My problem is related to [1]. I do not dispute the solution in [1], however, it is not helping me to understand the problem that I face. I am working through Example 10.5 in Modern Electrodynamics by Zangwill. This problem pertains to the double-curl equation.

In this example, we integrate $\boldsymbol{\nabla}\times\left(\boldsymbol{\nabla}\times \mathbf{A}\right)$ to find $\mathbf{A}$ for a long straight cylindrical wire with radius $a$ which carries a uniform current density $\mathbf{j} = j\,\hat{\mathbf{z}}$. The problem progresses until we arrive to $\mathbf{A} = A_z(\rho)\,\hat{\mathbf{z}}$, where is given by the case structure in the following equation in terms of an as yet to be determined constant $C$. \begin{align} A_z(\rho) = \begin{cases} -\frac{1}{4}\,\mu_0\,j\,\rho^2, &\rho\leq a;~\text{and} \\ -\frac{1}{4}\,\mu_0\,j\,a^2 +C \,\ln{\frac{\rho}{a}}, &\rho\geq a. \end{cases} \end{align} Zangwill states that, the matching condition $$\hat{ \mathbf{n}}_2 \times \left[ \mathbf{B}_1 - \mathbf{B}_2 \right] = \mu_0\,\mathbf{K}(\mathbf{r}_S) . \tag{10} $$ applied at $\rho = a$ fixes $C$. Since, in general, $\mathbf{B}(\mathbf{r}) = \boldsymbol{\nabla} \times \mathbf{A}(\mathbf{r}) $, therefore, in this case, $$ \mathbf{B}(\mathbf{r}) = \begin{cases} + \frac{\mu_0\,j}{2}\,\rho \,\hat{ \boldsymbol{\varphi}} , & \rho \leq a;~\text{and} \\ - \frac{ C }{ \rho} \,\hat{ \boldsymbol{\varphi} } , & \rho \geq a. \end{cases} $$ Then, since $\mathbf{K}(\mathbf{r}_S) = K(a,\varphi,z)\,\hat{\mathbf{z}} $, therefore Eq. 10 evaluated at $\rho = a$ results in $$ \hat{ \boldsymbol{\rho} } \times \left[ - \frac{ C }{ a} - \frac{\mu_0\,j\,a }{2 } \right]\hat{ \boldsymbol{\varphi} } = \frac{ - 2\,C - \mu_0\,j\,a^2 }{2\, a} \,\hat{ \mathbf{z} } = \mu_0\,K(a,\varphi,z)\,\hat{\mathbf{z}} \tag{20}. $$ From Eq. 20, we find that the matching condition gives the following equation for $C$: $$ C = - \frac{1}{2} \,\mu_0\,j\,a^2 - a \mu_0\,K(a,\varphi,z) . \tag{30} $$ Now, Zangwill claims that $$ C = - \frac{1}{2} \,\mu_0\,j\,a^2 . \tag{40} $$ This means that either the surface-current density is identically zero, or that I have made a mistake along the way.


(1) Smaller question: Have I properly applied the matching conditions so to obtain the correct formula for $C$ (e.g., perhaps I have a sign error).

(2) Larger question: How does one obtain the surface-current density, $\mathbf{K}$, from the current density, $\mathbf{j}$?

My attempts

Option 1 (the correct option): On the one hand, I would say that the existence of a current density in the bulk of a material does not necessarily imply the existence of surface-current density.

Option 2:

Since $I = \int d\mathbf{S}\cdot \mathbf{j}$ therefore $$I = \int_{0}^{2\pi}\int_0^\infty \delta(\rho-a)\,\rho \,d\rho\,d\varphi \,\hat{\mathbf{z}}\cdot j\,\hat{\mathbf{z}}= 2\,\pi\, a \, j. $$ This has the outstanding issue that the units are inconsistent on the left- and right-hand side of the equation. Bearing this in mind I come to option 3.

Option 3:

$$K = \int_{0}^{2\pi}\int_0^\infty \delta(\rho-a)\,\rho \,d\rho\,d\varphi \,\hat{\mathbf{z}}\cdot j\,\hat{\mathbf{z}}= 2\,\pi\, a \, j. $$ This has its own outstanding issue. Namely, upon substitution into Eq. 30, I find that $$ C = \mu_0\,j\,a^2\left[ - \frac{1}{2} - 2\,\pi\, \right], $$ which is a result that is inconsistent with Zangwill's answer (i.e., Eq. 40).

Option 4:

I adopt from joshphysics in [1], that
\begin{align} \mathbf j = \delta{\left(\rho -a\right)} \,\mathbf K . \end{align} However this leaves me scratching my head and asking for relief.


[1] Surface current and current density

[2] Zangwill, Modern Electrodynamics, pp. 31-2, 310, 323-4.


1 Answer 1


(1) You are correct.

(2) Not possible unless it has a Dirac delta part that essentially is just a specification of K. Typically the questions will give it to you, one way or another.

  • $\begingroup$ I understand from your answer that I did not make a sign error or other error. For emphasis, I repeat that the example statement does not---in any way---specify $\mathbf{K}$. Rather, the example statement only specifies $\mathbf{j}$. So do you think then that $\mathbf{K} = \mathbf{0}$? $\endgroup$ Apr 19 at 17:56
  • 1
    $\begingroup$ If you derived Equation (30), then by going to Equation (40) it is clear that Zangwill is telling you that K is zero. I was saying that either they give you K as part of j, a Dirac delta containing part, or they will give it to you as a boundary condition on a surface, and you have to manually deal with it. Or they give you A or B and you differentiate to get K. There is very little chance that you would have to determine it yourself. $\endgroup$ Apr 19 at 18:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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