Fiber Optics

1Introduction to fiber optics¶

This chapter introduces the field of fiber optics. In comparison to free space optics considered so far, fibers confine light to a small volume, which prevents power loss by diffraction. As such, optical signals can propagate over large distances enabling, among others, fast and reliable communication all over the world.

In this chapter we will restrict ourselves to the cylindrical step-index fiber to confine light and to introduce fiber optics concepts. Such a fiber consists of a core (refractive index $n_1$) and a cladding (refractive index $n_2$, $n_2<n_1$) as shown in Figure 1. A single-mode silica communication fiber typically has a core diameter of several micrometres and a cladding diameter of about $125\ \mu\mathrm{m}$; a protective coating and jacket surround the glass. Other fiber types have different dimensions and are protected by a (reinforced) jacket around the cladding, as observed in the same figure.

This chapter can be subdivided in three parts in which we move gradually from theory to application. Based on the step index fiber, this chapter first discusses light confinement by total internal reflection. Then, in Section 3, the concept of modes of light propagating through an optical fiber is introduced. This gives rise to certain figures of merit and leads to the important distinction between ‘‘single mode’’ and ‘‘multimode’’ fibers, discussed in Section 4. Although fiber optics has advantages over free space optics in terms of light propagation, propagation is not ideal. The effects and causes of dispersion and loss are therefore introduced and discussed in Section 5 and Section 6 respectively.

After this theoretical view on optical fibers, we zoom out a little to introduce various other fiber types in Section 7, since the step-index fiber is not the only implementation of optical fibers (although it is the most common). In Section 8, we discuss how optical fibers can be connected to the outside world. What this outside world may consist of is briefly discussed in Section 9. This section introduces common fiber optic components used to manipulate light propagating through fibers enabling us to build useful setups and applications. Some of these applications are elaborated on in Section 10. This section also provides an outlook in the possible directions the rich field of fiber optics is headed.

2Total internal reflection¶

Let us first concern ourselves with the question of how light can be confined in fibers. This happens by total internal reflection (TIR), which can be well explained using ray optics. Typically, fibers have a silicon-oxide ($\text{SiO}_2$) cladding, whereas in the $\text{SiO}_2$ core small amounts of germanium-oxide ($\text{GeO}_2$) are ‘‘dissolved’’. This doping increases the index of refraction $n_1$ slightly, thus $n_1>n_2$, as in Figure 1. The step in refractive index $n_1-n_2$ is indeed small, in the order of 10^-3.

Now consider Snell’s law for the core-cladding boundary in Figure 1,

For angles $\theta_{\text{i}}$ in excess of the angle $\theta_{\text{i,c}}=\arcsin(n_2/n_1)$ refraction no longer occurs (the maximum value of a sine is 1) and light is totally reflected back into the core. Therefore we will refer to $\theta_{\text{i,c}}$ as the internal critical angle. For meridional rays entering from air, only a limited external cone of directions is accepted; its half-angle is $\bar{\theta}_{\text{e,c}}$. This angle will be further discussed in Section 4. For $\text{SiO}_2$, in which $n_2\approx 1.444$ at telecom wavelengths ($\lambda=1550~\text{nm}$), TIR occurs if $\theta_{\mathrm i}>85.7^{\circ}$ when the core has $n_1=1.448$. Because this $\theta_{\text{i}}$ is close to $90^{\circ}$, fibers with $n_1\approx n_2$ are also referred to as ‘‘weakly guiding’’.

Since the angle of reflection equals the angle of incidence upon reflection, TIR occurs at every core-cladding interface and the light remains confined to the core (strictly speaking, this argument only holds for straight fibers. The influence of bends in fibers is discussed in Section 6). Because low-loss silica fibers can have attenuation of a few tenths of a decibel per kilometre,^[1] see Section 6), this implies fibers can easily carry optical signals over distances in the order of several kilometers without considerable loss.

3Fiber modes¶

Although some of the properties of optical fibers can be understood from ray optics, the exact propagation of light through fibers is described by Maxwell’s equations. These give rise to guided modes: electromagnetic fields whose transverse pattern retains its shape while a phase factor accumulates along the fiber. All possible EM-fields in fibers can be described as a superposition of fiber modes, like a musical tone played by an instrument can be described using a superposition of a note and its overtones.

To introduce the concept of fiber modes, we consider the planar mirror waveguide in Figure 3. This waveguide consists of two (perfect) planar mirrors placed in the $(x,z)$-plane in vacuum at positions $y=\pm d/2$ (hence their separation is $d$). Monochromatic light rays with wavelength $\lambda_0$ enter the waveguide under an angle $\bar{\theta}$. They propagate through the waveguide while constantly bouncing back and forth between the mirrors. This waveguide is a simplification of the optical fibers we are considering in this chapter, but it is more straightforward to study. Modes in optical fibers follow from the same considerations and the necessary generalizations are touched upon toward the end of this section.

