Solutions of Maxwell‘s Equation

Reference
Slides of EE4C05, TUD
Orfanidis S J. Electromagnetic waves and antennas[J]. 2002.
Jin J M. Theory and computation of electromagnetic fields[M]. John Wiley & Sons, 2011.

Fundamentals of EM Wave Radiation

Energy transfer by EM field

An intuitive impression: Electric field in a horn antenna

Inside antenna: (electrostatic field)

• Electric field is created by and linked to charges;
• At a large distance from a source the electric field decreases with distance as $1/R^2$;

Outside antenna: (electromagnetic wave/radiated field)

• Electric field lines are closed curves;

• Electric field is not linked to charges and exists independently from them;

• At a large distance from a source the electric field decreases with distance as $1/R$;

  Why $1/R$? [Electromagnetic wave/radiated field](# Electromagnetic wave/radiated field)

---

Electrostatic field

Maxwell equations know three sources: electric and magnetic currents and electric charges
$$\begin{array}{l}\nabla \times \vec{e}(\vec{r},t)=-\frac{\partial \vec{b}(\vec{r},t)}{\partial t}-\vec{m}(\vec{r},t)\\ \nabla \times \vec{h}(\vec{r},t)=\frac{\partial \vec{d}(\vec{r},t)}{\partial t}+\vec{j}(\vec{r},t)\\ \nabla \cdot \vec{d}(\vec{r},t)=q(\vec{r},t)\\ \nabla \cdot \vec{b}(\vec{r},t)=0\end{array}$$
However physically only one source exists. Let us look carefully at the field created by this source.

Electric field of electrical charges

For a point like charge $q$:
$$\vec{e}(\vec{r},t)=\frac{q{\hat{R}}_q}{4\pi {\epsilon }_0{R}_q^2}$$
The field strength decreases with distance as $1/R_q^2$

For distributed charge $Q$ with dipole moment $\vec{p}$ is described by a multiple series:
$$\vec{e}(\vec{r},t)=\frac{1}{4\pi {\epsilon }_0}\left[\frac{Q{\hat{R}}_q}{R_q^2}+\frac{3(\vec{p}\cdot {\hat{R}}_q){\hat{R}}_q-\vec{p}}{R_q^3}+O\left(\frac{1}{R_q^4}\right)\right]$$
The field strength decreases with distance as $1/R_q^2$ and higher orders

These fields are electrostatic fields:

• Decreases in amplitude with distance from the source as $1/R^2$;
• The electric field lines are connected to charges or currents.

---

Electromagnetic wave/radiated field

Assume that a localized source creates power $W$, which spreads outside the source and propagates away from it. Let us consider how this power flow crosses a few closed surfaces with spherical shape, where $r$ is sphere radius.

As power propagates in all directions simultaneously, and due to ENERGY CONSERVATION LAW (free space, no dissipated loss, use the real part of Poynting’s theorem for complex phasors) the total power remains the same at each spherical surface.

Note that the intensity at surface of sphere can also be expressed by Poynting vector $\vec{S}=\frac{1}{2}\vec{E}\times {\vec{H}}^{\ast }$, the electric and magnetic fields wh
ich transfer power AWAY from a source have specific amplitude dependence on the distance from the source
$$\vec{E}=\vec{A}\frac{e^{-jk{R}_q}}{R_q},\vec{H}=\vec{B}\frac{e^{-jk{R}_q}}{R_q}$$
This is an indication that this field is completely different from electrostatic field and is called electromagnetic wave or radiated field:

• Decreases in amplitude with distance from the source as $1/R$;
• The electric field lines are closed loops and are not connected to charges or currents.

How to create radiated field?

Basic radiation mechanisms

We have seen before that static charges do not produce radiate field, how about moving charge?

---

moving charge with constant velocity
$$\vec{e}(\vec{r},t)=\frac{q}{4\pi {\epsilon }_0}{\left[\frac{\left(1-v^2/c^2\right)\left({\hat{R}}_q-\overrightarrow{\mathbf{v}}/c\right)}{R_q^2{\left(1-{\hat{R}}_q\cdot \overrightarrow{\mathbf{v}}/c\right)}^3}\right]}_{t_r},t_r=t-R_q(t_r)/c\text{retarded time}$$

Electric field decreases in amplitude with distance from the source as $1/R^2$, and attached to the charge.

Therefore, charge moving with constant speed does not radiate.

---

accelerating electrical charge
$$\vec{e}(\vec{r},t)=\frac{q}{4\pi {\epsilon }_0}{\left[\frac{\left(1-v^2/c^2\right)\left({\hat{R}}_q-\overrightarrow{\mathbf{v}}/c\right)}{R_q^2{\left(1-{\hat{R}}_q\cdot \overrightarrow{\mathbf{v}}/c\right)}^3}\right]}_{t_r}+\frac{q}{4\pi {\epsilon }_0{c}^2}{\left[\frac{{\hat{R}}_q\times \left[\left({\hat{R}}_q-\overrightarrow{\mathbf{v}}/c\right)\times \overrightarrow{\mathbf{a}}\right]}{R_q{\left(1-{\hat{R}}_q\cdot \overrightarrow{\mathbf{v}}/c\right)}^3}\right]}_{t_r}$$
Having a term decreases in amplitude with distance from the source as $1/R$. Electric field leaves the charge and propagates in space as wave.

Therefore, charge moving with non-constant speed does radiate (creates an electromagnetic wave).

The acceleration $\vec{\mathbf{a}}$ can be colinear to $\vec{\mathbf{v}}$ or be orthogonal to $\vec{\mathbf{v}}$

---

Summary

To create EM wave there must be:

• Accelerated (or decelerated) charges;
• Time-varying imposed currents over a limited propagation path (terminations of the current will radiate);
• Charges moving with superluminal speed (Cherenkov radiation).

Hertzian dipole

Let us consider a very short current as a basic radiation element

A source of current $I_s(t)$ at the bottom of the element creates a traveling wave of current (a pulse) which leaves the source and propagates along the element at the speed of light, $c$ , until it reaches a perfect termination at the top of the element, where it is totally absorbed. Thus the distribution for the axial current along the element
$$\begin{gathered} I(z,t)=I_s(t-z/c)[U(z)-U(z-h)], \\ \text{where }U(z)-U(z-h)=\{\begin{array}{rl}& 1,0\le z\le h\\ & 0,\text{other}\end{array} \end{gathered}$$
The charge per unit length on the element is
$$Q(z,t)=Q_s(t-z/c)[U(z)-U(z-h)]+q_0(t)\delta (z)+q_h(t)\delta (z-h)$$

• $Q_s(t-z/c)[U(z)-U(z-h)]$: charges over the line
• $q_0(t)\delta (z)$: charge leaves the source
• $q_h(t)\delta (z-h)$: charge enters the termination

Created electric field equals:
$$\begin{gathered} \vec{e}(\vec{r},t)=\frac{1}{4\pi {\epsilon }_0}\left[\frac{q_0(t-r/c)}{r^2}\hat{r}+\frac{q_h\left(t-r_h/c\right)}{r_h^2}{\hat{r}}_h+\frac{\cot (\theta /2)I_s(t-r/c)}{cr}\hat{\theta } \\ -\frac{\cot \left({\theta }_h/2\right)I_s\left(t-h/c-r_h/c\right)}{c{r}_h}{\hat{\theta }}_h\right] \end{gathered}$$
The first two terms are Coulomb fields, they describe the field which stays around charges and does not leave them.
$${\vec{e}}_{Coulomb}(\vec{r},t)=\frac{1}{4\pi {\epsilon }_0}\left[\frac{q_0(t-r/c)}{r^2}\hat{r}+\frac{q_h\left(t-r_h/c\right)}{r_h^2}{\hat{r}}_h\right]$$
Two other terms describe the electric field which leaves charges and propagate away from them. This is an electromagnetic wave. After being created, this field becomes independent from these charges. The direction of ${\vec{e}}_{rad}$ is in the same plane with the current, and thus the direction of the magnetic field is perpendicular to the plane.
$${\vec{e}}_{rad}(\vec{r},t)=\frac{1}{4\pi {\epsilon }_0}\left[\frac{\cot (\theta /2)I_s(t-r/c)}{cr}\hat{\theta }-\frac{\cot \left({\theta }_h/2\right)I_s\left(t-h/c-r_h/c\right)}{c{r}_h}{\hat{\theta }}_h\right]$$
Created electric field far away from the element $(r/ct\to \infty )$ is
$$\vec{e}(\vec{r},t)=\frac{{\mu }_{0}c\sin \theta }{4\pi r(1-\cos \theta )}\left[I_s(t-r/c)-I_s\left(t-(h/c)(1-\cos \theta )-r_h/c\right)\right]\hat{\theta }$$

• A spherical wave front $W_1$ , centered at $z=0$, is produced when the pulse leaves the source, and a second spherical wave front $W_2$ , centered at $z=h$, is produced when the pulse is absorbed by the termination. These wave fronts travel outward from the ends of the element at the speed of light.

• The grey areas around the source and load are Coulomb fields.

• Wave takes energy away from the dipole, while static field does not.

