Novel ultrashort pulse fiber lasers for

RIKEN Review No. 43 (January, 2002): Focused on 2nd International Symposium on Laser Precision Microfabrication (LPM2001)

Novel ultrashort pulse fiber lasers for micromachining applications Heinrich Endert,∗1 Almantas Galvanauskas, Gregg Sucha,∗2 Raj Patel, and Michelle Stock IMRA America, Inc., USA

The development of ultrashort pulse laser technology will have a strong impact on the advancement of laser machining. Ultrashort laser pulses can reduce the heat-affected zone and the shock-affected zone, resulting in much cleaner cuts, and therefore higher precision. Also, ultrashort laser pulses have shown remarkable promise for processing transparent materials such as glass, fused silica, and sapphire. However, acceptance of ultrafast technology is hindered by the size, cost, and complexity of ultrafast lasers. In this paper, we describe recent progress in fiber-based ultrafast laser technology which promises to be sufficiently compact, rugged, and potentially low-cost.

Introduction: Applications of femtosecond lasers There is growing interest in using ultrashort laser pulses for specialized commercial applications, with particular emphasis on laser-based micromachining applications. Here, we define “ultrashort” or “ultrafast” pulses as laser pulses with a duration of 1 picosecond (10−12 sec) or less. Because they are thousands of times shorter than pulses from conventional Qswitched lasers, ultrashort laser pulses have unique capabilities, enabling applications that simply cannot be done with the present generation of industrial lasers. In particular, ultrashort laser pulses have the ability to cleanly ablate a wide variety of materials with a minimum of thermal or shock damage to the surrounding material. In many cases, this results in cleaner cuts when compared with conventional lasers. The higher precision, reduced heat-affected zone (HAZ) and reduced shock-affected zone (SAZ), are all by now well documented.1) Many exciting results have been demonstrated for micromachining with femtosecond optical pulses. Most of the initial results were obtained using millijoule laser systems, which are most suitable for heavier tasks like deep drilling (e.g., drilling fuel injection nozzles (FIN)) to depths of a few millimeters.2, 3) In fact, the initial application results were tailored around the more widely available millijoule systems. Ultrashort pulse lasers have also found good use in biomedical fields.4) They have demonstrated superior tissue interaction characteristics compared with conventional surgical techniques or longer pulse laser systems. Ultrashort pulse lasers have been shown to ablate hard tissues such as dental and middle ear tissues very effectively.5) The most advanced and well-known use of ultrashort pulse lasers is as a laser microkeratome in the field of ophthamology. In this procedure an ultrashort pulse laser is used instead of a mechanical blade to cut open the corneal flap. In the USA, a company called Intralase has received a FDA 510 (k) approval for use of their system in laser microkeratome.6, 7) A few other examples of the benefits of using ultrashort pulse lasers for heart and spinal surgery have also been reported.8) Pulse energies ∗1 e-mail address: [email protected] ∗2 e-mail address: [email protected]

in the range of 1 to 100 microjoules and repetition rate in the order of hundreds of kHz (e.g. 100–200 kHz) seem to be desirable for many of the biomedical applications. At still higher repetition rates and lower pulse energies, it is possible to modify the refractive index of glass and other transparent materials by inducing microexplosions.9) Laser pulse parameters in these cases are in the range of 10’s of nanojoules, with repetition rates of 1–80 MHz. This technique shows great promise as a method for writing waveguides in transparent optical materials such as glass for integrated optics applications. Also at this power/energy level, there are other applications such as mask repair, in which the thin metallic films on photomasks get ablated by the laser pulses.

