In plasma etching, ion energy is only part of the process picture. The ion angle distribution at a substrate also affects how ions reach the bottom and sidewalls of microscopic features. For engineers and process teams working to control trench and via profiles, understanding this distribution helps connect reactor conditions with feature shape and etch anisotropy.
Measurements in a low-pressure, high-density argon plasma showed that ions arriving at a grounded silicon surface had approximately Gaussian angle distributions. Their width changed with plasma power and pressure. The findings also demonstrate how energy and angle measurements can help evaluate plasma behavior and inform process models.
Why ion incidence angles affect feature profiles
During plasma etching, energetic ions enhance surface reactions and contribute to material removal. Their energy influences reaction rates and can affect etch rate and selectivity relative to a mask or underlying film. Their direction of travel influences where ion flux reaches a feature.
When most ions arrive close to the substrate normal, a larger share can reach the bottom of a trench or via. A broad distribution sends more ions along oblique trajectories, changing the flux along feature walls. This can contribute to mask undercutting, profile bowing, and reduced anisotropy. The result is that etch profile control depends not just on ion energy, but also on the spread of incidence angles.
Ion trajectories matter in deposition processes as well. In high-density plasma chemical vapor deposition, ion energy and angle can influence how topography evolves as gaps between metal lines are filled with silicon dioxide or fluorinated silicon dioxide.
Measuring ion angle distribution in a plasma reactor
Ion energy distributions had been studied extensively, while angular measurements were less common and often had limited resolution. The described retarding-field energy analyzer combined energy discrimination with position-sensitive detection to measure both energy and incidence angle. Its reported angular resolution was better than 0.5 degrees.
A small circular aperture in a silicon membrane sampled ions from the plasma-facing surface into a differentially pumped chamber. The analyzer used retarding grids to reject ions below a selected energy threshold. Ions that passed through were accelerated toward a microchannel plate detector. A phosphor screen and camera recorded where they arrived, preserving information about their trajectories.
In this arrangement, an ion’s impact position on the detector corresponded to its incidence angle relative to the substrate normal. The recorded image therefore showed the angular distribution of ions above the selected energy threshold. Changing the retarding voltage made it possible to examine energy-dependent distributions as well as the overall angle profile.
The measurements were made in the downstream region of a helicon-wave-excited argon plasma. The sampling aperture was 50 micrometers in diameter, and the analysis chamber was maintained at low pressure to limit collisions after ions passed through the aperture. The study focused on a low-aspect-ratio opening, allowing the measured distribution to approximate the ions arriving at the substrate surface.
What determines the width of the distribution?
For collisionless motion through the sheath—the region near the surface where the plasma potential accelerates ions toward the substrate—the distribution width depends on the balance between directed energy gained in the sheath and random ion energy in the plasma. A stronger directed component relative to the ions’ random energy corresponds to a narrower distribution.
Under the conditions described, the angle profiles were approximately Gaussian. The relationship between their width and the energy balance offers a useful way to interpret measurements: changes in the distribution can reflect changes in sheath potential, ion temperature, or both.
How power and pressure influenced ion angles
In the measured low-pressure argon plasma, the angle distribution broadened as RF power increased. The study attributed this trend to increased ion heating and higher ion temperature. Although power can affect multiple plasma properties, the observed change illustrates why power settings should not be treated as a direct proxy for ion directionality.
The pressure response was not a simple, continuously increasing broadening. The measured distribution width reached a maximum within the tested range of approximately 0.5 to 4 millitorr, then decreased as pressure rose further. This behavior was counterintuitive if one assumed that more pressure would simply cause more collisions and spread ion trajectories.
However, the ion mean free path remained longer than the estimated sheath thickness across the conditions considered, so collisions within the sheath did not explain the observed trend. Instead, pressure affects both the sheath potential and ion temperature. The balance between those changes determines the distribution width. Gas heating and variations in neutral density within the reactor can further complicate interpretation, making simple predictions difficult.
Implications for process development and modeling
For process engineers, the practical lesson is to consider ion energy and angle together when diagnosing profile behavior. A narrow angular spread favors directional bombardment, but its width depends on interacting plasma conditions rather than a single control setting. Measurements across power and pressure conditions can help distinguish changes in ion directionality from changes in energy.
Energy- and angle-resolved diagnostics can also provide data for checking plasma simulation models. Models that predict ion distributions as operating conditions change can help connect reactor settings with feature-scale outcomes. The reported analyzer was designed for grounded surfaces; extending the method to RF-biased substrates requires accounting for the electrode’s RF potential when setting analyzer voltages.
Conclusion
Ion incidence angle is a key variable in plasma etching because it influences how ion flux is distributed across feature bottoms and sidewalls. In the studied argon plasma, distributions were approximately Gaussian, broadened with increasing RF power, and showed a pressure-dependent maximum rather than a simple monotonic trend. For technical teams seeking predictable feature profiles, the takeaway is to evaluate ion energy, angle, sheath conditions, and operating pressure as related parts of the process—not as isolated parameters.