Abstract

Frequency combs (FC) have revolutionized the fields of spectroscopy and metrology [1], and are therefore of interest at any frequency region of the electromagnetic spectrum. Basis of any FC is a broadband laser, which until recently was missing e.g. at THz frequencies. Recent results have shown that THz quantum cascade lasers (THz QCL) can operate in a very broadband regime covering more than one octave in frequency [2]. In addition it has been shown that QCLs can operate as FC [2-4]. So far FC based on THz QCL have the drawback that they are either limited in frequency or that they have non-uniform power distribution over the individual comb teeth [2,4]. Therefore our work is two-fold. One goal is to provide THz FC spanning over more than one octave using the active region (AR) design presented in ref. [2]. The second goal is to further improve the bandwidth and power uniformity of the underlying AR. To achieve an octave spanning FC, one has to compensate for the dispersion present in the octave spanning lasers. We choose a similar approach as presented in ref. [4], using double-chirped mirrors. In a first step the dispersion present in the laser cavity needs to be measured. We do that by fully exploiting FTIR measurements of the laser. The recorded interferogram is zero-padded before performing the Fourier transform. This results in a smoothing of the frequency spectrum (fig.1a) and a preciser knowledge of the individual cavity mode positions. Comparing the spacing between the individual modes to FC modes results in an estimation of the group delay dispersion (GDD). A genetic optimization algorithm has been developed which calculates possible designs for double-chirped mirrors which compensate for the measured dispersion of the laser. The measured dispersion along with the calculated dispersion of the mirrors are displayed in Fig.1b. To fabricate lasers with double-chirped mirrors the laser ridge is partially deep-etched and planarized with Benzocyclobutene (BCB) to get an alternation in refractive index. The position and lengths of those deep-etchs are chosen according to our simulations. Fig.1d shows the dry-etched mirrors before BCB planarization. The processed lasers will be measured both at optical and radio frequencies. In order to improve the AR design we use the ability to stack different types of AR designs within one waveguide. This is not limited to just 3 different designs as in ref. [2]. We have developed a laser which consists of 4 different active region designs, having central frequencies of 2.3, 2.6, 2.9, 3.5 THz. The lower 3 stacks are similar to those from ref. [2] slightly optimized to have an almost identical alignment voltage. The structure has been grown by molecular beam epitaxy, and processed in double-metal waveguides. The pulsed lasing spectrum of a 1.5 mm × 150 μm laser is shown in fig.1c, spanning from 1.88 THz to 3.82 THz in frequency. This is an improvement in bandwidth of 230 GHz from the previous design to almost 2 THz. In addition the laser has 17% more peak power, which is important for future spectroscopy experiments.

© 2015 IEEE