Requirements for ultrafast laser technology Versatility of laser system requirements Although regeneratively amplified laser systems comprise the most widely adopted type of ultrafast lasers, the millijoule energies and low repetition rates of these lasers are not necessarily the best set of parameters for all femtosecond micromachining, as we have seen in the previous section. This is true, for example, in the case of patterning of thin films. To obtain feature sizes which are smaller than the diffraction limit of light, it is necessary to operate very near the ablation threshold while focusing very tightly.10) Under these circumstances, millijoule pulse energies are orders of magnitude larger than necessary and must be strongly attenuated in order to avoid the detrimental HAZ’s and SAZ’s which occur when machining so far above the ablation threshold. For very fine submicron patterning of thin films, it is desirable to use pulses with energies ranging from 10’s of nanojoules up to 10 microjoules. What is more, it has been postulated that machining processes using laser pulse trains with higher repetition rates can take advantage of certain slow thermal effects which are actually beneficial to clean ablation.11) This can only be exploited by using repetition rates of 100’s of kilohertz or higher. Additionally, throughput is also a critical factor for commercial applications, thus favoring higher repetition rates in many cases.

23

Because of the broad spectrum of applications and the various specific conditions involved, laser systems must be developed which can “cover the spectrum” in terms of not only wavelength, but in terms of pulse energy, repetition rate, and pulsewidth as well. Requirements for “user-friendliness” and practicality It is clear that ultrafast lasers will revolutionize micromachining. However, in spite of these possibilities, ultrafast lasers have not yet been widely adopted into the micromachining industry due to the high cost and complexity of most commercial ultrafast laser systems. The current generation of ultrafast lasers simply is not suitable for use on the factory floor. It is a minimum requirement that ultrafast lasers be as rugged and “hands-free” as industrial lasers. With the promise of improvements in micromachining, there is great incentive to develop new generations of femtosecond technology which will be compact, robust, low-maintenance, and easy to operate. High-power fiber laser technology is considered a prime candidate for transforming ultrafast lasers from a scientific tool into an industrial tool. In fact, highpower ultrafast fiber lasers which are suitable for ablation have now been introduced into the commercial market. Laser systems employing doped-fibers provide certain advantages over conventional solid-state lasers. First of all, fibers are flexible and can be rolled up into a tight space or spooled, making it possible to construct fiber lasers in a very compact package. Another advantage is that the laser light is confined to the core of the fiber, providing a well-controlled beam shape and superior pointing stability. Also, thermal management is easier because the long, thin fiber has much more surface area per unit volume than a bulk laser crystal. Finally, many kinds of fiber optic devices are rugged enough for difficult applications (for example, even deployment in harsh undersea conditions in telecommunications cables). These features have made fiber lasers useful in a number of industrial applications outside of telecommunciations such as thermal printing, marking, and materials processing.

Ultrafast fiber lasers Fiber lasers are well-suited to ultrafast applications as they have a large amplification bandwidth, so that they can also be mode-locked to generate ultrashort pulses. In commercial systems, Erbium-doped fibers serve as the gain medium for ultrafast fiber lasers, providing short pulses at wavelengths near 1560 nm, and in some cases at the second harmonic, 780 nm. These simple systems have output powers ranging from 10’s to 100’s of milliwatts at a repetition frequency of 50 MHz. In one particularly rugged design (IMRA Femtolite) single-mode diode pumping and polarization control result in highly stable operation, even with a change in environmental temperature of a ± 15◦ C, which is a much larger range than for most ultrafast lasers. It is possible to obtain higher average powers from ultrafast fiber lasers by using Yb-doped fiber, due to the higher gain and higher power efficiency. To take advantage of Yb:fiber, the wavelength of the ultrashort pulses must lie somewhere in the wavelength range of 1030–1060 nm. This can be accomplished by Raman shifting the pulses from an Er: fiber laser from 1560 nm out to near 2100 nm, and then frequency doubling in a nonlinear crystal of periodically poled Lithium

24

(a)

(b) Fig. 1. (a) Wattlite ultrafast Yb-fiber laser system. Produces fs pulses near 1030–1060 nm with 100 duration at 50 MHz. (b) Schematic of Wattlite. Ultrashort pulses at 1550 nm from an Er: fiber laser are Raman-shifted, and then frequency-doubled to 1030 or 1064 nm, making them compatible with Yb- or Nd-doped amplifiers. After amplification in a cladding-pumped Yb-doped fiber, the output power is about 0.5–1.0 Watts.

