The Stokes-Einstein rotational diffusion relation tells us that we can write down a rotational diffusion coefficient for a sphere as:

$D_r \approx \frac{k_B T}{\zeta_f} \approx \frac{k_B T}{(8 \pi \eta)(r)^3}$

Where $k_B$ is Boltzmann's constant, $T$ is the temperature in Kelvin, $\zeta_f \approx (8 \pi \eta)(r)^3$ is the friction, $\eta$ is the viscosity of the medium (e.g. $\approx 1 \space cP$ in pure water), and $r$ is the radius of the sphere.

For an example calculation of $D_r \approx 0.2 \space rad^2/s$ (or $Hz$) of a $\approx 1 \space \mu m$ radius sphere in pure water at room temperature, please see here: http://www.microurl.info/YOtwp (this shortened URL is going to WolframAlpha).

My question is the following: Take a point on the sphere undergoing rotational diffusion in solution, and select a random point on its surface. Define a vector $v_1$ between this random point and the point at the center of the sphere. Let $v_2$ be another randomly oriented vector. What probability distribution do we have for the time until a minimum angle $\theta$ between $v_1$ and $v_2$ is achieved?

Alternatively, if we start with the vectors $v_1$ and $v_2$ being within a minimum angle $theta$, what time distribution do we have until the minimum angle becomes larger than $\theta$?


If I have understood you correctly, your rotational diffusion problem is equivalent to the problem of a particle diffusing on the surface of a sphere (angular diffusion constant $D_r$), starting at some polar angle $\theta_0\in(0,\pi)$ and first crossing the angle $\theta\in(0,\theta_0)$ at time $t$. This is a classic problem in first-passage-times, see for example S. Redner, A guide to first-passage processes, or arXiv:1101.5043.

The mean first passage time is $$\langle t\rangle=\frac{2}{D_{r}}\ln\left(\frac{\sin(\theta_0/2)}{\sin(\theta/2)}\right).$$

If you choose the starting point at random, uniformly on the sphere, then you would perform one more average over $\theta_0$ (with weight $\sin\theta_0$), producing the average

$$\langle t\rangle=-\frac{2}{D_{r}}\left[\ln[\sin(\theta/2)]+\frac{1+\cos\theta}{4}\right].$$

The full probability distribution of the first passage time seems more problematic, I have not found it in the literature.

  • $\begingroup$ I agree with your interpretation of the problem. However, I'm a little confused about your expression for $<t>$ provided random uniform initiation, shouldn't there for a $D_r$ term somewhere? $\endgroup$ – Ayn Feb 17 '14 at 15:02
  • $\begingroup$ Also, is the expression for $<t>$ starting from $\theta_0 \in (0, \pi)$ from the linked arXiv paper? I can't seem to find the derivation for it in the 3D case section. $\endgroup$ – Ayn Feb 17 '14 at 15:05
  • $\begingroup$ This answers my question! However, if you could point out where $<t>$ was derived (unless you did it yourself) that would be very helpful for me in terms of understanding. $\endgroup$ – Ayn Feb 17 '14 at 15:18
  • $\begingroup$ typo (missing $D_r$) corrected; the expression for $\langle t\rangle$ is the unnumbered equation after Eq. 57 in arXiv:1101.5043, for $\omega=0$ (no bulk diffusion, only surface diffusion). $\endgroup$ – Carlo Beenakker Feb 17 '14 at 15:19
  • $\begingroup$ Ok, I see. Regarding the unnumbered equation after Eq. 57 in arXiv:1101.5043: we're assuming the unit radius sphere, otherwise it's: $\langle t\rangle=\frac{2R^2}{D_{r}}...$, we're dropping the terms involving contributions from $D_2$ (related to bulk diffusion) (which seems OK to me), and we're neglected the latter summation/integral term as a small positive constant that does not scale with the diffusion coefficient. $\endgroup$ – Ayn Feb 17 '14 at 15:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.