15 Feb 2008
One of today's most active research areas is exploiting ultrafast fibre lasers that produce picosecond or femtosecond pulses. Frank Lison and Thomas Renner of TOPTICA provide a back-to-basics look at the technology and review the emerging applications that could benefit.
Despite the fact that ultrafast modelocked lasers producing optical pulses in the picosecond and femtosecond regimes are established technologies, their main usage is still restricted to laboratories and high-end applications. In order to move forward, hands-off operation must be improved and costs need to be reduced. A key factor to achieving both of these objectives is switching from solid-state to fibre technology.
Short pulses for large markets
Applications in science and industry that use ultrafast lasers exploit a number of physical advantages. For example, materials processing, the largest market for lasers, is currently moving away from thermal (using continuous-wave [CW] or quasi-CW light) to non-thermal ablation. Here, the energy deposition and consequently the laser pulse duration must be shorter than the heat dissipation timescale. Typical heat dissipation times of metals, ceramics, semiconductors and polymers are on the order of several picoseconds meaning that picosecond or sub-picosecond laser pulses are essential.
Another major scientific application is time-resolved analysis, which typically uses a pump-probe set-up. A key requirement is that both the pump and the probe pulses must be shorter than the time constant of the process being studied. In biophotonics, fluorescent dyes can have lifetimes ranging from nanoseconds to picoseconds whereas in solid-state and physical chemical analysis the decay times can vary from a few femtoseconds up to 1 ns.
Emerging applications such as nonlinear microscopy and terahertz generation benefit from the high peak powers produced by ultrafast lasers. For example, nonlinear contrast techniques such as multiphoton excitation and coherent anti-Stokes Raman scattering both rely on high peak powers. The terahertz waves used in homeland security, food inspection or medical analysis can be generated when ultrafast laser pulses are focused onto a suitable semiconductor material.
The common base for all of these examples, and indeed a much larger variety of applications, is that they all require pulse durations either in the picosecond or femtosecond regime. While some of these applications are still under development, many are already established in laboratories and use modelocked Ti:sapphire or other solid-state lasers. The ability to expand out of the laboratory has to date been limited by the high cost per watt of the light emitted by the laser and the lack of industrialization/hands-off operation.
Reliability and cost
The telecoms revolution has benefited many types of lasers, including modelocked ultrafast. Since this variety of laser normally operates in the low- and medium-power regimes, the lifetime and costs of many optical components, including pump diodes, fibre splitters, combiners and polarization elements, have profited directly from telecom industry derived experiences.
Using only very short free-space set-ups, or avoiding them altogether, leads to much higher mechanical and thermal stabilities compared with conventional Ti:sapphire ultrafast systems or other diode-pumped solid-state lasers. With most of the laser system being fibre-based, the cavity can be packed into compact geometries and/or adapted to specific OEM customer requirements for further integration into the complete laser product.
Turnkey operation is achieved by using semiconductor pump sources and alignment-free fibre optics. Even the cooling of several-watt pump laser diodes is achievable by convection and conduction only.
Most fibre lasers have non-fibre components inside the cavity that are in principle bulk optical elements coupled to the fibre's input and output. An even wider definition also includes systems with free-space dispersion compensation, but to be classed as a fibre laser the gain medium should at least be all-fibre based.
Today, fibre lasers cover the complete time regime from CW through to modelocked femtosecond pulses. CW fibre lasers are available in power levels up to several kilowatts and are attractive economic alternatives to Nd:YAG, Yb:Disc or high-power CO2 lasers for materials-processing applications such as welding and cutting. Such fibre lasers are currently replacing conventional gas or solid-state lasers in a number of applications and are available from vendors such as IPG and SPI. Products are also coming through from Trumpf and Rofin and there have been announcements and rumours about future developments at Newport/Spectra Physics and Coherent.
Electrically pulsed fibre lasers (kilohertz repetition rate, 1 ms pulse regime) can enter the millijoule energy region and are currently the first choice in higher repetition-rate, medium peak-power applications such as the micro-cutting of stents.
Q-switched fibre lasers (microjoule to millijoule power level, 10...400 kHz repetition rate) have substantially changed the laser marking world. Such Q-switched systems emit pulses ranging from a few to hundreds of nanoseconds.