---

Assume the Hertzian dipole has following properties:

• Monochromatic (single frequency excitation);
• Time dependence as $\exp (i\omega t)$;
• Constant current over the whole segment;
• Geometry as below

Then the electromagnetic field:
$$\begin{gathered} \vec{E}(\vec{r})=i\eta \frac{kh{I}_0\sin \theta }{4\pi r}\left[1+\frac{1}{ikr}-\frac{1}{(kr{)}^2}\right]\exp (-ikr)\hat{\theta }+\eta \frac{h{I}_0\cos \theta }{4\pi r^2}\left[1+\frac{1}{ikr}\right]\exp (-ikr)\hat{r}; \\ \vec{H}(\vec{r})=i\frac{kh{I}_0\sin \theta }{4\pi r}\left[1+\frac{1}{ikr}\right]\exp (-ikr)\hat{\varphi }; \\ \eta =\sqrt{{\mu }_0/{\epsilon }_0};k=\omega /c \end{gathered}$$
Far away from dipole, the total field has only wave component
$$\begin{gathered} \vec{E}(\vec{r})=i\eta \frac{kh{I}_0\sin \theta }{4\pi r}\exp (-ikr)\hat{\theta }; \\ \vec{H}(\vec{r})=i\frac{kh{I}_0\sin \theta }{4\pi r}\exp (-ikr)\hat{\varphi }; \\ \eta =\sqrt{{\mu }_0/{\epsilon }_0};k=\omega /c \end{gathered}$$

Example:

• Compute total power radiated by a Hertzian dipole with a length of $1\mathrm{c}\mathrm{m}$ with current of $1\mathrm{A}$ at the frequency $1\mathrm{G}\mathrm{H}\mathrm{z}$.

  The radiation field of Hertzian dipole can be expressed as
  $$\begin{array}{l}\vec{E}(\vec{r})=i\eta \frac{kh{I}_0\sin \theta }{4\pi r}\exp (-ikr)\hat{\theta }\\ \vec{H}(\vec{r})=i\frac{kh{I}_0\sin \theta }{4\pi r}\exp (-ikr)\hat{\varphi }\\ \eta =\sqrt{{\mu }_0/{\epsilon }_0}\end{array}$$
  and the total power can be calculated by
  $$\begin{array}{rl}P& =\frac{1}{2}\Re \left\{\iint \vec{E}(\vec{r})\times \vec{H}(\vec{r})d\vec{S}\right\}\\ & =\iint \frac{\eta k^2{h}^2{I}_0^2{\sin }^2\theta }{32{\pi }^2{r}^2}\hat{\mathrm{r}}d\vec{S}\\ & ={\int }_0^{\pi }{\int }_0^{2\pi }\frac{\eta k^2{h}^2{I}_0^2{\sin }^2\theta }{32{\pi }^2{r}^2}\hat{\mathrm{r}}\cdot \hat{\mathrm{r}}{r}^2\sin \theta d\theta d\varphi \\ & =\frac{\pi \eta I_0^2{h}^2}{4{\lambda }^2}{\int }_0^{\pi }{\sin }^3\theta d\theta \\ & =\frac{\pi \eta I_0^2{h}^2}{3{\lambda }^2}\end{array}$$
  Note that $\lambda =c/f=0.3\mathrm{m}$, $I_0=1\mathrm{A}$, $h=0.01\mathrm{m}$, ${\epsilon }_0=8.854\times 10^{-12}\mathrm{f}\mathrm{a}\mathrm{r}\mathrm{a}\mathrm{d}/\mathrm{m}$ and ${\mu }_0=4\pi \times 10^{-7}\mathrm{h}\mathrm{e}\mathrm{n}\mathrm{r}\mathrm{y}/\mathrm{m}$, plug them into the equation above, we obtain
  $$P=0.4384\mathrm{W}$$

Uniform Plane Waves

mainly from Orfanidis S J. Electromagnetic waves and antennas[J]. 2002. Chapter 2.1-2.5

Uniform plane waves in lossless media

The simplest electromagnetic waves are uniform plane waves propagating along some fixed direction, say the $z$-direction, in a lossless medium $ϵ,\mu$.

The assumption of uniformity means that the fields have no dependence on the transverse coordinates $x,y$ and are functions only of $z,t$. Thus, we look for solutions of Maxwell’s equations of the form:
$$\vec{e}(x,y,z,t)=\vec{e}(z,t)\text{and}\vec{h}(x,y,z,t)=\vec{h}(z,t).$$
Because there is no dependence on $x,y$, we set the partial derivatives ${\partial }_x=0$ and ${\partial }_y=0$. Then, the gradient, divergence, and curl operations take the simplified forms:
$$\nabla =\hat{z}\frac{\partial }{\partial z},\nabla \cdot \vec{e}=\frac{\partial e_z}{\partial z},\nabla \times \vec{e}=\hat{z}\times \frac{\partial \vec{e}}{\partial z}=-\hat{x}\frac{\partial e_y}{\partial z}+\hat{y}\frac{\partial e_x}{\partial z}$$
Then the source-free M.E. become:
$$\begin{array}{rlrl}& \nabla \times \vec{e}=-\mu \frac{\partial \vec{h}}{\partial t}& & \hat{z}\times \frac{\partial \vec{e}}{\partial z}=-\mu \frac{\partial \vec{h}}{\partial t}\\ & \nabla \times \vec{h}=ϵ\frac{\partial \vec{e}}{\partial t}& \Rightarrow & \hat{z}\times \frac{\partial \vec{h}}{\partial z}=ϵ\frac{\partial \vec{e}}{\partial t}\\ & \nabla \cdot \vec{e}=0& & \frac{\partial e_z}{\partial z}=0\\ & \nabla \cdot \vec{h}=0& & \frac{\partial h_z}{\partial z}=0\end{array}$$
Taking the dot-product of Ampere’s law with the unit vector $\hat{z}$, and using the identity $\hat{z}\cdot (\hat{z}\times A)=0$, we have
$$\hat{z}\cdot (\hat{z}\times \frac{\partial \vec{h}}{\partial z})=ϵ\hat{z}\cdot \frac{\partial \vec{e}}{\partial t}=0\Rightarrow \frac{\partial e_z}{\partial t}=0$$
Because also ${\partial }_z{e}_z=0$, it follows that $e_z$ must be a constant, independent of $z,t$. We may take this constant to be zero, i.e., $e_z=0$. Similarly, we have $h_z=0$. Thus, the fields have components only along $x,y$ directions:
$$\begin{array}{rl}& \vec{e}(z,t)=\hat{x}{e}_x(z,t)+\hat{y}{e}_y(z,t)\\ & \vec{h}(z,t)=\hat{x}{h}_x(z,t)+\hat{y}{h}_y(z,t)\end{array}\text{\hspace{1em}(transverse fields)}$$
These fields must satisfy Faraday’s and Ampere’s laws. We rewrite these equations in the speed of light and characteristic impedance $c,\eta$ by replacing $ϵ,\mu$ by
$$ϵ=\frac{1}{\eta c},\mu =\frac{\eta }{c},\text{ where }c=\frac{1}{\sqrt{\mu ϵ}},\eta =\sqrt{\frac{\mu }{ϵ}}$$
Then
$$\begin{array}{rl}& \hat{z}\times \frac{\partial \vec{e}}{\partial z}=-\frac{1}{c}\eta \frac{\partial \vec{h}}{\partial t}\\ & \eta \hat{z}\times \frac{\partial \vec{h}}{\partial z}=\frac{1}{c}\frac{\partial \vec{e}}{\partial t}\end{array}$$
Crossing the first equation with $\hat{z}$, we have
$$(\hat{z}\times \frac{\partial \vec{e}}{\partial z})\times \hat{z}=\frac{\partial \vec{e}}{\partial z}(\hat{z}\cdot \hat{z})-\hat{z}(\hat{z}\cdot \frac{\partial \vec{e}}{\partial z})=\frac{\partial \vec{e}}{\partial z}$$
where we used $\hat{z}\cdot {\partial }_z\vec{e}={\partial }_z{e}_z=0$ and $\hat{z}\cdot \hat{z}=1$. Then we obtain the equivalent system:
$$\begin{array}{rl}& \frac{\partial \vec{e}}{\partial z}=-\frac{1}{c}\frac{\partial }{\partial t}(\eta \vec{h}\times \hat{z})\\ & \frac{\partial }{\partial z}(\eta \vec{h}\times \hat{z})=-\frac{1}{c}\frac{\partial \vec{e}}{\partial t}\end{array}\text{\hspace{1em}(equivalent system)}$$
Differentiating the first equation with respect to $z$ and using the second, we obtain the one-dimensional wave equation:
$$\left(\frac{{\partial }^2}{\partial z^2}-\frac{1}{c^2}\frac{{\partial }^2}{\partial t^2}\right)\vec{e}(z,t)=0\text{\hspace{1em}(wave equation)}$$
and similarly for $\vec{h}$. The equivalent system can be decoupled by introducing the so-called forward and backward electric fields defined as the linear combinations:
$$\begin{array}{rl}& {\vec{e}}_{+}=\frac{1}{2}(\vec{e}+\eta \vec{h}\times \hat{z})\\ & {\vec{e}}_{-}=\frac{1}{2}(\vec{e}-\eta \vec{h}\times \hat{z})\end{array}\text{\hspace{1em}(forward and backward fields)}$$
Component-wise, these are:
$$e_{x\pm }=\frac{1}{2}(e_x\pm \eta h_y),e_{y\pm }=\frac{1}{2}(e_y\mp \eta h_x)$$
Plugging the forward and backward fields into the equivalent system, we verify
$$\frac{\partial }{\partial z}(\vec{e}\pm \eta \vec{h}\times \hat{z})=-\frac{1}{c}\frac{\partial }{\partial t}(\eta \vec{h}\times \hat{z})\mp \frac{1}{c}\frac{\partial \vec{e}}{\partial t}=\mp \frac{1}{c}\frac{\partial }{\partial t}(\vec{e}\pm \eta \vec{h}\times \hat{z})$$
i.e.,
$$\begin{array}{rl}& \frac{\partial {\vec{e}}_{+}}{\partial z}=-\frac{1}{c}\frac{\partial {\vec{e}}_{+}}{\partial t}\\ & \frac{\partial {\vec{e}}_{-}}{\partial z}=+\frac{1}{c}\frac{\partial {\vec{e}}_{-}}{\partial t}\end{array}\text{\hspace{1em}(decoupled system)}$$
The most general solutions for the forward and backward fields are:
$$\begin{array}{r}{\vec{e}}_{+}(z,t)=\vec{f}(z-ct)\\ {\vec{e}}_{-}(z,t)=\vec{g}(z+ct)\end{array}\text{\hspace{1em}(general solutions)}$$
with arbitrary functions $\vec{f}$ and $\vec{g}$, such that $\hat{z}\cdot \vec{f}=\hat{z}\cdot \vec{g}=0$.

