2.1

Current advancement status

The fiber laser sources developed at ONERA generally follow a MOPA fiber architecture (Master Oscillator Power amplifier), schemed in Figure 1. The Master Oscillator (MO) is typically a continuous-wave, low-power, linearly polarized, spectrally-narrow laser emitter at the appropriate wavelength (most of the time a Distributed Feedback Diode or DFB). It is followed by an intensity modulator (IM) for temporal pulse-shaping. It can be for instance a Mach-Zehnder Modulator (MZM), a binary optical switch, or an Acousto-Optic Modulator (AOM). In coherent lidar architectures, the AOM is often the preferred solution, since it can provide at the same time a slight frequency shift (tens of MHz) that serves as an intermediate frequency for heterodyne lidar measurements. The IM output delivers low-energy laser pulses with the desired temporal profile, tailored so as to pre-compensate the pulse distortion induced by the subsequent amplifier stages. The Power Amplifier (PA) is often divided in several stages, in order to distribute the gain over different pieces of fiber, optimize fiber properties (length, diameter…) as the pulse energy is growing, and prevent excessive heat and noise accumulation. As the peak power is growing, Stimulated Brillouin Scattering (SBS) becomes a limiting process, and SBS-mitigation techniques must be employed to circumvent this limit. Working on special fibers (glass composition, large mode-field diameters, distributed mechanical strains…), ONERA has reported state-of-the-art high-peak power single-frequency fiber-based laser sources [4][5]. Table 1 summarizes the performances of three fiber MOPA sources built especially for GHG monitoring.

Figure 1

General scheme of a fiber Master Oscillator Power Amplifier (MOPA) architecture

The 1.58μm HEPILAS fiber laser was demonstrated in 2014 under specifications from the European Space Agency (ESA), and features a three-stage Er-doped power amplifier [6]. Though the 2 mJ-objective for the ON-line wavelength could not be met during the study, the laser exceeded the 0.2 mJ-objective for the OFF-line wavelength, which was a very encouraging step.

The 2.05 μm HELENA laser source features a four-stage Tm-doped MOPA architecture. Based on a former laser source [7], it has been very recently upgraded and optimized in the perspective of CO₂ lidar monitoring [8]. The source has been integrated into a coherent lidar architecture, and first CO₂ measurements have just been realized (reported in [9]). In both HELENA and HEPILAS lasers, Large Mode Area (LMA) fibers and distributed mechanical strains have been used efficiently as SBS-mitigation techniques.

The 1.64 μm VEGA fiber laser, completed in 2018, has been built for methane leaks monitoring from a ground-based lidar system, in partnership between ONERA and TOTAL. Since there are no efficient doping ions at 1.64 μm, the VEGA laser features a two-stage Raman amplifier [10]. In that case, the fiber core diameter must remain very narrow to enhance Raman non-linearity, and using LMA fibers was not an option. Therefore, only the distributed strain technique has been applied for SBS-mitigation at the signal wavelength. Moreover, since Raman amplifiers require high peakpower pumping lasers, SBS-mitigation also had to be implemented for pump lasers (using frequency chirps techniques). VEGA has been operated successfully with coherent detection to monitor industrial methane leaks in range-resolved mode, and it has also been tested in IPDA mode (Integrated Path Differential Absorption) on a hard target located 2.4 km away [11].

Table 1.

Fiber-based laser sources developed at ONERA.

2.2

Going further in amplification

In their current states, these fiber-based laser sources already open very attractive perspectives to build innovative and robust ground based lidars dedicated to GHG monitoring at local scale (gas leak detection on industrial sites, urban areas GHG monitoring…). Our current efforts on that way especially go to HELENA and VEGA lidar systems (CO₂ at 2.05 μm and CH₄ at 1.64 μm). But of course further amplification will be needed to fulfill requirements for future space-borne lidars. Several ways are currently considered at ONERA to amplify further our fiber-based laser sources (submitted patents). At 2.05 μm, the HELENA project is especially promising. After having developed the source reported in Table 1, the goal now is to amplify the output in a Ho:YLF crystal. This experiment is currently led by ONERA and the Laboratoire de Météorologie Dynamique (LMD). It is briefly described in [9]. The expected crystal gain depends of various parameters (repetition frequency, crystal length, beam focus…) but simulations indicate that several mJ could be obtained with a repetition rate higher than 1 kHz (5-10W average power). The general idea is to combine the benefits of fiber technology (settings versatility, pulse control, no alignments…) and crystal technology (very high peak power) into a single hybrid architecture (hence the name HELENA - Hybrid Laser Emitter for Environmental Applications), while minimizing the laser complexity in terms of free-space optics alignments (no free space cavity). This hybrid combination could be a breakthrough for future airborne and space-borne applications.

