4

Consider a point charge enclosed by some surface, using spherical coordinates, and taking $\hat a$ to be the unit vector in the direction of the surface element, flux is $$\oint\vec E\cdot d\vec A = \frac q{4\pi}\oint\frac{\hat r}{r^2}\cdot r^2d\phi \sin\theta d\theta\hat a \\ ‎\\=\frac q{4\pi}\oint\sin\theta d\theta d\phi (\hat r\cdot\hat a)$$ How to show that this integral is equal to $4\pi$ when the charge is inside and $0$ otherwise? I know that this can be written in terms of solid angle, but my acquaintance with that topic is very low, thus it would be appreciated if one refrains from using that to answer the question.

PS: I am working in HL system which for our purposes means that $\epsilon_0=1$ (though this doesn't affect the question at hand since it's about solving the integral, but is said just for the sake of completeness).

2 Answers2

3

PRIMER:

Note that in regions that are absent of charge, $\nabla \cdot E=0$. Let $V$ be a region, bounded by the surface $S$, that contains a point charge. Furthermore, let $V'$ be a region bounded by the surface $S'$, contained in $V$.

If the point charge is in $V'$, then it is outside of the region $V\setminus V'$ and from Gauss's law and the Divergence Theorem we have

$$\begin{align} \int_{V\setminus V'}0\,dV&=\int_{V\setminus V'}\nabla\cdot\vec E\,dV\\\\ &=\oint_{S} \vec E\cdot \hat n\,dS-\oint_{S'} \vec E\cdot \hat n\,dS \end{align}$$

from which we conclude

$$\oint_{S} \vec E\cdot \hat n\,dS=\oint_{S'} \vec E\cdot \hat n\,dS$$

That is, the value of the surface integral is independent of the surface over which the integration is taken.

If the point charge is in $V\setminus V'$, then $\nabla \cdot E=0$ in $V'$ and $\oint_{S'}\vec E\cdot \hat n\,dS=0$.

We now will show explicitly that if $S'$ is any sphere that contains a point charge, that $\oint_{S'}\vec E\cdot \hat n\,dS=q/\varepsilon_0$ (We could have even chosen a sphere for which the point charge is at the origin.). To that end we proceed.


The electric field $\vec E(\vec r')$ due to a point charge at $\vec r$ at points on the surface of a sphere, centered at $0$ and with radius $a$ is

$$\vec E(\vec r')=\frac{q(a\hat r'-\vec r)}{4\pi\varepsilon_0|\vec r-a\hat r'|^3}$$

So, we wish to compute the integral

$$\oint_{|\vec r'|=a} \vec E(\vec r')\cdot \hat n'\,dS'=q\int_0^{2\pi}\int_0^\pi \left(\frac{a\hat r'-\vec r}{4\pi \epsilon_0|\vec r-a\hat r'|^3}\right)\cdot\hat r'\,a^2\,\sin(\theta')\,d\theta'\,d\phi'$$

Due to the spherical symmetry, without loss of generality, we align the polar axis so that the point charge lies at $\vec r=r\hat z$. Then,

$$\begin{align} \oint_{|\vec r'|=a} \vec E(\vec r')\cdot \hat n'\,dS'&=\frac{qa^2}{2\varepsilon_0}\int_0^\pi \frac{a-r\cos(\theta')}{(r^2+a^2-2ar\cos(\theta'))^{3/2}}\sin(\theta')\,d\theta'\\\\ &=\frac{qa^2}{2\varepsilon_0}\int_{-1}^1 \frac{a-rx}{(r^2+a^2-2arx)^{3/2}}\,dx\tag1 \end{align}$$


The integral on the right-hand side of $(1)$ is

$$\begin{align}\int_{-1}^1 \frac{a-rx}{(r^2+a^2-2arx)^{3/2}}\,dx&=-\left.\left(\frac{r-ax}{a^2(r^2+a^2-2arx)^{1/2}}\right)\right|_{-1}^1\\\\ &=\frac{r+a}{a^2(r+a)}-\frac{r-a}{a^2|r-a|}\\\\ &=\begin{cases} 0&, r>a\\\\ \frac 2{a^2}&,r<a\tag2 \end{cases} \end{align}$$

Using $(2)$ in $(1)$ we find the coveted result

$$\oint_{|\vec r'|=a} \vec E(\vec r')\cdot \hat n'\,dS'=\begin{cases}\frac{q}{\varepsilon_0}&,r<a\\\\ 0&,r>a\end{cases}$$

as was to be shown!

