1470 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 6, NO.

6, NOVEMBER/DECEMBER 2000

A Brief History of High-Power Semiconductor Lasers

David F. Welch, Senior Member, IEEE

Invited Paper

Abstract—The following is a historical perspective of the signif- II. MATERIALS TECHNOLOGIES

icant technological demonstrations that shaped the history of high
power semiconductor lasers. This article is not meant to be a re- By the late 1970s, semiconductor lasers had advanced to
view article, as there are much better review articles and reviewers where double-hetereostructure lasers had been developed
available, nor would the article try to cover all of the contributions resulting in reduced threshold continuous wave (CW) emission
to such a rich technology. Nonetheless, this article will, anecdotally, [5], [6]. In addition advances in laser design included the
present a perspective on the technological advances that resulted
in the enabling technology of high power semiconductor lasers for
breakthrough realization of distributed feedback lasers. Liquid
applications such as fiber optic communications, data storage, and phase epitaxy (LPE) was used to fabricate these lasers [7]–[10],
material processing. however, the performance was limited by the inability of LPE
Index Terms—Optical communications, semiconductor lasers.
to grow uniform thin epitaxial layers and accurately tailored
doping profiles. Nonuniform materials from LPE-grown wafers
resulted in current crowding and optical self-focusing, thus
I. INTRODUCTION limiting the aperture size of the laser to a few micrometers,
and the thick active regions ( 0.5 m) were lossy limiting the
T HE SEMICONDUCTOR laser, first discovered in 1962
[1]–[4], was thought to be a breakthrough invention that
would revolutionize industry. As early as the late 1960s and
efficiency of the laser. The consequence was a laser that could
only operate reliably to a few milliwatts in output power and a
early 1970s there were patents and articles proclaiming the manufacturing process that resulted in low yields.
utility of this technology for optical data storage and fiber The first key technology advancement necessary for the re-
optic and free space communications. However in its early alization of high-power lasers was the development of two new
form, the simple p-n homojunction device was a long way from growth technologies: metallorganic chemical vapor deposition
realizing the dreams of these early inventors. To realize the (MOCVD) and molecular beam epitaxy (MBE). These two key
capability of semiconductor lasers and specifically high-power technological advancements, developed nearly simultaneously,
semiconductor lasers a convergence of many technologies had created a tool that enabled the laser designer to control the
to be realized. Advances in crystal growth technologies, the crystal deposition to atomic layer accuracy which resulted
development of double heterostructure lasers and subsequently in two benefits: uniform material deposition and ultimately
quantum well lasers, materials passivation technologies, quantum well active layers.
heatsinking technologies, pseudomorphic materials; break- Uniform material deposition is critical to both laser perfor-
throughs in device designs including single-mode lasers, laser mance and yields. The more uniform epitaxial layers enabled
arrays, distributed feedback lasers, and the simultaneous devel- the development of large aperture laser structures, the first such
opment of complementary technologies, the most significant device was the evanescently coupled laser array. Large aperture
of which is the rare earth doped fibers for fiber amplifiers devices dramatically broke through the barriers of power output
and fiber lasers, all contributed to one of the most enabling from a single laser. Output powers well in excess of several watts
technological industries today, that of high-power semicon- were demonstrated. Over the course of the next decade, higher
ductor lasers. In many ways semiconductor lasers are second and higher output powers will be developed where outputs ex-
only to the transistor and integrated circuit as to their impact ceeded 10 W CW from a 200- m laser source [11]–[16].
on today’s high-technology market place. The semiconductor The second effect of the conversion from LPE crystal growth
laser is the conduit in which the internet became economically technology to that of MBE and MOCVD was the ability to grow
feasible and is the backbone of which the information age of thin active layers on the order of 10 nm. LPE was typically lim-
tomorrow will depend. The following paragraphs will address ited to active layers on the order of 0.1–0.5 m. Thin active
specifically the technologies associated with the development layers were very difficult to grow by this technique. The con-
of high-power semiconductor lasers from an historical and sequence of a thick active layer laser structure was that there
tutorial perspective. was a high overlap of the optical mode with the free carriers
in the active layer. The propagation loss within the laser, being
dominated by the free carrier absorption in the active layer, was
therefore inherently high and consequently the threshold, effi-
Manuscript received October 9, 2000.
The author is with SDL, Inc., San Jose, CA 95134 USA. ciency of the laser, and the length of the laser cavity is limited.
Publisher Item Identifier S 1077-260X(00)11622-3. Fig. 1 depicts the impact of propagation loss on the efficiency of
1077–260X/00$10.00 © 2000 IEEE
WELCH: A BRIEF HISTORY OF HIGH-POWER SEMICONDUCTOR LASERS 1471

