Workpiece atoms penetrating tool

Hi, I am trying to simulate elliptical vibrated cutting of CuZr using diamond tool.
Workpiece: CuZr
Tool: C

Picture 1 (orthogonal cutting):

Picture 2 (elliptical vibrated cutting):

I realised that there was no problem at all with orthogonal cutting but when I applied elliptical vibration to my tool, workpiece atoms started to penetrate my tool.

Cu-Zr: eam
C-C: rigid (no potentials)
C-Cu, C-Zr: lennard-jones

I tried using morse potentials for C-Cu, C-Zr as well but it didnt help.
From what I observed, I suspect that it is the cutting speed that causes the penetration, as vibrating speed is definitely way higher than tool speed during orthogonal cutting.
I have discussed with someone familiar with LAMMPS but his feedback was to try different potentials. So far, I find it not useful.

It would be great if someone could provide some thoughts/guidance. Thank you in advance.

This is more likely due to the way how you move your tool.

Hi Dr @akohlmey, thank you fo your reply.

Here is a comparison between elliptically and diagonally cutting the moment the tool comes in contact with the workpiece.

I used command velocity for orthogonal/diagonally cutting, and fix move for elliptically cutting.
I guess this is what you meant and I believe this is the root cause as well.

orthogonal tool motion: velocity tool set -1.0 0.0 0.0 sum yes units box
elliptical tool motion:
variable Vx equal …
variable Vy equal …
fix 1 tool move variable NULL NULL NULL v_Vx v_Vy NULL

May I ask for suggestions on what options do I have, in order to realise elliptical motion and also avoiding atoms penetration at the same time?
Thank you very much.

There is not enough information here to make any specific statements.

Hi Dr @akohlmey , my problem has been solved, and it was due to error: one or more atoms are time integrated more than once. Thank you very much.

Lesson: always take warnings in your output seriously and only ignore them if you are dead certain they may be ignored in your specific case.

Noted. Thank you very much, Dr @akohlmey .