A permanent cylindrical magnet with diameter $D$ and length $L$ possesses an uniform magnetization $\mathbf{M}$ going in the $z$-direction in a cylindrical coordinate system. The magnet is placed with its centre in the origin, with its length going along the $z$-axis.


Given the expression for the magnetic field along the $z$-axis $$\mathbf{B}(z)= \mathbf{a_z} \cdot \frac{\mu_0M}{2} \Bigg[ \frac{z+ \frac{L}{2}}{\sqrt{\big(z+\frac{L}{2} \big)^2+ \big(\frac{D}{2} \big)^2}}- \frac{z- \frac{L}{2}}{\sqrt{\big(z-\frac{L}{2} \big)^2+ \big(\frac{D}{2} \big)^2}} \Bigg] $$ arrive at an expression for the $\mathbf{H}$-field along the $z$-axis. What is the direction of $\mathbf{H}$ inside and outside the magnet?

My attempt

Given the simple relation $\mathbf{H}=\frac{\mathbf{B}}{\mu_0}-\mathbf{M}$ we get

$$\mathbf{H}(z)=\mathbf{a_z} \cdot \frac{M}{2} \Bigg[ \frac{z+ \frac{L}{2}}{\sqrt{\big(z+\frac{L}{2} \big)^2+ \big(\frac{D}{2} \big)^2}}- \frac{z- \frac{L}{2}}{\sqrt{\big(z-\frac{L}{2} \big)^2+ \big(\frac{D}{2} \big)^2}} \Bigg]-M \cdot \mathbf{a_z} $$

But I'm not sure how to answer what direction $\mathbf{H}$ has inside and outside the magnet.


I found this example in my book, that revolves around the exact same scenario, a cylindrical magnet with uniform magnetization $\mathbf{a_z}M_0$. In the example, the author writes about the relationship about $\mathbf{H}$ and $\mathbf{B}$. He also talks about their direction, which I have marked with yellow.

enter image description here

So if the scenario is the same, doesn't the example answer also answer my problem completely?

  • $\begingroup$ Well, judgning by the answers in those two links it seems the example included in my edited question is exactly what I need to fully solve my problem. @RobJeffries $\endgroup$
    – Carl
    Oct 27, 2020 at 20:10
  • $\begingroup$ What book are you using? $\endgroup$ Oct 30, 2021 at 9:01
  • 1
    $\begingroup$ @user1700890 Field and wave electromagnetics, by Cheng I think. $\endgroup$
    – Carl
    Oct 30, 2021 at 13:53

1 Answer 1


I will give a more qualitative answer than just formulas. Considering the given proposal one has to be very careful since $\mathbf{H}$ not only has to fulfill Maxwell-equations and material equations, but also the boundary conditions which are not the same for $\mathbf{B}$ and $\mathbf{H}$.

The Maxwell equations in case of a permanent magnet without any currents are those of magnetostatics:

$\nabla \times \mathbf{H}= 0 \quad\quad \text{and} \quad\quad \nabla\cdot \mathbf{B}=0$

and there is the relevant material equation (assuming cgi-units): $\mathbf{B} = \mathbf{H} + 4\pi \mathbf{M}$.

As we want to know $\mathbf{H}$ we eliminate $\mathbf{B}$ and get:

$$\nabla \cdot \mathbf{H} = -4\pi \nabla \cdot\mathbf{M} \quad {and} \quad \nabla \times \mathbf{H}= 0$$

This system of equations resembles to the equations of electrostatics:

$$ \nabla \cdot \mathbf{E} = 4\pi \rho \quad {and} \quad \nabla \times \mathbf{E}= 0$$

So the solution will resemble an already well-known case of electrostatics.

In the following we also consider that the permanent magnet is directed along the z-axis, the only non-zero component of the magnetisation is $M_z$.

The only locations where the divergence of the magnetisation is non-zero are the ends of the permanent magnet. At the borders along the magnet, the z-component of the magnetisation $M_z$ also suffers an abrupt change, but in radial direction, so it does not matter since changes of $M_z$ in radial direction do not contribute to the divergence (only $\frac{\partial M_z}{\partial z}$ contributes).

So the permanent magnet resembles to two plates of opposite electrical charges at each end of the magnet, i.e. the $\mathbf{H}$-field looks like that of a cylindrical electrical condensator.

The equivalent charge of such a system is: $\rho = -\frac{\partial M_z}{\partial z}$ and assuming that the abrupt change at each end of the magnet is the same (respectively symmetrical) both charge densities $\rho_1 = - \rho_2$ as it is for a condensator. If the diameter $D \rightarrow 0$, the field line picture would resemble to that of two opposed point charges $+$ and $-$. So in order to know the $\mathbf{H}$ field in space as a function of coordinates just refer to the $\mathbf{E}$ field in space for a condensator.

  • $\begingroup$ Hi Frederic, thank you for your answer but I am not really sure how to follow it from there. However, I made an edit to the question, would you mind looking at it? $\endgroup$
    – Carl
    Oct 27, 2020 at 16:16
  • $\begingroup$ I accepted your answer, in hopes of getting an additional response. @FredericThomas $\endgroup$
    – Carl
    Oct 28, 2020 at 9:52
  • 1
    $\begingroup$ @Carl: Thank you for the green check mark. In the comments there are references to posts which explain even better what I said. In particular you find pics for the directions of the field lines of B and H, so I wonder what you still need more. A formula ? Anyway if I gave a formula it would be one based on the electric condensator picture. $\endgroup$ Oct 28, 2020 at 9:57
  • $\begingroup$ I saw the other posts that were linked and they also helped me, but I just want to be sure that I understand it correctly. Can you confirm that, in my case that: 1. Outside the magnet the H and B-field have the same direction and are proportional. 2. Inside the magnet H and B point in opposite direction and are related by the equation $\mathbf{H}=\mathbf{B}/\mu_0 -\mathbf{M}$? $\endgroup$
    – Carl
    Oct 28, 2020 at 10:10
  • 1
    $\begingroup$ @Carl: Locally this relation is correct. Inside the magnet it is like that $\mathbf{H}=\frac{\mathbf{B}}{\mu_0} -\mathbf{M}$. But outside of the magnet $\mathbf{M}=0$. So the formula given in your post only "holds" (I did not check the part concerning the $\mathbf{B}$-field though) inside the magnet (well outside one can set M=0 of course). $\endgroup$ Oct 28, 2020 at 11:46

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.