Fig. 1. Effect of distributed loss on a laser with reflectors of 4% and 100%,

an internal conversion efficiency of 100%, and a distributed loss of 15 cm
(typical for thick active layer LPE materials) and 1.5 cm (potential loss for Fig. 2. Relative electrical to optical conversion efficiency.
quantum well lasers).

cleaved facet causing a depletion of charge at the crystal surface.

a hypothetical laser with a device with 4% and 100% reflectors The depleted bands absorb the laser emission that, when the ab-
and an internal conversion efficiency of 100%. sorption is sufficient, causes thermal run away effect, melting
The consequence of a high-propagation loss in the laser cavity the end facet of the laser. COD is especially prevalent in Al
not only impacted the laser efficiency, but it also impacted the containing materials and as a result greatly impacted the relia-
ability to fabricate long cavity length lasers and therefore the bility of AlGaAs lasers, and therefore inhibited the introduction
thermal resistance of the laser. The thermal resistance of the of high-power AlGaAs lasers into applications like communi-
laser is a critical design consideration for high-power lasers. cations and optical data storage.
The short cavity length of LPE grown lasers, typically less than Extensive research by several laboratories was performed
250 m, increased the thermal resistance of the laser, limiting on both materials technology and passivation techniques to
the ability to dissipate power and ultimately the output power of eliminate COD. Several research organizations were able to
the laser. come up with process technology necessary to eliminate COD,
The advent of MOCVD and MBE changed the way the re- each organization having a different approach to the problem.
search community could think about laser designs. The ability Ultimately COD was eliminated as a reliability limitation for
to control the crystal deposition on an atomic scale in a fashion most laser diodes. It was these various techniques that enabled
that resulted in uniform epitaxial growth was the first major high-power lasers for both the communications market and
breakthrough for the development of high-power lasers. With the recordable optical data storage products. Furthermore,
this new tool the research community went to work and the reduction in the impact of COD on reliability greatly improved
first conceptual breakthrough was that of quantum well lasers. the performance of high-power lasers for DPSS applications,
Quantum-well lasers, lasers with active layer thicknesses on the enabling greater penetration of DPSS lasers into the material
order of 10 nm, resulted in a number of advantages including a processing and thermal printing markets.
dramatic reduction in threshold current, a reduction in the free
carrier loss, and a reduction in the temperature sensitivity of the IV. PSEUDOMORPHIC MATERIALS
threshold current. All of these effects increased the efficiency of
The next major technology breakthrough for semiconductor
the laser and the ability to make lasers with longer cavities and
lasers and specifically high-powered semiconductor lasers was
therefore lower thermal resistance. Furthermore, as the overlap
the conceptual development and experimental realization of
of the optical mode with the active layer was much less, the
psuedomorphic materials [17]–[22], otherwise referred to as
power limitation caused by catastrophic optical damage of the
strained layer materials. Up to this time crystal growth was
laser was significantly improved. The net effect was the ability
limited to the material systems that were lattice matched to a
to demonstrate higher output power lasers.
common substrate. In the mid-1980s, a series of conceptual
A significant metric of semiconductor lasers as compared to
developments occurred that resulted in the realization that a
other laser systems, and ultimately the determining factor in
layer need not be lattice matched if its thickness was less than
much of the performance of these lasers in ultrahigh-power ap-
the critical thickness, at which point the material would no
plications such as welding, is the ability to demonstrate elec-
longer be single crystal. Layers that were on the order of 10 nm
trical to optical conversion efficiencies of greater than 60%.
could be grown in the midst of a lattice match layer structure
No other laser media can approach this type of efficiency (see
where the lattice mismatch could be significant. This concept
Fig. 2) and therefore, in high-power applications where heat
lead quickly to the incorporation of In into AlGaAs quantum
generation and removal are the limiting factors, semiconductor
well material structures and was applied to all semiconductor
lasers have the distinct advantage.
laser material structures.
InGaAs active regions in AlGaAs layer structures resulted
III. CATASTROPHIC OPTICAL DAMAGE (COD) in several key benefits: higher gain from the materials, lower
The next limitation to output power for semiconductor lasers threshold current operation, higher efficiency, extension of the
after the initial advances in crystal growth technology was that emission wavelength to longer wavelengths, and higher relia-
of catastrophic optical damage (COD) to the end facet of the bility. For AlGaAs lasers grown on GaAs substrates, emission
laser. COD is a result of surface recombination sites on the wavelengths could then be extended from less than 780 nm to
1472 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 6, NO. 6, NOVEMBER/DECEMBER 2000