In order for a light ray to represent a mode, we impose a self-consistency condition that guarantees the invariance of distribution and polarization under propagation. In practice, this boils down to the wave repeating itself after every second bounce. With reference to Figure 3, the self-consistency condition thus imposes that the phase acquired by the reflected wave traveling from A to C equals the phase acquired by the ‘‘non-reflected’’ wave traveling from A to the virtual point B up to an integer multiple of $2\pi$. Of course, as the mirrors are considered perfect, both reflections add a phase shift of $\pi$, such that

or

Using elementary geometry and the relation $\cos(2\bar{\theta})=1-2\sin^2(\bar{\theta})$, it can be shown that $|AC|-|AB|=2d\sin(\bar{\theta})$ and thus the self-consistency condition becomes

We thus find that only certain bouncing angles $\bar{\theta}_m$ are allowed by the self-consistency condition and these are referred to as the $m$ modes of propagation. From Eq. (4) we can draw the important conclusion that the number of modes in the waveguide is limited, because $\max(\sin(\bar{\theta}_m))=1$. This implies that:

• the maximum number of modes $M$ possibly propagating through the waveguide equals

where $M$ counts the positive integers $m$ for which $m\lambda_0/(2d)<1$. Away from an exact cutoff, this equals $\lfloor2d/\lambda_0\rfloor$; at cutoff, the mode has no forward propagation;

• for $M=1$, the waveguide is a single mode waveguide, and if $M>1$ the waveguide is said to be multimode;

• if $M=0$, this particular $x$-polarized mirror-waveguide family has no forward-propagating mode. Its lowest-order cutoff is $\lambda_{\mathrm{c}}=2d$; the statement does not apply to every possible polarization of a parallel-plate waveguide.

To obtain the field distributions of the EM-fields propagating through the waveguide, consider a wave propagating along the waveguide in $z$-direction, which is polarized in $x$-direction. It should be realized that such a wave actually consists of two superposed waves, one in the upward and one in the downward direction due to internal reflection of the wave fronts, see Figure 4. In case the up-propagating wave is described by

the down-propagating wave is described by

Here, the wavenumbers are $k_{y,m}=2\pi\sin(\theta_m)/\lambda_0$ and $k_{z,m}=2\pi\cos(\theta_m)/\lambda_0$. The phase shift in the down-propagating wave of $(m-1)\pi$ follows from the condition that after superposing the two fields, their magnitude at the mirrors (at $y=\pm d/2$) should vanish^[2]. Therefore, the total field is of the form

in which $E_0$ is the field’s amplitude and the mode distributions are

The first few distributions are plotted in Figure 4; the mode index $m$ counts the transverse half-wave variations between the mirrors. The pre-factors $\sqrt{2/d}$ in Eq. (9) are chosen such that the functions are orthonormal, meaning

In practice this implies that all possible light pulses transmitting through the waveguide can be written as a linear combination, or superposition, of modes. As stressed before, this is analogous to tones played by a musical instrument.

This concludes our discussion on fiber modes in the planar-mirror waveguide. To generalize this discussion to optical step-index fibers, two considerations are added. First, in optical fibers the EM-fields are not bounded by the core. Rather, the fields also extend partly into the cladding, where their amplitude diminishes rapidly, or evanescent. Second, a cylindrical fiber has angular structure as well as radial structure, illustrated schematically in Figure 5 and by the field patterns in Figure 6. This implies fiber modes are labeled by two indices, generally $m$ (linear self-consistency) and $l$ (circumferential self-consistency). A selection of fiber mode field distributions is depicted in Figure 6 (b). A distinction is made between single mode fibers (SMF) and multimode fibers (MMF). In the former only the $(m,l)=(1,0)$ mode can propagate, whereas the MMF supports more or even many modes.

4Fiber parameters¶

Optical fibers can be characterized by a number of parameters. Here we list a selection of these numbers.

• $\Delta$-parameter -- The $\Delta$-parameter is directly related to the relative difference in core and cladding refractive index, see Figure 1. It is defined as

The approximation, which entails the relative difference of $n_1$ and $n_2$, holds for the weakly guiding fibers under consideration in this chapter. It follows from setting $n_2=n_1-\Delta n$ and neglecting terms of order $\Delta n^2$. For $n_1=1.448$ and $n_2=1.444$, the exact expression gives $\Delta\approx2.76\times10^{-3}$.

