In recent years, mid-infrared (IR) sources operating in the 3–5 μm spectral region have become ubiquitous tools in both industry and research, finding applications in areas including spectroscopy, materials processing, and defense [1,2]. Depending on the laser parameters required, a choice of sources exists that emit in this portion of the “molecular fingerprint region.” These include quantum-cascade (QC) semiconductor lasers [3], Cr/Fe-doped II-VI chalcogenide solid-state lasers [4,5], and parametric frequency conversion sources [6]. Of the two direct emission routes, QC lasers are available across a wide spectral range (3–25 μm), but power scaling opportunities are limited. In contrast, Cr:ZnSe/S lasers have been demonstrated with output powers $>10\mathrm{W}$ [4], but their gain bands do not extend beyond 3.1 μm [5]. While Fe:ZnSe/S crystals provide gain in the 4–5 μm window, they often require cryogenic cooling to lase efficiently and demand complex pumping schemes [4]. Similarly, Cr:CdSe/S lasers offer the potential for wide tuning at wavelengths $>3\mathrm{\mu m}$, but remain a relatively underdeveloped technology [7,8]. Alternatively, parametric wavelength conversion offers high average powers, supports large pulse energies, and wide spectral tunability dependent on the combination of pump source and nonlinear crystal [6,9].

Parametric sources based on a ${\chi }^{(2)}$ nonlinearity can be realized using distinct architectures: optical parametric oscillators (OPOs), optical parametric amplifiers (OPAs), and optical parametric generators (OPGs). In particular, ytterbium (Yb) fiber laser pumped OPOs are capable of producing multiwatt-level average powers, from the CW to femtosecond regime, often accompanied by a wide spectral tuning range [10–12]. However, OPOs have a number of drawbacks inherent in resonant cavity based systems, including: the need for optics with specialist broadband transmission coatings, precise alignment, and intracavity spectral tuning, while their repetition rate is fixed by the cavity length. OPG, OPA, and difference-frequency generation (DFG), however, provide single-pass amplification, simplifying the optical configuration, and removing the constraint of a resonant cavity. Unfortunately, OPG can result in broad signal and idler linewidths, while the required high pump energies of an unseeded scheme can approach the damage threshold of the crystals used, leading to long-term reliability issues or catastrophic damage. While the distinction between OPA and DFG is often unclear, here we use the term DFG to describe a three-wave process (${\omega }_{\mathrm{pump}}={\omega }_{\mathrm{signal}}+{\omega }_{\mathrm{idler}}$) involving a pump, signal, and idler, where the strength of the signal relative to the pump is significant (i.e., a high-power signal regime).

Recent demonstrations of high-average power mid-IR DFG sources include: $>3.5\mathrm{W}$ CW by mixing two high-power Yb and Er fiber lasers [13]; $>1\mathrm{W}$ of average power by mixing a high-pulse energy nanosecond Yb master oscillator power fiber amplifier (MOPFA) and a 1.55 μm CW laser diode [14]; and 1.7 W of average power using a filtered ASE source at 1.541 μm and a high-energy picosecond Yb-MOPFA [15]. In this Letter, we utilize synchronized Yb and Er picosecond MOPFA systems. The advantage of our approach compared to a pulsed pump and CW signal scheme is threefold: first, temporal tuning can be realized by strobing the pump pulse through the signal pulse; second, the pump pulse-energy/peak-power requirements for efficient conversion are relaxed due to the intensity of the signal; this is particularly important in PPLN crystals that exhibit relatively low damage thresholds (typically quoted in the range $0.1–1\mathrm{GW}/{\mathrm{cm}}^2$ at 1.064 μm [14,15] for similar peak/average powers and pulse durations to those used in this work [manufacturer and crystal specific]); third, extremely high conversion efficiencies can be achieved. We note that the use of synchronized sources for single-pass DFG has been demonstrated before, both in the femtosecond and picosecond temporal regimes [16–18]. However, here we fully exploit for the first time the very high nonlinear conversion achievable using such a scheme, reporting record efficiencies.

The experimental setup is shown in Fig. 1. The pump arm consists of an actively mode-locked external cavity laser diode (1.063 μm ECLD), with feedback provided by a polarization-maintaining fiber Bragg grating (PM-FBG), and driven by an electrical pulse generator (EPG). The DC-bias (DC-B) voltage allows for optimization of the pulse duration and extinction ratio. The 150 ps pulses [Fig. 2(b)] have a repetition rate of 39.945 MHz, half the fundamental cavity frequency of the ECLD. The pulses are then amplified in two Yb-doped fiber amplifiers (YDFAs), with interstage amplified spontaneous emission (ASE) suppression provided by a 1 nm full-width half-maximum (FWHM) tunable bandpass filter (TBP). Due to the use of isotropic gain fiber, polarization control comprising a quarter/half/quarter waveplate combination (WPS) is required to correct for polarization rotation in the amplifier stages. The power level of the Yb-MOPFA after the Faraday isolator (ISO) is $\sim 23\mathrm{W}$, with the spectral and temporal characteristics of the pump after amplification shown in Fig. 2(a) and Fig. 2(b), respectively. A beam expander (B-EXP––L1, $f=50\mathrm{mm}$; L2, $f=81\mathrm{mm}$) then resizes the pump beam for optimal spatial overlap with the signal beam in the nonlinear crystal. Finally, a half-wave plate (HWP) and polarizing beam splitter (PBS) are used to control the pump power delivered to the nonlinear crystal.