Fig. 3. Electroluminescent images of the surface of a broad-area laser operating at 810 and 915 nm. For this experiment defects were intentionally introduced in
the chip and in the case of 810-nm lasers (no In) these defects are seen to propagate through the active region [22].

the realization of high-power, high-efficiency lasers operating

at 630 and 680 nm.
Pseudomorphic concepts have since been applied
to most semiconductor material systems including
GaInAsP–InP laser where efficiency, power, and
polarization effects have been optimized, AlGaInN lasers
for efficient operation in the 380 to 470 nm region and in
AsSb-based materials for lasing properties in the mid-IR.
The above discussed material advances and the resultant laser
design advances created the ground work for a number of critical
applications that were enabled by high-power semiconductor
lasers. As each relevant application is discussed, high power is
qualified as a relative term to the state of the industry prior to
these advances. As an example, high power in the realm of in-
Fig. 4. Far field patterns from a BA amplifier for output powers from 2.8 to dustrial lasers is measured in watts and kilowatts, while high
61 W CW with an injected power of 180 mW. power for optical data storage is discussed as the advances nec-
essary to take the technology from the few milliwatts level to 30
longer than 1100 nm, easily reaching the emission wavelengths mW and beyond. Often the impact of achieving high power di-
necessary for pumping of Er doped fiber amplifiers (discussed in rectly correlates the overall reliability improvement of the laser.
more detail below). It was further shown that the incorporation
of In in the active region of an AlGaAs laser inhibited the migra- V. SINGLE-MODE SEMICONDUCTOR LASERS
tion of defects in the material thus improving the reliability of Advances in materials technology and process technology
the material. From these developments came high-power, highly as discussed above have made dramatic advancements in the
reliable lasers operating at 980 nm and the first short wavelength ability to efficiently generate power at high reliability and across
laser that could meet the 20-year lifetime required for commu- extensive wavelength coverage in semiconductor lasers. There
nication systems. is an entirely different branch of semiconductor laser develop-
Pseudomorphic materials were critical to the development ment associated with the generation of multiwatt output power
of another class of high-power lasers, that of AlGaInP lasers in a single spatial mode. Standard techniques for generating
for the emission between 630 and 680 nm. AlGaInP lasers had single-mode waveguides have been demonstrated in high-power
been demonstrated for operation in the 680-nm region, however, materials resulting in the demonstration of greater than 1 W of
these lasers when lattice matched resulted in high-threshold cur- single-mode operation from a few micrometers aperture. The
rents and limited their output power to a few milliwatts. With motivation to achieve output powers in excess of 1 W from
the introduction of pseudomorphic materials to the design of a monolithic device was driven largely by the perceived need
AlGaInP materials, the gain could be significantly increased for transmitters for free-space communications and direct diode
resulting in lower threshold current densities. The result was material processing applications.
WELCH: A BRIEF HISTORY OF HIGH-POWER SEMICONDUCTOR LASERS 1473

Fig. 5. Infrared image of the top of a broad-area gain region illustrating the effect of filamentation.

