In semiconductor plasma processing, a simulation’s usefulness depends in part on how well it represents the chamber and the way power reaches the discharge. A 3D ICP plasma simulation can account for geometric features that a two-dimensional axisymmetric model cannot, including the arrangement of gas inlets and the shape of the antenna. It can also help engineers examine how electromagnetic field calculations and antenna impedance affect predicted plasma properties.
A study using COMSOL Multiphysics explored these issues with a three-dimensional, asymmetric model of an inductively coupled plasma (ICP) discharge. Its central finding is relevant to process modeling: impedance estimates depend on how the electric and magnetic fields are represented, and those estimates influence the calculated power transferred to the plasma.
Why represent the chamber in three dimensions?
ICP sources are used in microelectronics manufacturing, including semiconductor and display fabrication. Their high plasma density, low-pressure operation, and uniformity can support etching and deposition processes. Modeling can help examine relationships between operating conditions and plasma behavior that may be difficult to isolate experimentally.
Many simulations use a two-dimensional axisymmetric geometry to reduce computational demands. That simplification is less suitable when important features break the chamber’s rotational symmetry. In the model considered here, four gas inlet ports were positioned at 90-degree intervals, and the antenna had a distinct power-input and grounded-output arrangement. A three-dimensional representation allowed the analysis to include the effects of these features rather than treating the chamber as symmetric around its axis.
The modeled plasma region was 35 cm in diameter. A 2 cm-high aluminum oxide ceramic window separated the plasma from a three-turn copper antenna, and an air-filled cap protected the coil. The simulation used argon at 20 mTorr, with a 100 A current applied to the antenna at 13.56 MHz. These are the conditions of this particular model, not general operating recommendations.
What a fluid model can—and cannot—show
The researchers used a fluid-based plasma model. Instead of tracking every particle individually, a fluid model calculates averaged quantities such as electron density, electron energy, and species transport. This approach can reduce computational requirements, an important consideration for three-dimensional simulations.
The trade-off is that a fluid description simplifies particle behavior. At low pressures, plasma phenomena are influenced by the motion and interactions of individual particles, which an averaged model does not resolve in the same way. Results should therefore be interpreted in light of the model’s assumptions and the questions it is intended to answer.
In this simulation, electron density and energy were calculated using drift-diffusion relationships, while heavy-species transport was represented with a mixture-averaged method. The reaction set included argon excitation, ionization, and other collision processes. Together, these components provided a framework for examining how electromagnetic power input relates to plasma characteristics.
Why impedance depends on electromagnetic field treatment
Current in the antenna creates a magnetic field. The time-varying field induces an electric field that transfers energy to electrons in the plasma. The antenna’s impedance affects its voltage, and the resulting field distribution is connected to how much power is absorbed by the discharge.
The study compared three calculation approaches. One used magnetic field physics with a coil analysis feature; another included both magnetic and electric field physics while retaining that coil analysis; the third calculated impedance from the electromagnetic fields using Poynting’s theorem. The key distinction was whether the electric field included only the inductive contribution or also the capacitive contribution, and whether the coil analysis represented the antenna’s boundary conditions and geometry appropriately.
In the field-based approach, Poynting’s theorem was used to evaluate power flow and derive resistance and reactance from field quantities. This offers a way to account for electromagnetic energy in the modeled volume and at relevant boundaries. It is particularly pertinent when the antenna geometry and electrical connections make simplified coil assumptions unsuitable.
Geometry and boundary conditions affect the result
The comparison identified limitations in the coil geometry analysis approach for this configuration. The model required a grounded antenna output, but the coil analysis feature did not provide the needed ground connection there. The simulated output consequently showed a nonzero potential, contributing to a numerical error. The method also assumes a coil with a consistent cross-section; changes or bottlenecks in the antenna geometry can make current calculations at bends less reliable.
These issues matter because coil voltage and current feed into impedance estimates. If the electromagnetic setup does not reflect the actual electrical connections and geometry, the calculated power coupled into the plasma may differ. The study’s results showed differences in calculated impedance, coil voltage, and absorbed power across the approaches. The field-based calculation produced the highest absorbed power among the three cases, accompanied by higher simulated electron density and temperature.
The three-dimensional model also showed plasma-property distributions skewed by the asymmetric chamber structure. Such spatial variation was not visible in a two-dimensional axisymmetric representation. This does not mean every process requires a 3D model, but it demonstrates why geometry simplifications should be evaluated against the features most likely to influence the question being studied.
Practical implications for process modeling
For engineers and simulation teams, the study points to several checks when building an ICP model:
- Match the geometry to the problem. Include asymmetric features such as inlet placement and antenna shape when their effects are under investigation.
- Review electrical boundary conditions. Confirm that power input, grounding, and antenna geometry are represented consistently with the modeled system.
- Consider both field contributions. Capacitive as well as inductive electric fields can affect energy calculations and the resulting impedance estimate.
- Understand model limitations. Fluid models offer computational advantages but rely on averaged descriptions of plasma behavior.
The work focused on electromagnetic modeling and plasma characteristics; it did not incorporate heat-transfer or flow physics to represent changes in enthalpy, pressure, and velocity. Those were identified as areas for future study, and they may be relevant when extending a model to address coupled thermal or gas-flow effects.
Conclusion
A 3D ICP plasma simulation can reveal how chamber asymmetry influences predicted plasma distributions, while careful electromagnetic modeling helps establish how antenna impedance relates to power absorbed by the discharge. The practical lesson is to align geometry, field physics, and boundary conditions with the system being represented. For semiconductor process analysis, that discipline can make simulation results more informative, while keeping the assumptions and limits of the model visible to engineering decision-makers.