Yb:YLF (${\rm Yb}:{\rm LiYF}_4$) is a well-known broadly tunable continuous-wave (cw) laser source at room temperature (RT) [1], with demonstrated tuning ranges covering the 998–1064 and 996–1076 nm regions for the ${E}//{a}$ $(E \perp c, \sigma )$ and ${E}//{c}$ ($\pi$) axis, respectively [2]. At RT, the ${E}//{c}$ axis of Yb:YLF has a higher gain and a broader overall bandwidth (e.g., see Fig. 2 in Ref. [3]), making it the ideal choice. The broad bandwidth of Yb:YLF has already been explored in mode-locking studies at RT [3,4]. Pulses centered at 1028 nm with pulse widths as short as 196 fs and average powers up to 120 mW were obtained from a multimode diode-pumped Yb:YLF laser at a pump power of 4.7 W [3]. The system was passively mode-locked with a saturable Bragg reflector (SBR) [5,6] at 55 MHz, and the corresponding pulse energies and peak powers were 2.2 nJ and 9.8 kW, respectively [3]. Later, via improved dispersion management, pulses as short as 87 fs and 35 mW average power were also achieved [4]. The average powers in these initial mode-locking studies were in the multi-milliwatt range due to the power handling limitations of thermo-mechanically weak Yb:YLF gain medium at RT. Similarly, to the best of our knowledge, the highest cw power demonstrated at RT is also only around 1.2 W [7], and quasi-cw operation was used to boost the output power to 4.5 W [2].

It is well known that operating Yb-based systems at cryogenic temperature allows [8–11] (1) a four-level laser structure, minimizing self-absorption losses and lasing thresholds, (2) improvement of thermal material parameters, and (3) increasing of the emission cross section (ECS) and hence gain (e.g., see Table 1 in [12] for comparison of relevant Yb:YLF parameters at cryogenic and RTs). Taking advantage of cryogenic operation, cw powers above 300 W [12,13], Q-switched ns pulses with around 150 W average power [14], regenerative amplifiers with 70 W average power [15], and multi-pass amplifiers with 100 W average power [16] were already demonstrated with Yb:YLF. At cryogenic temperatures, unlike the RT case, the emission bandwidth of the ${E}//{a}$ axis of Yb:YLF provides a broader and smoother gain profile (e.g., see Fig. 4 in Ref. [9] or Fig. 1 in Ref. [12]). Hence, despite the smaller gain, most of the amplification studies with Yb:YLF have been based on the ${E}//{a}$ axis [17]. In the ${E}//{a}$ axis, at cryogenic temperatures, a cw tuning range covering the 995–1020.5 nm region was also demonstrated [12], and amplification studies confirmed the possibility to amplify sub-250 fs pulses [8]. Hence, in principle, at cryogenic operation, the ${E}//{a}$ axis of Yb:YLF is suitable for direct generation of high-power sub-100 fs pulses around 1016 nm in mode-locked operation. Despite that, to the best of our knowledge, there is no report of mode locking of Yb:YLF at cryogenic temperatures.

In this Letter, we present the first mode-locking study with Yb:YLF gain media at cryogenic temperatures. Via passive mode locking with an SBR, 3–5 ps long pulses with up to 28 W average power were achieved at a repetition rate of 46.45 MHz. The achieved peak power (${ \gt }{120}\;{\rm kW}$) and pulse energies (${ \gt }{600}\;{\rm nJ}$) are two to three orders of magnitude higher than earlier RT results.

Figure 1 shows the schematic of the SBR mode-locked cryogenic Yb:YLF laser resonator. (The SBR is also known as the semiconductor saturable absorber mirror [SESAM]). A 2 kW fiber-coupled diode module at 960 nm was used as the pump source. The pump output from the fiber tip is first collimated with a 72 mm focal length lens (f1), and then reimaged to a diameter of 1.7 mm inside the Yb:YLF crystal using 250 (f2) and 150 mm (f3) focal length lenses. The pump beam had a full divergence angle of 25.2° and an estimated ${{M}^2}$ of 220. A 0.5% Yb-doped 20 mm long Yb:YLF crystal (${10}\;{\rm mm} \times {15}\;{\rm mm}$ cross section) with 3 mm long un-doped end caps diffusion bonded on both ends was used as the gain element. The ${E}//{a}$ axis of the crystal was employed in the lasing experiments, and the gain element was indium bonded from the top side to a cold head, which was cooled by boiling liquid nitrogen. The crystal had an estimated peak absorption coefficient of ${1.85}\;{{\rm cm}^{ - 1}}$ (960 nm, ${E}//{a}$ polarization) and effectively absorbed 65–80% of the incident pump, depending on saturation conditions and pump spectral overlap (pump FWHM: 4 nm). In the laser experiments, an X cavity with two curved dichroic mirrors (DMs) with a separation of 75 cm and a radius of curvature (ROC) of 10 m were used. An additional curved high-reflector mirror (CM) with a ROC of 1 m was employed to focus the intracavity laser beam on the SBR (BATOP Inc., SAM1040-1-1ps). The SBR had a company specified central reflectivity wavelength of 1040 nm, a modulation depth of 0.6%, a nonsaturable loss of around 1%, a saturation fluence of ${70}\;{\unicode{x00B5}}{{{\rm J}/{\rm cm}}^2}$, a damage threshold of $3\;{\rm mJ}/{\rm cm}^2$, and a relaxation time of 1 ps. In the experiments, output couplers with transmissions of 10%, 20%, and 40% were investigated. A 2 mm thick crystal quartz birefringent filter (BRF) and a thin-film polarizer (TFP) were inserted at Brewster’s angle for tuning experiments [18]. For the cold cavity, the spot size (${{1/e}^2}$ radius) inside the Yb:YLF crystal on the OC and on the SBR are estimated to be around 900, 600, and 200 µm, respectively. Note that the limited brightness of the high-power pump module required the usage of a relatively large spot size inside the crystal.

