The development of a new fiber laser design—
`
`the double-clad fiber laser—has impacted the
`
`telecommunications, graphic arts, medical
`
`technologies, and materials processing industries.
`
`This article looks at the properties of these
`
`relatively new devices and explains their impact
`
`on current technologies.
`
`Fiber lasers have been the
`subject of considerable
`activity over the 25 years
`since
`they were
`first
`demonstrated. Researchers
`around the world have
`explored their power, or
`pulse properties, and their ability
`to generate light at different wave­
`lengths. Fiber lasers have also
`been investigated for applications
`in areas such as telecommunica­
`tions, laser ranging, materials pro­
`cessing, imaging, and medicine
`using combinations of different
`gain mediums, pumping schemes,
`and cavity designs.1 Despite a
`detailed understanding of light
`coupling, generation, and propa­
`gation, fiber lasers are only now
`beginning to see commercial suc­
`cess. This is a result of the devel-
`
`By David J. DiGiovanni
`and Martin H. Muendel
`
`Energetiq Ex. 2051, page 1 - IPR2015-01300, IPR2015-01303
`
`

`
`is confined by the lower index of the outer cladding, and
`opment of a new type of fiber device, the double-clad
`is absorbed by optically active dopants (Yb3+ or N d 3 +)
`fiber laser. This article discusses the advantages of such
`in the core. The noncircular shape eliminates helical
`lasers and the various applications that they will impact.
`rays, which have poor overlap with the core. The scal­
`In principle, fiber lasers offer many advantages over
`loped and hexagonal shapes in Figure 2 combine good
`diode or bulk devices.2 Hair-thin glass fibers have
`mode coupling with a fabrication and low-loss device
`extremely high damage thresholds and their high sur­
`assembly similar to that of standard fiber. This low loss
`face to volume ratio provides excellent heat dissipation.
`is valuable because even small losses at fiber splices can
`They can be virtually limitless in length—allowing effi­
`lead to significant heating in multiwatt devices.
`cient use of pump light—and exhibit excellent beam
`quality, and wavelength and temperature stability. How­
`For high-power operation, the cladding's NA
`ever, conventional practical fiber lasers must be pumped
`should be as high as possible so that light
`with diode sources that couple light directly into the
`from the diodes is captured by a large
`single-mode core, which is 8 μm in diamter. Since this
`facet area. For silica-based fiber,
`diode pump power is currently limited to fractions of a
`materials constraints
`watt, the power levels generated by fiber lasers have been
`limit the
`relatively low and diode lasers continue to dominate
`NA if both
`the
`inner
`A critical weakness of diode lasers, though, is low
`and outer
`brightness, which is the product of beam area and diver­
`claddings
`gence. To increase output power, the diode facet area
`are glass.
`must be increased, because diode brightness is limited by
`Higher NA
`the damage threshold of the semiconductor material.
`This curtails some industrial applications since light
`requires the use of a low index polymer to define the
`from large area diodes has poor beam quality and cannot
`outer cladding. The low index polymer also protects the
`be focused to small sizes. It is also a severe limitation for
`glass surface from mechanical damage. If fluorinated
`telecommunications applications, which use only light
`acrylate polymers replace the standard fiber coating, an
`coupled into the single-mode core. The power of con­
`NA from 0.45-0.60 is achieved with no compromise in
`ventional fiber lasers can only be increased by combining
`mechanical properties.4 Alignment tolerances in cou­
`multiple pumps into a single fiber. One method that was
`pling to the large cladding are typically tens of microns
`used early in the development of optical amplifiers for
`rather than submicron for single-mode coupling. At the
`telecommunications was to polarization- and wave­
`fiber exit, on the other hand, the beam is diffraction-
`length-combine multiple diodes onto a single fiber. But,
`limited, and the fiber can be spliced to other fibers using
`practical implementation of this approach was limited
`conventional techniques.
`by the complexity and losses of the components, which
`increase in number exponentially.
`
`Figure 1. Double-clad fiber laser structure. Light from an area diode is
`launched into a noncircular inner cladding to pump Nd or Yb into the
`single-mode core. Fiber lases from the core at a high brightness level.
