Professional Documents
Culture Documents
Miroslaw Jonasz
MJC Optical Technology, Beaconsfield, Quebec, Canada
Handbook of Measuring System Design, edited by Peter H. Sydenham and Richard Thorn.
2005 John Wiley & Sons, Ltd. ISBN: 0-470-02143-8.
594 Measuring System Design
100
1.2.1 CW arc lamps
10
In arc lamps, optical radiation is generated by an electric
1 discharge (as an arc) through ionized gas.
The spectral output of an arc lamp (Figure 1) consists
0.1 of a broad, blackbody-type output and sharp peaks at
wavelengths that depend on the gas fill. Gases used in these
0.01
lamps are typically deuterium, mercury vapor (Hg), xenon
0.001 (Xe), and a Hg–Xe mixture. The deuterium lamp outputs a
smooth spectrum usable from about 160 (envelope limited)
0.0001 to about 400 nm with spectral peaks in an essentially
100 1000 10 000 100 000
Wavelength, nm unusable, owing to low spectral power density, visible
spectral range. The Hg and Xe lamps (envelope-limited
Deuterium Hg arc (peaks in UV and VIS)
TH TH envelope useful range of 250 to 2500 nm) have major peaks at 365.0,
N laser (337 nm) Ar-Kr laser (488-647 nm) 404.6, 434.8, and 546.1 nm (Hg) and between 750 and
CO2 laser (10200 nm) 1550 nm diode laser
1000 nm (Xe). Owing to a high gas pressure in these lamps
Xe arc (peaks in IR) (of the order of 10 to 100 atm at the operating temperature),
SiC
He-Ne laser (633 nm) the spectral peaks are significantly collision broadened (on
the order of several nanometers).
Figure 1. Spectra of representative light sources. Thermal
(broadband) light sources spectra are based on the data The generally small arc (on the order of 0.1 to 1 mm)
from a catalog of Oriel Instruments (now Spectra Physics, allows relatively efficient collimation or focusing of the
www.spectra-physics.com): 150-W Xe arc lamp, 200-W Hg arc lamp radiation. Compact Xe arc lamps containing an inte-
lamp, 30-W deuterium lamp, 100-W tungsten halogen (TH) lamp grated sealed-in paraboloidal reflector that produces a ‘col-
(dashed line above about 5000 nm represents the output of the limated’ beam with a moderate divergence are commer-
lamp envelope), and 50-W silicon carbide envelope-less infrared
element. Wavelength ranges of the lamps’ spectra are limited
cially available (CeraLux, Luxtel, www.luxtel.net). The
by the envelopes – in the case of a quartz envelope to a range arc position in an arc lamp varies in time (arc wander) and
of about 200 to between 2500 and 3500 nm. Spectral lines of may cause variations in the output power, especially when
few major fixed-wavelength laser types are indicated symbolically this output is collimated or focused.
with short dashed vertical lines. The vertical range of these lines
does not represent the magnitudes of the irradiances that are
many orders or magnitude higher than the maximum of the graph 1.2.2 Xe flash (pulsed) lamps
scale. For example, a typical 1-mW HeNe laser (λ = 633 nm,
λ = 0.0002 nm, beam diameter 1 mm2 ) produces irradiance on These arc lamps range from small flashlamps for instru-
the order of 5 × 109 mW m−2 nm−1 . ments, to large flashlamps such as those used as pump
Light Sources and Detectors 595
sources in lasers. The flash (of the order of microseconds a discrete wavelength and irradiance pattern in a plane
long) is produced by discharging a capacitor. Flash repeti- perpendicular to the optical axis of the cavity. The gain
tion rates are on the order of 100 pulses/s. medium is selected so that its absorption band overlaps with
the wavelength of at least one of the cavity modes and that
the radiative energy transition is the most probable transi-
1.3 Infrared incandescent sources tion corresponding to that absorption band. In lasers that
employ gain media with wide spectra, prisms or gratings
The radiating elements of these sources are made of high-
are used to select a subrange of the gain medium spectrum.