• Numerical aperture (NA) -- The NA relates the fiber and the (external) critical angle under which it accepts (or emits) light, $\bar{\theta}_{\text{e,c}}$. That is, the maximum value of $\bar{\theta}_{\text{e}}$ in Figure 1 such that TIR occurs. The NA for fibers is defined as

and equals approximately 0.108 if $n_1=1.448$ and $n_2=1.444$. With reference to Figure 1, this follows from

in which we use $\bar{\theta}_{\text{i,c}}=\arccos(n_2/n_1)$, such that $\cos(\bar{\theta}_{\text{i,c}})=n_2/n_1$. This expression assumes an external medium with $n_{\mathrm e}=1$. In general $n_{\mathrm e}\sin\bar\theta_{\mathrm{e,c}}=\sqrt{n_1^2-n_2^2}$ for meridional rays.

• $V$-number – The $V$-number, or fiber parameter, governs the number of modes supported in the fiber. It is defined as

  $$\begin{align*} V=\frac{\pi d}{\lambda_0}\mathrm{NA}. \end{align*}$$

  (14)

  For step-index fibers, the number of modes is approximately equal to $M=V^2/2$ if $V\gg 1$. If $V<2.405$, only the fundamental spatial mode is supported and the fiber is referred to as an SMF. If $V>2.405$, the fiber is multimode.

• higher-order-mode cutoff wavelength – For an ideal step-index fiber, $V=2.405$ marks the cutoff of the first higher-order mode. Its corresponding vacuum wavelength is

Above $\lambda_{\mathrm c}$ the ideal fiber supports only the fundamental mode; this cutoff does not extinguish it.

5Fiber dispersion¶

Light propagating through fibers often suffers from dispersion, or pulse broadening. The cause for dispersion is that transmitting waves (e.g. at different wavelengths) do not travel at the same velocity, which may cause information loss. In e.g. telecommunication, information is typically sent using short pulses of light, which may start to overlap in time upon broadening, see Figure 8. Substantial overlap makes symbols harder to distinguish and can increase errors unless the link uses suitable compensation. Here three main causes for dispersion are discussed, modal dispersion, material dispersion and waveguide dispersion.

Modal dispersion is relevant for MMFs, in which several modes of light are present simultaneously. As discussed in Section 3, light rays bounce through a fiber or any other waveguide. To introduce modal dispersion we refer back to the planar mirror waveguide depicted in Figure 3. Since the rays of higher-order modes are at larger angles $\bar{\theta}_m$ (see Eq. (4)) than lower-order modes, the higher-order modes have a longer path length. This means that the effective propagation velocity $v_z$, the velocity at which the wave travels in $z$ -direction, decreases with increasing mode number. For the planar mirror waveguide in vacuum the propagation velocity for the $m^{\text{th}}$ mode is

Here we used that $\cos^2(\theta)+\sin^2(\theta)=1$ along with Eq. (4).

Now consider an infinitely short pulse that consists of light in the $i^{\text{th}}$ and the $j^{\text{th}}$ mode only, with $i<j$. For the sake of the argument, we still presume that the light is monochromatic, although in reality this is not the case (see below). Then these modes travel in $z$-direction with a velocity difference

Due to the velocity difference, the difference in arrival time of the two modes at the end of the waveguide of length $L$ is

which is depicted in Figure 7 for several combinations of $i$ and $j$ at $\lambda_0/(2d)=0.15$. Thus, the infinitely short pulse that we started with has broadened to a ‘‘width’’ $\Delta\tau$ at the end of the waveguide. To avoid substantial overlap in a simple on–off link, the pulse period should be comfortably longer than this differential delay.

In reality, light pulses in multimode waveguides contain many modes and are not infinitely short. However, the same ideas hold: due to modal dispersion the pulse broadens, thereby limiting the rate of information transfer. This is also true for optical fibers. For a weakly guiding step-index fiber at large $V$, a ray estimate for the differential delay per unit length between near-axial and highest-angle paths is $\Delta\tau^{\mathrm{mod}}/L\approx(n_1-n_2)/c$. With $n_1=1.448$ and $n_2=1.444$, this scale is about $13.3\ \mathrm{ns/km}$. The precise delay depends on the supported modes and launch conditions.

Up to this point we have implicitly assumed that the light transmitted through optical fibers is monochromatic. This is not the case: in telecommunication information is sent through fibers by means of short pulses of light. However, the shorter the pulse, the broader its wavelength spectrum, as schematically depicted in Figure 9. Apart from this effect, any light source has an intrinsic width of its emitted wavelength spectrum.

