Hello Everyone,

Recently, I found it seems COMPASS forcefields parameters are not suitable for TiO2 surface or nano-cluster simulation.

I hope someone could kindly repeat the following steps (it will take only several minutes to complete all the steps):

Step 1: Import one TiO2 cell via: Import-->Structures-->metal-oxides Document. There are 3 types of TiO2 cells, i.e., TiO2_anatase/TiO2_brookite/TiO2_rutile, importing any one of them is OK.

Step 2: Ensure TiO2 cell document active. Open Forcite Calculation-->Energy tab, then choose "Forcefield" as "COMPASS" (you can also choose "COMPASSII" , or "COMPASSIII", any one is OK). And then, Click the "More..." button to open "Forcite Preparation Options dialog" dialog. Uncheck "Calculate automatically", then click the two "Calculate" buttons to set force field types and charges (See attached figure 1). Close all opened dialog.

Step 3: Ensure TiO2 cell document active. Cleave (0 0 1) crystal face and add a vacuum slab with a thickess of 30 A. See attached figure 2 for details. Thus a document named "TiO2_xx (0 0 1).xsd" is generated.

Step 4: Ensure "TiO2_xx (0 0 1).xsd"l document active. Use Forcite for geometry optimization. See attached figure 3 for setting details. Just leave other parameters settings not mentioned as default. Now, all steps completed.

After geometry optimization is completed. Checking the curve of Energy vs optimization step will find an abnormal high energy as shown in attached figure 4. Besides, an extremely high Electrostatic energy will be found in txt document (see figure 4). Moreover, check the optimized atoms on TiO2 surface will show an abnormal overlap between O atom and Ti atom (see figure 5).

Since the Electrostatic energy ∝ 1/R (R is distance from the positive charged atom to the negative charged atom), the overlap of atoms will result in high value of Electrostatic energy.Generally, when one atom gets very close to the other, the van der Waals force between them will prevent further closer contact. It seems this mechanism does not work in current simulation. So I guess the forcefield parameters related to the van der Waals force in COMPASS are not suitable for TiO2 surface simulation.

I have reproduced the similar results with COMPASS and COMPASSII forcefileds in MS 8.0, 2019 and 2020. Using COMPASSIII in MS 2020 also gave the similar results. Besides, similar results were found in TiO2 nano-cluster simulation.