Adding and subtracting forward and backward fields, and using the BAC-CAB rule and the orthogonality conditions ${\partial }_z{h}_z=0$, we obtain:
$$\begin{array}{rl}& \vec{e}(z,t)={\vec{e}}_{+}(z,t)+{\vec{e}}_{-}(z,t)=\vec{f}(z-ct)+\vec{g}(z+ct)\\ & \vec{h}(z,t)=\frac{1}{\eta }\hat{z}\times [{\vec{e}}_{+}(z,t)-{\vec{e}}_{-}(z,t)]=\frac{1}{\eta }\hat{z}\times [\vec{f}(z-ct)-\vec{g}(z+ct)]\end{array}$$
The term ${\vec{e}}_{+}(z,t)=\vec{f}(z-ct)$ represents a wave propagating with speed $c$ in the positive z-direction, while ${\vec{e}}_{-}(z,t)=\vec{g}(z-ct)$ represents a wave traveling in the negative z-direction.

To see this, consider the forward field at a later time $t+\Delta t$. During the time interval $\Delta t$, the wave moves in the positive z-direction by a distance $\Delta z=c\Delta t$. Indeed, we have
$$\vec{e}(z,t+\Delta t)=\vec{f}(z-c(t+\Delta t))=\vec{f}((z-\Delta z)-ct)=\vec{e}(z-\Delta z,t)$$
which means that the forward field at time $t+\Delta t$ is the same as the field at time $t$, but translated to the right along the z-axis by a distance $\Delta z=c\Delta t$.

The two special cases corresponding to forward waves only ($\vec{g}=0$), or to backward ones ($\vec{f}=0$), are of particular interest.

• For the forward case, we have:

$$\begin{array}{rl}& \vec{e}(z,t)=\vec{f}(z-ct)\\ & \vec{h}(z,t)=\frac{1}{\eta }\hat{z}\times \vec{f}(z-ct)=\frac{1}{\eta }\hat{z}\times \vec{e}(z,t)\end{array}$$

• For the backward case,
  $$\begin{array}{rl}& \vec{e}(z,t)=\vec{g}(z+ct)\\ & \vec{h}(z,t)=-\frac{1}{\eta }\hat{z}\times \vec{g}(z+ct)=-\frac{1}{\eta }\hat{z}\times \vec{e}(z,t)\end{array}$$

Note that in both cases the Poynting vector $\vec{s}=\vec{e}\times \vec{h}$ points in the propagation direction, i.e., the energy is being transported along the propagation direction. (more in [Energy density and flux](# Energy density and flux))

---

Monochromatic waves

Uniform, single-frequency, plane waves propagating in a lossless medium are obtained as a special case of the previous section by assuming the harmonic time-dependence:
$$\begin{gathered} \vec{e}(x,y,z,t)=\vec{e}(z)e^{j\omega t} \\ \vec{h}(x,y,z,t)=\vec{h}(z)e^{j\omega t} \end{gathered}$$
where $\vec{e}(z)$ and $\vec{h}(z)$ are transverse with respect to the $z$-direction.

Then the decoupled system may be solved easily by replacing time derivatives by ${\partial }_t\to j\omega$. Then
$$\frac{\partial {\vec{E}}_{\pm }(z)}{\partial z}=\mp jk{\vec{E}}_{\pm }(z),\text{where }k=\frac{\omega }{c}=\omega \sqrt{\mu ϵ}$$
with solutions:
$$\begin{array}{ll}{\vec{E}}_{+}(z)={\vec{E}}_{0+}{e}^{-jkz}& \text{ (forward) }\\ {\vec{E}}_{-}(z)={\vec{E}}_{0-}{e}^{jkz}& \text{ (backward) }\end{array}$$
where ${\vec{e}}_{0\pm }$ are arbitrary (complex-valued) constant vectors such that $\hat{z}\cdot {\vec{e}}_{0\pm }=0.$ The corresponding magnetic fields are:
$$\begin{array}{l}{\vec{H}}_{+}(z)=\frac{1}{\eta }\hat{z}\times {\vec{E}}_{+}(z)=\frac{1}{\eta }\left(\hat{z}\times {\vec{E}}_{0+}\right)e^{-jkz}={\vec{H}}_{0+}{e}^{-jkz}\\ {\vec{H}}_{-}(z)=-\frac{1}{\eta }\hat{z}\times {\vec{E}}_{-}(z)=-\frac{1}{\eta }\left(\hat{z}\times {\vec{E}}_{0-}\right)e^{jkz}={\vec{H}}_{0-}{e}^{jkz}\end{array}$$
where we defined the constant amplitudes of the magnetic fields:
$${\vec{H}}_{0\pm }=\pm \frac{1}{\eta }\hat{z}\times {\vec{E}}_{0\pm }$$
Therefore, we obtain the general solution for single-frequency waves
$$\begin{array}{rl}& \vec{E}(z)={\vec{E}}_{0+}{e}^{-jkz}+{\vec{E}}_{0-}{e}^{jkz}\\ & \vec{H}(z)=\frac{1}{\eta }\hat{z}\times [{\vec{E}}_{0+}{e}^{-jkz}-{\vec{E}}_{0-}{e}^{jkz}]\end{array}\text{\hspace{1em}(forward + backward waves)}$$
and
$$\begin{array}{rl}& \vec{e}(z,t)={\vec{E}}_{0+}{e}^{j(\omega t-kz)}+{\vec{e}}_{0-}{e}^{j(\omega t+kz)}\\ & \vec{h}(z,t)=\frac{1}{\eta }\hat{z}\times [{\vec{e}}_{0+}{e}^{j(\omega t-kz)}-{\vec{e}}_{0-}{e}^{j(\omega t+kz)}]\end{array}$$

• $\omega :$ number of oscillations per unit of time $\omega =2\pi /T$

• $k:$ number of oscillations per unit of space $k=2\pi /\lambda =\omega /c=2\pi f/c$ (free space), which is determined by the frequency (property of the source) and the velocity (properties of the environment).

  For a medium with complex-valued relative permittivity ${ϵ}_r$ and relative permeability ${\mu }_r$, $k$ can be a complex number $k=\beta -j\alpha$, where $\beta >0$ is the phase constant and $\alpha >0$ the attenuation constant.

• $v_p:$ the phase velocity determines how rapidly a point of constant phase moves, and is given by
  $$v_p={\frac{dx}{dt}|}_{\omega t-\beta x=const}=\frac{\omega }{\beta }$$
  In a dispersive medium $\beta =\beta (\omega )$, different signal frequencies will travel with different velocities, so how can we define an “effective signal velocity”?

• $v_g:$ the propagation of information of energy.
  $$v_g=\frac{d\omega }{dk}(=\frac{1}{dk/d\omega })$$
  The group velocity does not depend on frequency.

