Syafiq,
It is not an error in the input plane but in the design of your structure. You have a waveguide consisting of PMMA on top of Si sitting on a substrate of Si. Where precisely do you expect the light to be localized because the power injected into the waveguide will leak into the substrate and as such the mode solver is having issues finding the mode. Do you perhaps mean to have the substrate as SiO2?
As a secondary concern you are using a mesh resolution that is far too coarse. The auto setting ensures that you are getting a mesh resolution of 1/10 lambda to give you numerical stability for the source, this is the coarsest you should ever use. If you are going to use a custom setting it must be a mesh size smaller than the auto setting.
Scott