 

Fig. 1. Mid-IR DFG configuration. See body text for abbreviation definitions.

Download Full Size | PDF

 

Fig. 2. (a) and (b): Pump; and (c) and (d): signal; spectral and temporal characteristics at MOPFA outputs (23.0 W and 2.1 W, respectively).

Download Full Size | PDF

The signal arm consists of a tunable ECLD (1500–1580 nm), pulsed by a Mach–Zehnder amplitude modulator (MZAM) driven by a second EPG, producing 400 ps pulses [Fig. 2(d)] at a repetition rate of 39.945 MHz. The DC-B voltage allows for optimization of the pulse duration and extinction ratio. The two EPGs are synchronized using a common clock. The signal is then amplified in two Er-doped fiber amplifiers (EDFAs), with a 2% tap-coupler (TAP) to monitor the pulse extinction ratio. The output of the second EDFA is collimated with a lens (L3, $f=12.5\mathrm{mm}$) chosen to match the spot sizes of the pump and signal beams in the crystal. The Er-MOPFA provides 2.1 W of average power at the output of L3. A quarter/half WPS is used to linearize the output polarization from the non-PM EDFAs. The spectral and temporal characteristics of the signal after amplification are shown in Fig. 2(c) and Fig. 2(d), respectively.

The pump and signal are combined using a beam splitter (BS, highly reflective at 1.55 μm and highly transmissive at 1.06 μm). The B-EXP and L3 allow the ratio of the beam diameters of the pump to signal to be adjusted to 1.06/1.55, ensuring equal focal spot sizes in the center of the crystal. Lens 4 (L4, $f=150\mathrm{mm}$) is then used to focus the spatially and temporally overlapped pump and signal beam into the MgO:PPLN crystal. The pump beam is focused to a $1/{\mathrm{e}}^2$ beam diameter of 150 μm, measured using a scanning-slit beam profiler. The intensity of the pump at the focal spot in the crystal, at the maximum available pump power, is estimated to be $28\mathrm{MW}/{\mathrm{cm}}^2$, well below the $0.1–1\mathrm{GW}/{\mathrm{cm}}^2$ damage threshold range—one of the benefits of a synchronously pumped DFG system.

The MgO:PPLN crystal is mounted in a copper oven, capable of maintaining crystal temperatures in the range $20–250\pm 0.1°\mathrm{C}$. Both the input and output faces of the crystal are antireflection (AR)-coated for pump, signal, and idler wavelengths. The crystal is 40 mm long with an aperture of $1\times 10\mathrm{mm}$, and contains five poling periods in the range 29.52–31.52 μm. For the results presented here, a track with a period of 29.98 μm is selected. The corresponding phase-matching curve for this track is shown in Fig. 3(d), calculated using the Sellmeier equations and temperature-dependent corrections given in Refs. [19,20].

 

Fig. 3. (a) Maximum idler average powers across tuning range. (b) Idler wavelength tuning through signal wavelength scanning. (c) Idler spectrum at 3.334 μm. (d) Calculated phase-matching curves (blue) for signal/idler wavelengths at a pump wavelength of 1.063 μm and a grating pitch of 29.98 μm, experimental signal/idler wavelengths overlaid (orange circles). (e) Typical idler power stability.

Download Full Size | PDF

The pump, signal, and generated idler are collimated using lens 5 (L5, $f=100\mathrm{mm}$, uncoated ${\mathrm{CaF}}_2$) before being spatially dispersed using an uncoated ${\mathrm{CaF}}_2$ prism. Initially, a liquid nitrogen-cooled InSb detector is used to record the idler power and optimize the pump/signal overlap, both spatially and temporally in the crystal. The EPGs provide electrical delay control, enabling facile temporal overlap of the pump and signal pulses. An electrical rather than an optical delay also negates problems associated with beam misalignment when changing the optical path length of the pump relative to the signal.

Figure 3(b) shows the generated idler spectra, while tuning the signal wavelength over the range 1.535–1.570 μm, measured using a scanning monochromator. The temperature of the crystal is tuned over the range 130°C–210°C to maintain phase-matching. The tuning range of the signal is limited by the gain bandwidth of the Er-MOPFA. Greater than 3.4 W of average idler power is maintained across the full tuning range [Fig. 3(a)]. The experimental signal/idler wavelengths (orange circles) are shown in Fig. 3(d), plotted as a function of the phase-matched crystal temperature; excellent agreement between theory and measured values is observed. Figure 3(c) highlights the spectral shape of the idler. The double peak structure is attributed to the initial profile of the pump spectrum [Fig. 2(a)]. The idler output power exhibits excellent power stability [Fig. 3(e)], with a root-mean-square power deviation of less than 0.4% over a 90 min period.