Now we consider a signal pulse propagating through a medium. The pulse has a central wavelength $\lambda_0$ and a spectral width $\Delta\lambda$. If the index of refraction is wavelength-dependent, $n=n(\lambda)$, such as in SiO$_2$ , the spectral width of the signal propagating through a fiber results in dispersion. This is due to the fact that the pulse’s (group) propagation velocity is imposed by the group refractive index $n_{\text{g}}$ as $v_{\text{g}}=c/n_{\text{g}}$, which in turn depends on $n(\lambda)$ as

Hence all wavelength components present in the pulse travel at slightly different velocities and dispersion is the result. From Eq. (19) we see, however, that the effect only occurs whenever $\mathrm{d}^2n/\mathrm{d}\lambda^2\neq 0$ (NB: if $n$ is constant or of the form $n=a\lambda+b$, Eq. (19) still yields a constant group refractive index). This is the case for SiO$_2$ as can be observed in Figure 10.

Due to material dispersion, the change in pulse width after fiber length $L$ is given by

However, in practice often the dispersion is reported as

in units of $\text{ps}/(\text{km} \cdot \text{nm})$. The magnitude $|D^{\mathrm{mat}}|L\Delta\lambda$ estimates differential group delay across a narrow spectrum; the sign of $D$ indicates which wavelengths arrive later. Its value depends strongly on wavelength.

A second effect of the spectral width of light pulses is that they disperse even if light propagates in a single mode. As observed from Eq. (4), the angle $\bar{\theta}_m$ not only depends on the mode $m$, but also on the wavelength of the light. Therefore, all wavelength components of the light will propagate at a slightly different $\bar{\theta}_m$. This will cause dispersion in the same fashion as discussed before for modal dispersion. Waveguide dispersion can matter in both single-mode and multimode fibers; it is particularly important when a single mode is used and intermodal delay is absent.

For a guided mode, its effective refractive index $n_{\mathrm{eff}}(\lambda)$ depends on both material indices and the waveguide geometry. Its total chromatic-dispersion parameter is

The contribution caused by wavelength-dependent mode confinement is called waveguide dispersion. It depends on core size and index profile; there is no universal formula depending only on $n_1-n_2$ and $\lambda$.

In a conventional single-mode silica fiber, material and waveguide contributions can partly cancel near $1310\ \mathrm{nm}$, as illustrated in Figure 11. This is why that wavelength is useful when low chromatic dispersion matters. The exact zero-dispersion wavelength depends on the fiber design. Around $1550\ \mathrm{nm}$, silica attenuation is generally lower, while ordinary single-mode fiber has appreciable chromatic dispersion. Dispersion-shifted designs can move the zero-dispersion wavelength by changing the waveguide profile; low dispersion does not inherently require higher attenuation.

6Fiber loss mechanisms¶

So far, we have implicitly neglected fiber losses. However, in reality, losses do occur while light propagates through a fiber. This causes the light to be attenuated following the exponential decay

where $P(z)$ is the light power at position $z$ along the fiber and $\alpha$ is the loss coefficient per unit length. Taking the $\text{km}$ as unit for $z$, this value can be obtained from the loss coefficient in dB/km as $\alpha=-\ln(10^{-\alpha_{\text{dB}}/10})$. For SiO$_2$ the loss coefficient is plotted in Figure 12. In this section we discuss the main loss mechanisms in optical fibers: Rayleigh scattering, absorption, bending and coupling losses.

Rayleigh scattering arises from small refractive-index fluctuations frozen into the amorphous silica during manufacture. These fluctuations scatter a fraction of the guided light. The average loss factor from Rayleigh scattering is given by

where $\alpha_0^{\text{R}}$ is the loss factor experimentally measured at a wavelength $\lambda_0^{\text{R}}$. For SiO$_2$, $\alpha_0^{\text{R}}$ is approximately $0.15~\text{dB/km}$ at $\lambda_0^{\text{R}}=1550~\text{nm}$. This makes Rayleigh scattering the dominant loss mechanism in SiO$_2$ for lower wavelengths.

At larger wavelengths infrared absorption becomes the dominant loss mechanism in SiO$_2$ fibers. Infrared light may excite vibrational states of SiO$_2$, and the excitation energy is subsequently dissipated as heat in the fiber. The loss coefficient for infrared absorption is given by

For SiO$_2$ fibers $\alpha_0^{\text{IR}}$ is on the order of 10¹² dB/km and $\lambda_0^{\text{IR}}$ equals approximately $50\mu \text{m}$.

Additionally, the three attenuation peaks at approximately 975, 1225 and $1400~\text{nm}$ in Figure 12 are due to absorption. These absorption bands are caused by the presence of hydroxyl (OH) groups stemming from water vapor dissolved in the SiO$_2$ during fabrication. Apart from these bands, other bands may be present due to other contaminants such as Copper (Cu), Iron (Fe) and Nickel (Ni) (not depicted in Figure 12).

