FIR - ACRONYM

FIR - Far InfraRed, the long-wavelength end of the infrared spectrum. As is typical of infrared bands, definitions vary.

When Arthur Schawlow and Charles Townes proposed extending the maser principle to frequencies well above the 22 gigahertz (GHz) of the ammonia microwave maser reached in 1954, they targeted frequencies three orders of magnitude higher, near 300 terahertz (THz), corresponding to one micrometer (1 µm) in the near-infrared.  Their choice reflected the technological reality of the time, the long wavelength end of the infrared was a terra incognito of the electromagnetic spectrum, little explored because few sources and detectors were available. One reason that wavelengths longer than about 15 µm came to be called "far-infrared" in the early days of lasers probably was that that part of the spectrum seemed far out of reach.

Technology has come a long way since then, but the far-infrared remains beyond the well-developed parts of the infrared; although, as with other parts of the infrared, the specified wavelengths differ among definitions.

Wikipedia's Infrared article lists multiple definitions of the far-infrared. The first is 15 to 1000 µm, putting it thermal infrared where blackbody emission peaks around room temperature, longer than the mid-infrared and some definitions of the long-wavelength IR. It also cites the International Standardization Organization's definition of 50 to 1000 µm, a definition also used in the McGraw-Hill Dictionary of Scientific and Technical Terms. The infrared article notes that astronomers have their own definition, with 25 to 40 µm the short end and 200 to 350 µm the long end.

Oddly, the bands specified for "far-infrared lasers" differ. Wikipedia says their wavelengths range from 30 to 1000 µm, close to the 40 to 1000 µm I used in The Laser Guidebook. The McGraw-Hill Dictionary lists a far more limited range, from "well above 100 µm" to 500 µm.

A couple decades ago, such inconsistent definitions didn't matter much, because wavelengths longer than 30 µm were a sparsely inhabited part of the spectrum, largely absorbed by air, hard to detect, and even harder to use. Now new technology is upscaling the unfashionable far-infrared neighborhood and redefining it as the terahertz band, which Wikipedia defines as from 100 to 1000 µm, or 300 GHz to 3 THz.

Plot of atmospheric opacity shows the strength of atmospheric absorption in the far-infrared. (Wikipedia art, modified)

LWIR - ACRONYM

LWIR - Long-Wavelength InfraRed: an infrared band at wavelengths longer than the mid-infrared. One common definition is from 8 to 15 micrometers, also known as the thermal infrared, but there is no generally accepted standard.

The infrared spectrum sprawls from the edge of the visible, nominally 0.7 µm, to about one millimeter, such a broad range that it demands subdivision. The definitions of mid- and long-wavelength bands may have grown from the atmospheric transmission windows at 3-5 µm, and 8-14 µm. Those bands generally require different detectors, and also were a handy division between the blackbody peaks of "hot" objects and those of "body-temperature" objects. The strongly absorbed wavelengths in between didn't matter much as long as the infrared was mostly used for looking through the air. Longer wavelengths were lumped as the "far-infrared," a vast region extending to about one millimeter that seemed of little use because atmospheric transmission was spotty and instrumentation was poor.

Atmospheric transmission has not changed, but new infrared detectors and sources have opened up previously little-used regions of the infrared, and satellites have opened the whole infrared spectrum to astronomers. New applications have emerged, such as LWIR monitoring of beehives. That has made drawing dividing lines problematic, particularly on the ends of the LWIR. Should the ends be defined by the atmospheric windows or at some other points? One suggestion was to define each band as an octave wide, spanning a factor of two in wavelength or frequency, but that logical idea failed a crucial practical test because it could not fit both the 3-5 µm and 8-14 µm bands in adjacent octaves. So we're stuck with informal definitions that depend on things like atmospheric windows, and detector ranges, and differ between fields like lasers, astronomy, and night vision.

It could be worse. Geologists built their time scale for the Earth's history on the boundaries between solid rocks, then found that their calendar changed every time a better way was found to date the rocks.

MIR - ACRONYM

MWIR or MIR - Mid-Wave-InfraRed or Mid-InfraRed, a range of wavelengths nominally in the middle part of the infrared spectrum, defined in a variety ways.

The terms hiding behind acronyms sometimes can be more confusing than the acronyms themselves. The diverse definitions for "mid-infrared" illustrate the problem. The name clearly implies that MWIR/MIR should be  the middle of the infrared spectrum, but the infrared is a vast region, sprawling from 700 nm at the long end of the visible range to around 1 mm at the upper end of the radio/microwave band. What is the middle?

On a logarithmic scale, it seems simple enough--10 to 100 µm fits right in the middle. But nobody uses that definition. Wikipedia, today's great arbiter of popular culture, defines the MWIR as 3 to 8 µm, but that section of the "infrared" article bears a warning dating from July 2006 that it needs to be cleaned up! The 3–8 µm definition follows the recommendations of the International Commission on Illumination, which considers the MWIR as the short end of the IR-C band from 3 µm to 1 mm.

