Efficient, High-Power Mid-Infrared Laser for National Security and Scientific Applications

Leily Kiani (16-ERD-021)


The Lawrence Livermore National Laboratory's Fiber Laser Group developed a unique short-wave-infrared, high-pulse energy, high-average-power fiber based laser. This unique laser source has been used in combination with a nonlinear frequency converter to generate wavelengths useful for remote sensing and other mid-wave infrared applications. Sources with high average-power and high efficiency in this mid-wave infrared wavelength region are not yet available with the size, weight, and power requirements or energy efficiency necessary for future deployment. The Laboratory-developed Fiber-Laser Pulsed Source (FiLPS) design was adapted to erbium-doped silica fibers for 1.55 µm pumping of cadmium silicon phosphide (CSP). We demonstrated for the first time an optical parametric amplification of 2.4 µm light via difference frequency generation using CSP with an erbium-doped fiber source. For efficiency-comparison purposes, we also demonstrated direct optical parametric generation and optical parametric oscillation (OPO).

Background and Research Objectives

Short-wave infrared (2–3 µm) and mid-wave infrared sources (3–5 µm) currently enable the remote detection and quantification of sub-parts-per-million detection of gaseous species such as methane (CH4). Long-wave infrared sources (8–11 µm) have shown potential for standoff detection and identification of microgram quantities (a literal fingerprint) of explosive solid residue (e.g., TNT, PETN, and TATP) (Fuchs et al. 2010). Pulsed coherent light in mid-infrared atmospheric transmission windows is currently generated by using either (1) a traditional architecture based on a crystalline laser host such as neodymium-doped yttrium aluminum garnet (Nd:YAG) pumping a nonlinear optical parametric oscillator (which is limited in overall efficiency), or (2) through the direct use of quasi-continuous wave sources, such as quantum cascade lasers (QCL), which are limited in power scalability (Yao et al. 2012). In pursuit of improving deployable sensing systems capable of both high efficiency (i.e., greater than 20 percent) and standoff detection greater than one kilometer, the goal of this project was to demonstrate the utility of a fiber-based laser source that is the simplest, most efficient, and most power-scalable architecture available. The pulsed fiber-laser technology would likely be more robust than competing solid-state solutions (e.g., Tm:YAG, Er:YAG, Cr:ZnSe, and Fe:ZnSe). Such an improvement in efficiency and peak power (enabling long detection distances) would enhance the utility of remote-sensing systems.

We developed a high-efficiency, frequency-agile pulsed source using Lawrence Livermore National Laboratory's Fiber-Laser Pulsed Source (FiLPS) architecture and parametric converters for a long-pulse (nanosecond) laser at wavelengths ranging from 2 to 7 µm, in addition to a short-pulse (femtosecond) laser at the same wavelengths. We proposed to do so by enhancing the Laboratory's existing fiber-laser system to operate at higher energy, average power, and repetition rate and subsequently demonstrating parametric frequency conversion in solid-state candidate materials with desirable optical properties for mid-infrared generation.

We assembled and demonstrated the potential of the fiber-based Raman excitation source in the long-pulse regime. We scaled up the system to 1.7 mJ of pulse energy and 36 W of average power, thus exceeding our proposed goals of 1 mJ pulse energy and 20 W average power. In collaboration with our partners at BAE Systems and the Air Force Research Labs Materials Directorate, this new excitation source also was used to demonstrate parametric gain in CSP, a new candidate crystal for mid-infrared generation.

Scientific Approach and Accomplishments

The Laboratory's fiber-based Raman source was adapted to 1.55 µm range by employing erbium-doped gain fiber. The pulsed nature of this source (more than one megawatt of peak power with average-power scalability) made it attractive for use in nonlinear conversion. BAE Systems provided the CSP, a nonlinear optical crystal with low-loss in a broad range spanning the mid-infrared spectrum (0.7 to 8 µm). The subsequent demonstration of seeded amplification of a 2.4-µm diode laser source (a representative wavelength for mid-infrared generation) demonstrated the feasibility of a fiber-based alternative to competing solid-state systems.

In FY16, we (1) demonstrated the activation of an all-fiber, laser-based system operating in the shortwave infrared with an energy-scalable architecture; (2) demonstrated narrowband and tunable seeding in the short-wave infrared without the need for galvanometers or other motion-based tuning systems; (3) developed the concept of a potentially high-efficiency parametric conversion scheme based on the characteristics of our all-fiber, laser-based system; and (4) demonstrated 1.7 mJ of pulse energy and 36 W of average power at a high repetition rate of 25 kHz.

In FY17, we (1) demonstrated the pulse-width and repetition rate versatility high energy FiLPs source—0.5–1.5 ns, 1kHz–25 kHz at constant energy 1.7 mJ; (2) demonstrated nonlinear conversion via optical parametric amplification of light at 2.4 µm, which agreed with a modeling-based expected gain; and (3) developed understanding of CSP mid-infrared frequency-conversion performance in optical parametric amplification, generation, and oscillation configurations.

Our collaborators supplied two crystals of CSP whose dimensions were 5.3 mm x 5.7 mm x 16.7 mm. Anti-reflective coatings were added to these crystals to transmit light at 1,553, 2,400, and 4,400 nm with the lowest loss possible. Losses of less than one percent were achieved at all three wavelengths.