  Derivation: Consider an amplitude modulated wave
  E y ( 0 , t ) = E 0 { 1 + m cos ⁡ ( ω m t ) } cos ⁡ ( ω c t ) = E 0 { cos ⁡ ( ω c t ) + m 2 [ cos ⁡ ( [ ω c + ω m ] t ) + cos ⁡ ( [ ω c − ω m ] t ) ] } \begin{aligned} E_y(0,t)&=E_0\left\{1+m\cos (\omega_m t)\right\}\cos(\omega_c t)\\ &=E_0\left\{\cos(\omega_c t)+\frac{m}{2}\big[\cos ([\omega_c+\omega_m] t)+\cos ([\omega_c-\omega_m] t)\big]\right\} \end{aligned} Ey​(0,t)​=E0​{1+mcos(ωm​t)}cos(ωc​t)=E0​{cos(ωc​t)+2m​[cos([ωc​+ωm​]t)+cos([ωc​−ωm​]t)]}​
  The wave in space away from the $x$ plane
  $$\begin{array}{rl}& E_y(x,t)=E_0\left\{\cos \left({\omega }_{c}t-k_R\left({\omega }_c\right)x\right)+\frac{m}{2}\left[\begin{array}{c}\cos \left(\left[{\omega }_c+{\omega }_m\right]t-k_R\left({\omega }_c+{\omega }_m\right)x\right)\\ +\cos \left(\left[{\omega }_c-{\omega }_m\right]t-k_R\left({\omega }_c-{\omega }_m\right)x\right)\end{array}\right]\right\}\\ & \begin{array}{r}{E}_y(x,t)=E_0\left\{\begin{array}{r}\cos \left({\omega }_{c}t-k_R\left({\omega }_c\right)x\right)+m\cos ({\omega }_{m}t-\frac{1}{2}\underset{\text{Effective wave number for modulation }}{\underbrace{\left[k_R\left({\omega }_c+{\omega }_m\right)-k_R\left({\omega }_c-{\omega }_m\right)\right]x}})\\ \times \cos ({\omega }_{c}t-\frac{1}{2}\underset{\text{Effective wave number for carrier }}{\underbrace{\left[k_R\left({\omega }_c+{\omega }_m\right)+k_R\left({\omega }_c-{\omega }_m\right)\right]x}})\end{array}\right\}\end{array}\end{array}$$
  Using Taylor expansion
  $$\begin{array}{l}{k}_R\left({\omega }_c+{\omega }_m\right)-k_R\left({\omega }_c-{\omega }_m\right)\approx 2{\omega }_m\frac{d{k}_R\left({\omega }_c\right)}{d\omega }\\ k_R\left({\omega }_c+{\omega }_m\right)+k_R\left({\omega }_c-{\omega }_m\right)\approx 2k_R\left({\omega }_c\right)\end{array}$$
  and finally
  $$E_y(x,t)=E_0\left\{1+m\cos \left({\omega }_m\left[t-\frac{d{k}_R\left({\omega }_c\right)}{d\omega }x\right]\right)\right\}\cos \left({\omega }_c\left[t-\frac{k_R\left({\omega }_c\right)}{{\omega }_c}x\right]\right)$$
  The group velocity results from the time shift of the envelope:
  $$v_g={\left(\frac{d{k}_R\left({\omega }_c\right)}{d\omega }\right)}^{-1}$$

---

Polarization

Consider a forward-moving wave and let ${\vec{E}}_0=\hat{x}{A}_{+}+\hat{y}{B}_{+}$ by its complex-valued phasor amplitude, so that $\vec{E}(z)={\vec{E}}_0{e}^{-jkz}$. The time-varying field is obtained by restoring the factor $e^{j\omega t}$:
$$\vec{e}(z,t)=(\hat{x}{A}_{+}+\hat{y}{B}_{+})e^{j\omega t-jkz}$$
Polarization is the direction of the time-varying real-valued field $\vec{\mathcal{E}}(z,t)=\Re [\vec{e}(z,t)]$.

The polarization properties of the plane wave are determined by the relative magnitudes and phases of the complex-valued constants $A_{+},B_{+}.$ Writing them in their polar forms $A_{+}=A{e}^{j{\varphi }_a}$ and $B_{+}=B{e}^{j{\varphi }_b},$ we obtain:
$$\vec{e}(z,t)=\left(\hat{x}A{e}^{j{\varphi }_a}+\hat{y}B{e}^{j{\varphi }_b}\right)e^{j\omega t-jkz}=\hat{x}A{e}^{j\left(\omega t-kz+{\varphi }_a\right)}+\hat{y}B{e}^{j\left(\omega t-kz+{\varphi }_b\right)}$$
Extracting real parts and setting $\vec{\mathcal{E}}(z,t)=\mathrm{Re}[E(z,t)]=\hat{x}{\mathcal{E}}_x(z,t)+\hat{y}{\mathcal{E}}_y(z,t),$ we
find the corresponding real-valued $x,y$ components:
$$\begin{array}{l}{\mathcal{E}}_x(z,t)=A\cos \left(\omega t-kz+{\varphi }_a\right)\\ {\mathcal{E}}_y(z,t)=B\cos \left(\omega t-kz+{\varphi }_b\right)\end{array}$$
For a backward moving field, we replace $k$ by $-k$ in the same expression. To determine the polarization of the wave, we consider the time-dependence of these fields at some fixed point along the $z$ -axis, say at $z=0$ :
$$\begin{array}{l}{\mathcal{E}}_x(t)=A\cos \left(\omega t+{\varphi }_a\right)=A[\cos \omega t\cos {\varphi }_a-\sin \omega t\sin {\varphi }_a]\\ {\mathcal{E}}_y(t)=B\cos \left(\omega t+{\varphi }_b\right)=B[\cos \omega t\cos {\varphi }_b-\sin \omega t\sin {\varphi }_b]\end{array}$$
By defining the relative phase angle $\varphi ={\varphi }_a-{\varphi }_b$, we find
$$\begin{array}{rl}\cos \omega t\sin \varphi & =\frac{{\mathcal{E}}_y(t)}{B}\sin {\varphi }_a-\frac{{\mathcal{E}}_x(t)}{A}\sin {\varphi }_b\\ \sin \omega t\sin \varphi & =\frac{{\mathcal{E}}_y(t)}{B}\cos {\varphi }_a-\frac{{\mathcal{E}}_x(t)}{A}\cos {\varphi }_b\end{array}$$
Thus
$${\left(\frac{{\mathcal{E}}_y(t)}{B}\sin {\varphi }_a-\frac{{\mathcal{E}}_x(t)}{A}\sin {\varphi }_b\right)}^2+{\left(\frac{{\mathcal{E}}_y(t)}{B}\cos {\varphi }_a-\frac{{\mathcal{E}}_x(t)}{A}\cos {\varphi }_b\right)}^2={\sin }^2\varphi$$
This simplifies into:
$$\frac{{\mathcal{E}}_x^2}{A^2}+\frac{{\mathcal{E}}_y^2}{B^2}-2\cos \varphi \frac{{\mathcal{E}}_x{\mathcal{E}}_y}{AB}={\sin }^2\varphi \text{\hspace{1em}(polarization ellipse)}$$

• Linear polarization: set $\varphi =0$ or $\varphi =\pi$, corresponding to ${\varphi }_a={\varphi }_b=0$, or ${\varphi }_a=0,{\varphi }_b=-\pi$
  $$\frac{{\mathcal{E}}_x^2}{A^2}+\frac{{\mathcal{E}}_y^2}{B^2}\mp 2\frac{{\mathcal{E}}_x{\mathcal{E}}_y}{AB}=0\Rightarrow {\left(\frac{{\mathcal{E}}_x}{A}\mp \frac{{\mathcal{E}}_y}{B}\right)}^2=0$$
  representing the straight lines:

• Circular polarization: set $A=B$ and $\varphi =\pm \pi /2$. In this case, the polarization ellipse becomes the equation of a circle:
  $$\frac{{\mathcal{E}}_x^2}{A^2}+\frac{{\mathcal{E}}_y^2}{B^2}=1$$
  If $\varphi =\pi /2$:

if $\varphi =-\pi /2$:

Left circular or right circular: IEEE convention

Curl the fingers of your left and right hands into a fist and point both thumbs towards the direction of propagation. If the fingers of your right (left) hand are curling in the direction of rotation of the electric field, then the polarization is right (left) polarized.

• Elliptical polarization: If $A\ne B$ and $\varphi$ is arbitrary

$$\frac{{\mathcal{E}}_x^2}{A^2}+\frac{{\mathcal{E}}_y^2}{B^2}-2\cos \varphi \frac{{\mathcal{E}}_x{\mathcal{E}}_y}{AB}={\sin }^2\varphi \text{\hspace{1em}(*)}$$
It can be shown that the tilt angle $\theta$ is given by:
$$\tan 2\theta =\frac{2AB}{A^2-B^2}\cos \varphi$$
The ellipse semi-axes $A^{\prime },B^{\prime },$ that is, the lengths $OC$ and $OD,$ are given by:
$$\begin{array}{l}{A}^{\prime }=\sqrt{\frac{1}{2}\left(A^2+B^2\right)+\frac{s}{2}\sqrt{{\left(A^2-B^2\right)}^2+4A^2{B}^2{\cos }^2\varphi }}\\ B^{\prime }=\sqrt{\frac{1}{2}\left(A^2+B^2\right)-\frac{s}{2}\sqrt{{\left(A^2-B^2\right)}^2+4A^2{B}^2{\cos }^2\varphi }}\end{array}$$
where $s=\mathrm{sign}(A-B).$ These results are obtained by defining the rotated coordinate system of the ellipse axes:
$$\begin{array}{l}{\mathcal{E}}_x^{\prime }={\mathcal{E}}_x\cos \theta +{\mathcal{E}}_y\sin \theta \\ {\mathcal{E}}_y^{\prime }={\mathcal{E}}_y\cos \theta -{\mathcal{E}}_x\sin \theta \end{array}$$
and showing that $(\ast )$ transforms into the standardized form:
$$\frac{{\mathcal{E}}_x^{\prime 2}}{A^{\prime 2}}+\frac{{\mathcal{E}}_y^{\prime 2}}{B^{\prime 2}}=1$$
The polarization ellipse is bounded by the rectangle with sides at the end-points $\pm A,\pm B,$ as shown in the figure.