temperature-resistant materials such as silicon carbide (Glo-
By ‘pumping’ the gain medium with an electric dis-
bar), and a sintered mixture of oxides of zirconium, yttrium,
charge, current, or light, one excites electrons of that
thorium, or erbium (Nernst glower). These elements, with
medium to an energy state. Electrons falling (initially spon-
sizes on the order of several millimeters, are heated resis-
taneously) from that state to a lower one emit light of
tively to a temperature on the order of 1000 to 2000 K. The
various radiation modes. Competition, via stimulated emis-
radiating elements are operated in the air with no envelope,
sion, of modes supported by the cavity for energy supplied
limiting the spectral ranges of these sources (Table 1).
to the gain medium quickly leaves a few dominant modes
Incandescent IR sources have spectral outputs (Figure 1)
inside the cavity. The simplest and commonly provided
similar to that of the blackbody, but their wavelength- mode, TEM00 , has an axially symmetric Gaussian irradi-
dependent emissivity is less than unity, typically on the
ance profile that is frequently associated with the laser beam
order of 0.9. Blackbody sources operable at temperatures in general. The process of generation of light by a laser is
ranging from 400 to 3000 K are brighter (emissivity >0.99), more akin to that used in an electronic frequency generator
but an order of magnitude more expensive alternatives to than to optical noise generation characteristic of a thermal
these IR emitters. light source.
The cavity is deliberately made slightly leaky (<1% loss)
2 LIGHT-EMITTING DIODES (LEDs) in order to let laser radiation out. In pulsed lasers that use
Q-switching technique for generation of high-power mod-
The light-emitting diodes are relatively narrowband erately short pulses (on the order of nanoseconds long),
(20–50 nm), low-voltage (2 V), and low-power sources the cavity quality (Q-factor) is spoiled for a sufficient time
(Table 1) available in a wide spectral range spanning the to build a significant population of electrons in an excited
visible and near IR. White-light LEDs that use a phosphor state and then it is temporarily restored to generate a short
to convert the base emission wavelength range into a broad pulse of radiation. Trains of much shorter pulses (several
range perceived as white light are also available. LEDs femtoseconds) can be obtained by mode locking, that is,
convert electric current (∼20 mA) to light by recombining introducing a fixed phase relationship between the cavity
charge carriers in semiconductors. The semiconductors used modes. The pulse train is a result of interference between
include SiC, GaN, ZnSe, GaInN (blue), GaP (green), GaP:N the locked modes.
(yellow), and AlInGaP, GaAsP (red). The cooperation of the resonant cavity properties, gain
LEDs are available in a wide range of packages, ranging medium absorption profile, and the stimulated emission
from plastic encapsulation to TO-type metal transistor gives the laser beam its directionality, high monochromatic-
cans with flat or lensed windows. The emitting area is ity, and coherence. Stabilized frequency lasers have line
on the order of a fraction of 1 mm in size. Plastic and widths <1 × 10−11 nm (3 MHz), and coherence lengths on
glass lensed packages provide rough collimation of the the order of 1 km. Key characteristics of representative laser
output, but a better result in collimating or focusing a types are summarized in Table 2. Wavelengths of major
LED output can frequently be obtained with flat-window laser types are shown in Figure 1.
packages that generally provide a nearly Lambertian spatial
radiation pattern.
3.1 Gas lasers
Table 2. Basic data on major laser types listed according to the increasing first wavelength of the spectral range. CW denotes continuous output. This table is mainly
based on the Laser FocusWorld 2003 Buyer’s Guide. Additional data obtained from specifications of selected manufacturers have been also used. Users of this table should
exercise caution when applying the data given.
