In the previous article, I briefly explained and compared two types of optical sources used in transmitters: LED and LD. Today, I am going to discuss what happens at the other end of a fiber link -- detectors. Optical detectors, as the name implied, can detect the amount of light received. Our very own eyes are a pair of detectors as they can receive light information with the retina and transmit that light data to our brain. In the visible light spectrum, our eyes are great detectors to inspect fiber break or light leakage. However, most fiber works in the invisible wavelength spectrum where human eyes won't be able to see. That is the where the optical detectors come in.
Photoelectric effect
It is impossible to explain how optical detectors work without mentioning the photoelectric effect. To put it simply, the metal will release electrons if it is hit by photons1
This phenomenon was first observed by German scientist Heinrich Hertz who only published his observations. It was Albert Einstein who later studied this effect and quantified the discrete light energy as photons in one of his famous papers that won him a Nobel Prize in 1921. Vacuum photodiodes and photomultipliers take advantage of this technology and can convert the light signal back to electric signals. One critical parameter for characterizing detector is responsivity. It is the ratio of output electric current to the optical input power, with the unit A/W.
In the end, we will compare the responsivity of different detectors and choose wisely based on each application.
Vacuum Photodiode and Photomultiplier
A vacuum photodiode (or phototube) is mainly comprised of a cathode and an anode. When the cathode detects photons, electrons are emitted according to the photoelectric effect, and current will go through the circuit since electrons are attracted to the anode. The following sketch shows how vacuum photodiode works2.
2 https://kids.britannica.com/students/article/photoelectric-device/276409/286736-toc
The limitation of a vacuum tube is that it is physically too big and operates in a wavelength range lower than what fiber communication requires. Another issue is that it also involves much voltage to power it. The typical responsivity of a vacuum photodiode is in the magnitude of mA/W.
Photomultiplier, on the other hand, works more efficiently because of its built-in gain mechanism. In addition to the anode and cathode, it also has a series "dynodes" for accelerating the electrons. The following illustration shows the simplified circuitry of a photomultiplier3.
Just like in vacuum tube, electrons are radiated after photons got absorbed by the cathode. However, the emitted electrons are attracted by intermediate dynodes which have very high voltage. What is so good about dynodes is that there can be more than one electron gets emitted when only one electron is attracted to it. This is called secondary emission caused by high kinetic energy electrons possess. Each electron now becomes more than one electron after hitting each dynode, causing a series of multiplying which eventually leads to electric signal amplification.
The gain at each dynode is about 5, so if there are 3 dynodes in the tube, the total increase will 125 (5x5x5). In reality, there are typically 5 to 10 dynodes in each photomultiplier, so the actual gain is in the magnitude of millions. Photomultiplier tubes are high-speed but also consume hundreds of voltage to power each dynode. It is heavy and big, almost the size of a hand grenade4. Unfortunately, photomultipliers are not suitable for the fiber optic communications.
Semiconductor Photodiode
Just like optical transmitters, the most efficient optical detectors for fiber optic transmission is also made of semiconductor materials. The detailed mechanism of how they work is very complicated, so I am going to omit electric-chemistry and focus on how photons are converted to electrons.
When light strikes onto the p-n junction (a fundamental component of any semiconductor) in the form of photons, each photon must have enough energy to free an electron. A hole will be left behind after the electron departs, and it will move towards anode while the electron moves toward cathode -- electric current is therefore produced. In the image below, each white circle represents a hole, and
The area in between the p-region and n-region is called depletion zone, a high resistance region where photons being absorbed. When some photons hit the depletion region, current is immediately created. Some
PIN photodiode
Between p-region and n-region, another layer named intrinsic layer is added to widen the depletion zone, increasing the probability that photons directly being absorbed in this layer. The following diagram 5shows the extended intrinsic layer between p-region and n-region.
These semiconductor photodiodes are typically made out of Silicon or Germanium and have peak responsivity of 0.5-1 A/W which is considered a huge improvement compared to vacuum tubes. They are also versatile with the optical wavelength ranging from 300nm - 1700nm. The size of each semiconductor is similar to that of an LED which can be easily soldered to a printed circuit board. Also, you don't have to worry about fiber coupling like you do in LED because the detector area is significantly larger than fiber's diameter, leading to almost lossless photon reception.
Avalanche Photodiode
Imagine how nice it would be if you could combine the high gain capability of photomultiplier with PIN diode's responsivity and small footprint? You can check with Avalanche Photodiode or APD which is a variation of PIN diodes with internal gain. Unlike photomultipliers, however, which uses multiple dynodes to accelerate and duplicate electrons, APD's gain mechanism works by applying a large voltage to the electron-hole pairs converted from photons so that they collide with other atoms, knocking more electrons out of them. These new electrons can further collide with more atoms, hence rapidly generating more current in the Avalanche process. Below is a simplified circuit of APD5.
The gain of such APD, although won't be in the millions like a photomultiplier, varies from tens to hundreds. So if a Silicon PIN photodiode has the responsivity of 0.5A/W, a Silicon APD's responsivity will be 75A/W assuming the gain is 150 over the same optical wavelength range.
Summary
Both PIN photodiodes and APDs are ideal for fiber optical transmission, with the PIN being less expensive than APDs. PIN photodiodes can handle most of the situations where the fiber link is short/medium. For longer distances or weak source signal, APD would be the choice since it has greater sensitivity. When choosing an ideal detector, it is important to consider all factors such as working wavelength, light source, fiber length, fiber type, and even fiber material. Below is a selection chart6:
6 Palais, J. (2008). Fiber Optic Communication. 5th ed. Pearson, p.235
600-800nm800-900nm1200-1700nmSemiconductor
The table above is a wavelength guide for choosing a photodetector. Other things to keep in mind are losses, attenuation and the actual amount of power can be detected at the photodiode. In general, the sensitivity of the photodiode determines the type.
This will wrap up the two-part introductory articles on common optical sources and detectors, as well as how to properly choose one for each application and how do they work. M2 Optics is dedicated to providing fiber simulation and latency platforms optimized for different sources and detectors, as well as tools for measuring power and losses.
Since 2001, M2 Optics has been an established manufacturer and innovator of professional optical fiber platforms for fiber network simulation, latency / optical time delay, training, and demonstration applications. Our customer base includes many of the world's most recognized communications service providers, equipment manufacturers, data centers, web service providers, financial institutions, research institutions, and government agencies.