The angle $x$ subtended by the major to minor ellipse axes is given as follows:
$$\sin 2\chi =\frac{2AB}{A^2+B^2}|\sin \varphi |,-\frac{\pi }{4}\le \chi \le \frac{\pi }{4}$$
that is, it can be shown that $\tan \chi =B^{\prime }/A^{\prime }$ or $A^{\prime }/B^{\prime },$ whichever is less than one.

---

• The MATLAB function ellipse calculates the ellipse semi-axes and tilt angle, $A^{\prime },B^{\prime },\theta$ given the parameters $A,B,\varphi$. It has usage:

  [a,b,th]=ellipse(A,B,phi) % polarization ellipse parameters
  

• Finding the sense of polarization – a short-cut method

  1. Find out the (unit) vector in the direction of propagation, $\hat{n}$
  2. Write down the time domain expression $\vec{e}(\vec{r},t)=\Re [\vec{E}(\vec{r})e^{j\omega t}]$ at a fixed point in space. You can take $r=[xyz]=[000],$ which gives $\vec{e}(0,t)$.
  3. Compute $\vec{e}(0,t)$ at two different time instants $t_1$ and $t_2.$ For simplicity, you can assume $\omega t_1=0$ and $\omega t_2=\pi /2.$ Find the fields $\vec{e}\left(0,t_1\right)$ and $\vec{e}\left(0,t_2\right).$
  4. If $\vec{e}\left(0,t_1\right)\times \vec{e}\left(0,t_2\right)$ is in the direction of $\hat{n},$ the sense is right-handed $(\mathrm{R}\mathrm{H}).$ If it is opposite to $\hat{n},$ the sense is left-handed (LH).

EM wave scattering from an interface

Reflection from a PEC (metal) surface

For PEC surface, ${\vec{E}}_1={\vec{H}}_1=0$, we have boundary condition:
$$\begin{array}{rl}& \hat{n}\cdot {\vec{D}}_2={\varrho }_{e,s}\\ & \hat{n}\cdot {\vec{B}}_2=0\\ & \hat{n}\times {\vec{E}}_2=0\\ & \hat{n}\times {\vec{H}}_2={\vec{J}}_s\end{array}$$
where $\hat{n}$ points from medium 1 (PEC) to medium 2.

The incident EM wave creates electric currents on the PEC surface and these currents re-radiate an EM wave.

• Microwave does not penetrate into perfectly conducting material (metal), since ${\vec{E}}_1={\vec{H}}_1=0$.
• The re-radiated (scattered) wave has the same frequency as the incident wave.
• According to the boundary condition, the tangential component of the total electric field at the surface will be equal to zero. Therefore, the re-radiated (scattered) wave has the same magnitude and opposite phase as the incident wave, and ${\theta }_i={\theta }_r$.

---

Polarizations

Vertical (Parallel, P, TM) polarization – electric field component parallel to the plane of incidence

Horizontal (Perpendicular, S, TE) polarization – electric field component perpendicular to the plane of incidence

• The linear polarization of a plane microwave by reflection from a plane metal surface remains unchanged: TM remains TM and TE remains TE. For instance, for TM waves, $\vec{H}$ is perpendicular to the plane of incidence. According to the boundary condition $\hat{n}\times \vec{H}=\vec{J}$, we have $\vec{J}$ is parallel to the plane of incidence. $\vec{J}$ then re-radiated an EM wave, and the electric field ${\vec{E}}_r$ is in the same plane with the current. Therefore, TM remains TM.
• A circularly polarized plane microwave by reflection from a flat metal surface changes direction of rotation: left-polarized becomes right-polarized and vice versa. This is because $\vec{E}$ has a $180^{\circ }$ phase shift and $\vec{H}$ does not have phase shift.

---

For example, consider TE wave and normal incidence:
$$\begin{array}{rl}\text{ Incident wave }& {\vec{E}}_i=\hat{y}{E}_{yi}{e}^{-j\beta z}\\ \text{ Reflected wave }& {\vec{E}}_r=\hat{y}{E}_{yr}{e}^{j\beta z}\end{array}$$
According to the boundary condition,
$$-\hat{z}(\hat{y}{E}_{yi}{e}^{-j\beta z}-\hat{y}{E}_{yr}{e}^{j\beta z}){|}_{z=0}=0$$
Then $E_{yi}=-E_{yr}$, i.e., amplitude of reflected wave is equal to amplitude of incident wave, but reversed in phase.

If $E_{yi}$ is chosen to be real,
$$\begin{array}{rl}{\vec{e}}_T(z,t)& =\text{ incident wave }+\text{ reflected wave }\\ & =\hat{y}\Re (E_{yi}{e}^{j(\omega t-\beta z)}+\hat{y}{E}_{yr}{e}^{j(\omega t+\beta z)})\\ & =\hat{y}\Re (E_{yi}\left(e^{-j\beta z}-e^{j\beta z}\right)e^{j\omega t})\\ & =-\hat{y}2j{E}_{yi}\sin \beta z\sin \omega t\end{array}$$
Then incident and reflected wave combine to produce a standing wave.

Now we prove that $\vec{H}$ does not have phase shift. Since incident wave ${\vec{E}}_i=\hat{y}{E}_{yi}{e}^{-j\beta z}$, we have ${\vec{H}}_i=-\hat{x}{H}_{xi}{e}^{-j\beta z}$. Then
$${\vec{E}}_i\times {\vec{H}}_i^{\ast }=\hat{z}{E}_{yi}{H}_{xi}$$
For the reflected wave,
$${\vec{E}}_r\times {\vec{H}}_r^{\ast }=-\hat{z}{E}_{yi}{H}_{xi}$$
Since ${\vec{E}}_r=\hat{y}{E}_{yr}{e}^{j\beta z}$, ${\vec{H}}_r=-\hat{x}{H}_{xi}{e}^{-j\beta z}={\vec{H}}_i$ and the magnetic field is reflected without change in phase.

A more general proof can be obtained using the conclusion in Section [Reflection from a dielectric interface](# Reflection from a dielectric interface) by setting ${\eta }_2=0$ and obtaining reflection and transmission coefficients, then plugging them back to the expression for plane wave of the incident, reflected, and transmitted electric fields.

---

Reflection from a dielectric interface

mainly from Jin J M. Theory and computation of electromagnetic fields[M]. John Wiley & Sons, 2011. Chapter 4.4

Since any uniform plane wave can be decomposed into two orthogonal linearly polarized waves, we need to consider two linearly polarized incident waves so that their combination can yield the solution to any kind of incident plane wave.

Perpendicular Polarization (TE)

(Note: The directions are undetermined. They are specified in the figure for the convenience of derivation and may be changed if relevant coefficients are negative.)

Based on the expression for a plane wave, we can write down the expressions of the incident, reflected, and transmitted electric fields as
$$\begin{array}{rl}& {\vec{E}}^i=\hat{y}{E}_0{e}^{-j{\beta }^i\cdot \mathbf{r}}=\hat{y}{E}_0{e}^{-j{\beta }_1\left(x\sin {\theta }_i+z\cos {\theta }_i\right)}\\ & {\vec{E}}^r=\hat{y}{R}_{\perp }{E}_0{e}^{-j{\beta }^r\cdot \mathbf{r}}=\hat{y}{R}_{\perp }{E}_0{e}^{-j{\beta }_1\left(x\sin {\theta }_r-z\cos {\theta }_r\right)}\\ & {\vec{E}}^t=\hat{y}{T}_{\perp }{E}_0{e}^{-j{\beta }^{\prime }\cdot \mathbf{r}}=\hat{y}{T}_{\perp }{E}_0{e}^{-j{\beta }_2\left(x\sin {\theta }_t+z\cos {\theta }_t\right)}\end{array}$$
where $E_0$ and ${\theta }_i$ denote the amplitude and the angle of the incident field, which are known; $R_{\perp }$ and $T_{\perp }$ denote the unknown reflection and transmission coefficients; and ${\theta }_r$ and ${\theta }_t$ denote the unknown angles of the reflected and transmitted waves measured from the normal of the interface. The goal of analysis is to determine $R_{\perp },T_{\perp },{\theta }_c$, and ${\theta }_t$. For this, we first use $\nabla \times \vec{E}=-j\omega \mu \vec{H}$ to find the corresponding magnetic fields as
$$\begin{array}{l}{\vec{H}}^i=\left(-\hat{x}\cos {\theta }_i+\hat{z}\sin {\theta }_i\right)\frac{E_0}{{\eta }_1}{e}^{-j{\beta }_1\left(x\sin {\theta }_i+z\cos {\theta }_i\right)}\\ {\vec{H}}^r=\left(\hat{x}\cos {\theta }_r+\hat{z}\sin {\theta }_r\right)\frac{R_{\perp }{E}_0}{{\eta }_1}{e}^{-j{\beta }_1\left(x\sin {\theta }_r-z\cos {\theta }_r\right)}\\ {\vec{H}}^t=\left(-\hat{x}\cos {\theta }_t+\hat{z}\sin {\theta }_t\right)\frac{T_{\perp }{E}_0}{{\eta }_2}{e}^{-j{\beta }_2\left(x\sin {\theta }_t+z\cos {\theta }_t\right)}\end{array}$$
Next, we apply the boundary conditions that require the continuity of the tangential field components, $\hat{z}\times \vec{E}(z=0_{+})=\hat{z}\times \vec{E}(z=0_{-})$ and $\hat{z}\times \vec{H}(z=0_{+})=\hat{z}\times \vec{H}(z=0_{-}),$ which yield
$$\begin{array}{rl}{e}^{-j{\beta }_1\sin {\theta }_{i}x}+R_{\perp }{e}^{-j{\beta }_1\sin {\theta }_{r}x}& =T_{\perp }{e}^{-j{\beta }_2\sin {\theta }_{t}x}\\ \cos {\theta }_i\frac{1}{{\eta }_1}{e}^{-j{\beta }_1\sin {\theta }_{i}x}-\cos {\theta }_r\frac{R_{\perp }}{{\eta }_1}{e}^{-j{\beta }_1\sin {\theta }_{r}x}& =\cos {\theta }_t\frac{T_{\perp }}{{\eta }_2}{e}^{-j{\beta }_2\sin {\theta }_{t}x}\end{array}\text{\hspace{1em}(RT.1)}$$
These two equations have to be satisfied for any values of $x(-\infty <x<\infty )$. It can be shown mathematically that this is only possible when the associated phases match in each term. Consequently,
$${\beta }_1\sin {\theta }_i={\beta }_1\sin {\theta }_r={\beta }_2\sin {\theta }_t\text{\hspace{1em}(RT.2)}$$
which is known as phase matching. Since ${\beta }_1=\omega \sqrt{{\mu }_1{ϵ}_1}$ and ${\beta }_2=\omega \sqrt{{\mu }_2{ϵ}_2}$, we obtain
$${\theta }_r={\theta }_i,\frac{\sin {\theta }_t}{\sin {\theta }_i}=\sqrt{\frac{{\mu }_1{ϵ}_1}{{\mu }_2{ϵ}_2}}\triangleq \frac{n_1}{n_2}\text{\hspace{1em}(RT.3)}$$
which are called Snell’ s laws of reflection and refraction.

