Solution at angle (phase) in RF module background field / scattered field formulation

Hi, I'm using the background field / scattered field formulation to do RF frequency domain simulations.
Basically I have two steps.
In step 1, I use a port to generate a planar EM wave. My simulation domain has multiple dielectric layers.
In step 2, I use the result of step 1 as the background field, and put a lossy scatterer in the simulation domain to solve for the scattered field.
Then when I want to see the result, I find for the result of step 1, which is the background field, I can specify the value of "solution at angle (phase)" to see the E-field (Ex, Ey or Ez) changes. I find it is equivalent to multiplying the E-field (Ex, Ey or Ez) by exp(i*phase), which makes sense.
However, if I specify the value of "solution at angle (phase)" for the E-field obtained in step 2, the effect is no longer equivalent to multiplying the E-field (Ex, Ey or Ez) by exp(i*phase). I know this should make sense as well. However, does anyone know how comsol determine the change of the E field obtained in step 2, when I change the "solution at angle (phase)"? There must be an equation, or algorithm that comsol uses but I can't figure out what that is.
Any comments will be appreciated!