2.3

Frequency stabilization

In classical IPDA lidar concepts for space-borne GHG monitoring, the emitted frequencies (especially the ON-line frequency) must be stabilized in order to avoid uncontrolled frequency drifts, which would convert into measurement bias on the gas concentration. For CO₂, the absolute stability requirement is in the range of 100-200 kHz (Allan deviation) in 10 s, while it is relaxed down to a few MHz for CH₄ (indeed, for CH₄ at 1.645 μm, the absorption crosssection at the ON-line wavelength is located in a trough with low frequency sensitivity). MOPA architectures are particularly convenient to install frequency-stabilization loops directly at the output of the Master Oscillator, since power amplifiers usually do not add any frequency jitter. We have designed a stabilization loop based on the Frequency Modulation Spectroscopy (FMS) technique [12]. Schemed in Figure 2, this technique is particularly efficient for high-precision frequency locking at the center of an absorption line in a gas cell. The demodulation signal (error signal for locking) looks like the absorption first derivative, and presents a maximal slope at the line center (the slope also depends of the lock-in phase setting, but the latter can be adjusted). Figure 3 shows the build I/Q demodulation circuit (radiofrequency components), as well as the demodulation signals that can be observed in channels I, Q, and Z=(I²+Q²)^1/2 (phase-independent). These experimental signals, obtained during tests conducted with a C₂H₂ gas cell at 1539 nm are in excellent agreement with expectations. We are currently implementing this technique to stabilize the VEGA lidar at 1645 nm, using a gas cell filled with CH₄. Work is ongoing to assess the obtained frequency stability (usually, this requires duplicating the setup and measuring the Allan deviation of the beat note between two stabilized lasers).

Figure 2

Scheme of the frequency-modulation technique used to lock a DFB laser at the center of an absorption line in a gas cell.

Figure 3

Left: I/Q demodulation stage of the FMS-based stabilization loop. Right: Locking signals observed on channels I, Q, and Z.

The FMS technique uses an external modulation mean, such as an electro-optic phase modulator (EOM), and is efficient for locking a laser at the center of an absorption line (it can also be slightly tuned off-center but the stability decreases since the error locking signal slope decreases). However, for space-borne observation of CO₂, especially at 2.05μm, the On-line frequency would be ideally quite far aside from the line center (typ. 2 GHz), so that the measurement weighting function in the troposphere (where sources and sinks are located) could be favored. In that case, the scheme of Figure 2 can still be used to lock a master oscillator on the line center of a gas cell, and a slave-oscillator can then be locked on the master oscillator in the radiofrequency domain (note that the slave-oscillator is the main oscillator of the MOPA architecture in that case). Here again the fiber technology is an asset for realizing such master-slave locking circuits. Eventually, the fiber technology can also be used to replace the gas cell by itself. For instance, using a hollow-core gas-filled Kagome fiber, Curtis et al. [13] have reported very compact architectures and excellent stability results at 2.05 μm using the FMS locking technique.

3.1

Principle and previous results

The general idea of EO-DCS is to probe an atmospheric absorption line simultaneously at several frequency points, distributed along the absorption line, following a regular and stable short-comb pattern (between 5 and 15 teeth typically). Compared to the traditional and sequential two-frequency pattern (ON-OFF) involved in current IPDA spaceborne concepts (such as MERLIN), the measurement simultaneity at several frequencies would offer significant advantages. It would ensure that all the probed frequency points travel through exactly the same atmospheric column (cancelling any bias induced by the time lag between sequential On-Off patterns) and it would also relax the frequency stability and knowledge requirement (discussed in the precedent section). Indeed, a multi-frequency measurement should permit to retrieve continuously not only the atmospheric gas column-content, but also the frequency center of the absorption line (thus reducing bias induced by an inaccurate knowledge of its position). Potentially this could avoid adding a sophisticated frequency-stabilization sub-system in a future frequency-comb based space-borne lidar.