`
`Double-clad fibers offer a clever solution to increas­
`ing the amount of pump power in a fiber laser.3 By
`turning the entire glass cladding into a second wave­
`guide surrounding a rare-earth doped core, low bright­
`ness pump light can be injected into a large area rather
`than into the small single-mode core. This construction
`is shown in Figure 1. Since the light overlaps with the
`core to some degree, it pumps the core as it propagates
`down the fiber. Therefore, laser emission generated in
`the core can have a brightness level thousands of times
`higher than that of pump diodes. This increased bright­
`ness is the significant advantage of double-clad fiber
`over both diodes and other types of fiber lasers. Addi­
`tionally, powers exceeding 100 W have been efficiently
`coupled from two InGaAs diodes into high numerical
`aperture (NA) double-clad fibers only 200 μm in diame­
`ter, which is over two orders of magnitude higher than
`for single-mode guides. Using tens of meters of double-
`clad fiber, cw powers in excess of 50 W have been gener­
`ated in a single-mode beam at 1100 nm, and devices
`approaching 10 W have run for thousands of hours.
`
`Double-clad fiber design
`The defining difference between double-clad devices
`and conventional active fibers is the provision of a
`cladding waveguide, often noncircular (see Fig. 2).
`Pump light coupled into the noncircular inner cladding
`
`Double-clad laser operation
`A fiber laser can be configured by defining a resonant cav­
`ity using Bragg gratings written directly into the core of
`the fiber, or by using mirrors deposited on or attached to
`the fiber ends.5 Since the pump energy is distributed
`across the inner cladding, pump intensity is relatively low.
`To allow population inversion, the absorption of the core
`at the lasing wavelength must be low. This limits opera­
`tion to four-level laser systems, and also diminishes the
`amount of pump
`absorbed per unit
`length, which scales
`with the core to clad
`area ratio. However,
`because the core
`absorption at the
`pump wavelength is
`typically several dB/
`cm, laser lengths are
`less than 100 m.
`Nd-doped
`lasers
`pumped at 810 nm
`with GaAlAs diodes
`have generated over
`30 W at 1060 n m .6
`Typical slope effi­
`ciencies are around
`50%, with power
`
`Figure 2. Cross-sectional view of a double-clad fiber with
`a Yb-doped single-mode core (about 7-μm-diameter),
`noncircular pump waveguide, and low-index polymer.
`
`Optics & Photonics News/January 1999
`1047-6938/99/1/0026/05-$0015.00 © OSA
`
`27
`
`Energetiq Ex. 2051, page 2 - IPR2015-01300, IPR2015-01303
`
`

`
`levels limited only by the available pump power. Yb-
`doped lasers may be pumped from 900-975 nm using
`InGaAs diodes. Given the broad Stark splitting in Yb,
`quasi-four-level operation allows lasing from 1060-1120
`nm, though this range can be extended with effort. Slope
`efficiency approaches 70% and, again, output power is
`limited only by the pump. In practice, coupling losses
`reduce the total efficiency to about 40% of the diode facet
`power.
`
`Applications in telecom
`High-power amplifiers are of great interest as commu­
`nications networks expand in capacity and distance. The
`explosion of Internet traffic has fueled the development
`of WDM systems with an ever-increasing number of
`channels, each with around 1 mW of power. Moreover,
`as components such as dispersion compensators and
`gain equalizers are added to cope with system impair­
`ments, power budgets continue to increase. Double-clad
`fiber lasers are obviously compatible with fiber devices
`used for optical communication. However, an interme­
`diate stage must be used to convert laser light (typically
`from 1060-1120 nm) to the wavelengths of interest for
`telecom, 1300-1600 nm. This can be accomplished
`using rare-earth doped or Raman devices and can be
`either a complication or a benefit, as discussed below.
`
`power to more than 10 W at 1550 nm. In addition to
`pumping a single-mode Er/Yb fiber amplifier with a dis­
`crete double-clad laser, an amplifier may be configured
`directly in a double-clad fiber with an Er/Yb doped core.8
`Though double-clad amplifiers have fewer components and
`higher efficiency, coupling of both the pump into the
`cladding and the signal into the core is a complication.