Semiconductor lasers, being highly nonlinear devices and ciencies of 30%–50%, the double-clad fiber lasers could operate
having a large coupling between the gain of the laser and at optical to optical conversion efficiencies of 60%–80%. Al-
the index of refraction, made it difficult to fabricate a large though fiber lasers do not result in the broad wavelength cov-
aperture single-mode laser. Several conceptual techniques erage of semiconductor only solutions, fiber lasers have meet
were pursued [23]–[38]: 1) coupled laser arrays through either the power and scaling requirements of most potential multi-
evanescent or direct coupling of individual laser elements; 2) watt single-mode applications. Currently, fiber lasers are used
surface-emitting laser arrays where serial injection locking in marking systems, thermal printing systems, and Raman am-
of multiple cavities were studied; 3) master oscillator power plifiers.
amplifier configurations on a monolithic chip; 4) externally
injection locked lasers; 5) asymmetric gain profiles in large
VI. DIODE-PUMPED SOLID-STATE (DPSS) LASERS
aperture lasers; 6) multimode large aperture devices with
highly differentiated modal gain profiles; 7) external cavity The first use of MOCVD and MBE was in the fabrication
lasers; and 8) unstable resonator lasers with large emitting of AlGaAs lasers operating between 780 and 860 nm. From
apertures. Several general issues were barriers to a number this material system came the first application of high-power
of these potential solutions; first the output power needed to semiconductor lasers, that of pumping Nd : YAG lasers at wave-
be efficiently coupled into a single emission radiation lobe, lengths around 810 nm. The use of diode pumping of Nd : YAG
the radiation patter, or far field pattern, needed to remain lasers enabled a dramatic reduction in size and a significant
stable over all power and temperature operating conditions, the increase in operating efficiency as compared to flash lamp
discrimination between modes of operation had to be sufficient pumped solid-state lasers. In later years as the semiconductor
to insure single-mode operation as the gain uniformity and lasers became more reliable, so did the solid-state lasers. A
index uniformity changed during operation of the laser and the steady progression in DPSS technology over the past 15 years
laser had to be able to be made reproducibly with high yield. has transformed the laser-based material processing industry to
All of these conditions became very problematic in a material where DPSS lasers now compete with high-power CO laser
system that was highly nonlinear and the index and gain were systems for cutting and welding applications, and a significant
highly coupled. Finally and more importantly, the need for fraction of the market is the sale of DPSS lasers.
an application that is economically large enough that would High-power semiconductor lasers used for pumping of
justify the expense associated with the migration from a DPSS Nd : YAG lasers were first commercially introduced in 1984
solution to a semiconductor manufacturing solution. To date at output powers of 100 mW CW. Nd : YAG lasers have
some exceptional work by a variety of research organizations several distinctions compared to semiconductor lasers; first
has resulted in the demonstration of single spatial mode laser at they have a long excited state lifetime and can therefore store
output powers in excess of 5-W CW. energy and are applicable to -switched operation resulting in
In parallel with the development of monolithic single spatial high-peak power applications, and second they have the ability
mode lasers came the development of double-clad rare earth to efficiently convert multimode light into single-mode light.
doped fiber. As discussed below, rare earth doped fiber orig- The designers of DPSS lasers looked to use of high-power
inated around the need for optical amplification in the fiber semiconductor lasers to replace that of flash lamp pumping
where Er doped fiber amplifiers (EDFA) were developed. From for the advantages of greater efficiency and higher reliability.
the technology of Er fibers, which were originally developed This gave great design flexibility to the DPSS laser designer
for single-mode fibers requiring single-mode pump lasers, came in the choice of format for the high-power semiconductor
the conceptual development of double-clad fibers. Double-clad laser. Quickly, the market moved to a design of a 1-cm-long
fibers have a single-mode core surrounded by a second mul- monolithic semiconductor laser array that could be stacked to
timode optical cladding layer. In this configuration multimode create a two-dimensional (2-D) emitting aperture, Fig. 5. The
light is injected into the outer cladding layer and the the light 1-D and 2-D arrays were used in either CW or what was noted
propagates over several tens of meters while it is absorbed by as quasi-CW mode, where the quasi-CW operation was a series
the rare earth dopant in the single-mode core of the fiber. This of long pulses used to match the upper state lifetime of the
process has demonstrated to be a very efficient mechanism of solid state laser, on the order of a few hundred microseconds to
conversion of the multimode radiation from the semiconductor a few milliseconds. Today, monolithic laser arrays have been
laser to a single-mode output. While the Nd : YAG laser, dis- demonstrated at output powers approaching 200 W CW where
cussed below, will result in optical to optical conversion effi- reliable operation of 60 W is commercially available.
1474 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 6, NO. 6, NOVEMBER/DECEMBER 2000

Emerging from the DPSS systems were first lasers operating

in the few watts average power region. Applications for these
lasers included marking on electronic packages, thermal or
ablative printing on plates for offset printing applications,
micro- welding for various applications in the hard disk and
semiconductor industry, heat treatment of micromechanical
components, and many others. A second class of applications
arose from the ability to efficiently convert the output of
the DPSS to green and UV for application to photosensitive
materials.
The second class of products, which is just beginning today,
are the applications that require high average power operation
including cutting and welding applications where the average
powers are in the range of 100–1000 W.