Because of $(RT.2)$, $(RT.1)$ can be reduced to
$$\begin{gathered} 1+R_{\perp }=T_{\perp } \\ \cos {\theta }_i\frac{1}{{\eta }_1}-\cos {\theta }_r\frac{R_{\perp }}{{\eta }_1}=\cos {\theta }_t\frac{T_{\perp }}{{\eta }_2} \end{gathered}$$
whose solution yields the reflection and transmission coefficients
$$R_{\perp }=\frac{{\eta }_2\cos {\theta }_i-{\eta }_1\cos {\theta }_t}{{\eta }_2\cos {\theta }_i+{\eta }_1\cos {\theta }_t},T_{\perp }=\frac{2{\eta }_2\cos {\theta }_i}{{\eta }_2\cos {\theta }_i+{\eta }_1\cos {\theta }_t}\text{\hspace{1em}(RT.4)}$$
If we define the wave impedance looking into the $z$ -direction as $Z_z=-E_y/H_x,$ then the wave impedances for perpendicular polarized waves in the two regions are
$$Z_{z1}=\frac{{\eta }_1}{\cos {\theta }_i},Z_{z2}=\frac{{\eta }_2}{\cos {\theta }_t}\text{\hspace{1em}(RT.5)}$$
As a result, $(RT.4)$ can be written as
$$R_{\perp }=\frac{Z_{z2}-Z_{z1}}{Z_{z2}+Z_{z1}},T_{\perp }=\frac{2Z_{z2}}{Z_{z2}+Z_{z1}}\text{\hspace{1em}(RT.6)}$$
which has the same form as the solution in a transmission line.

Parallel Polarization ™

The incident, reflected, and transmitted electric fields can be written as (To derive the reflection and transmission coefficients, we can also write the incident, reflected, and transmitted magnetic fields, which have the same form as the electric fields in TE)
$$\begin{array}{rl}& {\mathbf{E}}^i=\left(\hat{x}\cos {\theta }_i-\hat{z}\sin {\theta }_i\right)E_0{e}^{-j{\beta }_1\left(x\sin {\theta }_i+z\cos {\theta }_i\right)}\\ & {\mathbf{E}}^r=\left(\hat{x}\cos {\theta }_r+\hat{z}\sin {\theta }_r\right)R_{\parallel }{E}_0{e}^{-j{\beta }_1\left(x\sin {\theta }_r-z\cos {\theta }_r\right)}\\ & {\mathbf{E}}^t=\left(\hat{x}\cos {\theta }_t-\hat{z}\sin {\theta }_t\right)T_{\parallel }{E}_0{e}^{-j{\beta }_2\left(x\sin {\theta }_t+z\cos {\theta }_t\right)}\end{array}$$
By following the same procedure as for the perpendicular polarization case, we obtain the same Snell’s laws of reflection and reflection as expressed in $(RT.2)$ because of phase matching, and the reflection and transmission coefficients as
$$R_{\Vert }=\frac{{\eta }_2\cos {\theta }_t-{\eta }_1\cos {\theta }_i}{{\eta }_2\cos {\theta }_t+{\eta }_1\cos {\theta }_i},T_{\Vert }=\frac{2{\eta }_2\cos {\theta }_i}{{\eta }_2\cos {\theta }_t+{\eta }_1\cos {\theta }_i}\text{\hspace{1em}(RT.7)}$$
Again, if we define the wave impedance looking into the $z$ -direction as $Z_z=E_x/H_y,$ then the wave impedances for parallel polarized waves in the two regions are given by
$$Z_{z1}={\eta }_1\cos {\theta }_i,Z_{z2}={\eta }_2\cos {\theta }_t\text{\hspace{1em}(RT.8)}$$
and $(RT.7)$ can be written in the forms similar to $(RT.6)$. Therefore, the results in transmission line theory can be used to obtain solutions for plane wave problems as long as we define wave impedances correctly.

---

Before we discuss the results derived in this section further, we note that when the right half-space is occupied by a perfect electric conductor such that ${\eta }_2=0,$ we find that for both polarizations, the total tangential component of the magnetic field on the surface of the conductor (the $xy$ -plane) is
$$\hat{n}\times \mathbf{H}=\hat{n}\times \left({\mathbf{H}}^i+{\mathbf{H}}^r\right)=2\hat{n}\times {\mathbf{H}}^i$$
This result is valid for any plane wave since any plane wave can be decomposed into a perpendicular and a parallel polarized wave. Therefore, for a large and smooth conducting surface, the induced surface current for any incident wave is ${\vec{J}}_s\approx 2\hat{n}\times {\vec{H}}^i$, which is called physical optics approximation.

---

Matrix which relates vector of the scattered and incident EM waves is called the scattering matrix (h->TE, v->TM)

---

Total Transmission and Total Reflection

• Total transmission - Brewster angle

  For perpendicular polarization (TE),
  $$R_{\perp }=\frac{{\eta }_2\cos {\theta }_B-{\eta }_1\cos {\theta }_t}{{\eta }_2\cos {\theta }_B+{\eta }_1\cos {\theta }_t}=0⟹{\eta }_2\cos {\theta }_B={\eta }_1\cos {\theta }_t$$
  This equation can be solved together with Snell’s law of refraction to yield
  $$\sin {\theta }_B=\sqrt{\frac{{ϵ}_2/{ϵ}_1-{\mu }_2/{\mu }_1}{{\mu }_1/{\mu }_2-{\mu }_2/{\mu }_1}}$$
  Obviously, this phenomenon can happen only when ${\mu }_1\ne {\mu }_2$ and cannot occur at an interface between two different non-magnetic media.

  For parallel polarization ™,
  $$R_{\Vert }=\frac{{\eta }_2\cos {\theta }_t-{\eta }_1\cos {\theta }_B}{{\eta }_2\cos {\theta }_t+{\eta }_1\cos {\theta }_B}=0⟹{\eta }_2\cos {\theta }_t={\eta }_1\cos {\theta }_B$$
  which can be solved together with Snell’s law of refraction to yield
  $$\sin {\theta }_B=\sqrt{\frac{{ϵ}_2/{ϵ}_1-{\mu }_2/{\mu }_1}{{ϵ}_2/{ϵ}_1-{ϵ}_1/{ϵ}_2}}$$

• Total reflection - critical angle

  Rewrite Snell’s law of refraction as
  $$\sin {\theta }_t=\sqrt{\frac{{\mu }_1{ϵ}_1}{{\mu }_2{ϵ}_2}}\sin {\theta }_i$$
  Clearly, when $\sin {\theta }_i=\sqrt{{\mu }_2{\epsilon }_2/{\mu }_1{\epsilon }_1},\sin {\theta }_t=1$ or ${\theta }_t=\pi /2.$ Consequently, $R_{\perp }=1$ and $R_{\Vert }=-1,$ which indicate that the entire incident wave is reflected. This phenomenon is called total reflection. The corresponding incident angle is called the critical angle, which is given by
  $${\theta }_c={\sin }^{-1}\sqrt{\frac{{\mu }_2{ϵ}_2}{{\mu }_1{ϵ}_1}}={\sin }^{-1}\frac{n_2}{n_1}$$

Wave Propagation in Guiding Structures

Guided propagation

We are looking for a solution of Maxwell equation in a source free area for a wave propagating along $z$ direction along infinitely long structure.