A representative evolution of the generated power is shown in Fig. 4(a), for a pump/signal wavelength of 1.063/1.560 μm. The data represents the average powers generated in the DFG process, with the input signal power (1.87 W) subtracted from the generated signal. A maximum idler power of 3.66 W is obtained at a wavelength of 3.334 μm and a pump power of 17.1 W/2.5 kW (av./pk.). In all the power metrics presented, we consider the loss contribution from optics after the crystal; thus the data represents the power measured directly at the exit face of the crystal.

 

Fig. 4. (a) Signal (blue), idler (orange), and total (signal plus idler—green) powers generated for a given pump average (bottom axis)/peak-power (top axis), with pulse energy shown on right-hand axis. (b) Signal, idler, and total pump power conversion. Connecting lines to guide the eye only, circles represent experimental data.

Download Full Size | PDF

The corresponding conversion efficiency of the process is shown in Fig. 4(b). We define pump conversion as the percentage of the total input pump power converted to either the idler, the amplified signal, or the total generated DFG power. The conversion efficiencies reach a maximum of $\sim 26\%$ for the idler, $\sim 52\%$ for the signal, and $\sim 78\%$ for the combined power (amplified signal + idler). Beyond a pump power of 8 W, we observe a roll-off in the conversion efficiency, attributed to back-conversion of the signal and idler power to the pump beam, evidence of which is presented in streak camera traces of the pump pulses at increasing pump power levels [Figs. 5(a)–5(e)]. In Figs. 5(a)–5(c), the center of the pump pulse is initially depleted due to increasing parametric conversion of the pump power to the signal and idler wavelengths, resulting in an effective increase in the pump pulse duration (compare to the undepleted pump pulse duration of 150 ps [Fig. 2(b)]). In Fig. 5(d), the point of maximum pump conversion efficiency, the center of the pulse has hollowed out due to extreme pump power conversion. Then in Figs. 5(e)–5(f) the center of the pulse reappears as the pump light is back-converted from the signal and idler wavelengths.

 

Fig. 5. (a)–(f) Temporal evolution of pump pulse with increasing pump power after the crystal. Annotations in each figure indicate the pulse duration (FWHM) and average pump power level at the input crystal face. All pulse intensities are normalized.

Download Full Size | PDF

Pump back-conversion notwithstanding, we note that the reported efficiencies are, to the best of our knowledge, significantly higher than comparable state-of-the-art results using a CW-seeded single-pass DFG/OPA scheme. We anticipate, with improved focusing conditions and/or optimization of the pump peak-power, it should be possible to maintain the high conversion efficiencies, even at high pump powers (e.g., $>8\mathrm{W}$), allowing significant power-scaling of the pump, and corresponding power-scaling of the mid-IR idler radiation. We also note that as we are operating beyond the point of the maximum pump depletion [see Fig. 4(b)], we expect the beam quality of the signal and idler to be degraded due to the pump back-conversion. Again, through shifting the point of maximum pump conversion to higher average powers, we expect to avoid such signal and idler beam quality degradation issues in the future. Finally, we add that at no point did we observe any photorefractive damage or green-induced IR absorption effects occurring due to either the high peak or average intensities in the crystal.

Due to the non-polarization maintaining design of the synchronous MOPFAs, in order to extract the maximum mid-IR power while minimizing output power fluctuations, we operate the source in the heavily saturated signal power regime. Figure 6(a) shows the dependence of the generated idler and amplified signal powers on the input signal power at a fixed pump power of 17.1 W. Both idler and amplified signal saturate after 0.5 W of signal power. The corresponding gain of the amplified signal is shown in Fig. 6(b). Due to fluctuations of the input signal across the gain band of the Er-MOPFA, operating in the saturated regime also improves the output power stability when tuning the source, as confirmed by the idler power stability curve [Fig. 3(e)].

 

Fig. 6. (a) Generated idler and amplified signal powers against input signal power; (b) signal gain against input signal power. In both figures, the maximum available pump power of 17.1 W was used. Connecting lines to guide the eye only.

Download Full Size | PDF

In summary, we have presented a high average power ($>3.6\mathrm{W}$ at 3.334 μm), high conversion efficiency (maximum pump-to-DFG power conversion 78%) picosecond source of mid-IR radiation tunable from 3.28 to 3.45 μm. Using a synchronous signal and pump pulse lowers the peak intensity requirements of the pump laser, and thus avoids the need to operate the source close to the damage threshold of PPLN to achieve high efficiency. This approach prolongs crystal life, allows greater average power-scaling potential of DFG-based pulsed-pump sources, and supports record high conversion efficiencies. We anticipate that the added freedom, e.g., repetition-rate selectability, and reduced complexity of nonresonant, single-pass parametric sources will make them increasingly attractive systems for mid-IR applications. Ongoing work is aimed at shifting the point of maximum conversion to higher average power levels, to extract the maximum possible power from single-pass DFG systems while maintaining the excellent beam quality that fiber pump sources can inherently provide.