Wavelengths (nm)m Laser type Power (W) or pulse Pulse repetition Pulse width (s) Beam Beam
energy (J) rate (1 s−1 )l sizea divergence
(mm) (mrad)
Measuring System Design
154–532f , (1064), 2950 Nd:YAGc (pulsed) 0.001–180 J 1–50 × 103 20 × 10−12 –10 × 10−3 1–14 0.5–130
157, 193, 248, 308, 351 Excimerb (pulsed) 0.001–0.2 W 20–2 × 103 0.01–30 × 10−9 2.5 × 22 0.3–4
190–18000h,k Tunable solid state (pulsed) 0.1 × 10−9 –100 J 0–90 × 106 17 × 10−15 –100 × 10−9 3–10g 0.5–1.5g
197–4500h Tunable dye (pulsed)g 30 × 10−6 –0.2 J 1–10 × 103 5 × 10−12 –20 × 10−9 1–3 × 6 5
202–1090i (514) Argonb (CW) 2 × 10−3 –25 W – – 0.6–3.5 0.05–3
266–532f (1064), 1320 Nd:YAGc (CW) 0.01–5.4 × 103 W – – 0.7–6 1–200
(337) Nitrogenb,g (pulsed) 0.1 × 10−3 –0.5 × 10−3 J 1–100 ∼1 × 10−9 1×2 2×3
375–760h Alexandritec,g (CW) 0.01–0.15 W – – 1 0.1
375–2900h (500–1550)h Semiconductor diode (CW) 0.1 × 10−3 –4 × 103 e W – – – 1–17 ×
1–17j,d ,
87 × 870d
380–1935h (600–1550)h Semiconductor diode (pulsed) 5 × 10−12 –102 e J 100–100 × 106 15 × 10−12 –0.8 × 10−3 – 0.3j , 87 × 785d
543, (632.8), 1152, 3392 HeNeb 0.1 × 10−3 –35 × 10−3 W – – 0.5–2.7 0.7–4.5
690–2000h Ti: sapphirec,g (CW) 0.05–3.5 W – – 0.6 1.7
694 Rubyc,g (pulsed) 0.08–20 J 0.005–0.1 20 × 10−9 –2.5 × 10−3 7–25 0.2–1
3000–25 000 Lead salt diodeg (CW) 0.1 × 10−3 W – – – 870
5200–6500h COb,g (CW) 10–1 × 103 W – – 5–16 4
9000–11 400 (10 600) COb2 (CW) 0.04–12 × 103 W – – 0.5–22 0.5–9
9000–11 400 (10 600) COb2 (pulsed) 0.5–60 J 0–130 × 103 0.3 to over 10−3 1.8–30 × 30 1–10
a
If one value is given, it is the TEM00 (Gaussian) beam diameter.
b
Gas laser.
c
Solid-state laser.
d
In two perpendicular dimensions.
e
The upper end of the range refers to the output of laser stacks.
f
Through harmonic frequency conversion.
g
Few data.
h
Continuously tunable, may be tunable in a subrange of the stated range.
i
Line selection.
j
Diode lasers equipped with collimating optics.
k
Line widths from several GHz down to 1 kHz.
l
A repetition rate of 0 denotes a single-shot mode.
m
Dominant wavelengths in parentheses.
Light Sources and Detectors 597
3.4.1 Fixed-wavelength lasers In a thermal detector, the active element is heated via
absorption of light. This heating induces electrically sensed
Solid-state lasers (SSLs) utilize a solid gain medium, per- changes in that element’s properties. Thermal detectors
mitting high-quality beams, and various pumping mech- have a flat spectral response in a broad spectral range, deter-
anisms. The first solid-state laser used the ruby crystal mined by the absorption spectrum of the sensing element
as the gain medium and a flashlamp as the pump. SSLs material or its absorptive coating that converts the energy
are also pumped by other lasers and by semiconductor of the incident radiation to heat. The detector spectral range
diode lasers (diode-pumped solid-state lasers – DPSSL). is limited/modified by the detector window/filter. A weak
In addition to ruby, the gain media include neodymium- thermal link to a heat sink allows the detector to correctly
doped (Nd:YAG, Nd:YLF, Nd:YVO4 ) and holmium-doped respond to the incident radiation power. Thermal detectors
(Ho:YAG, Ho:YLF) crystals and glasses, alexandrite and are generally less expensive than photon detectors, and do
titanium-doped sapphire (Ti:sapphire). As in dye lasers, the not require cooling when operated in the infrared. However,
wide absorption bands of the solid gain media support the thermal detectors are typically slower and have a lower
generation of extremely short light pulses. sensitivity than photon detectors.