`One-watt amplifiers are of immediate interest for
`broadcast analog transmission where the signal power
`per channel must be high enough (around 1 mW) to
`maintain adequate carrier-to-noise power across 80
`channels. Meanwhile, the amplifier's cost must be kept
`very low to create a financially viable local loop network
`that penetrates to the home. High-output power allows
`massive power splitting so that the cost of an amplifier
`may be shared by many users.
`Multiwatt amplifiers are of interest for laser-based
`satellite communication. As data capacity increases,
`microwave, the current mode of transmission, becomes
`prohibitively large and costly. A fiber-based alternative
`combines the benefit of high-speed electronics devel­
`oped for terrestrial applications (greater than 10 Gb/s)
`with low weight and power consumption. In the pro­
`posed schemes, up and down links to the ground may
`be optical or microwave while optical signals are passed
`satellite-to-satellite in an orbiting constellation. For this
`application, immunity to ionizing radiation must be
`considered in addition to high efficiency.
`
`Rare-earth doped systems
`Conventional Er-doped optical amplifiers are limited by
`the amount of power coupled from highly reliable
`Raman devices
`single-mode diodes with wavelengths in the pump bands
`Interaction of high optical intensity with vibrational
`of erbium, namely 980-1480 nm. An early effort to
`modes of the glass results in stimulated Raman scatter­
`exploit diode-pumped solid-state (DPSS) lasers such as
`ing. The light generated is downshifted in frequency by
`Nd:YLF at 1060 nm required removal of the constraint
`about 450 cm- 1, with a very broad spectrum caused by
`on wavelength. This was accomplished using energy
`the random nature of the glass host. Although the non­
`transfer from Yb to Er through co-doping.7 The high
`linear cross-section is small, low-loss fiber allows use of
`long interaction lengths, and amplification of a signal can
`exceed 40 dB over fiber distances less than 1 km. Lasers
`and amplifiers using this phenomenon have been
`explored since the mid-1980s, but suffered from the lack
`of practical pump sources. The advent of double-clad
`fiber lasers has altered this situation. Laser output at 1120
`nm can be Stokes shifted over five orders to 1500 nm.9
`The efficiency of wavelength conversion can be
`greatly increased by capturing the intermediate orders
`within a resonant cavity defined by Bragg gratings or
`fiber couplers. The configuration and output spectrum
`of such a cascaded Raman resonator (CRR) is shown in
`Figure 3. Through the careful design of the resonant
`structure, conversion efficiency to 1480 nm can exceed
`50%, which is remarkable for a nonlinear device.10 Giv­
`Figure 3. A cascaded Raman resonator composed of up to 1 km of optical fiber between
`nested sets of fiber Bragg gratings. Light generated by Raman scattering is confined by
`en the broad gain spectrum, CRR output can range
`gratings and resonantly enhanced. The graphs show the residual light at intermediate
`from 1200-1600 nm, with linewidths around 2 nm,
`Stokes frequencies (left) and laser emission (right).
`controlled by the gratings.
`An obvious use of high-power at 1480 nm is to pump
`conventional Er-doped fiber. With an appropriate fiber
`design, output powers can exceed several watts without
`rollover, due to excited state absorption. A more interest­
`ing application is to locate the Er-doped fiber remotely,
`say 100 km offshore in an unrepeatered submarine link.
`Double-clad lasers enabled demonstration of communi-
`
`absorption cross-section of Yb allows pumping from
`800-1100 nm with subsequent transfer of this energy to
`Er for signal amplification. A suitable glass host, such as
`p-doped Si, ensures efficient transfer. Slope efficiencies
`from pump to signal of 40% are typical.
`Double-clad fiber lasers enable an almost 50-fold
`increase beyond DPSS-pumped systems in amplifier
`
`28 Optics & Photonics News/January 1999
`
`Energetiq Ex. 2051, page 3 - IPR2015-01300, IPR2015-01303
`
`

`
`cation over 529 km without amplification. The avail­
`ability of high-power from 1200-1600 nm opens many
`possibilities for optical amplification, which is timely since
`the demand for capacity is doubling every two years or so.
