A likely reason for the deformed mesa shape is anisotropic lateral etching due to crystallographic orientation of the InP/InGaAs substrate. Wet etchants like HCl:H₃PO₄:DI or H₂SO₄:H₂O₂:DI often exhibit crystal-plane-dependent etch rates, especially on III–V semiconductors. This can cause initially circular features to evolve into diamond-like shapes, reflecting faster etching along specific directions (e.g., ).

Additionally, photoresist undercutting may occur if the etch rate approaches or exceeds the resist thickness, especially if the resist is not fully cross-linked or chemically resistant to the etchant. Even with AZ 4330, swelling or edge lift-off can contribute to shape distortion.

To minimize this:

• Try aligning patterns to crystal orientation to anticipate anisotropy.
• Use a hard mask (e.g., SiO₂ or Si₃N₄) instead of photoresist for better etch resistance.
• Optimize etch time and agitation to reduce lateral etching.

Thanks for the well explained answer, Yana Suchikova

Yana Suchikova, I have follow-up question, due to this deformed mesa, can we expect deviation in the device performance? I have been facing an issue for the current after the breakdown. It seems my device does not reach the compliance sharply (please see the attached image as a reference). I tried using several metal stacks but still I see the same trend in current. Also, the forward resistance is high.

My other question is: As I need to do wet etching, is there any wet etchants that do not exhibit crystal-plane-dependent etch rates?

Thank you

Thank you for your follow-up questions!

Impact of deformed mesa on device performance: Yes, the deformation of the mesa can definitely impact device performance. Non-uniform mesa shapes can cause:

• Local electric field enhancement at sharper corners (e.g., diamond tips instead of smooth circles), leading to earlier or less uniform breakdown.
• Non-uniform current spreading, increasing series resistance and causing higher forward resistance.
• Inhomogeneous contact between metal and semiconductor if the surface is no longer ideally flat, which could also contribute to degraded current-voltage characteristics.

The behavior you observe after breakdown, where the current does not sharply reach compliance, could very well be influenced by local breakdown at the mesa edges rather than a uniform, bulk breakdown. This can result in gradual rather than sharp current increases.

Thus, the deformed mesa is a strong candidate for explaining the non-idealities you see, along with high forward resistance.

Wet etchants and crystal-plane dependence: Unfortunately, almost all wet etchants for III–V materials, including InP and InGaAs, show some crystallographic anisotropy. This is inherent to the material's atomic structure. However, you can reduce the impact by:

• Using more isotropic etchants like H₃PO₄:H₂O₂:H₂O, which are softer and etch more slowly but tend to show less strong anisotropy compared to HCl-based solutions.
• Using shorter etch times and refreshing the solution often to maintain uniformity.

In summary:

• Your mesa deformation is very likely contributing to both the strange post-breakdown behavior and high forward resistance.
• True isotropic wet etchants for InP/InGaAs are rare, but milder etchants like H₃PO₄-based solutions might reduce — not eliminate — the anisotropy.

Yana Suchikova I really appreciate your time and detail discussion. I will change the etchants and take other recommended steps to have somewhat better uniformity. Thanks once again!