As can be observed in Figure 12, silica-fiber attenuation has a low-loss window near $1550~\text{nm}$. This favors that band for long links, subject to dispersion and system-design constraints.

Besides the intrinsic losses of Rayleigh scattering and absorption, losses also occur as a result of extrinsic factors, such as (excessive) bending and coupling.

In case an optical fiber bends, losses may occur if the bending radius is too small. This can be understood from considering the light rays in the fiber. As these travel on straight paths, a bend changes the angle under which the rays hit the core-cladding boundary, see Figure 14. As such, the ray may reach this boundary under an angle $\theta_{\text{i}}$ below the internal critical angle. As a result some of the light refracts into the cladding and is lost.

Bending may occur over a visible radius, as in Figure 13, or through small local deformations, as in Figure 14. The latter may occur as a result of improper fiber handling, e.g. when the fiber is strained excessively.

To prevent macro-bending losses, the critical radius of fibers should be noted. This parameter is listed in the fiber datasheet. Making bends tighter than this critical radius results in macro-bending losses.

Apart from internal fiber losses, losses may also occur while coupling fibers, see Figure 15. This happens when fiber cores are not aligned properly (they are shifted with respect to each other, placed under an angle or there is a gap in between the fibers), when a higher-NA fiber is coupled to a fiber with lower NA, or when the core size of the two fibers is different. In case of a mismatch in NA, the lower-NA fiber will support fewer modes (see Eq. (14)) and the higher order modes will be lost. Apart from these issues with connecting fibers, internal reflections between different fiber cores cause losses. These reflections occur whenever the index of refraction changes. For each type of coupling loss, the loss factor can be easily a few dB, as you can estimate yourself in one of the problems at the end of this chapter. Therefore care must be taken to couple fibers properly.

7Fiber types¶

Apart from the step index fibers that we have considered up to this point, other types of optical fibers are available. The only condition for constructing an optical fiber is that light remains confined to the fiber core. In step index fibers this occurs due to the discontinuous decrease of the refractive index at the core-cladding boundary giving rise to TIR. In this section we briefly introduce different types of optical fibers and color coding.

In graded index fibers, the refractive index changes in a continuous fashion over the fiber diameter. At a cost of a more complex fabrication process, the advantage of GRIN fibers over step index fibers is that modal dispersion is less severe.

Another fiber type is the so-called holey fiber, see Figure 16. Instead of changing the refractive index using dopants, the cladding of holey fibers contains a carefully designed array of holes containing air, which change the refractive index of the ‘‘cladding’’ (part of the fiber with holes) with respect to the ‘‘core’’ (part of the fiber without holes). These holes run over the whole length of the fiber. This implies holey fibers can be constructed from a single material.

A photonic-bandgap fiber also uses a periodic microstructure, although Figure 16 shows a solid-core holey fiber rather than a bandgap-guided hollow core. However, in PBFs these holes are arranged in a fashion that a bandgap is created in the cladding. Such a bandgap does not allow light of certain wavelength(s) to transmit through the cladding, thus confining light of this wavelength to the core. This is fundamentally different from fibers based on changes in refractive index. Photonic-bandgap guidance works in designed spectral bands, whereas ordinary step-index guidance is governed by total internal reflection. The band width depends on the particular microstructure.

Apart from wavelength, light also carries the property of polarization. In the fiber types discussed so far, no measures are taken to preserve the polarization of the light transmitting through the fiber. Therefore, crosstalk may occur between light transmitted in different polarizations as a result of, e.g., bends in the fiber. This is not a problem if the fibers are used in applications in which only the intensity of the transmitted light is considered, e.g., in telecommunication, in which information is stored in binary pulses that are either on (logical 1) or off (logical 0). However, in other applications such as quantum key distribution, information is stored in the polarization of the light: horizontal or vertical.

In such applications, the light’s polarization needs to be preserved throughout the fiber’s length for which polarization maintaining fibers have been developed. When light is launched along one of their principal axes, polarization-maintaining fibers reduce coupling to the orthogonal axis and preserve that axis approximately over the specified operating conditions. Polarization-maintaining fibers are usually chosen when controlling polarization justifies their added cost and handling requirements.

The final fiber type to be discussed is the plastic fiber. Plastic fibers are generally made with a PMMA (polymethyl methacrylate, $n_1=1.49$) core, whereas the cladding is fluorinated. This lowers the refractive index to approximately $n_2=1.40$, implying plastic fibers have a large NA. Apart from such step-index plastic fibers, GRIN plastic fibers are available. Plastic fibers are low-cost and more resilient to bending and handling than SiO$_2$ fibers. This is due to plastic fibers having a core diameter of typically $1~\text{mm}$, instead of the fragile few-micrometer core diameter of SiO$_2$ fibers. Using plastic instead of SiO$_2$ comes, however, at a cost of higher losses in the order of $1~\text{dB/m}$. Due to these properties, plastic fibers can be found in, e.g., home, company and car data networks.