However, other definitions abound. In its tabulation of optical spectrum bands, ISO, the International Standardization Organization, defines MWIR as 3-50 µm. NASA's Infrared Processing and Analysis Center at Caltech defines the astronomical mid-IR as stretching from 5 µm to 25 or 40 µm. Military system designers traditionally define 3–5 µm as MIR, the window used by heat sensors on missile guidance systems. The McGraw-Hill Dictionary of Scientific and Engineering Terms does not list mid-infrared, but defines "intermediate infrared radiation" as from 2.5 to 50 µm.

Who's right? It depends on your viewpoint. In my recent mid-infrared lasers and applications webcast, I spoke from the laser industry viewpoint set the boundaries as 2 to 12 µm--starting beyond the telecommunications band and extending to include the carbon-dioxide laser band, as shown in the image. But when I wrote about uncooled infrared cameras in the April issue, I wrote from the detector viewpoint, and called the thermal imaging band from 7.5 to 14 µm "long-wave infrared."  In truth, detectors used in the 3–5 µm and 7.5- to 14-µm bands differ much more than the laser used. But on reflection I have to wonder why the spectral bands should differ between the light source and the detector.  

APD - ACRONYM

An avalanche photodiode (APD) is a semiconductor photodetector in which incident light generates a photocurrent, which is then multiplied by an avalanche process to give a stronger signal.

An avalanche photodiode is a single device that incorporates two distinct semiconductor stages. The first is a photodiode detector, in which light with energy above the bandgap of a semiconductor delivers enough energy to valence electrons for them to enter the conduction band. The electron leaves behind a hole in the valence band, which also functions as a current carrier. Application of a voltage across the device pulls the electrons and holes in opposite directions.

The second stage applies a strong reverse bias across the semiconductor to accelerate electrons. When electrons reach high enough velocities, they can ionize other atoms in the semiconductor, producing an avalanche of electrons. This multiples the original photocurrent and produces a much stronger response than a simple photodiode. Typically, silicon APDs are biased with about 100 V to multiply photocurrent by around a factor of 100. Further increasing the reverse voltage increases the dark current as it approaches the ionization threshold of the semiconductor. Breakdown occurs at a reverse bias of about 150 to 200 V for silicon, depending on device design, and at different voltages in other semiconductors.

You can think of APDs as solid-state counterparts of photomultiplier tubes (PMTs), but they are far from plug-in replacements. As you expect from solid-state devices, APDs are much smaller and their bias voltages are much lower--about a tenth of those in PMTs. However, PMTs are less subject to noise, their gain does not depend as strongly on bias voltage, and they can be engineered to respond to different wavelengths than APDs, so they are among the few survivors of the vacuum-tube era.

New designs and new materials are extending the range of APDs for applications ranging from single-element fiber-optic detectors to focal-plane arrays for imaging. Germanium/silicon ADPs have reached a gain-bandwidth product of 105 GHz, attractive for high-speed optical interconnects. APD elements designed for single-photon counting can be assembled in arrays to make a single-photon counting camera.

Gain or multiplication factor of an APD increases sharply as voltage approaches breakdown, but so does dark current, limiting usable gain. (From Jeff Hecht, Understanding Fiber Optics: 5th edition [Pearson Prentice-Hall, 2006])

PMT - ACRONYM

PMT - Photomultiplier Tube, a vacuum tube light sensor in which input photons cause a cathode to emit electrons that are amplified through a chain of electron amplifiers called a multiplier. Invented in 1934, PMTs are still in use. They offer high gain, low noise, reasonably fast response, and a large collecting area, all important for detecting faint signals.

First observed in 1887 by Heinrich Hertz, the light-induced emission of electrons became an important puzzle because classical physics could not explain why electron emission occurred only for wavelengths shorter than a threshold value rather than depending on light intensity. Albert Einstein won the Nobel prize in 1921 for showing that the photoelectric effect was due to photons needing to have a threshold energy to free electrons from atoms.

The first photoemissive detectors were vacuum photodiodes, in which light illuminated a metal cathode, freeing electrons collected by an anode when a voltage was applied across the tube. Alkali metal cathodes were used to detect visible light. In the days before semiconductor electronics, these devices were called photodiodes or photocells. They were too insensitive for use in early electronic television cameras, so engineers added amplification stages, called dynodes, inside the tube; electrons collided with the dynodes, producing additional or secondary electrons. A series of acceleration stages and dynodes multiplied the photocurrent, thus earning the name photomultiplier. Developed in the 1930s, PMTs became the detectors of choice for applications demanding high sensitivity, low noise, and high speed.

