6

Why do RDF plots have the probability at a particular distance >1. How can probability be more than one?

I am asking this because I want to understand what do the peaks of different length but at the same position on the x axis mean

The below is an example of a RDF plot: enter image description here

Ian Bush
  • 3,039
  • 1
  • 15
  • 28
fireball.1
  • 460
  • 4
  • 15

1 Answers1

12

The radial distribution function isn't a probability or a probability density. The usual definition is $$g(r)=\frac{\rho(r)}{\rho_0}$$ where $\rho(r)$ is the number density at a particular distance $r$ away from some specified atom and $\rho_0$ is the bulk density of the material.

It can be larger than one if the local density is greater than the bulk (which you see in the first peak of your graph), so it can't be a probability. It also can't be a probability density, as integrating the RDF from $0$ to $r$ should give you the number of particles (divided by the bulk density) in that spherical volume, which should continue to increase even up to macroscopic distances (in contrast to a probability density which should integrate to $1$).

So what does the RDF tell you? Well if you integrate it over a spherical shell and multiply by the bulk density, you obtain the number of atoms in that shell. One particular shell we are often interested in is the first coordination sphere. Using the RDF, we can define the number of particles in the first coordination sphere as $$n(r')=4\pi\rho_0\int_0^{r'}g(r)r^2dr$$ where $r'$ is the location of the first minimum of the RDF. We can similarly determine the number of atoms in the $n^{\text{th}}$ coordination sphere by integrating from the$(n-1)^{\text{th}}$ minimum to the $n^{\text{th}}$ minimum.

Tyberius
  • 11,686
  • 10
  • 42
  • 86
  • https://chem.libretexts.org/Bookshelves/Physical_and_Theoretical_Chemistry_Textbook_Maps/Book%3A_Quantum_States_of_Atoms_and_Molecules_(Zielinksi_et_al.)/08%3A_The_Hydrogen_Atom/8.02%3A_The_Wavefunctions defines RDF for bound electrons. – Karsten Jul 13 '20 at 10:13
  • This is not about electrons – Ian Bush Jul 14 '20 at 09:03
  • Why at longer distances converges to 1, that is it behaves like bulk density? Is it because when we have larger spherical shells we approximate better the bulk? – Anton May 13 '22 at 15:50
  • @Anton considering very long (i.e. macroscopic) distance, we know it has to converge to the bulk density because eventually we are including the whole solution. For moderate microscopic distances, it should already start to approach 1, as you start to include more of the bulk structure rather than just individual local clusters. – Tyberius May 13 '22 at 16:20
  • @Tyberius When you say we are including the "whole solution" you mean that the shell dimensions are so big that it mimics the bulk right? Because $ρ(r)$ refers to density between $r$ and $r+dr$ and not $0$ and $r$ (the whole solution can be thought as the limit of $r$ going to infinity). – Anton May 13 '22 at 18:10
  • @Anton Looking over this again, I realize my statements about $g(r)$ converging are incorrect. While the integrated RDF should eventually converge to the bulk density, the local density doesn't in general have to approach the bulk and consequently $g(r)$ doesn't have to converge to 1. A simple example case would be a sphere of constant density $\rho_0$, surrounded by a large shell of constant density $\rho_1$. The bulk density is some average of these two, but the bulk density is always one or the other. – Tyberius May 13 '22 at 18:50