niobate (PPLN). These pulses are then injected into a length of Yb:fiber, and are amplified to average power levels of 0.5– 1.0 Watts, which makes this comparable to solid-state ultrafast laser systems in terms of performance.12) This system (IMRA Wattlite) is shown in Fig. 1. This laser provides turnkey operation and excellent amplitude stability (0.1% RMS relative intensity noise). Because of the all-fiber architecture, the pointing stability (Fig. 2) is on the order of 10–20 microradians in both dimensions which is at least an order of magnitude better than for solid-state femtosecond lasers with similar power levels.13) Additionally, the laser is cooled by conduction and convection, thus eliminating the need for water cooling.

High power ultrafast fiber lasers Previous generations of fiber lasers were limited in output power because they required single-mode pumping—and

Fig. 3. Large core fibers (left) can be scaled up to 50 micron core diameter, giving power scaling by a factor of at least 40 or more over standard single-mode fibers (right) which have core diameters of 8 microns. Although these are multimode fibers, they can still maintain single mode propagation under proper launch conditions.

times. By careful design, even these large-core, multimode fibers can transmit optical pulses in only a single transverse mode,16) giving high quality beam profiles, which are essential for precision applications such as micromachining. Large-core fibers have enabled a new generation of ultrashort pulse lasers which operate in the sub-picosecond regime. IMRA researchers have amplified ultrashort pulses in largecore Ytterbium-doped fibers. In these systems, fiber tapers are used to protect the fiber from the very high peak power levels which would otherwise damage it.

Fig. 2. Pointing stability of Wattlite in vertical (19 microradians) and horizontal (8 microradians) directions over a period of 80 hours. Turn-on transients are evident, followed by long periods of high stablity (courtesy S. Biswal, Lawrence Livermore National Laboratory).

single-mode pump diodes are, in turn, limited in power to only a few hundred milliwatts. In the last few years however, fiber lasers with much higher output powers have been developed and are being offered as commercial products. The breakthroughs in higher power have come thanks to a variety of developments. The first breakthrough is the technique of cladding pumping. Double-clad fibers can be designed to provide single-mode wave-guiding in the core (for the signal) and multimode wave-guiding in the inner cladding (for the pump).14) This structure makes it possible to use highpower diode bars and arrays as pump sources, which can provide 100’s of Watts of pump light. Cladding pumped fiber lasers are capable of generating significant amounts of power, especially when using highly efficient Yb-doped fibers. Researchers have recently demonstrated that they can obtain over 100 Watts of CW power from a Yb-doped fiber laser.15) Ultrashort pulse amplification in large core fibers In another major technological development, researchers have been able to scale up the fiber core diameter from the standard 8 microns up to 25, 50, and even 100 microns (Fig. 3). This in turn, facilitates scaling up the output power by many

Fiber Chirped Pulse Amplification (FCPA) Scaling up the pulse energy of ultrafast fiber lasers is not simply a matter of using larger core sizes. Because ultrashort laser pulses have such high peak powers (10’s of Megawatts), nonlinear optical effects in the fiber’s core can severely distort an optical pulse, resulting in very pathological pulse intensity profiles–including such effects as ringing, pedestal, satellites, intensity modulation, and prepulse–if the system is not designed properly. These effects can negate many of the advantages provided by clean, ultrashort optical pulses. Therefore special techniques are required to overcome the problems which are unique to amplifying high-power ultrashort pulses. The most commonly employed technique is chirped pulse amplification (CPA).17) Here, a short pulse is stretched in time (up to durations ranging from 10’s of ps to ∼ 1 ns) to lower the peak power. After being amplified, the pulse is recompressed to its original length of less than 1 ps. This has been applied in solid-state, ultrafast lasers for several years. The CPA technique can also be applied to ultrafast fiber lasers–called Fiber Chirped Pulse Amplified, or FCPA lasers—in which a stretched pulse is amplified in a gain fiber and then recompressed to very short duration. Laser systems based on this technique afford tremendous flexibility in terms of repetition frequency (1 Hz to 1 MHz) and pulse energy. This type of technology forms the basis of a commercial product prototype (IMRA FCPA-2) which can produce ultrashort pulses with energies of 2 microjoules at a repetition frequency of 250 kHz. At the “low” end of repetition rate scale, FCPA systems are capable of producing energies of 100

25

Fig. 4. Ultrafast FCPA laser system using 1055 nm all-fiber source, and large-core. Yb: fiber amplifier in the last stage. Pulse energies of 100 microjoules are generated.