These lasers emit in the red and near IR range. Radiation The basic types of thermal detectors are:
in the near-UV and blue/green region of the visible spec-
trum can be generated by the harmonic up-conversion (for • thermocouples
example, 532 nm in an ND:YAG laser). • thermopiles
598 Measuring System Design
Table 4. The first order of magnitude incident radiation power ranges that can be measured by the various detector types. The minimum
power is the detector noise equivalent power at the wavelength of the maximum responsivity. PMT = photomultiplier, APD = avalanche
(semiconductor) photodiode.
Approximate power rangeh , W Detector type Approximate spectral rangeb
Minimum Maximum
10−19 a 10−12 PMT: photon countinge 100–900 nm
10−19 a 10−12 APD: photon countinge 200–2500 nm
10−15 10−9 c PMT: analog detectione 100–900 nm
10−13 10−6 APD: analog detectione 200–2200 nm
10−12 10−1 Semiconductor photodiodee 0.1 nm–5 µm
10−12 1 Semiconductor photoconductore 400 nm–150 µm
10−9 10−3 Thermopilesf 200 nm–200 µm
10−9 10−3 Pyroelectricf 200 nm–1 mm
10−8 d 1 Bolometerf 0.1 nmg –5 mm
a
This power corresponds to ∼1 photon/s in the visible and near infrared.
b
Various detectors of the stated type (except thermal detectors) may have their maximum sensitivity located at different wavelengths in this range and
be sensitive each in a subrange of this range.
c
A much higher maximum power (about 7 mW) was measured with short (∼1 µs) light pulses (Lasher and Redhead, 1963) at the cost of high mortality
of the PMTs.
d
down to 10−17 in the far infrared at a temperature close to 0 K.
e
Photon detector.
f
Thermal detector.
g
Can be used as a photon counter/spectrometer for X rays.
h
The dynamic range of a detector may be smaller than the min–max range.
that collect and multiply the photocurrent generated at the the MCP multiplier, the imaging of the input electron irra-
cathode. A high voltage of the order of 1000 V is dis- diance distribution is ensured by restricting each electron
tributed between the electrodes of a PMT by a voltage cascade to its own microchannel and by placing the photo-
divider network. A photon striking the photocathode ejects cathode, MPC, and the anode-phosphor screen close to each
an electron with the quantum efficiency (QE) of less than other (<1 mm). This creates compact devices with imaging
one-fourth. That electron is accelerated by the potential dif- resolutions of the order of 50 line pairs per mm. Photon
ferences between the cathode and the following electrode gains up to the order of 109 are possible with devices that
and – upon impacting that electrode – ejects several next- employ stacks of two or three MCPs.
generation electrons. This electron-multiplication process The IMI is by itself an optical-to-optical converter and
continues for each following dynode until the anode, where requires an array detector such as a CCD (discussed
electrons are collected, and results in a total electron gain on in Section 6.2) to convert an optical image to electron-
the order of 106 to 108 . A high gain and a low dark current ically processable image. Phosphor-less versions of an
on the order of 1 photoelectron/s (and less when cooled) image-intensified CCD are also available. In these imaging
enable the detection of single photons of the incident light. detectors, the phosphor screen is replaced with a CCD
Classical PMTs are available generally in two forms: embedded inside the vacuum tube.
Table 6. Basic characteristics of typical photovoltaic semiconductor detectors at room temperature, unless indicated otherwise.
