The ideal way to learn TCAD...
First, it is very useful (I would say - it's mandatory) to know (or learn) the device physics of the device that you are planning to simulate.
Learn what Ohmic contact is, doping, p-n junction, carrier lifetime, electric field, basic parameters of the semiconductor, potential, the difference between potential and voltage, Fermi (or quasi-Fermi) level, electrons and holes, etc.
Learn how does device operate, what are the design parameters, optimization goal, ideal characteristics, parasitic effects, etc.
There are some books on that.
Second, you need to learn how to apply TCAD software to simulate the device.
The most valuable resource would be your knowledgeable, senior colleagues - but that seems to be not available to you...
Next is tutorial test cases from your TCAD vendor.
Take a test case for solar cell, run the default simulations. Simulate dark I-V curve. Simulate illumination conditions.
Learn how to read and analyze the text output.
Learn how to use graphical visualization - to see what's going on inside the device - various characteristics (electron and hole concentrations, potential, photogeneration rate), various representations (2D plots, 1D cross-sections, etc.).
Run simulations by changing various parameters - lifetime, temperature, illumination power, wavelength, etc.
See the impact of various parameters on the results. Get a feel for the problem and your device system.
See how that agrees with the zero- or fist-order theory from the textbooks.
See where is the difference, and why it is there.
Play around with the structure of the solar cell (geometry, contacts, material parameters, etc.).
See how the results change due to change in the structure.
Then go to the process simulation, if you need to do this (Athena).
This is much less physics-based than device simulation - a lot of empirical models and empirical fitting parameters.
Go over several iterations.
Re-read the manual - manuals contain a lot of useful info, but they are to thick to be read all at once.
You need to refer to the manuals when looking for a specific piece of info, or at the next spiral of your learning, to learn new things.
TCAD is a very powerful tool, but it has a limited applicability for process/device simulations.
Many effects and phenomena can not be simulated from the first principles.
Learn about the limitations of the TCAD.
Go to the next level.
Learn how to build a complete simulation flow (process + device simulation + post-processing of the results) - so that you can get the required results (for example - responsivity, or open-circuit voltage) with a push of a button.
Learn what parameterization is. Implement it and optimize it.
Try doing 1D and 2D simulations - analyze the difference, look at the performance, etc.
...
These are just general suggestions...