 

Fig. 1. Experimental setup of the diode-pumped SBR mode-locked cryogenic Yb:YLF laser resonator. DM, dichroic mirror; CM, curved high-reflector mirror; f1–f3, pump beam shaping optics; BRF, birefringent tuning filter; OC, output coupler.

Download Full Size | PDF

Figure 2 shows the variation of Yb:YLF laser output power with absorbed pump power at different output coupling values. Here, as an example, we will elaborate on the laser performance with the 20% OC. Using the 20% OC, the lasing threshold was 95 W and, at pump powers up to around 110 W, the laser generated unstable cw output. Between 110 and 120 W pump power, ${q}$-switched mode-locked operation was observed. Stable cw mode locking was achieved at pump powers above 120 W. Note that stable cw mode-locked operation points are marked by filled markers in Fig. 2. At these operation points, mode-locked operation was self-starting and robust against environmental fluctuations, and could be sustained for about an hour. (A liquid nitrogen auto-refill system needs to be implemented for longer operation periods.) A cw mode-locked average power of 12.2 W was attained at an absorbed pump power of 135 W (${\rm incident}\;{\sim}{200}\;{\rm W}$). The optical spectrum of the pulses was centered at 1018.8 nm and had a FWHM of around 0.45 nm (Fig. 3). The mode-locked laser output was TM polarized. The measured autocorrelation indicated a pulse width of 3 ps (assuming ${{\rm sech}^2}$ pulse shape), resulting in a time-bandwidth product of 0.39, a value slightly above the ideal (0.3148 for ${{\rm sech}^2}$). The repetition rate is measured as being 46.45 MHz, and an RF spectrum is shown in Fig. 4, which confirms clean mode-locked operation. The corresponding pulse energy and peak power are then around 262.65 nJ and 77.2 kW, respectively. Note that increasing the pump power further provided little benefit (3.4 ps long pulses with 12.5 W average power at 145 W). A further increase in pump power resulted in pulse breakup instabilities, increased pulse widths, and a decrease in output power. The estimated peak fluence on the SBR was around ${2}\;{{{\rm mJ}/{\rm cm}}^2}$ (${\sim} \times 30\;{\rm times}$ SBR saturation fluence) [3], explaining the observed limitation in obtainable pulse energies via SBR reflectivity rollover [19–22].

 

Fig. 2. Measured laser output average power versus absorbed pump power for the SBR mode-locked cryogenic Yb:YLF laser. The data are taken using 10%, 20%, and 40% output couplers. The empty markers denote cw or ${q}$-switched mode-locking regimes, whereas the filled markers represent stable cw mode locking. Inset figure: measured near-field beam profile at the laser output with the 40% output coupler (at 28 W average power level).

Download Full Size | PDF

 

Fig. 3. (a) Measured spectrum and (b) autocorrelation trace for the 3 ps, 262 nJ pulses with 12.2 W of average power taken using the 20% OC.

Download Full Size | PDF

 

Fig. 4. Measured microwave spectrum indicating clean mode locking at 46.45 MHz repetition rate.

Download Full Size | PDF

A similar trend is observed for the 10% and 40% output couplers. With the 10% OC, the increased intracavity power put an earlier onset on obtainable pulse energies, and we could only achieve 3.4 ps long pulses with 4.8 W average power at an absorbed pump power of 80 W (pulse energy: 103 nJ, peak power 26.8 kW). Employing the 40% OC, the system generated 4.2 ps long pulses with up to 28 W average power around a central wavelength of 1017.6 nm (absorbed pump power: 330 W). The pulses were nearly transform limited with a time-bandwidth product of 0.36 (Fig. 5). The corresponding pulse energy and peak power were 602 nJ and 126.5 kW, respectively. The inset picture in Fig. 2 shows the beam profile at 28 W power level, confirming ${\rm TEM}_{00}$ operation. The ${{M}^2}$ value of the output beam was estimated to be better than 1.1 in both axes. As a side note, for the 4.2 ps long 28 W pulses in our system, due to the usage of large output coupling, estimated intracavity peak powers are only around 0.3 MW and, for the estimated SPM coefficient of around 10 mrad/mW, this results in an estimated round-trip phase shift of only around 3 mrad (includes the Yb:YLF crystal, the Dewar windows, the TFP, and BRF).

