Questions and Answers

0 Like 0 Dislike


Core-shell simulation and material size in nm

I am working on Ag core-Au shell nanoparticles and I am using DDSCAT+ software to simulate the extinction spectrum of the materials. I did follow the tutorial video on youtube and knew how to use the software. However, I have some questions as follow:


1)    The materials size is at nm, but the measurement system in DDSCAT+ is either no unit or metric. Thus, what can I do to set up the nm-level size for 3-D model?


2)    I am working on the core-shell formation in Blender, but I don’t know the proper way to design that structure. What I did was that I constructed two UV sphere meshes and I increased the thickness of one shape to form the shell. Is there any proper way to design a core-shell formation in Blender? All the parameters and screenshots are attached in this e-mail, so please take a look.


3)    In the last step of simulation, the software has been running for about an hour and 30% of work has been done. I stopped the simulation to check the results, but the simulation is not complete and the near field could not be calculated (shown in the screenshots). Therefore, could you please let me know the normal amount of time that the simulation needs?


Thank you so much for your consideration and I am looking forward to hearing from you!

Report abuse

Chosen Answer

  1. 0 Dislike

    AbderRahman N Sobh

    Hi, there are a few notes I will make about the simulation you are trying to create:

    1) The easiest way to do elliptical (i.e. spherical) shells is to use the preset option. Skip using Blender and DDAConvert and simply choose the "Concentric Ellipsoids" shape option in nanoDDSCAT. This will allow you to easily allocate the size of your system directly in nanometers.

    2) If you want to make other types of shapes, i.e. rectangular prism shells or something that is not a preset: Place the shapes inside of each other and in DDAConvert select the option to have separate dielectric files for each shape. By default, overlapping shapes will give the overlapped area to the smaller of the 2 shapes. There is another option that you can choose in DDAConvert if instead you want to allocate that overlapped area to the larger shape (this is the "Inverted Point Allocation" option).

    3) The progress bar is a little bit rigid (it costed too much resource to keep polling the program during simulation). If it looks "frozen" it is probably still running. Our program is pretty good about crashing when something goes wrong. Sitting at 30% is normal behavior, once the simulation is complete it will jump to 70% or something like that.

     In general, simulations take more time if you ask for more wavelengths, use a more dense point-population for your shape, or want to calculate a nearfield. Periodic simulations take an extremely long amount of time (2D periodic simulations with a nearfield may take over a week to complete). Unfortunately, we do not currently have a clear algorithm for prediction of how long simulations will take. My best advice is to set a large time limit and run the simulation remotely if you are running a dense shape (i.e. 100k+ dipoles), large wavelength sweep with many divisions, and electric field. Additionally, you can try simulating a broad wavelength sweep before doing more specific tests (i.e. doing 500-600nm with 6 steps: 500,520,540,560,580,600 then maybe zeroing in on 520-540nm with 20 steps: 520,521,522,...540). Or even just a single wavelength, without a nearfield, to confirm that your parameters are working and to get a timing benchmark before attempting the larger jobs.

    Do note: there is a 12 hour time limit of the Virtual-Local simulation option. If you choose "Remote" submission, you must choose your own time limit which is basically your best guess of how long the simulation will take. If the time limit is too low, the simulation will fail. Raising the time limit may increase the queue time before the simulation is run, but there is less of a chance of failure due to not enough time.

    AbderRahman (Abby)

    Reply Report abuse

    Please login to answer the question.

0 Responses

No other responses made.