(a)

Fig. 6. 3D modeling and finite element analysis are applied to optomechanical components in ultrafast fiber lasers. Upper figure shows mount for nonlinear crystal. Lower figure shows simulated temperature differences on laser diode heatsink.

amplification (PCPA), which uses a parametric amplifier pumped by a fiber-amplified microchip laser.21) The parametric amplifier crystal consists of Periodically-poled Lithium Niobate (PPLN) which provides very high nonlinearity and engineerable phase matching. One of the chief advantages of this system is that the parametric amplification produces very high gain (∼ 80 db) in a single stage, thus eliminating the need for complex multi-stage, or multi-pass amplification. Such systems can provide 10’s of microjoules, and appear to be scalable to 100’s of microjoules, using the pump-source desribed earlier (i.e., a Yb-fiber amplified microchip laser). Additionally, by adding a solid-state amplifier to the pump source, the PPLN crystal can be pumped with several millijoules of energy.22) In this configuration, the PCPA system produces pulses with over 1 mJ of energy, and duration of 1.5 ps.

Engineering ultrafast fiber lasers

(b) Fig. 5. (a) Fiber chirped pulse amplification (FCPA) system employing large-core amplifier fiber produces pulse energies up to the millijoule level. (b) Fundamental mode profile of pulsed output from 50 micron core fiber.

microjoules, at repetition rates of 5 kHz, with pulsewidths of 250 fs (Fig. 4).18) And recently, pulse energies up to the millijoule level have been demonstrated by using a final power amplifier stage with a 50-micron core (Fig. 5a).19) Even with this large core size, the output beam profile has excellent properties, with an M 2 value of 1.16 (Fig. 5b). At the high repetition rates of 50 MHz, FCPA systems can generate up to 13 Watts of average power.20) Parametric Chirped Pulse Amplification (PCPA) Another promising new technique is parametric chirped pulse

26

The requirements for ruggedness and reliability have not been met by conventional ultrafast laser technology. As mentioned earlier, fiber lasers offer particular advantages over solidstate lasers, particularly since many of the fiber optic components used to make fiber lasers are off-the-shelf telecommuciations components which are already engineered to meet telecom requirements for operating conditions in harsh environments. However, while using telecomm components is an important step toward ruggedization, this in itself is not sufficient. Sound optomechanical engineering principles and methods must be applied to the design of ultrafast lasers just as they would for any other type of laser or optical system for industrial use. Components and subsystems are designed and analyzed for thermal expansion, mechanical distortions and vibration—all with an eye toward making the laser systems rugged, resistant to environmental effects, and as trouble-free as possible (Fig. 6). Fiber lasers have additional advantages which contribute to reliability and trouble-free operation. The high gain values (10–20 dB) of the doped-fibers make the lasers less sensitive to small optical losses and thus more resistant to contamination of the optics, which has very strong detrimen-

tal effects on solid-state lasers where the single-pass gain is on the order of 10%. Power efficiency is another advantage, especially for lasers using Yb-doped fibers. Ti:sapphire lasers require pumping by a green laser, which requires a 4stage pumping scheme: (laser diode → Nd: YVO → SHG → Ti: sapphire). Since Yb-fiber lasers are directly diode pumped, and have optical-to-optical conversion efficiency of over 50%, the wallplug efficiency is about 10-times better than for Ti:sapphire lasers, which not only reduces power consumption, but also reduces the heat load which must be removed from the laser package. This in turn, eliminates the need for water cooling of the package, which even further reduces the physical plant requirements, maintenance, and thus the operating costs. Finally, rugged construction and insensitivity to small optical losses allows for laser units to be shipped to the user’s location and installed by the user directly out of the box without requiring alignment or a service call for installation.

