I am working on Cu2ZnSnSe4 semiconductor thin film on glass substrate. The thickness of absorber is around 650 nm-720 nm.
I roughened the back side of the substrate to suppress the effect of back side reflection, and I work with 55-75 degree (5-degree interval).
The optical model I used: Srough (EMA=50%void+50%toplayer)
Absorber(Cu2ZnsnSe4) (I used GenOsi to produce the optical constant)
Intermix (to account the inhomogeneities and/or mixture effect of Absorber and Substrate)
Substrate (Glass)
Here my question, All the thicknesses are in the range of expected values and MSE is around 19. However, when the thickness of intermix layer jump to a huge number (from 3nm to 320nm), the MSE becomes smaller 10 and the fit quality becomes better. But, The value of intermix layer thickness is too much. What kind of effect could it be a reason?
p.s. I work with 240 nm-1700nm wavelengths. We 're considering the back side reflection and hampering it, But what if there is a reflection effect between absorber and substrate and the existence of intermix layer in the optical model is not enough to define the material properties.
Thanks for in advance.
This indicates high correlation between parameters chosen in the model. Do check the fit statistics to ensure that the parameter correlation is less than +/- 0.92 . The model can fit with absurd values but you have to be sure that it is physically acceptable. If possible go for cross section SEM and see the interface for better correlation.
Multiple angles are only required if you get difference in pseudo dielectric constants otherwise you will not get any new information. You can include depolarization data in your sample. This will allow you to include thickness non uniformity in the model.
On experimental side, you can deposit with thickness close to 50 nm or 100 nm so that you accurately model the interface layer.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Material Characterization