Short frequency combs can be generated in fiber technology using electro-optic phase modulators [14]. It is known that applying a radiofrequency (RF) sine modulation in an EOM converts a single-frequency input into a Bessel frequency-comb output. In [3], it has also been mentioned that by applying a RF frequency plus its second and third harmonics with convenient power levels, it is possible to generate a 9-teeth or even 13-teeth flat frequency-comb. Finally, it has also been presented that probing an absorption line with two distinct frequency combs with slightly different frequency steps and frequency offsets allows transferring the absorption line measurement from the optical domain to the RF domain [15][16]. This is the basis of what we refer as EO-DCS in this paper, schemed in Figure 4.

Figure 4

Concept of simultaneous sampling of an absorption line in the RF domain using EO-DCS

3.2

Amplified and pulsed EO-DCS

The results reported in [3] have been obtained in the continuous-wave regime and with relatively low power (<0.4 W). However in the perspective of a space-borne system, it is necessary to evaluate EO-DCS principle in pulsed mode, and with significant pulse energy. Therefore, we have made an experiment, schemed in Figure 5, where the double-comb is amplified into two EDFAs (Erbium-Doped Fiber Amplifiers). The first one operates in cw-mode (same as in [3]), and the second operates in pulsed regime. The pulse shape is adjusted by an AOM located between the EDFAs (it also brings a +40 MHz shift). The pulse profile generated by the AOM is calculated so as to pre-compensate the pulse distortion that occurs in the pulsed EDFA. The output pulses are square-shaped or smooth-square-shaped, with a duration of 2 μs. The pulsed EDFA delivers a maximum power of 100μJ at 10 kHz. It was operated here up to 92 μJ. The frequency-combs are 13-teeth long. Their respective modulation frequencies are f_MEO1=497 MHZ and f_MEO2=497.5 MHz (Δf_MEO=500 kHz), and the total offset frequency is f_c=25.5 MHz (=f_AOM3-(f_AOM1-f_AOM2)). The optical wavelength is set at the middle of an HCN absorption line at λ_c=1545.23 nm, noted v_c in frequency. Consequently, the probe comb is located at v_k=v_c+kf_MEO, k={-6, +6} in the optical domain, with a 6 GHz total span, and at f_k=f_c + k Δf_MEO in the RF domain, with only a 6 MHz span. The signal can thus be detected and digitized with a very limited bandwidth.

Figure 5

Pulsed and amplified EO-DCS experiment scheme. (DFB: Distributed Feedback Laser, EOM: Electro-optic modulator, AOM: Acousto-optic modulator, CPL: fiber coupler, EDFA: Erbium-Doped Fiber Amplifier, P: glass plate, M: mirror, L: lens, DET: Detector)

Figure 6 (left) shows the time profile of an output pulse, after amplification. The interference pattern of the two combs is clearly visible. The 2 μs duration is chosen accordingly with the 500 kHz frequency step of the comb in the RF domain. This choice ensures that the first zero of the sinc function associated with the square pulse profile coincides with the adjacent comb tooth. In that way, each tooth is efficiently isolated from its neighbor. We have measured the power associated with each tooth at the pulsed amplifier output, compared to the input teeth power distribution. It appeared that the pulsed EDFA gain is relatively homogeneous inside the comb, within 2 dB from the 1^st tooth to the 13^th. The Figure 6 (right) shows the computed transmission profile of the HCN gas cell using the EO-DCS technique in pulsed mode at 76 μJ. It is computed by integrating the power of each tooth in the frequency domain, and by performing the ratio between the power measured in the reference channel and in the gas cell channel (DET-2/DET-1 in Figure 5). Because the detector was slightly saturated here, the repeatability was not at its best for this measurement, but still the measured absorption line agrees very well with the expectation. It is thus demonstrated that EO-DCS is feasible in pulsed mode with significant pulse energy. In 2021, a similar experiment has been led with 500 μJ pulses, using a “home-made” highpeak power amplifier from ONERA, and will be reported soon.

Figure 6

Left: Time profile of a double-comb emitted pulse; Right: Measured transmission of the HCN line at 1545.23 nm in pulsed mode with 76μJ, 10 kHz, 2μs square-shaped pulses.