Summary Recent advances in fiber-based laser technology show promise as industrial laser systems of the future. In contrast to early technology, current fiber lasers can generate 10’s of watts of average power; while 100 W lasers, and even kilowatt fiber lasers are under development. The combination of fiber technology with ultrafast laser techniques is creating completely new application opportunities, which will qualify ultrafast lasers as real industrial solutions. In particular, it is now becoming increasingly clear that ultrafast laser technology would be a valuable tool in various industries such as microelectronics, automotive, textiles, telecommunications, and medical device manufacturing. Clearly, advances are being made to make ultrafast lasers rugged, compact, reliable, and easy to use. Fiber-based laser technology, with its many advantages, is one of the new promising alternatives to solid-state lasers.

References 1) P. S. Banks et al.: Proc. Conf. Lasers and Electro-optics (CLEO) (1998 OSA Technical Digest Series) (Optical Soci-

ety of America, Washington DC, 1998), p. 510. 2) S. Nolte, B. N. Chichkov, H. Welling, Y. Shani, K. Lieberman, and H. Terkel: Opt. Lett. 24, 914 (1999). 3) X. Liu, D. du, and G. Mourou: IEEE J. Quantum Electron. 33, 1706 (1997). 4) J. Neev: Photonics Spectra, 35, 44 (2001). 5) M. H. Niemz: Proc. SPIE 3255, 84 (1998). 6) F. Loesel: Laser Focus World, 37, 143 (2001). 7) T. Juhasz, F. H. Loesel, R. M. Kurtz, C. Horvath, J. F. Bille, and G. Mourou: IEEE J. Sel. Top. Quantum Electron. Lasers Med. Biol. (1999). 8) C. Seaton and S. Butcher: Laser Focus World, 5, 137 (1999). 9) E. N. Glezer et al.: Opt. Lett. 21, 2023 (1996). 10) P. Pronko, S. K. Dutta, J. Squier, J. V. Rudd, D. Du, and G. Mourou: Opt. Comm. 114, 105 (1995). 11) R. S. Marjoribanks, Y. Kerachian, P. R. Hermann, S. Camacho-Lopez, and M. Nantel: Proc. Conf. Lasers and Electro-optics (CLEO) (2001 OSA Technical Digest Series) (Optical Society of America, Washington DC, 2001), p. 575. 12) M. E. Fermann, M. L. Stock, A. Galvanauskas, A. Hariharan, G. Sucha, D. Harter, and L. Goldberg: Proc. Conf. Lasers and Electro-optics (CLEO) (2000 OSA Technical Digest Series) (Optical Society of America, Washington DC, 2000), p. 83. 13) These values are an upper limit, determined mostly by the measurement technique. 14) G. Sucha and H. Endert: Industrial Laser Solutions (2000). 15) V. Dominic et al.: Proc. Conf. Lasers and Electro-optics (CLEO) (Optical Society of America, Washington DC, 1999), paper CPD-11. 16) M. E. Fermann: Opt. Lett. 23, 52 (1998). 17) D. Strickland and G. Mourou: Opt. Commun. 56, 219 (1985). 18) G. C. Cho et al.: Proc. Conf. Lasers and Electro-optics (CLEO) (2000 OSA Technical Digest Series) (2000), paper CMW2. 19) A. Galvanauskas, Z. Sartania, and M. Bischoff: Advanced Solid State Lasers (2001). 20) A. Galvanauskas and M. E. Fermann: Proc. Conf. Lasers and Electro-optics (CLEO) (2000 OSA Technical Digest Series) (Optical Society of America, Washington DC, 2000), paper CPD3. 21) A. Galvanauskas et al.: in Ultrafast Phenomena XI, edited by T. Elsaesser, J. G. Fujimoto, D. A. Wiersma, W. Zinth (Springer-Verlag, Berlin, 1998), p. 60. 22) A. Galvanauskas, A. Hariharan, and D. J. Harter: Ultrafast Phenomena XII (Springer, Berlin, 2001).

27