VII. OPTICAL DATA STORAGE

The second application that has been impacted by high-power
semiconductor lasers has been optical data storage. Read-only Fig. 6. Power output as a function of drive current for a diode-pumped fiber
laser.
applications within optical data storage have existed for a
number of years at 830 and 780 nm. It has been the advances
in high-power laser technology that hve pushed the reliable
output powers to greater than 30 mW that has enabled the
ability to write on optical discs. Initially, this was introduced
to that market at 830 nm, followed closely behind by 780 nm
and more recently 650–680-nm lasers, the movement to shorter
wavelength for the benefits of higher storage densities. These
devices are all single spatial mode operation.
Other applications that have benefited from higher power
semiconductor lasers in the early years of their development
include free-space/satellite communications, where extensive
work and a number of demonstrations were successful at
secure free-space optical links, direct diode material processing
applications including heat treatment of metal surfaces, medical
applications including photodynamic therapy, hair removal,
and other therapeutic applications.

VIII. TELECOMMUNICATIONS
Fig. 7. Photograph of a 2-D array of laser diode bars; maximum output power
In parallel with the development of high-power semicon- of 2.56-kW average power.
ductor lasers was the development of rare earth doped optical
fiber. Rare earth doped fiber in conjunction with the advances
in high-power lasers are, in this author’s perspective, the true
enabler to WDM communication systems. It was the ability
to amplify light in fiber that eliminated the need for expensive
regeneration every 100 km, thus enabling an economic solution
for WDM fiber optic communication. The development of rare
earth fiber has had two primary areas of impact: the first being
Er: doped fiber for amplification at 1550 nm.
The first demonstration of Er doped fiber amplifiers (EDFA)
was in conjunction with 1480-nm semiconductor lasers. The
conversion efficiency of the 1480 pump lasers to 1550 ampli-
fication of the signal source is in the range of 70%. Rapidly, the
communication system designs pushed for output powers from
the amplifiers on the order of 20 mW (13 dBm), and for a variety Fig. 8. FIT rate for 980-nm laser chips as a function of output power.
of reasons due to multichannel inputs and multistage amplica- requiring greater than 30 dBm and therefore further driving the
tion, these amplifiers required pump powers of greater than 50 output power requirements from the emiconductor lasers. The
mW. As the channel count further increased over time the am- research community implemented some of the technologies and
plifier output power exceeded 23 dBm with some applications designs discussed above, increasing the cavity length of the laser
WELCH: A BRIEF HISTORY OF HIGH-POWER SEMICONDUCTOR LASERS 1475

Fig. 9. Equivalent noise figure as a function of pump power for a Raman pump laser for both SMF28 and LEAF fiber.

Fig. 10. Impact of Raman amplification on the SNR degradation in system performance. For a system designed for 35-dB degradation in SNR, the incorporation
=
of Raman amplification in complement with EDFA (NF 5 dB) extends the distance between repeaters from approximately 500 to 3000 km. For this example
the amplifier spacing is 100 km.

EDFA was deployed in an undersea communication link in the

early 1990s.
The second absorption band of EDFAs was at 980 nm. The
advantage of 980-nm pumping was that the noise figure was
much lower than 1480-nm pumping. As the development of the
EDFA was nearly simultaneous with the development of pseu-
domorphic materials and the processing developments that lead
to the elimination of COD, the next high-power application to
develop was that of 980-nm laser for EDFA pumps. There are
several inherent tradeoffs between 980-nm lasers and 1480-nm
lasers: 1) 980-nm pumping results in a lower noise figure from
the amplifier than 1480-nm pumps; 2) the optical conversion of
the pump laser to 1550-nm light is more efficient in 1480-nm
lasers; 3) the drive current requirements are higher for 1480-nm
Fig. 11. Power output as a function of current for a fiber coupled 1455 nm lasers, an important parameter in the design of undersea equip-
laser [39].
ment; and 4) the initial reliability of 1480-nm lasers was greater
than that of 980-nm lasers. The result was that the communica-
to 1 mm and beyond in research devices to handle the thermal tion industry wanted to use 980-nm lasers in the EDFA, how-
dissipation of the high-power lasers. The communication in- ever, the reliability of the 980-nm pump laser needed to be im-
dustry had significant data on the reliability of InP-based lasers proved to the same level of performance as the 1480-nm laser.
in the form of DFB lasers, and InP lasers did not demonstrate Over a number of evolutionary growth, processing, and pack-
COD, therefore the communications industry adopted 1480-nm aging developments, several organizations were able to develop
lasers as the first high-power laser for communications. The first highly reliable 980-nm pump lasers that met both the terres-
1476 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 6, NO. 6, NOVEMBER/DECEMBER 2000

Fig. 12. Laser marketplace (Laser Focus World) [40].