To distinguish fibers from each other, jacket colors often follow deployment conventions. For example, many multimode cables are orange or aqua and many single-mode cables are yellow, but color alone does not establish the fiber specification. The manufacturer marking or test documentation should be checked. In cables with multiple fibers and MPO connectors, individual fibers may use a separate sequence of colors to identify their positions. Such markings help technicians trace and connect the intended fibers.

8Fiber connections¶

In order to couple light into and out of optical devices and components, fiber connections need to be made. This also holds in case one wants to couple two fibers together. These connections are made using fiber connectors for which many designs have been put forward over the years. Ideally, the connection is loss-less. However, in practice all connections induce some (sub-dB) loss. This section discusses the most common manners in which fibers are connected currently, by splicing or using a fiber connector. The most common fiber connectors are depicted in Figure 17.

In fiber splicing, two fibers are permanently joined together by fusion or mechanical means. In fusion splicing the fibers are fused together by aligning the cores and claddings of both fibers and heating the joint shortly. Fusion splicing often gives low loss and a mechanically robust joint. In mechanical splicing a mechanical fixture is used to align and fixate the fibers. Although such splices are less strong and increase loss with respect to fusion splices, the equipment for making mechanical splices is less expensive. Fusion splices are commonly used where a durable, low-loss permanent joint is needed, including submarine links. Apart from this, splicing can be used as a means to repair broken optical fibers.

There are many types of connector. We report on the main ones here.

Lucent connector (LC) -- The lucent connector, also referred to as the little connector, is a widely used compact connector. In the tip of the connector, the fiber is protected by a so-called ferrule (Figure 18), which for LCs is only $1.25~\text{mm}$ in diameter. The connector snaps into place and can be detached using a push-pull motion. This makes installation an easy procedure. Due to its small footprint, it is ideal for high-density fiber applications such as in data centers.

Subscriber connector (SC) -- The square-bodied connector was an early attempt to standardize fiber connectors. With a $2.5~\text{mm}$-diameter ferrule, it is larger than the previous LC connector, and is used in many passive networks and equipment connections. The SC connector uses a push-pull latch, making its installation easy.

Straight tip (ST) connector -- The straight tip connector has a $2.5~\text{mm}$-diameter spring-loaded ferrule. This implies that the ferrule is pushed onto the device it is connected to thereby improving the transmission. Typically used for multimode fibers, it has a bayonet mount, which secures itself after turning it by a half twist.

Ferrule connector (FC) -- The FC connector is typically used for SMF and high precision applications, such as optical time domain reflectometry (OTDR). Its spring-loaded ferrule is $2.5~\text{mm}$ in diameter. It has an alignment key ensuring the connector is always inserted to the device in the same orientation. Secondly, the connector is secured using a threaded collet. Although this connector is thus more complex in installation, as well as in manufacturing, the repeatable accuracy with which it can be installed gives the precision in aforementioned applications.

Multi-position optical (MPO) connector -- The multi-position optical connector, with MTP as one trade name, can terminate multiple fibers in one ferrule, commonly in arrays of 12 or more. Such fibers are often used for high-bandwidth optical parallel connections in, e.g., data centers and servers.

Physical contact (PC) -- All of the above connectors are available in three forms of physical contact of the ferrule, see Figure 18. PC indicates that the connectors are designed such that fibers are brought into physical contact upon coupling a connector to a photonic device or another connector. As a result of polishing, the air gap in a fiber connection is diminished, thus improving the transmission of light through the connection (and reducing the back-reflection). Ordinary PC connectors have a back reflection below $-40~\text{dB}$. As variations, ultra-physical contact (UPC, connector colored blue) and angled physical contact (APC, colored green) connectors are available. UPC connectors have a finer polish which decreases back reflection to below $-50~\text{dB}$. APC connectors are polished under an angle of $8^{\circ}$, which changes the direction of the reflected light. As such, back-reflection of less than $-60~\text{dB}$ is achieved. Improving physical contact comes with increased cost: PC connectors are cheapest, and connector cost depends on construction and required return loss. Importantly, it should be noted that PC and UPC connectors are compatible, whereas APC connectors are incompatible with both. Connecting PC and APC fibers leads to a large air gap between the fibers, see Figure 15-(c), and thus to large losses.