 

Fig. 5. (a) Measured spectrum and (b) the autocorrelation trace for the 4.2 ps, 602 nJ pulses with 28 W of average power taken using the 40% OC.

Download Full Size | PDF

We note that cw power levels above 100 W were easily available from the same system (by replacing the SBR with an HR mirror); hence, the observed power limitation is not due to thermal effects, but due to the high fluences on the SBR. It is very well-known that, for SBR mode-locked lasers, the critical intracavity pulse energy required for stable cw mode locking ($E_{p,c}$) is proportional to the square root of the product of the effective laser mode area inside the gain medium and effective laser mode area on the SBR [19–21]. Unfortunately, the low brightness of the diode pump source at hand prevented us from implementing pump spot sizes smaller than 1.8 mm in this Letter. (For tighter focusing, the mode matching between the pump and cavity modes deteriorates.) To achieve stable and clean mode locking without any ${Q}$-switching instabilities, one then needs a tight spot on the SBR to reduce $E_{p,c}$ to a reasonable level (measured as 85, 115, and 125 nJ for the 10%, 20%, and 40% output couplers, respectively). As an unavoidable consequence, at high pumping levels, this limited the achievable pulse energies due to SBR reflectivity rollover (also known as inverse saturable absorption) [19–21]. As a further drawback, the SBR that was employed in this Letter had a design wavelength of 1040 nm and suffered from a quite large nonsaturable loss; it was actually optimized for relatively low-power Yb:YAG/Nd:YAG systems [23]. Hence, we strongly believe that, in future studies, usage of (1) proper SBRs with the correct design wavelength and characteristics optimized for high-power operation, and (2) higher brightness pump diodes with smaller spot sizes on the crystal, could significantly improve the performance above what is already achieved here [22].

In mode-locked tuning experiments, the BRF and the TFP were inserted into the cavity at Brewster angle with around 10% power penalty. Tuning was performed by adjusting the face normal rotation of the birefringent plate, and for each wavelength the incident pump power was also adjusted to attain the best mode-locking performance. The BRF-TFP pair provided an effective filtering action with a modulation depth of 100%, a free-spectral-range of around 50 nm, and a FWHM of 15 nm [18]. Using the 10% OC, the central wavelength of the mode-locked pulses could be continuously tuned from 1013.5 to 1019 nm (Fig. 6), except around 1014 nm, where there is a known dip in the ECS curve. As the laser was tuned, the pulse widths and laser average powers varied between 3–5 ps and 3–4.5 W, respectively. Due to the increased losses of the SBR at hand and the decrease in the ECS (Fig. 6), mode locking was not possible at wavelengths shorter than 1013.5 nm. Tuning of the mode-locked pulses was also feasible with the 20% and 40% OCs, with the benefit of higher average output powers, but the tuning range was limited to a 1016.5–1019 nm region. Note that the BRF used in this Letter was a simple on-surface optic axis BRF, and a properly designed off-surface optic axis BRF could significantly improve the tuning performance in a mode-locked regime [18,24]. Furthermore, as the tuning results, and the measured ECS shows, the Yb:YLF gain medium ideally possesses a gain bandwidth wide enough to enable generation of sub-100 fs pulses. For comparison, the narrow bandwidth of the well-known Yb:YAG limits the obtainable pulse widths to around 10 ps in mode-locked operation [25], whereas demonstrated mode-locked average power levels are still relatively low in alternative broadband materials such as Yb:YGAG [26]. In this initial Letter, we have focused on average power scaling of Yb:YLF mode-locked oscillators, rather than ultrashort pulse operation; hence, the pulse widths are kept relatively long on the picosecond range via operating the laser in the positive dispersion regime. Via usage of suitable well-designed negative dispersion mirrors, and implementing soliton based pulse shaping, shorter pulses should be feasible, at the expense of increased difficulty in managing the intracavity nonlinearities.

 

Fig. 6. Example spectra from the mode-locked Yb:YLF laser showing tunability of the central wavelength of the laser from 1013.5 to 1019 nm. The measured ECS of Yb:YLF for the ${E}//{a}$ axis and small signal reflectivity of the SBR are also shown.

Download Full Size | PDF

In summary, we have presented, to the best of our knowledge, (1) the first mode-locked operation of Yb:YLF at cryogenic temperatures, (2) the first report of mode-locked tuning in Yb:YLF, and (3) record average and peak powers in mode-locked Yb:YLF systems. The initial results presented in this Letter clearly show that with further progress, femtosecond high-power Yb:YLF laser systems could be developed that could become useful sources for a variety of applications including pumping of high-energy and average power optical parametric amplifiers [27], seeding of further amplifier chains [17], strong-field terahertz generation [28], spectral broadening, and compression of high-energy pulses and ultrafast x-ray generation [29,30].