Mark Viola
  • 179,405
  • Sorry man, I didn't wanted to ignore you, it's just that your answer requires a lot of attention on my part which I sadly can't afford right now, but tomorrow morning the first thing I'll do is read your answer. Your answer looks really detailed and I appreciate that, I am sure it will help me understand better, thanks for that. – GedankenExperimentalist Oct 10 '22 at 22:25
  • You're welcome! My pleasure. – Mark Viola Oct 10 '22 at 22:33
  • Thank you for the answer, can we generalize this for any arbitrary closed surface not just a sphere? – GedankenExperimentalist Oct 11 '22 at 06:47
  • Yes, of course. Just exploit the fact that $\nabla \cdot \vec E=0$ in regions without charge. – Mark Viola Oct 11 '22 at 12:38
  • I meant the integral approach without resorting to divergences and Dirac delta's – GedankenExperimentalist Oct 11 '22 at 12:51
  • https://math.stackexchange.com/q/1866709/1053268 This question over here is essentially the same, but the answer given there is way beyond my mathematical capabilities at the moment. It seems like I have to be satisfied with the divergence approach in the meantime. – GedankenExperimentalist Oct 11 '22 at 12:55
  • @GedankenExperimentalist My solution only relies on direct integration, which you almost assuredly have the background to understand. If you can point to the part of the solution that is giving you a problem, I would be happy to walk you through it. – Mark Viola Oct 11 '22 at 13:25
  • Over here you assumed the sphere to integrate over right? I am asking for the case where one proves the Gauss theorem directly by integration for any arbitrary surface, not just a sphere. One way I can think to do this in a less rigorous way is to construct any arbitrary surface out of many small spheres, and then apply your proof on each of them. – GedankenExperimentalist Oct 11 '22 at 14:29
  • @GedankenExperimentalist Inasmuch as $\nabla \cdot \vec E=0$ in regions that are absent of charge, we are free to do the following. Let's call the surface over which you want to integrate $S$ and the region that it bounds $V$. The point charge is in $V$. Now, choose a spherical region inside $V$ and call the surface that bounds it $S'$. Note suppose the charge is inside $S'$. Then $\oint_S \vec E\cdot \hat n ,dS=\oint_{S'} \vec E\cdot \hat n ,dS$ since $\nabla \cdot E=0$ in the region bounded by $S$ and $S'$. Therefore, we can proceed by integrating over $S'$, which is what we did here – Mark Viola Oct 11 '22 at 15:59
  • @GedankenExperimentalist I've edited the solution with a primer section that shows the result is correct for any surface that contains a point charge. – Mark Viola Oct 11 '22 at 16:22
2

This answer is only helpful if you are familiar (and comfortable) with the divergence theorem and Dirac delta distribution.

By the divergence theorem we have $$ \oint_{\partial V} \vec E\cdot d\vec A = \int_V \nabla\cdot\vec E\, dV. $$ It is very useful to know that $\nabla\cdot\left(\frac{\hat r}{r^2}\right)= 4\pi \delta(\vec r)$. Here is a link. This gives the divergence of $\vec E$ of a charge at position $\vec r_0$ to be $\nabla \cdot \vec E= \frac{q}{4\pi}\times 4\pi\,\times \delta(\vec r- \vec r_0)$. So we have: $$ \oint_{\partial V} \vec E\cdot d\vec A = q\int_V \delta(\vec r- \vec r_0)\, dV=\begin{cases}q&\vec r_0 \in V\\0&\vec r_0 \notin V\end{cases}. $$

stochastic
  • 2,560
  • 1
    Thanks for the answer, I know this method but I was looking for a more direct approach, nevertheless, the method that you mention is indeed the most practical one from the perspective of physics. – GedankenExperimentalist Oct 10 '22 at 20:03