Although waves appear in many forms in nature, such as water waves, acoustic waves, and seismic waves, I believe that nothing represents the wave nature of reality more clearly than electromagnetic waves. If someone asked, “What is light?”, our answer today would be very different from what scientists might have said more than 300 years ago, when Newton published one of the most influential works in the history of optics, Opticks. Much of this change is due to James Clerk Maxwell, who showed that light can be understood as a form of electromagnetic radiation. In this description, light is represented by two vector fields: the electric field \( \mathbf{E} \) and the magnetic field \( \mathbf{B} \). These fields depend on both space and time, especially when charges are in motion. We are usually introduced to Maxwell’s equations in their integral form. Maxwell’s equations in empty space with no charges nor currents are as follows,

Gauss Law:
\[ \oint_S \mathbf{E}\cdot d\mathbf{S}=0 \]
No Monopoles:
\[ \oint_S \mathbf{B}\cdot d\mathbf{S}=0 \]
Faraday’s Law:
\[ \oint_C \mathbf{E}\cdot d\boldsymbol{\ell} = -\frac{d}{dt} \oint_S \mathbf{B}\cdot d\mathbf{S} \]
Ampère-Maxwell:
\[ \oint_C \mathbf{B}\cdot d\boldsymbol{\ell} = \mu_0\varepsilon_0 \frac{d}{dt} \oint_S \mathbf{E}\cdot d\mathbf{S} \]

However, a more insightful way of describing these equations is by using the curl and divergence operators. With the help of the constitutive relations, we can write

\[ \mathbf{D}=\varepsilon\mathbf{E}, \qquad \mathbf{B}=\mu\mathbf{H} \]

and obtain Maxwell’s equations in differential form:

\[ \nabla\times\mathbf{E} = -\mu_0\frac{\partial\mathbf{H}}{\partial t} \]
\[ \nabla\times\mathbf{H} = \varepsilon_0\frac{\partial\mathbf{E}}{\partial t} \]
\[ \nabla\cdot\mathbf{E}=0 \]
\[ \nabla\cdot\mathbf{H}=0 \]

Using these equations, the development of the electromagnetic wave equation is quite straightforward. In empty space, Faraday’s law can be written in terms of \(\mathbf{H}\) as

\[ \nabla\times\mathbf{E} = -\mu_0\frac{\partial\mathbf{H}}{\partial t}. \]

Next, we apply the curl operator to both sides:

\[ \nabla\times(\nabla\times\mathbf{E}) = -\mu_0 \nabla\times \frac{\partial\mathbf{H}}{\partial t}. \]

Assuming that the magnetic field is sufficiently smooth, the spatial curl and the time derivative can be interchanged:

\[ \nabla\times \frac{\partial\mathbf{H}}{\partial t} = \frac{\partial}{\partial t} (\nabla\times\mathbf{H}). \]

Thus,

\[ \nabla\times(\nabla\times\mathbf{E}) = -\mu_0 \frac{\partial}{\partial t} (\nabla\times\mathbf{H}). \]

We now use the vector identity

\[ \nabla\times(\nabla\times\mathbf{u}) = \nabla(\nabla\cdot\mathbf{u}) – \nabla^2\mathbf{u}. \]

Applying this identity to the electric field gives

\[ \nabla(\nabla\cdot\mathbf{E}) – \nabla^2\mathbf{E} = -\mu_0 \frac{\partial}{\partial t} (\nabla\times\mathbf{H}). \]

Since there is no charge in empty space, Gauss’s law gives

\[ \nabla\cdot\mathbf{E}=0. \]

Therefore, the first term vanishes:

\[ \nabla^2\mathbf{E} = \mu_0 \frac{\partial}{\partial t} (\nabla\times\mathbf{H}). \]

Now we substitute the Ampère-Maxwell law. Since there is no conduction current in empty space, \(\mathbf{J}=0\), so

\[ \nabla\times\mathbf{H} = \varepsilon_0 \frac{\partial\mathbf{E}}{\partial t}. \]

Hence,

\[ \nabla^2\mathbf{E} = \mu_0 \frac{\partial}{\partial t} \left( \varepsilon_0 \frac{\partial\mathbf{E}}{\partial t} \right). \]

Since \(\mu_0\) and \(\varepsilon_0\) are constants, this becomes

\[ \nabla^2\mathbf{E} = \mu_0\varepsilon_0 \frac{\partial^2\mathbf{E}}{\partial t^2}. \]

Finally, using

\[ c^2=\frac{1}{\mu_0\varepsilon_0}, \]

we obtain the electromagnetic wave equation:

\[ \boxed{ \nabla^2\mathbf{E} = \frac{1}{c^2} \frac{\partial^2\mathbf{E}}{\partial t^2}. } \]

In this way, we arrive at a partial differential equation that depends on both space and time. This derivation is quite standard in physics. What is sometimes overlooked, however, is that by applying separation of variables, one can extract a time-independent equation that describes only the spatial structure of the wave.