This high performance has made the PMT a remarkably durable technology, one of the last vacuum tubes that is still a standard product in the age of solid-state photonics. Continuing refinements in design and packaging have adapted PMTs for modern applications such as single-photon counting. PMTs continue to new challenges, such as silicon photomultipliers, containing a hundred to several thousand tiny avalanche photodiodes (APDs) connected in parallel for single-photon detection, also called multichannel APDs. But PMTs just keep plugging along.

Modern metal-channel dynode PMT, shown in cutaway. (Image courtesy of Hamamatsu)

WDM - ACRONYM

WDM: Wavelength-division multiplexing, transmission of separate signals simultaneously at separate wavelengths through the same transmission medium, usually an optical fiber.

Multiplexing combines two or more signals for simultaneous transmission through the same medium. It was first used to increase capacity of 19th century telegraph wires. Later, frequency-division multiplexing divided the radio spectrum into separate broadcast channels. Each station was assigned a fixed transmission frequency, and listeners tuned the frequency of their receiver to select a station. Frequency-division multiplexing let cable television networks pack many video channels into frequency slots for transmission through copper cable.

Optics specialists think of wavelength rather than frequency, so when Bell Labs tried sending multiple laser wavelengths through the same hollow light pipe in the 1960s, they called it wavelength-division multiplexing. The appeal was understandable, and the Bell System tried WDM again in 1980 when it designed its first high-capacity fiber-optic system along the Northeast Corridor from Boston to Washington. But that system had a fatal flaw; it used multimode fibers, so every seven kilometers it needed a repeater to separate the wavelengths, detect the signals separately, amplify each one electronically, and drive separate transmitters that had to be multiplexed together. Single-mode fiber won hands down.

Wideband erbium-doped fiber amplifiers (EDFAs) revived interest in WDM because they could amplify many separate laser signals near 1550 nanometers. For long-haul, high-speed transmission, optical channels were packed close together for dense-WDM or DWDM systems. However, a committee from the International Telecommunications Union, apparently dominated by radio engineers, specified DWDM channels in frequency units. Typically 50 gigahertz wide, those channels initially transmitted 2.5 or 10 gigabits per second, and now can carry up to 100 GHz using coherent transmission.

Coarse-WDM (CWDM) came later, allowing the use of lower-cost multiplexing and demultiplexing optics in lower-speed, shorter-distance WDM systems, such as dividing 10-Gigabit Ethernet traffic among four lower-speed CWDM channels. Optical engineers apparently won on the ITU committee handling CWDM, because they specified CWDM channels in wavelength, setting center wavelengths every 20 nm from 1270 to 1610 nm.

CWDM and DWDM grids. The wavelengths for the DWDM grid are approximate; the specification are defined in terahertz.

CPA - ACRONYM

CPA - Chirped Pulse Amplifier, an optical amplifier that generates very high peak powers in very short pulses by stretching the pulse duration before amplification, then compressing the pulse after amplification. Chirped pulse amplification can produce peak powers in the terawatt range from small systems, and is vital in building petawatt lasers.

Nonlinear effects inherently limit the amount of optical amplification possible in a gain medium. Effects such as Brillouin scattering reduce gain, and effects such as self-focusing can cause optical damage. These effects are proportional to the peak power in the medium, putting an upper limit on the gain possible.

Chirped pulse amplification circumvents this limit by spreading the energy in the pulse over a longer period of time, thus reducing the peak power throughout the longer pulse. This is done by sending the input pulse through a medium with a high wavelength dispersion, such as a pair of gratings or prisms, or a length of dispersive optical fiber. The pulse that emerges from the dispersive medium is chromatically dispersed, with the short wavelengths at one end and the long wavelengths at the other. The degree of dispersion depends both on the medium and the spectral width of the pulse. In practice, chirped pulse amplification works best with pulses lasting tens to hundreds of femtoseconds, which are inherently broadband.

The longer dispersed pulse is amplified in a broadband gain medium, then passed through a medium with dispersion of the opposite sign, so the wavelengths that passed first through the amplifier are delayed and those that passed through the amplifier later in the pulse can catch up. The pre-amplification and post-amplification dispersion do not have to cancel each other out, although the minimum pulse duration still depends on the spectral bandwidth.

Chirped pulse amplification also can be used in optical parametric amplifiers, which have broader bandwidth than laser oscillators and thus can be chirped more strongly to generate higher peak powers. Optical parametric chirped-pulse amplification (OPCPA) will be used in Europe's Extreme Light Infrastructure.

How a CPA works. A pair of gratings that delay the blue end of the spectrum stretches input pulses about a factor of 1000 in duration. Those pulses then pass through a broadband amplifier, and the higher-power output is compressed by a second pair of gratings that delay red wavelengths to produce a high-energy ultrashort pulse.