significant issues remain for the TMDSE process as well as room for improvement even though it has been both experimentally studied and modeled by a wide variety of researchers. This is because it is a highly complex process. Aspect ratio dependencies, selectivity, and the ability to use photoresist masks (instead of SiO2) are examples of remaining issues. The presently obtainable etch rates do not indicate efficient use of the etchant species. In this article, the authors focus on the deposition step in the TMDSE process. While prior research has generally assumed that the deposition step can be adequately modeled as being controlled by a reactive sticking coefficient, they have experimentally examined the deposition step of the process and found that the film growth is dominantly ion-enhanced. The r...

Chemistry in long residence time fluorocarbon plasmas

The densities of radicals and neutrals in fluorocarbon (FC) plasmas have been investigated in an inductively coupled plasma system to understand the predominant gain and loss mechanisms of dissociative products and their interaction with chamber surfaces. The input parameters varied in this experiment are the source to chuck gap and the F: C ratio of the feed gas. The densities of F, CF2, CF3, CF4, C2F4, SiF4, COF2, CO, and CO2 are measured and analyzed. In addition, two different forms of C4F8, the standard cyclic c and a radical linear l structures are observed in C4F8 containing plasmas. l-C4F8 is shown to be the primary dissociation product of c-C4F8 and, thus, cannot be neglected from calculations of the loss rate of c-C4F8 to electron collisions. This implies that the typically cited dissociative products of c-C4F8 (primarily C2F4) can have dual production channels: one from l-C4F8 and the other directly from c-C4F8. Furthermore, the measured density of CF4 shows strong correlation to the loss of F ...

Electronegative Plasma Structure

It is well understood that simply changing the feed gas from an electropositive to electronegative gas can result in a substantially different discharge structure. We find that the electron pressure (or energy density) can visibly vary for discharges with significant electron attachment, whereas it is much more uniform for electropositive glows. Such structure in the electron energy density is difficult to measure by using other experimental diagnostics; however, it is relatively easy to see by visual inspection.

Gain and loss mechanisms for neutral species in low pressure fluorocarbon plasmas by infrared spectroscopy

This article examines the chemical reaction pathways of stable neutral species in fluorocarbon plasmas. Octafluorocyclobutane (c-C4F8) inductively coupled plasma discharges were found to primarily produce stable and metastable products downstream from the discharge, including c-C4F8, C2F4, C2F6, CF4, C3F8, C4F10, C3F6, and CF2. A novel analysis technique allows the estimation of gain and loss rates for neutral species in the steady state as functions of residence time, pressure, and discharge power. The gain and loss rates show that CF4, C2F6, C3F8, and C4F10 share related gain mechanisms, speculated to occur at the surface. Further analysis confirms that CF2 is predominantly produced at the chamber walls through electron impact dissociation of C2F4 and lost through gas-phase addition reactions to form C2F4. Additionally, time-resolved FTIR spectra provide a second-order rate coefficient of 1.8 × 10−14 cm3/s for the gas-phase addition of CF2 to form C2F4. Finally, C2F4, which is much more abundant than CF...

Role of Surface Temperature in Fluorocarbon Plasma-Surface Interactions

This article examines plasma-surface reaction channels and the effect of surface temperature on the magnitude of those channels. Neutral species CF4, C2F6, and C3F8 are produced on surfaces. The magnitude of the production channel increases with surface temperature for all species, but favors higher mass species as the temperature is elevated. Additionally, the production rate of CF2 increases by a factor of 5 as the surface temperature is raised from 25 °C to 200 °C. Fluorine density, on the other hand, does not change as a function of either surface temperature or position outside of the plasma glow. This indicates that fluorine addition in the gas-phase is not a dominant reaction. Heating reactors can result in higher densities of depositing radical species, resulting in increased deposition rates on cooled substrates. Finally, the sticking probability of the depositing free radical species does not change as a function of surface temperature. Instead, the surface temperature acts together with an etch...

Nanofabrication of diamond-like carbon templates for nanoimprint lithography

Diamond like carbon (DLC) films were deposited on Si and then patterned to form 40 nm features as nanoimprint templates. A plasma enhanced chemical vapor deposition (PECVD) system with CH4 precursor was used to deposit DLC films on Si and quartz substrates. Then these films were characterized using Raman spectroscopy, atomic force microscopy (AFM), nanoindentation, and contact angle measurement. By varying the RF power and pressure of the PECVD, DLC films with good uniformity, smooth surfaces (<0.2 nm RMS), low surface energy (∼40 mJ/m2), and high hardness (∼22 GPa) were achieved. Nanoimprint lithography and liftoff process were used to pattern Cr mask on DLC films. An inductively coupled plasma (ICP) etching process was performed with CF4 to transfer the patterns into the DLC films to form nanostructured template for nanoimprint. Water contact angles on the patterned DLC templates were measured and it was stable at about 70° under thermal annealing at 180 °C for more than 12 hours. With these DLC templates, UV and reversal UV nanoimprint lithography were carried out on SU-8 at typical imprint conditions and then the fidelity of pattern-transfer was investigated. These experimental results indicate that DLC is an excellent material for nanoimprint templates because of its high wear resistance, robust low surface energy, UV transparency, and ease of patterning.