Photosensitive Spectral range Wavelength(s) of Maximum Maximum
material (µm) maximum responsivity, quantum
responsivity (µm) A/W efficiency
Si 0.2–1.2 0.8–0.95 0.7 0.9
GaAs 0.3–0.75 0.6–0.72 0.4 0.7
Ge 0.8–1.8 1.55 0.9 0.7
InGaAs 0.7–1.7 1.6 1 0.8
0.7–2.6 2.3a 1.1 0.6
InAs 1–3.1 3b 2 0.8
InSb 1–5.5 5.2b 2 0.5
a
At 253 K.
b
At 77 K (liquid nitrogen).
that is relatively insensitive to temperature changes, except quantum efficiency of the CCD array as compared to that
near the long-wavelength cutoff (Figure 2). of the photodiodes.
Photovoltaic detectors are available both as: Back-illuminated operation of the CCDs compensates
somewhat for this reduction. CCDs tend to suffer from
• unity gain devices (photodiodes, photodiode arrays, ‘blooming’ caused by spilling charge from an overfilled
CCDs, CMOS arrays) pixel to the surrounding pixels (Janesick et al., 1987).
• detectors with internal amplification: avalanche photo- Recently, electron-multiplying CCDs (EMCCDs) have
diodes (APDs) and APD arrays. been commercialized (e2v technologies, www.e2v.com).
These CCDs utilize, in the readout section, an electron-
In the latter detectors, the cascade multiplication of multiplication technique implemented in the APDs. This
photoelectrons is induced by applying a high voltage on allows high gain (single-photon counting) with the low-
the order of several hundred to several thousand volts. In noise characteristic of the CCDs.
the lower part of the high-voltage range, the number of The CMOS imaging arrays utilize silicon as photosensi-
electrons produced is proportional to the applied voltage, tive material and have a similar spectral response to that
as it is in the PMT. In the high part of the voltage range, of silicon CCDs. However, the CMOS arrays use a differ-
the cascade multiplication becomes free running. This latter ent approach to sensing the incident light power and also
mode of operation (Geiger mode), is used for single-photon different manufacturing techniques. In a pixel of a CMOS
counting. In that mode, the electron cascade is quenched to array, a photodiode typically discharges a storage capaci-
enable photon counting at a rate on the order of up to 106 tor, thus preventing blooming. The capacitor charge can be
photons/s (active quenching) limited by dead time and APD directly read out via an integrated amplifier.
heating (for example, Dravins, Faria and Nilsson, 2000). This allows addressing selected sections of the array and
Semiconductor array detectors are the mainstream of enables a high-speed readout. The CMOS arrays employ
imaging detectors. Arrays with pixel sizes on the order of technology used for microprocessors. This permits inte-
10 × 10 µm are available in the photodiode array, CCD, gration of preprocessing electronics, such as an analog-to-
and CMOS formats. The photodiode arrays have individu- digital converter, on the array chip. However, additional
ally addressable elements and offer high quantum efficiency semiconductor/metal structures implementing these func-
but suffer from a substantial read-out noise (for exam- tions take up a significant portion of the area available
ple, Johnson, Edwards and Mendenhall, 1994). CCD arrays for light collection. Complexity of these structures also
have a much lower readout noise but also a lower quan- increases the manufacturing defect rate. Such disadvan-
tum efficiency. In a CCD array, the photogenerated charges, tages have generally limited applications of CMOS arrays
instead of being in real time removed from the sensitive vol- to low-cost and low-quality consumer applications (Tay-
ume as it is done in the photodiode, are stored in a potential lor, 1998).
well until they are transported out with an extremely high
efficiency, along a column or row of pixels that perform
6.2.2 Photoconductive devices
the role of an analog shift register.