`Demand is being met by increasing the bandwidth and
`transmission rate as high as fiber nonlinearities allow. To
`ease these limits, Raman gain can be developed within the
`transmission fiber itself by injecting fractions of a watt at
`1450 nm to amplify signals at 1550 nm. In this configura­
`tion, lab demonstrations have indicated that Raman gain
`can quadruple system capacity.11
`To open other spectral windows for optical amplifi­
`cation, Raman devices can be configured as discrete
`amplifiers. For example, by truncating the nested grat­
`ings shown in Figure 3 to 1240 nm, signals at 1310
`nm—the next Stokes shift—can be amplified by more
`than 40 dB. Amplifiers using this principle have been
`demonstrated for both analog12 and digital systems13
`and at wavelengths of 1310 nm and 1400 nm. Given the
`broad spectral response of Raman scattering and the
`flexibility of fiber gratings, the entire low-loss window
`of silica—from 1300-1620 nm—can be exploited using
`Raman devices. The combination of a double-clad fiber
`laser pumping a cascaded Raman resonator in either a
`discrete or distributed amplifier has tremendous poten­
`tial for increasing transmission capacity.
`
`Applications in industry and medicine
`Double-clad fiber lasers have characteristics and capa­
`bilities that uniquely qualify them for several industrial
`and medical markets, including
`a very high diffraction-limited cw power
`high gain and high average power in amplifier
`configurations
`wavelength agility, reliability, efficiency, and economy.
`
`Figure 4. Schematic of the
`laser ablation transfer (LAT)
`process in which a fiber
`laser pulse is focused onto
`a rotating pigment donor
`sheet ablating the pigment,
`which is deposited on a
`donor plate or sheet.
`
`For exposure, thermal processing requires a much
`higher threshold fluence than photographic tech­
`niques—typically on the order of 100 mJ/cm2. More­
`over, thermal conduction makes the threshold intensity
`dependent; exposures in excess of tens of nanoseconds
`become progressively less efficient. In order to expose a
`standard-sized image (44" X 32") in a few minutes, cw
`laser powers on the order of 10 W are required. The
`double-clad fiber laser is close to ideal for this applica­
`tion. The fiber laser's higher brightness allows the use of
`hardware similar to that used in most current reproduc­
`tive systems (the "internal drum" architecture, in which
`only a scanning mirror moves). Because of the low
`brightness of laser diodes, diode-based machines require
`an alternative architecture, namely an external drum, in
`which the entire workpiece spins. There is considerable
`inertia to staying with internal-drum machines, thereby
`making fiber lasers the preferred choice.
`Compared with competing high-brightness sources,
`such as diode-pumped crystalline solid-state lasers (e.g.,
`Nd:YAG and Nd:YVO), the fiber laser offers superior
`thermal stability and efficiency, as well as the potential
`for internally modulated output using a master-
`oscillator/power-amplifier (MOPA) architecture.
`Presently, fiber lasers are used in the proofing and plate-
`setting machines made by most graphic arts hardware
`manufacturers. Several hundred high-power fiber lasers
`are in commercial operation in this market today.
`
`Graphic arts
`Double-clad fiber lasers represent an enabling technolo­
`gy for the new advanced thermal media that are becom­
`ing available in the graphic arts industry. Presently, most
`graphic arts houses (including publishers of magazines,
`newspapers, and books) produce their prepress images
`and printing plates using wet photographic techniques.
`Virtually all of the major graphics media suppliers,
`however, are now developing instant, thermally exposed,
`digital media for proofing and platesetting. The lack of
`chemical processing makes these "dry" media much
`more convenient for the user, as well as more environ­
`mentally benign. Widespread agreement exists in the
`graphics industry that the long-term trend is toward an
`Marking
`all-digital workflow using this type of media.
