Multi-parameter tunable ultrafast fiber lasers are driving many emerging areas of femtosecond biomedical photonics. Since solid-state ultrafast lasers are difficult to independently tune the three parameters of central wavelength, repetition frequency and pulse width with guaranteed output pulse energy, femtosecond biomedical photonics usually uses pulse-selective single fiber chirped pulse amplifiers (pp-FCPA) with optical parametric amplifiers (OPAs) as the driving light source. However, the complex spatial components of OPA greatly affect the beam quality and environmental immunity of the system, and the tedious routine maintenance is beyond the knowledge of life scientists. Therefore, to replace OPA technology and take advantage of the pp-FCPA system, the authors developed a wavelength-tunable femtosecond light source based on supercontinuum generation.
Figure 1 illustrates three common methods for generating the supercontinuum spectrum. Method 1 uses an all-fiber fusion architecture, which has the most compact structure and excellent environmental stability, but the transmission in the fiber is mostly picosecond lasers, commonly found in commercial lasers. Method 2 uses a commercial closure with fiber end caps and mode extensions as an additional nonlinear wavelength converter for titanium sapphire oscillators to support wavelength conversion of femtosecond pulses. Method 3 is similar to Method 2, but incorporates a pp-FPCA front-end with fiber advantages by coupling a high-energy femtosecond pulse into a section of photonic crystal fiber to produce a coherent supercontinuum spectrum. It is this third method that the authors use in this paper.
However, it was found that the fiber used for supercontinuum generation is usually damaged after about 100 hours of cumulative operation. Irreversible optical damage significantly limits the lifetime of the supercontinuum source. Therefore, it is necessary to determine the principle of this optical damage in order to find a means of circumventing it. If the photodamage is caused by airborne contaminants in a non-superclean room environment and/or spatial coupling of high peak power at the fiber endface, it can be addressed by commercially available photonic crystal fiber end caps or by collapsing specific apertures in the fiber endface.
Table 1 lists the three experimental schemes used by the authors to study the damage mechanism of optical fibers. Scheme 1 coupled an input pulse with a central wavelength of 1030 nm, a repetition frequency of 10 MHz, and a pulse width of 280 fs into a 25 cm section of LMA-PM-15 fiber, and after repeated experiments, all found that the fiber was damaged after 100 ± 40 hours of cumulative operation. Scheme 2 used a different drive source and photonic crystal fiber, but the peak power density coupled to the fiber end face remained the same as that of Scheme 1. Scheme 2, however, results in photodamage within 10 ± 2 hours. The location where the optical damage occurs differs between these two scenarios: the optical damage in scenario 1 is located <10 cm from the incident end of the fiber, whereas the optical damage in scenario 2 is located <1 cm from the incident end of the fiber. This difference indicates that the cause of fiber damage is not air contaminants in the environment or the high peak power density at the time of coupling, and that optical damage cannot be avoided by fiber end caps. Upon analysis, this fiber damage can be explained by the optical waveguide theory of long-period fiber gratings (LPFG). When a pulse is coupled into the fiber, part of the energy enters the core while the other part is transmitted into the cladding. When the light from the core mode and the cladding mode interfere with each other and generate standing waves, an LPFG is written into the fiber. the shorter the period of the LPFG, the more periods are contained in the same fiber length, and the more easily the fiber is damaged.
To verify this idea, the authors chose the LMA-PM-40-FUD fiber with a mode field diameter of 32 μm in Scheme 3. The LPFG period was calculated to be about 9 cm, and the fiber length of 9 cm is less than one cycle, so the fiber damage effect caused by LPFG will theoretically disappear. Experimentally, the optical system of scheme 3 also does remain stable after 2000 hours of cumulative operation.
Figure 2 shows the multiparametric tunable femtosecond light source built by the authors based on Scheme 3. The entire light source consists of a pp-FCPA system with a repeat frequency tunable from 1-10 MHz as the front end, and a photonic crystal fiber avoiding optical damage caused by LPFG as the unit generating the supercontinuous spectrum, i.e., a fiber nonlinear converter (FNWC). After spectral broadening the pulses are directed to a programmable pulse shaper. By selecting a specific filter window and dispersion compensation amount, the central wavelength can be tuned in the range of 950-1110 nm and the pulse width can be tuned in the range of 40-400 fs. In addition, the final output pulse can be transmitted with a section of low-dispersion Kagome hollow-core fiber patch cable, allowing this light source to be easily switched between different application modules.
In summary, the authors have developed a reliable accessory for femtosecond fiber lasers with substantial tunability in terms of re-frequency, wavelength, and pulse width, which explains and suppresses optical damage in the coupled fiber system, and whose corresponding integrated laser system is highly stable, promising to broaden the applications of tunable ultrafast lasers in biological and medical fields.