We assume that this is a plane wave (with flat phase front orthogonal to $z$ direction), lossless propagation with propagation wavenumber $\beta$ and with electric and magnetic fields in a form of progressive wave
$$\begin{array}{r}\vec{E}(x,y,z)=\vec{E}(x,y)e^{-j\beta z}\\ \vec{H}(x,y,z)=\vec{H}(x,y)e^{-j\beta z}\end{array}\text{\hspace{1em}(GP.1)}$$
The corresponding wavelength, called the guide wavelength, is denoted by
$${\lambda }_g=\frac{2\pi }{\beta }\text{\hspace{1em}(GP.2)}$$
We know that EM field in a source free area in a free (homogeneous) space satisfies Maxwell’s equations
$$\begin{array}{rl}& \nabla \times \vec{E}=-j\omega \mu \vec{H}\\ & \nabla \times \vec{H}=j\omega \epsilon \vec{E}\end{array}\text{\hspace{1em}(GP.3)}$$
Which also means that each field component of EM field satisfies Helmholtz equation
$$\begin{array}{r}\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+\frac{{\partial }^2}{\partial z^2}+k^2\right)\vec{E}=0\\ \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+\frac{{\partial }^2}{\partial z^2}+k^2\right)\vec{H}=0\end{array}\text{\hspace{1em}(GP.4)}$$
Substituting $(GP.1)$ into $(GP.4)$, we obtain
$$\begin{array}{r}\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}-{\beta }^2+k^2\right)\vec{E}\triangleq \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+k_c^2\right)\vec{E}=0\\ \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}-{\beta }^2+k^2\right)\vec{H}\triangleq \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+k_c^2\right)\vec{H}=0\end{array}\text{\hspace{1em}(GP.5)}$$
where
$$k_c^2=k^2-{\beta }^2=\frac{{\omega }^2}{c^2}-{\beta }^2={\omega }^2\epsilon \mu -{\beta }^2\text{\hspace{1em}(GP.6)}$$
is so-called cut-off wavenumber.

---

Longitudinal-transverse decompositions

Substituting $(GP.1)$ into $(GP.4)$ one also gets
$$\begin{array}{rlrl}& \frac{\partial H_z}{\partial y}+j\beta H_y=j\omega \epsilon E_x& & \frac{\partial E_z}{\partial y}+j\beta E_y=-j\omega \mu H_x\\ & -j\beta H_x-\frac{\partial H_z}{\partial x}=j\omega \epsilon E_y& & -j\beta E_x-\frac{\partial E_z}{\partial x}=-j\omega \mu H_y\\ & \frac{\partial H_y}{\partial x}-\frac{\partial H_x}{\partial y}=j\omega \epsilon E_z& & \frac{\partial E_y}{\partial x}-\frac{\partial E_x}{\partial y}=-j\omega \mu H_z\end{array}\text{\hspace{1em}(LT.1)}$$
In this system of six equations, there are only two independent unknowns: $E_z$ and $H_z$ and four other field components can be expressed through them:
$$\begin{array}{rl}{H}_x& =\frac{j}{k_c^2}\left(\omega \epsilon \frac{\partial E_z}{\partial y}-\beta \frac{\partial H_z}{\partial x}\right)\\ H_y& =\frac{-j}{k_c^2}\left(\omega \epsilon \frac{\partial E_z}{\partial x}+\beta \frac{\partial H_z}{\partial y}\right)\\ E_x& =\frac{-j}{k_c^2}\left(\beta \frac{\partial E_z}{\partial x}+\omega \mu \frac{\partial H_z}{\partial y}\right)\\ E_y& =\frac{j}{k_c^2}\left(-\beta \frac{\partial E_z}{\partial y}+\omega \mu \frac{\partial H_z}{\partial x}\right)\end{array}\text{\hspace{1em}(LT.2)}$$
Therefore, we can obtain $\vec{E}(x,y,z)$ and $\vec{H}(x,y,z)$ if we get the longitudinal components $E_z$ and $H_z$. They can be found as solutions of Helmholtz equations:
$$\begin{array}{r}\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+k_c^2\right)E_z=0\\ \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+k_c^2\right)H_z=0\end{array}\text{\hspace{1em}(LT.3)}$$
which satisfy some boundary conditions at the surface forming the waveguide cross-section.

---

TEM, TE, TM waves

Depending on whether both, one, or none of the longitudinal components are zero, we may classify the solutions as transverse electric and magnetic (TEM), transverse electric (TE), transverse magnetic ™, or hybrid:
$$\begin{array}{ll}{E}_Z=0,H_Z=0,& \text{ TEM modes }\\ E_z=0,H_Z\ne 0,& \text{ TE or H modes }\\ E_z\ne 0,H_z=0,& \text{ TM or E modes}\\ E_z\ne 0,H_Z\ne 0,& \text{ Hybrid or }\mathrm{H}\mathrm{E}\text{ or }\mathrm{E}\mathrm{H}\text{ modes}\end{array}$$
In a generic case, Maxwell equation has three independent (orthogonal) solutions: two curl-based solutions (two source free solutions with orthogonal polarizations) and one gradient-based (electrostatic, linked to charges) solution.

TE- and TM-waves are two such curl-based solutions with orthogonal polarizations and TEM wave is a charged linked solution.

Each of these solutions can be described with a single scalar function (known as potential). So from 6 unknowns in Maxwell equations (EM field components) maximally three can be fully independent.

• TEM: $E_z=H_z=0$

  From $(LT.2)$, since $E_z=H_z=0$, nontrivial solution exists only when $k=\beta$ or $k_c=0$. Substituting $E_z=H_z=0$ into $(LT.1)$, then
  $$\begin{array}{rlrl}& j\beta H_y=j\omega \epsilon E_x& & j\beta E_y=-j\omega \mu H_x\\ & -j\beta H_x=j\omega \epsilon E_y& & -j\beta E_x=-j\omega \mu H_y\\ & \frac{\partial H_y}{\partial x}-\frac{\partial H_x}{\partial y}=0& & \frac{\partial E_y}{\partial x}-\frac{\partial E_x}{\partial y}=0\end{array}\text{\hspace{1em}(TEM.1)}$$
  It can be derived from the first two rows that
  $$\begin{gathered} {\beta }^2={\omega }^2\epsilon \mu ,\text{ i.e.},\beta =k=\omega /c \\ \text{\hspace{1em}(TEM.2)} \end{gathered}$$
  and
  $$Z_{TEM}=\frac{E_x}{H_y}=\frac{-E_y}{H_x}=\sqrt{\frac{\mu }{\epsilon }}\triangleq \eta \text{\hspace{1em}(TEM.3)}$$
  $(TEM.2)$ shows that the propagation constant $k$ linearly depends on frequency. Since the phase velocity $v_p=\omega /k=c$ and the group velocity $v_g=(\frac{dk}{d\omega }{)}^{-1}=c$, they are frequency independent and equal to each other.

  $(TEM.3)$ is the wave impedance of a TEM wave. It is frequency independent and similar to those of a plane wave in a homogeneous medium.

  Since $k_c=0$, the Helmholtz wave equations for all transversal components will have a form
  $$\begin{array}{r}\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}\right)E_{x,y}=0\\ \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}\right)H_{x,y}=0\end{array}\text{\hspace{1em}(TEM.4)}$$
  In other words, both transversal electric and magnetic fields satisfy Laplace equation
  $${\nabla }_t^2{\vec{E}}_t(x,y)=0,{\nabla }_t^2{\vec{H}}_t(x,y)=0\text{\hspace{1em}(TEM.5)}$$
  Therefore, transverse field of a TEM wave is the same as the static field and can exist only between the conductors. At least two conductors are needed for existence of such field.

• TE: $E_z=0,H_z\ne 0$

  Simplify $(LT.2)$ with $E_z=0,H_z\ne 0$
  $$\begin{array}{rlrl}& H_x=\frac{-j\beta }{k_c^2}\frac{\partial H_z}{\partial x}& & H_y=\frac{-j\beta }{k_c^2}\frac{\partial H_z}{\partial y}\\ & E_x=\frac{-j\omega \mu }{k_c^2}\frac{\partial H_z}{\partial y}& & E_y=\frac{j\omega \mu }{k_c^2}\frac{\partial H_z}{\partial x}\end{array}\text{\hspace{1em}(TE.1)}$$
  With propagation constant different from one for free space
  $$k_c\ne 0⟹\beta =\sqrt{k^2-k_c^2}\text{\hspace{1em}(TE.2)}$$
  And impedance
  $$Z_{TE}=\frac{E_x}{H_y}=\frac{-E_y}{H_x}=\frac{\omega \mu }{\beta }=\frac{k\eta }{\beta }\text{\hspace{1em}(TE.3)}$$
  which is frequency dependent.

  The longitudinal component $H_z$ is a solution of Helmholtz wave equation
  $$\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+k_c^2\right)H_z(x,y)=0\text{\hspace{1em}(TE.4)}$$
  subject to boundary conditions of the specific geometry.
  In generic case, there are multiple nontrivial solutions of this equation for different values of $k_c$ (cut-off wavenumbers). These solutions are called (waveguide) modes
  $$H_z(x,y)=\sum _n{H}_{zn}(x,y)\Rightarrow H_z(x,y,z)=\sum _n{H}_{zn}(x,y)e^{-j{\beta }_{n}z}\text{\hspace{1em}(TE.5)}$$
  The values of $k_c$ for which non-trivial solutions do exist are called spectrum.