For example, consider the scalar wave equation

\[ \left( \nabla^2 – \frac{1}{c^2} \frac{\partial^2}{\partial t^2} \right) \psi(x,y,z,t)=0. \]

We assume that the wave function can be written as the product of a spatial function and a temporal function,

\[ \psi(x,y,z,t) = \psi^{(s)}(x,y,z)\psi^{(t)}(t). \]

Substituting this form into the wave equation gives

\[ \psi^{(t)} \nabla^2\psi^{(s)} = \frac{\psi^{(s)}}{c^2} \frac{d^2\psi^{(t)}}{dt^2}. \]

Dividing both sides by \(\psi^{(s)}\psi^{(t)}\), we obtain

\[ \frac{1}{\psi^{(s)}} \nabla^2\psi^{(s)} = \frac{1}{c^2\psi^{(t)}} \frac{d^2\psi^{(t)}}{dt^2}. \]

The left-hand side depends only on the spatial variables, while the right-hand side depends only on time. Therefore, both sides must be equal to the same constant,

\[ \frac{1}{\psi^{(s)}} \nabla^2\psi^{(s)} = \frac{1}{c^2\psi^{(t)}} \frac{d^2\psi^{(t)}}{dt^2} = K. \]

Thus, the separated equations are

\[ \nabla^2\psi^{(s)} = K\psi^{(s)} \]

and

\[ \frac{d^2\psi^{(t)}}{dt^2} = c^2K\psi^{(t)}. \]

For oscillatory wave solutions, we choose

\[ K=-k^2. \]

The separation therefore gives two independent equations: Helmholtz’s equation for the spatial variable \(\mathbf{r}\), and a second-order ordinary differential equation in time.

\[ \boxed{ \nabla^2\psi^{(s)} + k^2\psi^{(s)} = 0 } \]
\[ \boxed{ \frac{d^2\psi^{(t)}}{dt^2} + c^2k^2\psi^{(t)} = 0 } \]

For the one-dimensional case, the spatial part of the Helmholtz equation becomes

\[ \frac{d^2 \psi^{(s)}(z)}{dz^2}+k^2\psi^{(s)}(z)=0. \]

This equation can be solved using the ansatz method. We assume a solution of the form

\[ \psi^{(s)}(z)=e^{rz}. \]

Substituting this expression into the differential equation gives

\[ r^2e^{rz}+k^2e^{rz}=0. \]

Since \(e^{rz}\neq 0\), we obtain the characteristic equation

\[ r^2+k^2=0. \]

Solving for \(r\), the roots are

\[ r=\pm jk. \]

Therefore, the two complex exponential solutions are

\[ e^{jkz}, \qquad e^{-jkz}. \]

By the principle of superposition, any linear combination of these two solutions is also a solution. Thus, the complex form of the one-dimensional Helmholtz solution is

\[ \psi^{(s)}(z)=C_1e^{jkz}+C_2e^{-jkz} \]

Equivalently, using Euler’s formula, the same solution can be written in terms of real sinusoidal functions:

\[ \psi^{(s)}(z)=A_0\cos(kz)+B_0\sin(kz) \]

A less obvious consequence of the Helmholtz equation appears when the wave is forced to keep a transverse phase variation that is too large for the second medium to support as a propagating wave.

Starting from the homogeneous Helmholtz equation,

\[ \nabla^2\psi+k^2\psi=0, \]

consider a two-dimensional field of the form

\[ \psi(x,z)=e^{i\beta x}e^{ik_z z}. \]

Here, \(\beta\) is the transverse wave number along the interface, and \(k_z\) is the wave number in the direction normal to the interface.

Substituting into Helmholtz gives

\[ \frac{\partial^2\psi}{\partial x^2} + \frac{\partial^2\psi}{\partial z^2} + k^2\psi = 0. \]

Since

\[ \frac{\partial^2\psi}{\partial x^2} = -\beta^2\psi, \qquad \frac{\partial^2\psi}{\partial z^2} = -k_z^2\psi, \]

we obtain

\[ -\beta^2\psi-k_z^2\psi+k^2\psi=0. \]

Therefore,

\[ \beta^2+k_z^2=k^2. \]

Equivalently,

\[ k_z^2=k^2-\beta^2. \]

If

\[ \beta < k \]

then \(k_z\) is real:

\[ k_z=\sqrt{k^2-\beta^2}. \]

In this case, the solution is propagating in the \(z\)-direction:

\[ \psi(x,z)=e^{i\beta x}e^{ik_z z}. \]

However, if the transverse wave number is larger than the wave number of the medium,

\[ \beta>k, \]

then the longitudinal component \(k_z\) cannot remain real. This situation appears naturally at an interface. The tangential component of the wave vector must be conserved, so we write

\[ \beta=k_1\sin\theta_i. \]

In the second medium, the total available wave number is

\[ k_2=n_2k_0. \]

If \(n_1>n_2\) and the angle of incidence is larger than the critical angle, then

\[ \beta=k_1\sin\theta_i>k_2. \]

Therefore, in the second medium,

\[ k_{z}^2=k_2^2-\beta^2<0. \]

Since \(k_{z}^2\) is negative, the normal component of the wave vector becomes imaginary. We write

\[ k_{z}=i\alpha, \qquad \alpha=\sqrt{\beta^2-k_2^2}. \]

The transmitted field then becomes

\[ \psi_t(x,z)=e^{i\beta x}e^{-\alpha z}. \]

This is no longer a propagating wave in the \(z\)-direction. It oscillates along \(x\), but decays exponentially along \(z\):

\[ \boxed{ \psi_t(x,z)=e^{i\beta x}e^{-\alpha z} } \]

This is an evanescent wave.


Bibliography

Born, M., & Wolf, E. (1999). Principles of Optics: Electromagnetic Theory of Propagation, Interference and Diffraction of Light (7th Edition). Cambridge University Press. ISBN: 0521642221.

Fowles, G.R. (1968). Introduction to Modern Optics (3rd Edition). Dover Publications Inc., New York, NY.

Hafner, J. (2024). Physics of Light and Materials [MOOC]. Coursera.

Posted in

コメントを残す