Fig. 13. Revenue by market sector for semiconductor lasers (Laser Focus World) [41].

trial and undersea reliability requirements. Today, high-power fiber being used and the system implementation of the fiber and
980-nm lasers can be deployed with FIT rates less than 100 at ranges between 500 mW and 1.5 W.
powers of several hundred milliwatts. The alternative technology to fiber laser-based Raman ampli-
The combination of Er fiber, and high-power 1480- and fication pumps is the use of direct high-power semiconductor
980-nm lasers are the essential elements of the WDM long-haul lasers. For Raman amplification in the C and L transmission
communication systems. The 980-nm laser provides the noise bands of the optical fiber, the semiconductor lasers need to op-
performance, while the combination of 1480- and 980-nm erate in the 1450-nm range. To achieve the output power of 500
for power generation enable the multichannel architectures of mW to 1 W, organizations are both polarization and wavelength
today. Without these technologies there would be no high data multiplexing four–six pump lasers in to a single-fiber output.
rate communication systems capable of handling the traffic Raman amplification will require the output power of discrete
needed for internet applications. 1455-nm pumps to approach 300 mW and beyond in order to
The next generation of amplification technology for commu- optimize the performance and manufacturing cost.
nication networks is Raman amplification. Raman amplification Today, most next-generation long-haul transmission system
transforms the network from amplification at discrete points to designs will utilize Raman amplification. This technology is a
a network where the transmission fiber becomes part of the am- breakthrough technology as it is the enabler for ultralong-haul
plification network, resulting in a dramatically reduced noise transmission and high data rate ( Gb/s) transmission. Cur-
figure of the amplifier, thus enabling ultralong haul transmission rently, the Raman amplification is designed to complement the
and 40-Gb/s transmission. Raman amplification grew out of two EDFA.
different high-power laser technologies. The first deployment of
Raman amplification has come from the use of fiber laser-based
IX. CONCLUSION
pump sources. These amplifier pumps are currently being de-
ployed in undersea festooning application and dry-side-based Over the past 20 years since the first work for high-power
preamplifiers. As discussed above, the fiber laser configuration semiconductor lasers, the technology and market requirements
combines the technologies of high-power semiconductor lasers have advanced dramatically. The market for semiconductor
with that of double-clad fiber laser converting the multimode lasers has grown at staggering clips from 1980 to 2000 and is
semiconductor light to single spatial mode output and subse- currently growing at greater than 40% per annum, the largest
quent Raman shifting to an output wavelength of 1455 nm at a market being the communications market. The semiconductor
power in excess of 1 W CW. These light sources were initially laser is the largest market of all optical technologies (Figs. 12
tools of the research community to investigate the properties and 13). High-power lasers as a segment of the overall market
of Raman amplification in fiber optic transmission. Optimum are growing even faster at a rate in excess of 90% per annum.
pump powers for fiber is dependent on the type of transmission For high-power lasers the growth and impact is substantial
WELCH: A BRIEF HISTORY OF HIGH-POWER SEMICONDUCTOR LASERS 1477

