Simulating colloidal Ag nanowires

I have been trying to simulate the extinction (absorption + scattering) spectra of colloidally synthesized metallic nanoparticles, i.e. particles that can freely move in an aqueous solution. I took inspiration from simulations of incoherent unpolarized dipoles (Incoherent unpolarized dipole) and simulated the metallic nanostructure in three orthogonal orientations and then averaged all the spectra. This gives good agreement with experimental spectra for low aspect ratio structures.

However, I am now trying to simulate an Ag nanowire with diameter of 50 nm and a length of 10 000 nm (see attached simulations), which seems to be more challenging. The experimental and simulated spectra do not match well, see below. This is mostly due to the contribution of the simulation where the source impinges on the side of the nanowire (labeled ‘sourcexaxis’ in the attachment), which gives the peak at 445 nm.

Surprisingly, when we define the averaged spectrum as (1 * the longitudinal spectrum + 2 * the transverse spectrum)/3, i.e. we double the weight of the ‘pol90’ simulation and we ignore the ‘sourcexaxis’ simulation, the agreement with the experiment is perfect. Of course, I don’t really have a physical reason to define the spectrum this way, but I’d be surprised if it is a coincidence.

My questions now are as follows:

  • Would this method of averaging over three orthogonal orientations be the correct way of simulating the average behavior of a freely moving particle? If no, which method would you recommend?
  • Can you think of any other reason why the averaged spectrum does not agree with experiment, whereas the spectrum with the strange weighing does match?

Thanks in advance!

AgNW_50nm_diameter_pentagonal_water_3D_singleP_pol0.fsp (340.7 KB) AgNW_50nm_diameter_pentagonal_water_3D_singleP_pol90.fsp (340.7 KB) AgNW_50nm_diameter_pentagonal_water_3D_singleP_pol90_sourcexaxis.fsp (343.3 KB)

Hi @r.f.hamans,

Thank you for the question. I agree that the match between experiment and simulation in your second plot is very interesting. Unfortunately I haven’t seen this type of simulation before, and I don’t have any immediate answers for you right now. This is a difficult case so I will have to investigate this issue further and get back to you.

For now, I would recommend that you consult the literature in this field to see if and how this type of simulation has been done in the past. Please let me know if you come across anything.

Hi @kjohnson ,

Thanks for the reply! It’s maybe useful to add that, experimentally, the peak at 445 nm is never observed, even for wires of different diameters (see for example ACS Nano 2016, 10, 7892−7900 and Nano Lett. 2018, 18, 5329−5334).

My logic was that in an aqueous solution the wire can have any orientation and that all orientations can be decomposed into superpositions of three orthogonal orientations. However, it seems that the simulation where the source impinges on the side of the wire never contributes to the experimental spectrum, which is something that from a conceptual point of view seems weird.

I’ll let you know if I find anything else. Looking forward to your reply!

Edit (I accidentally closed the topic): if I normalize the cross sections by the area over which the source is injected the agreement indeed is much better. As the source has the dimensions 10500x500x500 nm, I simply multiplied the cross sections of the side of the wire by 500x500/(500x10500) before doing the averaging (I also used a slightly more accurate dielectric function here, which results in the narrower peak, but that’s fine):

Hi @r.f.hamans,

I’ve discussed this with a colleague and we believe that your general approach should work. The issue probably has to do with how the results are normalized before you combine them. Normalization for the TFSF source is a bit different than how normalization works for most sources. See these pages for more information:

You should take a close look at how the results are being normalized, and whether it makes sense to combine them in the way you mentioned above. You may have to normalize the results by their scattering cross-sections, for example. I would expect the ends of the wire to have a small scattering cross-section, which may be why its peak doesn’t appear in the experimental results.

Let me know if this helps. I will update you if I make any more progress on this.

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.