`Significant numbers of fiber lasers are also being used
`commercially in marking applications. In these applica­
`In this process, images are formed on a dry substrate
`tions, the laser's output is modulated then scanned across
`by scanning and modulating a laser across the substrate's
`a surface. Fiber lasers have been especially effective in
`surface, exposing individual pixels down to about 10 μm
`making small or micron-scale inscriptions on semicon­
`in size. In one approach, called laser-ablative transfer
`ductor chips and packages. They are also used for mark­
`(LAT), pigments or sensitizing agents from a donor sheet
`ing plastic and metal. The particular benefits of fiber
`are ablated by the laser and deposited on a receptor sheet
`lasers here are their beam quality and pointing accuracy.
`or plate (see Fig. 4). Alternatively, the laser may modify
`the adhesive properties of imaging agents in a multilay­
`ered sandwich that is pulled apart after exposure, sepa­
`rating the imaged from the non-imaged agents.
`
`Generation of pulses
`Double-clad fiber devices show considerable unrealized
`
`Optics & Photonics News/January 1999
`
`29
`
`Energetiq Ex. 2051, page 4 - IPR2015-01300, IPR2015-01303
`
`

`
`medical imaging of low-density tissue such as lungs.14
`promise in generating pulses at high average power lev­
`High-power 2-μm fiber sources may prove useful in
`els. They may become an efficient, versatile alternative
`microsurgical applications. Finally, medical spectro­
`to some Q-switched or mode-locked crystalline lasers,
`scopic applications in areas such as dermatology, diag­
`and may also be a key to realizing commercially viable
`nostic imaging, and cancer therapy should benefit from
`ultrafast systems.
`doubled fiber lasers serving as a replacement for 1-10 W
`Double-clad devices can have high average powers
`dye lasers in the visible range. There are doubtlessly a
`with moderate pulse energies. This is because the energy
`host of medical diagnostic and therapeutic applications
`that can be stored in a gain medium before self-lasing
`for fiber lasers that remain to be explored.
`begins is proportional to the gain cross-sectional area
`rather than the volume. A typical Yb-doped fiber, with a
`core area of 5 X 10-7 cm2, can only store up to about
`The future
`100 μJ. While the use of fibers with larger gain regions has
`Although the technique of cladding pumping is not
`raised the attainable pulse energy somewhat, it is clear
`new, commercial lasers and amplifiers are now available.
`that single-mode fiber devices will be limited to pulse
`Fiber lasers offer advantages over diode lasers for appli­
`energies below approximately 1 mJ. On the other hand,
`cations in which spot size and beam quality are impor­
`attainable average powers are still in the 10-100 W range,
`tant. For telecom applications, interest is mounting
`so repetition rates are typically on the order of MHz.
`rapidly as optical networks require greater power to
`This new regime of moderate energies and high rep­
`handle greater capacity. A significant hurdle for
`etition rates is currently adequate for some pulsed
`telecommunications devices has been the requirement
`applications. Double-clad devices are a particularly
`of high reliability. As long-term testing is completed,
`good match with ultrafast lasers, using the technique of
`these devices will begin to be deployed and will become
`chirped-pulse amplification to amplify nanojoule, fem­
`a significant component of the next generation of opti­
`tosecond pulses by 30 dB or more. Ultrafast lasers are
`cal networks. Applications of fiber lasers in graphic arts
`superior in materials-processing applications because
`are well established, and work is proceeding rapidly on
`they ablate material rather than melt it. Efficient and
`the development of fiber lasers as alternatives to existing
`inexpensive double-clad amplifiers may finally make
`lasers and as enablers for new applications.
`ultrafast systems competitive.
`
`Generation of desired wavelengths in the IR and the visible
`The broad gain bandwidth of rare-earth doped glasses,
`coupled with the high gain of fiber devices, makes
`double-clad devices well suited for tunable operation
`and generation of specific wavelengths. Fiber Bragg
`gratings can enforce laser operation at desired wave­
`lengths for use in pumping other laser mediums such as
`Pr:fluoride fibers (lasing at 1.3 μm), Tm:fluoride fibers
`(480 nm), and Tm- and Ho-doped materials (2 μm).