Charge integration and transfer are achieved at the rate Photogeneration of charges in semiconductors significantly
of an external clock signal by varying potentials of an array affects its conductivity. This is utilized in junctionless
of electrodes and semiconductor gates located at the CCD semiconductor detectors, photoconductors, for the detection
chip surface. It is these surface structures that reduce the of electromagnetic radiation. Typical semiconductors used
602 Measuring System Design
Table 7. Basic characteristics of typical photoconductive semi- Camparo, D.C. (1985) The Diode Laser in Atomic Physics.
conductor detectors at room temperature, unless indicated Contemporary Physics, 26, 443–477.
otherwise. Donati, S. (2000) Photodetectors: Devices, Circuits and Applica-
Photosensitive Spectral range Maximum tions, Prentice Hall, NJ.
material (max sensitivity) responsivity Dravins, D., Faria, D. and Nilsson, B. (2000) Avalanche Diodes
(µm) (V/W) as Photon-counting Detectors in Astronomical Photometry,
in Optical and IR Telescope Instrumentation and detectors,
CdS 0.4–0.75 (0.56) –b
(eds I. Masanori and A.F. Morwood); SPIE Proceedings, 4008
CdSe 0.6–0.9 (0.71) –b
298–307.
PbS 1–2.7 (2.2) 6 × 105
1–4.2 (3.1)a 9 × 105 a Hagins, W.A., Penn, R.D. and Yoshikami, S. (1970) Dark Current
PbSe 1–5 (4) 7500 and Photocurrent in Retinal Rods. Biophysical Journal, 10,
1–6 (5)a 55 000a 380–411.
HgCdTe 1–4a – 24a 105 a –200a Hamamatsu. (1994) Photomultiplier Tube: Principle to Applica-
a
At 77 K (liquid nitrogen). tions, Hamamatsu Photonics KK, Hamamatsu, Japan.
b
These detectors are used mainly as light switches, because of the Hecht, J. (1992) The Laser Guidebook, McGraw-Hill, New York.
sensitivity of their response to their light-exposure history.
Hudson, R.D. (1969) Infrared System Engineering, Wiley Inter-
Science, New York.
in these photoconductors are CdS and CdSe (visible), PbS Janesick, J.R., Elliot, T., Collins, S., Blouke, M.M. and Free-
and PbSe (near IR), and HgCdTe (IR) (Table 7). Note man, J. (1987) Scientific Charge-coupled Devices. Optical
Engineering, 26, 692–714.
that the semiconductor conductivity is also a significant
function of temperature, so that the detector temperature Johnson, J.B., Edwards, G. and Mendenhall, M. (1994) Low-cost,
High-performance Array Detector for Spectroscopy Based on
must be stabilized.
a Charge-coupled Photodiode. Review of Scientific Instruments,
The literature on optical engineering usually covers 85, 1782–1783.
sources and detectors. A useful general book is OSA (1993).
Lang, R.J. (2002) Semiconductor Lasers: An Overview of Com-
A useful reference work on infrared engineering is Hud- mercial Devices, in Photonics Handbook, Laurin Publishing,
son, 1969. Pittsfield, MA, (pp. H215–H218).
Lasher, D.R. and Redhead, D.L. (1963) High-current Operation
of RCA Type 2059 Photomultiplier Tubes. Review of Scientific
RELATED ARTICLES Instruments, 34, 115–116.
OSA. (1993) The Handbook of Optics, McGraw-Hill, New York.
Article 89, Optical Materials, Volume 2; Article 90,
Taylor, S.A. (1998) CCD and CMOS Imaging Array Technologies:
Optical Elements, Volume 2; Article 92, Optical Mea- Technology Review , Technical Report EPC-1998-106, Xerox
suring Instruments, Volume 2; Article 93, Testing Opti- Corporation, Cambridge, UK, (p. 15).
cal and Other Radiation Systems, Volume 2.
FURTHER READING
REFERENCES
Johnson, C.B. and Owen, L.D. (1993) Image Intensified Elec-
BI. (1980) Photomultiplier Handbook, Burle Industries, Inc, tronic Imaging, in The Handbook of Optics, McGraw-Hill, New
Lancaster. York, (pp. 21.1–21.32).
This article is based in part on a chapter by J. Prasad and G. Mitra originally published in Handbook of Measurement Science Vol 2,
P.H. Sydenham, Wiley, 1983.