  Due to different values of $k_c$ each mode has its own propagation constant and propagates with own phase velocity
  $${\beta }_i^2=k^2-k_{ci}^2=\frac{{\omega }^2}{c^2}-k_{ci}^2,i=0,1,\cdots \text{\hspace{1em}(TE.6)}$$
  The propagation constant is real only above so called cutoff frequency ${\omega }_c=c{k}_c$, and is purely imaginary below such frequency. This means that below the cut-off frequency the mode attenuates exponentially along $z$.

• TM: $E_z\ne 0,H_z=0$

  Simplify $(LT.2)$ with $E_z\ne 0,H_z=0$
  $$\begin{array}{rlrl}& H_x=\frac{j\omega \epsilon }{k_c^2}\frac{\partial E_z}{\partial y}& & H_y=\frac{-j\omega \epsilon }{k_c^2}\frac{\partial E_z}{\partial x}\\ & E_x=\frac{-j\beta }{k_c^2}\frac{\partial E_z}{\partial x}& & E_y=\frac{-j\beta }{k_c^2}\frac{\partial E_z}{\partial y}\end{array}\text{\hspace{1em}(TM.1)}$$
  With propagation constant different from one for free space
  $$k_c\ne 0⟹\beta =\sqrt{k^2-k_c^2}\text{\hspace{1em}(TM.2)}$$
  And impedance
  $$Z_{TE}=\frac{E_x}{H_y}=\frac{-E_y}{H_x}=\frac{\beta }{\omega \epsilon }=\frac{\beta \eta }{k}\text{\hspace{1em}(TM.3)}$$
  which is frequency dependent.

  The longitudinal component $E_z$ is a solution of Helmholtz wave equation
  $$\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}+k_c^2\right)E_z(x,y)=0\text{\hspace{1em}(TM.4)}$$
  subject to boundary conditions of the specific geometry.
  In generic case, there are multiple nontrivial solutions of this equation for different values of $k_c$ (cut-off wavenumbers). These solutions are called (waveguide) modes
  $$E_z(x,y)=\sum _n{E}_{zn}(x,y)\Rightarrow E_z(x,y,z)=\sum _n{E}_{zn}(x,y)e^{-j{\beta }_{n}z}\text{\hspace{1em}(TM.5)}$$
  The values of $k_c$ for which non-trivial solutions do exist are called spectrum.

  Due to different values of $k_c$ each mode has its own propagation constant and propagates with own phase velocity
  $${\beta }_i^2=k^2-k_{ci}^2=\frac{{\omega }^2}{c^2}-k_{ci}^2,i=0,1,\cdots \text{\hspace{1em}(TM.6)}$$
  The propagation constant is real only above so called cutoff frequency ${\omega }_c=c{k}_c$, and is purely imaginary below such frequency. This means that below the cut-off frequency the mode attenuates exponentially along $z$.

Waveguide wavelength

For TE and TM waves, the guide wavelength can be determined from wavenumber
$${\lambda }_g=\frac{2\pi }{\beta }=\frac{2\pi }{\sqrt{k^2-k_c^2}}=\frac{c}{f\sqrt{1-(\frac{{\omega }_c}{\omega }{)}^2}}=\frac{\lambda }{\sqrt{1-(\frac{f_c}{f}{)}^2}}$$
where $c$ is propagation velocity in unbounded medium.

• At the cut-off frequency the guide wavelength equals infinity.
• It is always longer than the wavelength in free-space.

Waveguide phase and group velocities

For TE and TM waves, the waveguide phase velocity is
$$v_p=\frac{\omega }{\beta }={\lambda }_{g}f=\frac{c}{\sqrt{1-(\frac{f_c}{f}{)}^2}}\ge c$$
As waveguide wavelength is longer than the wavelength in free space, the phase velocity in waveguide is higher that the speed of light.

The waveguide group velocity is
$$v_g=(\frac{d\beta }{d\omega }{)}^{-1}=(\frac{d}{d\omega }\frac{\sqrt{{\omega }^2-{\omega }_c^2}}{c}{)}^{-1}=c\sqrt{1-(\frac{f_c}{f}{)}^2}<c$$

---

Coaxial cable

Both electric and magnetic fields in TEM mode are subject of Helmholtz wave equations for all transversal
components
$$\begin{array}{r}\left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}\right)E_{x,y}=0\\ \left(\frac{{\partial }^2}{\partial x^2}+\frac{{\partial }^2}{\partial y^2}\right)H_{x,y}=0\end{array}\text{\hspace{1em}(TEM.4)}$$
Boundary conditions: (in cylindrical coordinates)
$$\begin{array}{r}\rho =a:E_{\phi }=0,E_z=0;\\ \rho =b:E_{\phi }=0,E_z=0;\end{array}\text{\hspace{1em}(CC.1)}$$
For TEM wave both fields can be derived via the scalar potential
$${\vec{E}}_t=-{\nabla }_t\Phi (\rho ,\phi )\text{\hspace{1em}(CC.2)}$$
The scalar potential is a solution of Laplace equation, which in cylindrical coordinates looks like
$$\frac{1}{\rho }\frac{\partial }{\partial \rho }\left(\rho \frac{\partial \Phi (\rho ,\phi )}{\partial \rho }\right)+\frac{1}{{\rho }^2}\frac{{\partial }^2\Phi (\rho ,\phi )}{\partial {\phi }^2}=0\text{\hspace{1em}(CC.3)}$$
with boundary conditions
$$\Phi (a,\phi )=V_0;\Phi (b,\phi )=0\text{\hspace{1em}(CC.4)}$$
The solution for the scalar potential is
$$\Phi (\rho ,\phi )=V\frac{\ln (b/\rho )}{\ln (b/a)};a\le \rho \le b;0\le \phi <2\pi \text{\hspace{1em}(CC.5)}$$
and for fields
$$E_{\rho }(\rho ,\phi )=\frac{V}{\ln (b/a)}\frac{1}{\rho };H_{\phi }(\rho ,\phi )=\frac{V}{\eta \ln (b/a)}\frac{1}{\rho }\text{\hspace{1em}(CC.6)}$$
The total current is
$$\begin{array}{rl}I& ={\iint }_S\vec{J}\cdot d\vec{S}={\iint }_S(\nabla \times \vec{H}-j\omega \epsilon \vec{E})\cdot d\vec{S}={\oint }_C\vec{H}\cdot d\vec{l}\\ & ={\int }_0^{2\pi }{H}_{\phi }(\rho ,\phi )\rho d\phi ={\int }_0^{2\pi }\frac{V}{\eta \ln (b/a)}\frac{1}{\rho }\rho d\phi =\frac{2\pi V}{\eta \ln (b/a)}\end{array}\text{\hspace{1em}(CC.7)}$$
With the circuit impedance for the coaxial line
$$Z=\frac{V}{I}=\frac{\eta \ln (b/a)}{2\pi }\text{\hspace{1em}(CC.8)}$$
For TEM, wave group and phase velocities are equal, thus
$$v_p=v_g=c=\frac{c_0}{\sqrt{{\epsilon }_r{\mu }_r}}\text{\hspace{1em}(CC.9)}$$
The series resistance per unit length
$$R\approx \frac{R_s}{2\pi }\left(\frac{1}{a}+\frac{1}{b}\right)\text{\hspace{1em}(CC.10)}$$
The conductance per unit length
$$G=\frac{2\pi \omega {\epsilon }^{\prime \prime }}{\ln (b/a)}\text{\hspace{1em}(CC.11)}$$
The attenuations:
$$\begin{array}{rlrl}& \text{conduction: }& & {\alpha }_c\approx \frac{R}{2Z_0}(\mathrm{N}\mathrm{p}/\mathrm{m})\\ & \text{dielectric: }& & {\alpha }_d\approx \frac{G{Z}_0}{2}(\mathrm{N}\mathrm{p}/\mathrm{m})\\ & \text{total: }& & {\alpha }_{tot}=8.686({\alpha }_c+{\alpha }_d)(\mathrm{d}\mathrm{B}/\mathrm{m})\end{array}\text{\hspace{1em}(CC.12)}$$
In addition to TEM the coaxial cables also support higher order TE– and TM–modes. The lowest cut-off frequency of a higher order mode corresponds to $T{E}_{1,1}$:
$$k_c=\frac{2}{a+b}⟹f_c=\frac{{\omega }_c}{2\pi }=\frac{k_{c}c}{2\pi }=\frac{c_0}{\pi (a+b)\sqrt{{\epsilon }_r}}\text{\hspace{1em}(CC.13)}$$
which is the highest usable frequency in coaxial cable.

Power handling:

A simple relationship between $P_{max}$ and maximum frequency for coaxial cable:
$$P_{max}=5.3\times 10^{12}(E_d/f_c{)}^2\text{\hspace{1em}(CC.14)}$$
By comparison, a similar formula for a hollow waveguide yields:
$$P_{max}=26\times 10^{12}(E_d/f_c{)}^2\text{\hspace{1em}(CC.14)}$$