Frequency and intensity noise of a grating-tuned external-cavity quantum cascade laser

TL;DR

The paper addresses achieving a broadly tunable, narrow-linewidth mid-infrared source by using a simple external-cavity configuration with a diffraction grating to force single-mode operation in a Fabry-Pérot QCL. The authors implement a Littrow-grade external cavity (9 cm) and characterize frequency and intensity noise using an unbalanced Mach-Zehnder interferometer, showing substantial reductions in intrinsic and full linewidth, along with maintained or slightly reduced RIN across 4.29–4.44 μm. Quantitatively, the intrinsic linewidth narrows by about a factor of five (e.g., from ≈960 Hz to ≈250 Hz at 4.36 μm), while emitted power increases and threshold currents decrease under external feedback. These improvements across the tuning range make the EC-QCL well suited for high-resolution spectroscopy, such as CO2 monitoring in the 4.2–4.3 μm region.

Abstract

Quantum cascade lasers (QCLs) are semiconductor-heterostructure devices known for their emission in the mid-infrared and THz spectral regions. Due to their operating regime, their intrinsic linewidth is significantly narrower compared to bipolar semiconductor lasers. Here, we demonstrate that by implementing an external-cavity (EC) configuration based on a commercial diffraction grating, we have successfully induced a Fabry-Perot QCL to emit on a single mode with a broadly-tunable wavelength in the range 4.29-4.44 μm. This very simple setup enhances the laser's performance in terms of threshold current and emitted power. We further prove that the EC configuration positively impacts the laser's noise properties. In particular, the intrinsic linewidth is substantially reduced, the full linewidth is also decreased (depending on the integration timescale), and the relative intensity noise is slightly reduced. These characteristics, which hold within the whole tuning range, make the EC-QCL a good candidate for spectroscopy applications where broad tunability and narrow linewidth are highly demanded.

Figures (4)

• Figure 1: (a) Experimental setup for measuring the frequency noise of the EC-QCL. The laser's output beam is sent to an unbalanced Mach-Zehnder interferometer. The interference between the two time-shifted beams coming from the two unequal arms results in intensity fluctuations acquired via the real-time spectrum analyzer (SA). These are converted into frequency noise power spectral density (FNPSD) via the conversion factor found by means of the oscilloscope. By blocking one arm, the intensity noise can be measured. This is done for all of the modes depicted in Fig. \ref{['fig:setup']}(b), whose spectrum is monitored via an optical spectrum analyzer (OSA). (BS: beam splitter, DC: detector's DC signal, AC: detector's AC signal) (b) Optical spectra of the laser operated at a bias current of 430 mA in free-running (blue trace) and EC configuration (other traces). Each peak actually represents the power of the respective mode. The blue trace represents the free-running laser's emission, centered at approximately 4.36µ m and with a power of 13.46mW. When the grating vertical angle is optimized for the central mode, the power reaches 31.5mW. By adjusting the grating's tilt angle, the laser emission wavelength is tuned accordingly. With a careful optimization of the vertical angle, a broad tuning range of the emission is obtained, covering 153.6nm or, equivalently, 80.6cm^-1, with a modal power ranging from 15.5 to 31.5 mW, as represented by the peaks in this figure.
• Figure 2: (a) FNPSD spectra of the free-running QCL (blue) and of the EC-QCL emitting at 4.36µ m (red). Here, the spectra are calculated as the single-sided Fourier Transform of the laser instantaneous frequency. The single spectrum is obtained as a superposition of two acquisitions according to the different Fourier frequency range. The FNPSD of the EC-QCL is generally lower for any mode of the tuning range. The peaks at low frequencies (50 Hz--5 kHz) are due to the mechanical vibrations of the grating. The spectra show an unphysical growth after 70 MHz, which is an artifact given by the compensating factor introduced to account for the interferometer's cutoff. The whole set of FNPSD, including the related background spectra, can be found in section E of Supplement 1. (b) Intrinsic linewidth of the QCL as a function of the wavelength. The intrinsic linewidth is computed via Eq. \ref{['eq:intrinsic_lw']}, where $h\mathrm{_0}$ is the level of white noise measured by averaging the spectra in Fig. \ref{['fig:FN-intrinsic-lw']}(a) in the frequency range 30--50 MHz. The intrinsic linewidth of the EC-QCL is significantly narrower than the one of the free-running QCL, for any of the wavelengths.
• Figure 3: Relative intensity noise of free-running (blue trace) vs EC-QCL operated at 4.36µ m (red trace). The RIN is slightly reduced in the EC-QCL, proving that the grating does not induce additional intensity noise, except for the few peaks in the low frequency range caused by the grating's vibrations.
• Figure 4: (a) and (b) show the lineshape of the free-running QCL (blue) and the one of the EC-QCL emitting at 4.36µ m (red), computed numerically via Eq. \ref{['eq:optical PSD']} using different starting integration frequencies $f_\mathrm{start}$: 5 Hz in (a), corresponding to an integration time of 0.2 s, and 1 kHz in (b), corresponding to an integration time of 1 ms. The linewidth of the free-running together with all EC modes is plotted against their respective wavelength in (c) for the case of an integration time of 0.2 s and in (d) for integration time 1 ms. While varying the integration time does not change substantially the free-running QCL linewidth, which stays around 1 MHz, it has a significant impact on linewidth of the EC-QCL modes which result between 100 and 200 kHz.