And finally, modelocked fibre lasers are currently available from medium-sized companies such as IMRA, Koheras, Fianium, Thorlabs and TOPTICA. Such lasers are finding a growing number of applications such as ophthalmology, metrology, biophotonics, telecommunications, terahertz generation, spectroscopy and micro-material processing. Due to their superb pulse performance, these modelocked lasers are also ideal for use as seed sources in high-power laser materials-processing systems.
Ultrafast fibre lasers
In a fibre laser, an optical waveguide confines the electrical field to a very small area of the fibre core. Although power levels within fibre lasers are comparable with or lower than other laser sources, the combination of power level and interaction length leads to nonlinearities.
Due to the long gain medium – typically in the range of tens of centimetres up to 1 m compared with about 1 cm in solid-state lasers – several fibre-laser parameters differ by orders of magnitude compared with solid-state laser technology. Crucial decisions defining the set-up of an ultrafast fibre-laser system are: the modelocking mechanism, dispersion management, short-pulse generation and output power requirements.
In principle, the modelocking mechanisms developed for solid-state lasers can be transferred to ultrafast fibre lasers although one has to keep in mind the much higher roundtrip gain and tolerable losses of fibre lasers. Today most commercially available ultrafast fibre-laser systems use saturable absorbing mirrors as a modelocking element.
It is important to remember that dispersion in a fibre laser will be more than one order of magnitude larger than in a solid-state laser. Dispersion compensation in erbium fibre lasers can be accomplished using pure fibre solutions, whereas ytterbium fibre technology requires the use of gratings – preferably chirped fibre Bragg gratings – or photonic crystal fibre (PCF).
With proper cavity dispersion management and careful selection of the out- coupling position, sub-50 fs pulses have been generated directly from an erbium fibre oscillator. For ytterbium, even shorter pulses of 33 fs have been demonstrated using a free-space prism-grating sequence inside the resonator for third-order dispersion compensation. More typical results in a standard system are in the range of 100 to a few hundred femtoseconds. These output pulses are then easily amplified and compressed in a nonlinear amplifier to power levels up to a few hundred milliwatts and pulse durations down to 30 fs.
Due to their compact size, high efficiency and low power requirements, ultrafast fibre lasers offer clear advantages when compared with solid-state technology. Scientific applications specifically benefit from the superior stability of fibre-laser systems, especially in the acoustic frequency range. Industrial applications appreciate hands-off operation, beam-pointing stability as well as low investment and running costs. Last but not least, users in biophotonics and healthcare appreciate the wide and easily accessible wavelength coverage and short coherence length.
Fibre lasers are the ideal way to generate reliable and cost-effective short pulses. Wavelength, pulse duration and repetition rate can all be adjusted modularly over a wide range (see table). The typical output powers of pure fibre oscillators are however in the single milliwatt region. A fibre-based amplification stage can boost the output power to several hundred milliwatts. Higher powers can currently not be achieved with core-pumped concepts due to the high peak powers and the corresponding optical nonlinearities.
Finding ways to reach power levels of up to 100 W is an active research field and current ideas include using large mode-area fibre amplifiers and chirped pulsed amplification. Alternatively, higher powers can be achieved by adapting existing master oscillator power amplifier (MOPA) concepts. Typical amplifier lasers are conventional disc or rod laser systems.
MOPA concepts with fibre oscillators are advantageous because for example, the power amplification and pulse generation are intrinsically decoupled and additional pulse pickers between oscillators and amplifiers can easily reduce the repetition rate into the kilohertz regime without changing the average power level.
Because of their high peak power, ultrafast fibre lasers are ideal pump sources for generating supercontinuum in the near-infrared and visible using PCFs, tapered fibres and highly nonlinear fibres (HNLF). The output spectrum of these fibres can be tailored to individual requirements by adjusting fibre parameters such as mode-field diameter, dispersion profile and length.
Erbium ultrafast fibre lasers emitting at 1550 nm can generate a supercontinuum in HNLFs spanning from 900 nm to well over 2 µm. This spectrum can then, for example, be frequency-doubled to generate tunable narrow linewidth visible light anywhere between 480 and 700 nm. If maximum output power per nm and power stability are not a priority, similar results can be achieved by generating a supercontinuum in a PCF with a ytterbium fibre laser emitting at 1030 nm. The advantage of this second approach is that individual lines can be extracted from the spectrum by using acousto-optical tunable filters.
• Frank Lison is vice-president of research and development, and Thomas Renner is vice-president of sales and marketing at TOPTICA Photonics AG. For more information, see www.toptica.com.
• This article originally appeared in the February 2008 issue of Optics & Laser Europe magazine.