Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Scattering Cross Section
Posted 1 nov 2009, 13:24 GMT-5 Modeling Tools & Definitions, Parameters, Variables, & Functions Version 5.2 25 Replies
Please login with a confirmed email address before reporting spam
I am trying to find the best way to calculate the scattering cross section in COMSOL for a 2-D cylinder. I eventually will move to 3-D spheres and cylinders. What I have is a circle of 30 nm in radius and my circular PML layer drawn 10 Lambda away, where at this boundary (far field) I want to calculate the scattering cross section. I am doing all of this in the RF Module->In Plane Waves->TE Waves->Scattering Harmonic Propagation.
Scattering cross section to me seems to be defined differently depending on what text book you are looking at. The one I see most is:
Csca = (1/|Eincident|^2) Integral (|Escattered|^2) dr
So here are my more specific questions.
1. Would you take Eincident to be E0iz_rfweh or Ez?
2. Is scEZ the Escattered filed I am looking for or do I have to subtract out the incident field?
Any input on your experiences calculating scattering cross sections would be greatly appreciated.
Regards,
Jacob
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
What I am trying to do first in 2-D (cylinder/circle) and will eventually move to 3-D like Yevgeniy's model. This should be easy I would think in 2-D, but I am missing something.
Here is a breakdown of what I am doing maybe someone on here will see something obvious I am missing. I have actually attached a COMSOL report with the details as well. I have a 30nm Ag cylinder (circle in 2-D) that I am using a Drude model to model the material. I want to have a TM wave incident on the cylinder and measure the far field scattering cross section at a wavelength range from 300-700nm. I know there should be a peak in the Scr at ~340nm and know what the surface plot should look like. I am using RF Module/TM-waves /Harmonic Propagation right now. I tried using Scattered Harmonic Propagation with the same results.
I have a PML, an intermediate ring of just vacuum where I will measure Scr and the Ag cylinder. Just looking at one wave length say 340nm (where I expect my resonance), I get a very strange result. When I plot my surface of the incident Ey wave (Ey_rfweh-scEy_rfweh) my incident field looks very distorted. I think this be my first problem. I attached a screen capture.
Any input would be greatly appreciated. I know I probably haven’t provided enough information as there is a lot that goes into each model. Hopefully some has had similar issues when they first started and can suggest things to check from experience.
Yevgeniy, have you ever done this in 2-D?
Thanks,
Jacob
Attachments:
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
Maybe I am doing this all wrong? Once again, all I am trying to do is to measure the scattering cross section of a 30nm (radius) Ag cylinder. The model I uploaded is only evaluating at 340nm where I expect a resonance, which should be visible in a surface plot of Ey and also Hz. Once the model is correct I will sweep frequencies between 300 and 700nm. I hope to generate a Cscatter plot which looks like my attachment (done with analytical approach in Maple)
Any help would be greatly appreciated.
Please login with a confirmed email address before reporting spam
But for sure decrease subdomain size, make mesh more fine (34nm for outersubdomains and even less for cylinder). And try other modes. And try to get good results with complex index of refraction for Ag, and then continue to Drude model.
Attachments:
Please login with a confirmed email address before reporting spam
I guess one last question is on the maximum mesh size. Is there a rule of thumb mesh based on geometry and wavelength used for excitation? I have ordered a FEM modeling book, but I am still waiting for it to arrive.
I am going to have a 100nm cylinder with a 10nm "jacket" around it of a different material (metal coated nanowire). I worry in the 10nm shell that my mesh maximum mesh size will need to be <1nm. I don't even know if that sounds feasible or not.
I guess we will see what happens! Thanks again.
Jacob
Please login with a confirmed email address before reporting spam
If you make your mesh finer and your solution changes dramatically then you probably need to further decrease mesh size, if however solution didn't change much, then your mesh probably OK.
Also always look at your solution if you see that it has artifacts then decrease size of the mesh. If it not converging then again decrease size of mesh.
Please login with a confirmed email address before reporting spam
I have got the message to ensure at least 5 nodes per wavelegth period, as a minimum. But it's as usual, ask yourself how fast is your solution modulating, what kind of gradient do you have, you must adapt the mesh to be able to catch correctly the gradients slope, or the slope of your changes, otherwise you might get aliasing and strange results.
In structural, and probably it's rather general, doubling the node numer and expect less than 10% change then you are generally in a good situation.
You have also the mesh refinement option in the solver, you might need to adapt the L2 optimisation function for your case.
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
In the end I want a extinction, and if I can calculate an absorption cross section then I can get there.
Cext = Cabs + Csca
Anyone every calculated the absorption or extinction?
Please login with a confirmed email address before reporting spam
Nanoshells to nanoeggs to nanocups: optical properties of reduced symmetry core/shell nanoparticles beyound the quasy-static limit, by Mark W. Knight and Naomi Halas, in New Journal of Physics...
You will see there that absorption cross section is proportional to integral of resisting heating over your metallic nanoparticles...
Cheers
Srdjan
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
Rule of thumb,
If you make your mesh finer and your solution changes dramatically then you probably need to further decrease mesh size, if however solution didn't change much, then your mesh probably OK.
Also always look at your solution if you see that it has artifacts then decrease size of the mesh. If it not converging then again decrease size of mesh.
Higher in the thread, someone indicated that you posted a 3D mie scattering model.
I don't find it in the Model Exchange - is it still available?
Thanks,
Lloyd
Please login with a confirmed email address before reporting spam
www.uk.comsol.com/showroom/gallery/3459/
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
i had done for silver particle.
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
very thankful.
Attachments:
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
However, the abs efficiency is on the order of 10^10 and the scattering efficiency on the order of 10^23. So i guess there is someting wrong with the constants in front of the integrals.
Did you applied these formulas straight away? Or did you make any change due to the fact that we are dealing with 2D simulations instead of 3D? (I guess tat the cross sections should have dimensions of m in 2D, instead of m^2 in 3D)
Any help is more than welcome!
Jorge
Please login with a confirmed email address before reporting spam
Please login with a confirmed email address before reporting spam
This thread seems to touch on some very old versions of the software. To address the general question of computing the cross section of an object illuminated by a plane wave we would suggest that you first start with this example:
www.comsol.com/model/computing-the-radar-cross-section-of-a-perfectly-conducting-sphere-10332
This example computes the radar cross section of a perfectly conductive sphere, and compares the computational result to analytic solutions and shows agreement. Most of the older questions in this thread are moot or addressed by this model.
Please login with a confirmed email address before reporting spam
Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.
Suggested Content
- FORUM Brillouin scattering
- KNOWLEDGE BASE Graphics Window Not Responding on Computer with Intel Iris Xe Graphics Chip
- BLOG Simulation Improves Electromagnetic IED Detection Systems
- KNOWLEDGE BASE What to do when a linear stationary model is not solving
- BLOG Electromagnetic Scattering in 2D Axisymmetric Models