as high-power lasers enable dramatic cost reduction and [12] M. Sakamoto, D. F. Welch, H. Yao, J. G. Endriz, and D. R. Scifres,
performance enhancements in the communications systems. Electron. Lett., vol. 26, p. 729, 1990.
[13] E. Wolak, M. Sakamoto, J. Endriz, and D. R. Scifres, LEOS ’92 Digest,
The trend is to shift value from the transmission products to the 1992, Paper No. DLTA 5.1, p. 175.
amplification products as a percentage of the overall systems [14] D. K. Wagner, R. G. Waters, P. L. Tihanyi, D. S. Hill, A. J. Roza Jr., H. J.
cost. This trend represents the ability to replace the expensive Vollmer, and M. M. Leopold, IEEE J. Quantum Electron., vol. QE-24,
p. 1258, 1988.
repeaters with optical amplifiers. [15] D. F. Welch, B. Chan, W. Streifer, and D. R. Scifres, Electron. Lett., vol.
High-power semiconductor lasers have become a pillar for a 24, p. 113, 1988.
number of markets including fiber optic communications, ma- [16] G. L. Harnagel, P. S. Cross, D. R. Scifres, and D. Worland, Electron.
Lett., vol. 2, p. 231, 1986.
terials processing/manufacturing technologies, printing, optical [17] G. Agrawal and N. K. Dutta, Long-Wavelength Semiconductor Lasers:
data storage, medical therapeutic and cosmetic applications, Van Norstrand Reinhold Company, 1986, ch. 9, p. 372.
[18] W. T. Tsang, Appl. Phys. Lett., vol. 38, p. 661, 1981.
free-space communications, and many others. Research in this [19] D. F. Welch, W. Steifer, C. F. Schaus, S. Sun, and P. L. Gourley, Appl.
area has shifted from basic research to applications research as Phys. Lett., vol. 56, p. 10, 1990.
these markets have matured. [20] D. P. Bour, D. B. Gilbert, G. L. Elbaum, and M. G. Harvey, Appl. Phys.
Lett., vol. 53, p. 2371, 1989a.
[21] H. K. Choi, C. A. Wang, D. F. Kolesar, R. L. Aggarwal, and J. N.
ACKNOWLEDGMENT Walpole, Photon. Tech. Lett., vol. 3, p. 857, 1991.
[22] R. G. Waters, P. K. York, K. J. Beernink, and J. J. Coleman, J. Appl.
The advancements of high-power lasers is the culmination Phys., vol. 67, p. 1132, 1990b.
of thousands of research, development, manufacturing, and [23] D. R. Scifres, R. D. Burnham, and W. Streifer, Appl. Phys. Lett., vol. 33,
p. 1015, 1978.
application engineers. To give adequate and definitive recog- [24] D. E. Ackley, Appl. Phys. Lett., vol. 42, pp. 152–154, 1983.
nition across this industry would be a monumental task and [25] W. Streifer, A. Hardy, R. D. Burnham, and D. R. Scifres, Electron. Lett.,
was not the purpose for this paper. Nonetheless a number of vol. 21, p. 118, 1985.
[26] H. Hosoba, M. Matsumoto, S. Matsui, S. Yano, and T. Hijikata, (in
key organizations have played a role in the development of Japanese)The Review of Laser Engineering, vol. 17, p. 32, 1989.
the technology that has resulted in the above advancements, [27] D. Botez and G. Peterson, Electron. Lett., vol. 24, p. 1042, 1988.
including (but not limited to): SDL, Spectra Physics, Coherent, [28] D. R. Scifres, W. Streifer, and R. D. Burnham, IEEE J. Quantum Elec-
tron., vol. QE-15, p. 917, 1979.
Sony, Hitachi, Mitsubishi, IBM, Uniphase, Lasertron, Pirelli, [29] D. F. Welch, W. Streifer, P. S. Cross, and D. Scifres, IEEE J. Quantum
U.S. Government, TRW, Hughes, Perkin Elmer, AT&T, Lucent, Electron., vol. QE-23, p. 752, 1987.
Nortel, Alcatel, Siemens, Philips, Polaroid, IPG, McDonnell [30] L. J. Mawst, D. Botez, M. Jansen, T. J. Roth, C. Zmudzinski, C. Tu, and
J. Yun, SPIE Proceedings, vol. 1634, p. 2, 1992.
Douglas, Sharp, Toshiba, NEC, NTT, EG&G, David Sarnoff [31] J. S. Major, D. Mehuys, and D. F. Welch, Electron. Lett., vol. 28, p. 1101,
Research Labs, Naval Research Labs, Philips Lab USAF, Rome 1992.
Lab USAF, Cornell University, University of New Mexico, [32] R. Lang, IEEE J. Quantum Electron., vol. QE-18, pp. 976–983, 1982.
[33] U. Koren, B. Miller, G. Raybon, M. Oran, M. Young, T. Koch, J.
Caltech, Stanford, Oregon Graduate Research Center, and DeMiguel, M. Chien, B. Tell, K. Brown-Goebeler, and C. Burrus, Appl.
many others. Phys. Lett., vol. 57, pp. 1375–1377, 1990.
[34] D. Welch, R. Waarts, D. Mehuys, R. Parke, D. Scifres, R. Craig, and W.
The author would like to thank all of the individuals in these Streifer, Appl. Phys. Lett., vol. 57, pp. 2054–2056, 1990.
and other organizations that have contributed to the technology [35] L. Goldberg, D. Mehuys, and D. Hall, Electron. Lett., vol. 28, pp.
and applications for these devices over the years. It is through 1082–1084, 1992.
[36] J. Walpole, E. Kintzer, S. Chinn, C. Wang, and L. Missagia, Appl. Phys.
the cumulative efforts of these people that this industry has be- Lett., vol. 61, pp. 740–742, 1992.
come a technology backbone of the information age. The author [37] D. Mehuys, D. Welch, and L. Goldberg, Electron. Lett., vol. 28, pp.
would like to acknowledge some of the many contributors that 1044–1046, 1992.
[38] S. O’Brien, R. Parke, D. Welch, D. Mehuys, and D. Scifres, Electron.
he has had the benefit of working with at SDL, many of whom Lett., vol. 28, pp. 1272–1273, 1992.
have contributed in one way or another to this data in this article [39] M. Ziari, OFC, 2000.
including, B. Streifer, D. Scifres, R. Parke, J. Major, R. Lang, S. [40] S. G. Anderson, “Review and forecast of laser markets: 1999—Part I,”
Laser Focus World, vol. 35, no. 1, pp. 80–100, 1999.
Sanders, R. Waarts, T. Tally, B. Gignac, R. Craig, V. Dominic, [41] R. V. Steele, “Review and forecast of laser markets: 1999—Part II,”
D. Mehuys, S. Grubb, R. Zanoni, M. Devito, M. Cardinal, D. Laser Focus World, vol. 35, no. 5, pp. 52–72, 2000.
Coblentz, M. Verdiell, and countless others.