`Nonlinear frequency conversion can greatly extend
`the wavelength coverage. As mentioned above, Raman
`shifting can efficiently convert Yb-fiber laser output to
`longer wavelengths. Additionally, direct frequency dou­
`bling of Yb-fiber lasers has yielded several hundred mil­
`liwatts at yellow and green wavelengths, while narrow-
`line tunable radiation at 585 nm has been demonstrated
`by mixing the output of a 1060-nm double-clad fiber
`amplifier with the output of a 1.3 μm Nd crystal laser in
`a periodically poled nonlinear crystal. Both pumps were
`about 5W.
`
`Medical applications
`Double-clad fiber devices are well suited to medical
`applications because they are compact, relatively inex­
`pensive, and free of gases, dyes, solvents, and special
`utility needs. Researchers have used an inexpensive,
`low-power fiber laser for confocal microscopy; even at
`only 1-2 W, such a laser has enough headroom in power
`that low-power imaging and higher-power therapeutic
`functions might be integrated in one device. Small fiber
`devices may also prove useful for optical coherence
`tomography. Multiwatt, narrow-line 1083-nm Y b3+
`MOPAs have been used to spin-polarize helium gas for
`
`References
`1. P. Urquhart, "Review of rare-earth doped fibre lasers and
`amplifiers," IEEE Proc. 135 Pt. J (6), 385-407 (1988).
`2. L. Zenteno, "High-power double-clad fiber lasers," J. Light­
`wave Tech. 11, 1435-1446 (1993).
`3. E. Snitzer et al., "Double-clad Offset Core Nd Fiber Laser,"
`Optical Fiber Communications Technical Digest PD5 (OSA,
`Washington, DC, 1988).
`4. A. Hale et al., "Polymer claddings for optical fiber," Am.
`Ceram. Soc. V216, 578 (1998).
`5. G.A. Ball and W.W. Morey, "Continuously tunable single-mod­
`er erbium fiber laser," Opt. Lett. 17 (6), 420-422 (1992).
`6. V. Reichel et al., "High-power single-mode Nd-doped fiber
`laser," Photonics West Laser '98 3265-24, San Jose, CA,
`(1998).
`7. J. Townsend et al., "Yb 3+ sensitized E r 3+ doped silica opti­
`cal fiber with ultrahigh transfer efficiency and gain," Elec­
`tron. Lett. 27 (21), 1958-1959 (1991).
`8. J.D. Minelly et al., "Diode-array pumping of Er/Yb co-doped
`fiber lasers and amplifiers," IEEE Photon. Tech. Lett. 5 (3),
`301-303 (1993).
`9. J. Auyeung and A. Yariv, "Spontaneous and stimulated
`Raman scattering in long low loss fibers," IEEE J. Quantum
`Electron. QE-14, 347-352 (1978).
`10. S.G. Grubb et al., "High-power 1.48 μm Cascaded Raman
`Laser in Germanosilicate Fiber," Optical Amplifiers and
`their Applications, Technical Digest SaA4 (OSA, Washing­
`ton, DC, 1995), pp. 197-199.
`11. P.B. Hansen et al., "Capacity Upgrade of Transmission Sys­
`tems by Raman Amplification," Optical Amplifiers and their
`Applications Technical Digest ThB4 (OSA, Washington, DC,
`1996).
`12. A.J. Stentz et al., "Raman Ring Amplifier at 1.3 μm with
`Analog Grade Noise Performance and Output Power of +23
`dBm," Optical Fiber Communications Technical Digest
`PD16 (OSA, Washington, DC, 1996).
`13. A.J. Stentz et al., "Raman Amplifier with Improved System
`Performance," Optical Fiber Communications Technical
`Digest TuD3 (OSA, Washington, DC, 1996).
`14. S.V. Chernikov et al., "1083 nm ytterbium doped fiber
`amplifier for optical pumping of helium," Electron. Lett. 33
`(9), 787-789 (1997).
`
`David J. DiGiovanni is technical manager in the Optical Fiber Research Department
`at Lucent Technologies, Murray Hill, NJ, and Martin H. Muendel is senior scientist
`at Polaroid Corp., Cambridge, MA.
`
`30 Optics & Photonics News/January 1999
`
`Energetiq Ex. 2051, page 5 - IPR2015-01300, IPR2015-01303