0
$\begingroup$

I have a uniformly polarized infinite cylinder of radius $a$ whose axis lies along the $z$ axis with polarization vector $\vec{P} = P \vec{u_x}$, where $\vec{u_x}$ is the unit vector towards the $x$ direction, as shown in the figure, and I want to know the electric field $\vec{E}$ along the $z$ axis.

enter image description here

I first attempted using Gauss Law and considering the cylinder as the superposition of two cylinders separated a distance $\vec{\delta}$ with volumetric charge densities $+\rho$ and $-\rho$ respectively, and I obtained that the electric field at any point along the $z$ axis is:

$$\vec{E} = -\frac{\vec{P}}{2\epsilon_0}$$

Well, along the $z$ axis and for all positions inside the cylinder. However, I want to solve this problem as well by direct integration of the potential $\phi$ and then take its gradient to obtain the electric field, so I first obtained the superficial bound charge density as follows:

$$\sigma_b = \vec{P}\vec{u_n} = \vec{P}\vec{u{\rho}} = P\cos{\phi}$$

Where $\phi$ here means the angle between $\rho$ and the $x$ axis. Then I take the integral over the potential:

$$\vec{r} = z \vec{u_{z}}$$ $$\vec{r'} = a \vec{u_{\rho}} + z' \vec{u_{z}}$$ $$\vec{R} = \vec{r}- \vec{r'}$$ $$R^2 = (a)^2 + (z-z')^2$$

$$\phi = \frac{1}{4\pi\epsilon_0}\int\int \frac{\sigma_b}{R}da = \frac{1}{4\pi\epsilon_0}\int^{+\inf}_{-\inf}\int^{2\pi}_{0} \frac{P\cos{\phi'}}{\sqrt{(a)^2 + (z-z')^2}}d\phi'dz'$$

But then, as the integral of $\cos{\phi'}$ is zero, then the whole integral gets zero and the potential is zero, so there is no electric field. I know that this is not correct because I have calculated the electric field before with Gauss Law, but I don't understand what I am doing wrong so that I get these results. Please, could somebody tell me where is the mistake?

Thanks in advance!

$\endgroup$
5
  • $\begingroup$ Find your whole question confusing. You agree (below) that there is no $E_z$, yet you also claim it is $-P/2\epsilon_0$. $\endgroup$
    – ProfRob
    Dec 2 '20 at 7:23
  • $\begingroup$ In every point along z axis $\vec{E} = -P/2\epsilon_0 \vec{u_x}$, because that's the $\vec{P}$ directrion and because this is an isotropic medium. So the $E_z$ and $E_y$ components are zero. $\endgroup$
    – conradDell
    Dec 2 '20 at 16:52
  • $\begingroup$ Even though $E_x$ exists inside the cylinder and its value is $E_x = -P/2\epsilon_0$ in every point of it, including the z axis. $\endgroup$
    – conradDell
    Dec 2 '20 at 16:58
  • $\begingroup$ I see. I have made a couple of small edits. $\endgroup$
    – ProfRob
    Dec 2 '20 at 18:05
  • $\begingroup$ Thanks! It's much more clear now. $\endgroup$
    – conradDell
    Dec 2 '20 at 19:56
1
$\begingroup$

I'm pretty sure you have over-simplified the integral.

The separation between a general point in the cylinder, $(\rho, \phi, 0)$, where we can set $z=0$ by symmetry, and a point on the surface of the cylinder, $(a, \phi', z')$, in cylindrical coordinates is: $$ R^2 = a^2+\rho^2-2a\rho\cos(\phi-\phi') + z'^2$$

Thus $$V(\rho, \phi) = \frac{1}{4\pi \epsilon_0} \int^{\infty}_{-\infty} \int^{2\pi}_{0} \frac{ P \cos \phi'}{(a^2+\rho^2-2a\rho\cos(\phi-\phi') + z'^2)^{1/2}}\ a d\phi'\ dz'\ .$$

Looks tricky, but presumably must give the solution $$V(\rho, \phi, z) = \frac{P}{2\epsilon_0} \rho\cos \phi$$

The potential at $\rho=0$ then is zero, because the integral collapses to what you have in your question (bar the fact that the surface element is $a d\phi'\ dz'$ and not just $d\phi'\ dz'$).

I believe it would be more usual to tackle this using the general solution to Laplace's equation in cylindrical coordinates.

$\endgroup$
1
  • $\begingroup$ Thanks! Yes, the only component is $\vec{E_x}$, which is the result I've got from Gauss. What I don't understand is why the scalar potential is zero from direct integration while $\vec{E}$ isn't. $\endgroup$
    – conradDell
    Dec 1 '20 at 20:58

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.