9Fiber components¶

Now that a solid background on the topic of optical fibers has been established, we move on toward optical fiber setups. These setups contain fibers as a means to transmit light, and contain fiber components to manipulate light. Examples of light manipulation are splitting or combining light, influencing its polarization or changing its intensity by attenuation or amplification. This section describes components commonly found in fiber setups.

In order to couple light into or out of fiber setups from free space, collimators are used as a gateway. Collimators contain a positive lens and are attached to a fiber with its end placed in one of its focal points. As such, a collimated beam of light, such as laser light, can be coupled into the fiber. The opposite process can be used to couple light out of the fiber setup into free space and yields a collimated beam. This is illustrated in Figure 19. In case the collimator is used to couple light out of a fiber, the beam width (half the beam diameter) of the resulting collimated beam can be reasonably approximated as

where $f_{\text{coll}}$ is the focal length of the collimating lens. $w_{\text{coll}}$ is in the order of $1~\text{mm}$ for a typical fiber NA of 0.1 and an $f_{\text{coll}}=1~\text{cm}$. On the other hand, it should be realized that upon coupling light into a fiber from a collimated beam, the ( incoming) beam width should be smaller than the value resulting from Eq. (26). If the beam is broader, some light will not couple into the fiber and is therefore lost. This results from $\bar{\theta}_{\text{e,c}}$ (see Figure 1) being exceeded by part of the light beam.

Light propagating in fibers can be split in (un)equal parts and light from multiple fibers can be combined into one fiber. Devices for these goals are called splitters and combiners. These belong to the general class of couplers, in which light from $N_{\text{in}}$ input channels is redistributed over $N_{\text{out}}$ output channels. Couplers are based on the phenomenon of evanescent fields. Although light rays are confined to fiber cores, their associated EM-fields extend in the fiber cladding, the evanescent field, as briefly mentioned in Section 3 and illustrated in Figure 20. If two fiber cores are brought in close proximity, energy from the evanescent field from light transmitting through the one fiber core enters the other fiber’s core and continues its path there. The coupling fraction depends on core separation, interaction length, wavelength, and mode properties. Wavelength-division multiplexing requires components designed to combine or separate wavelength channels; a wavelength-independent splitter alone does not perform that task.

In isolators, light is able to propagate in one direction, but not in the other. As such optical devices can be isolated from each other, with the result that light reflected from one device cannot back-propagate to another device. This is important to protect e.g. a laser from incoming light back-reflected from another device. A common free-space isolator combines a nonreciprocal Faraday rotator with polarizing elements: forward light passes, while reverse light is rejected or directed to an absorber.

Circulators also use nonreciprocal routing. These components can have three or more ports and light coupled into one port is only transmitted to one other port. Thus, in a three-port circulator, port 1 is coupled to port 2, 2 to 3 and 3 to 1. Using circulators, a single fiber can be used to transmit and receive a (reflected) signal. Both are then separated using this component, which is relevant for Fiber Bragg gratings (discussed in Section 10) and Michelson-Morley-type interferometers.

In some applications it is necessary to attenuate or amplify the light power transmitting through fibers. This would be the case if power levels are too high or low for a detector or for another component further downstream in the fiber network. Also, attenuators and amplifiers may be used to stress-test a network under (unwanted) power loss or gain, and amplifiers in particular can be used to compensate for fiber losses in long-distance fiber connections.

Optical attenuation can be achieved by the methods that have so far been considered as a negative influence on light transmission: scattering, absorption and air gaps, see Section 6. Attenuators can either be fixed or variable in their attenuation. The latter can be tuned continuously or step-wise by hand or electronically to the desired attenuation level. This allows for testing a network under variable attenuation level without making hardware changes. Reflection performance depends on the attenuator design; a return-loss specification is needed when back-reflections matter.

For optical amplification, an intense pump signal is coupled into an amplifier, along with the signal-to-be-amplified. In a rare-earth-doped fiber amplifier, an optical pump laser supplies this energy. Due to stimulated emission in the amplifier, some energy supplied by the pump is transferred to the signal-to-be-amplified, thus increasing (amplifying) its intensity. An erbium-doped fiber amplifier (EDFA) is commonly optimized for signals around 1530 to $1565~\text{nm}$, with some designs extending into the L band. Optical pumping excites erbium ions, enabling stimulated emission that amplifies telecom signals near $1550~\text{nm}$. However, fiber amplifiers based on other rare-earth dopants are also available, in particular neodymium, ytterbium, praseodymium, or thulium.

With polarization controllers the polarization of the light within an optical fiber can be influenced. These controllers consist of two or three ‘‘flaps’’ around which the fiber is wound, see Figure 21. By adjusting the position of these flaps, either manually or electronically, the polarization state of the propagating light is controlled as a result of stresses applied to the fiber. A suitable controller can transform the output polarization over a useful range; complete control depends on the number of paddles and their adjustment.