Figure 1(a) is a photograph of the optical parametric amplifier (OPA). A continuous-wave (CW) 2,400-nm seed (Brolis) was aligned into the crystal to be collinear with the pump. The output of the OPA was separated from the pump at a dichroic reflector (high reflection at 2,400 nm and 4,400 nm with high transmission at 1,550 nm, 45° angle of incidence, Twister). The signal pulse was measured at a photodetector (Thorlabs DET050) that was sensitive to wavelengths as long as 2.8 µm. The pulse energy was measured here as the integral of the detector response over 17.5 ns. The true signal pulse was expected to be closer to the duration of the pump pulse of about 1.3 ns due to the limited bandwidth of mid-wave infrared photodiodes. Comparison of pulse energy measured at 2,400 nm to the CW seed power in the duration of the pump pulse was taken to be the OPA gain.

Figure 1. Photograph (a) of the optical parametric amplifier (OPA) setup. A plot (b) of gain measured from the parametric amplifier for a single crystal (blue squares) and for two crystals (red circles).

In an effort to generate as much gain as possible as a starting point, we placed both available crystals in the beam path. This resulted in a dual crystal gain of 49 times for 0.274 mJ of pump energy. The best performance of the OPA with two crystals was less than the gain obtained using only one crystal. Removing the second crystal and tuning the crystal orientation axis of only the first crystal into an alternative alignment resulted in a gain of 121 times at a pump pulse energy of 0.27 mJ. This was due to the fact that the pump exited the first crystal with some displacement from its original path and a displacement (due to walkoff) from the signal beam. The interaction length for two crystals was basically no better than that of a single crystal. The gain was exponential and did not display saturation. Pumping power for this measurement was well below the maximum available power to maintain pump fluence below the damage threshold. An expansion of the pump beam size would allow more pumping power to be used. We also characterized the transmission of the CSP crystal itself, which exhibited lower absorption of ZnGeP2 and appeared to have a substantially higher damage threshold than other materials, such as AgGaS2.

In an optical parametric oscillation (OPO) configuration, a single crystal was placed in an oscillation cavity. The oscillation cavity provided feedback of the amplified signal into the gain medium so that it could further interact with the pump, resulting in increased frequency conversion. Mirrors with high reflection at 2,400 nm, high transmission at 1,550 nm, and normal incidence (Twinstar) were placed as close as possible to the gain crystal to maximize the duration of the interaction. The cavity length was 4 cm (+/-0.1 cm). When a pump pulse entered the crystal, a signal and idler were immediately generated. At the end of the cavity, the pump pulse exited while some fraction of the signal pulse was reflected back into the cavity. At 1.3 ns in duration, the pump pulse occupied about 39 cm in space. Taking the refractive index of the material portion of the cavity to be 2.2, the corresponding single-pass cavity length was 5.9 cm. The reflected signal could only make fewer than three trips around the cavity before the pump was no longer in the cavity and the amplification process died out. The additional gain from signal feedback was limited by the small number of interactive roundtrips and walkoffs of pump and signal beams, so the single-pass gain was the dominant contribution to the OPO performance. One way to address this inefficiency was to directly coat reflective surfaces onto the crystal-end faces. This represented the shortest possible cavity, and thus the longest pump interaction for a non-synchronously pumped OPO. As an alternative, a synchronously pumped configuration could have been arranged along the cavity length such that the round-trip signal time matched the repetition rate of the pump. However, this would have been infeasible for a source with a kilohertz-scale repetition rate as the cavity would have had to have been several kilometers long. However, this would be feasible if CSP were used in a traditional mode-locked cavity architecture operating near 100 MHz. Figure 2 (a) shows the beam profile for the OPO output with both sidebands.

Figure 2 (b) shows the OPO output power as a function of pumping power demonstrating a slope efficiency of 10%. The pump depletion shows that over 50% of the pump is depleted by the conversion combined with the crystal losses at the pump wavelength.

Figure 2. Image (a) of the combined 2.4 and 4.4 µm sideband output of the optical parametric oscillation (OPO). Pump power plots (b): the left axis is a plot of the total OPO output power as a function of pump power (linear fit shows a slope efficiency of 10 percent); right axis is a plot of pump depletion as a function of OPO pumping.

Impact on Mission

This research project supports the development of new efficient, scalable, pulsed-laser technology that will, in turn, support the Laboratory's core competency in lasers, optical science, and mid- and long-wave infrared technology and techniques. Long-pulse applications include active remote sensing, infrared countermeasures, and remote sensing exploitation. Short-pulse applications include high harmonic generation, pump-probe experiments, and sources for high-energy-density science. This advancement in laser development also supports the Laboratory's cyber security, space, and intelligence focus area with technology relevant to remote sensing.


We demonstrated a high-average-power (36 W), scalable, temporally agile (0.5–2 ns), erbium-doped fiber pump source. The architecture we developed can operate at 1 µm (Yb:silica), 1.55 µm (Er:silica), and 2 µm (Tm:silica). We demonstrated the first amplification at 2.4 µm (and 4.4 µm) using a pulsed 1.7 mJ, 1.3 ns output from a 1.55-µm fiber source. Our work in optical parametric amplification performance, characterization, and comparison with direct optical parametric generation, as well as optical parametric oscillation, will result in at least two publications.


Fuchs, F., et al. 2010. "Imaging Standoff Detection of Explosives Using Widely Tunable Midinfrared Quantum Cascade Lasers." Optical Engineering 49 (11): 111127. doi: 10.1117/1.3506195.

Yao, Y., et al. 2012. "Mid-Infrared Quantum Cascade Lasers." Nature Photonics 6 (7): 432—439. doi: 10.1038/nphoton.2012.143.

&nbsp &nbsp