REFERENCES
[1] R. N. Hall, G. E. Fenner, J. D. Kingsley, T. J. Soltys, and R. O. Carlson, David F. Welch (S’81–M’85–SM’90) received the
Phys. Rev. Lett., vol. 9, p. 366, 1962. B.S.E.E. degree from the University of Delaware,
[2] M. I. Nathan, W. Dumke, G. Burns, F. H. Dill Jr., and G. Lasher, Appl.
Newark, in 1981 and the Ph.D. degree from Cornell
Phys. Lett., vol. 1, p. 62, 1962.
[3] N. Holonyak Jr. and S. F. Bevacqua, Appl. Phys. Lett., vol. 1, p. 82, 1962. University, Ithaca, NY, in 1985.
[4] T. M. Quist, R. H. Rediker, R. J. Keyes, W. E. Krag, B. Lax, A. L. In January 1985, he joined SDL, Inc. as a Research
McWhorter, and H. J. Zeigler, Appl. Phys. Lett., vol. 1, p. 91, 1962. Scientist and was promoted to Vice President of
[5] I. Hayashi, M. B. Panish, W. Foy, and S. Sumski, Appl. Phys. Lett., vol. R&D in 1990. In 1996, he became Vice President
17, p. 109, 1970. of the Systems Business Group within SDL. His
[6] Zh. I. Alferov, V. M. Andreev, D. Z. Garbuzov, Yu. V. Zhilyaev, E. P. most recent position, as of September 1998, is Chief
Morozov, E. L. Portnoi, and V. G. Trofim, Sov. Phys. Semicond., vol. 4, Technical Officer and Vice President of Corporate
p. 1573, 1971. Development, where his responsibilities include
[7] H. Kressel and J. K. Butler, Semiconductor Lasers and Heterojunction overall technology direction and merger/acquisitions. He has authored more
LEDs, 1977. than 250 articles and more than 50 patents. He has worked on the development
[8] G. H. B. Thompson, Physics of Semiconductor Laser Devices, 1980. of advanced optical technologies for fiber optic communications while at SDL.
[9] T. P. Pearsall, Ed., GaInAsP Alloy Semiconductors, 1982. Dr. Welch received the 1992 Adolph Lomb Award from the OSA, the 1998
[10] J. K. Butler, Ed., Semiconductor Injection Lasers, 1980. Engineering Achievement Award from LEOS, the 1999 Fraunhoffer Award, and
[11] D. R. Scifres, C. Lindstrom, R. D. Burnham, W. Streifer, and T. L. Paoli, his contributions have been acknowledged in more than 17 product of the year
Electron. Lett., vol. 19, p. 169, 1983. awards. He is a Fellow of the OSA.