10Applications of fiber optics and future directions¶

Nowadays, fiber optics is applied in many fields, both in industry and in academia. In this section we give a (non-exhaustive) overview of these fields as well as a hint in which direction the field of fiber optics is moving.

The main field in which fiber optics is applied is the field of telecommunication. Fiber optics allows for reliable communication over large distances with little loss of power, as we have seen throughout this chapter. Fiber optics is replacing conventional electronic (copper) technology, which is subject to different bandwidth, reach, and power constraints, first in intercontinental communication lines and nowadays in local communication lines. Also data centers have transferred or are transferring from copper technology to fiber optics, which can change link power consumption depending on the application. Also, fiber optics can be used as a medium for quantum communication, which promises secure communication based on the laws of quantum physics.

Apart from telecommunication, fiber optics can be applied in sensing systems. Fiber Bragg Gratings (FBGs) have been developed to measure strain (due to mechanical, temperature and pressure loading) in structures and materials. In FBGs, the fiber core consists of periodically interdigitated sections with refractive index $n_1$ and $n_3$, see Figure 22, such that at each $n_1$-$n_3$ boundary some light is reflected. Reflections from successive index modulations add constructively near the Bragg wavelength $\lambda_{\mathrm B}=2n_{\mathrm{eff}}\Lambda$, where $\Lambda$ is the grating period and $n_{\mathrm{eff}}$ is the effective index of the guided mode. Upon attaching an FBG to a structure, it stretches or shortens due to strain and accordingly, the reflection peak shifts in wavelength. The same goal can be achieved using fiber-optic interferometers.

Another application of fiber optics in sensing is given by evanescent wave sensing. Evanescent waves have been briefly introduced in the context of couplers in Section 9. Tapering a fiber can increase the overlap of the guided mode’s evanescent field with its surroundings; changes in absorption or refractive index can then be detected through transmitted or reflected light. With this technique optical biosensors, water quality sensors and chemical sensors have been developed and the toolbox of evanescent sensors is still increasing.

Fiber lasers have been developed in which the gain medium is formed from a fiber doped with rare-earth elements, such as erbium. Since the gain medium is a fiber, it is flexible and therefore fiber lasers are easier to use as they allow for more straightforward delivery of power in a desired direction. This makes fiber lasers useful in processes such as cutting and welding. Their power and beam quality depend on the particular design and application. Fiber systems can produce narrowband or broadband output, including supercontinuum light in nonlinear fibers. The development of fiber lasers is still ongoing.

Other areas of current and future development of fiber optic applications are photonic integrated circuits that integrate electronic and optical devices. Multicore fibers are being developed to further increase information density and microstructured fibers such as holey fibers and PBFs are further investigated and developed in order to find designs with optimal properties in terms of loss and dispersion. The same holds for evanescent wave sensing and FBGs for which new applications are being explored continuously, e.g. in the sectors of biosensing and health monitoring (in particular point of care diagnostics).

With new applications of fiber optics arising in sectors such as agriculture, food, space, defense and energy, the field is steadily expanding. It holds promise for developments making processes in those sectors easier to measure, more cost effective and more energy efficient. This makes the field very interesting to follow in the upcoming years.

11Chapter Summary¶

• Total internal reflection confines light in optical fibers when $n_{core} > n_{cladding}$.

• Fiber modes are discrete propagation patterns satisfying self-consistency; determined by the V-number: $V = \frac{2\pi a}{\lambda}\text{NA}$.

• Single-mode fibers ($V < 2.405$) support only the fundamental mode; multimode fibers support many modes.

• Numerical aperture satisfies $n_{\mathrm e}\sin\theta_{\mathrm{e,max}}=\sqrt{n_1^2-n_2^2}$ for meridional rays; the usual air-side NA equals the right-hand side.

• Dispersion causes pulse broadening: modal dispersion (multimode), material dispersion, and waveguide dispersion.

• Fiber losses arise from Rayleigh scattering, absorption, bending, and coupling; minimized at $\lambda \approx 1550$ nm in silica.

• Step-index fibers have uniform core index; graded-index fibers have gradual index variation reducing modal dispersion.

• Photonic crystal fibers (holey fibers, photonic bandgap fibers) offer unique dispersion and single-mode properties.

• Fiber connections: Connectors (LC, SC, FC) for temporary connections; splicing (fusion or mechanical) for permanent joins.

• Fiber components: Couplers/splitters, isolators, circulators, wavelength-division multiplexers, and fiber amplifiers (EDFA).

• Applications: Telecommunications, fiber sensors (FBGs), and fiber lasers.