1. Introduction

In the recent years, mid-infrared (MIR) silicon photonics is attracting a lot of attention [1–3]. Taking advantage of the unique molecular absorption lines in the MIR range [4,5], silicon photonics has been proposed as a convincing solution for the development of high-performance and cost-effective MIR integrated sensors. A large number of applications are foreseen, for instance, real-time environmental monitoring [6,7], bio-sensing and medical diagnosis [8,9]. Interestingly Germanium (Ge) and Silicon Germanium (SiGe) alloys are strong candidates for extending the operation wavelength of silicon photonics in the mid-IR [10–16]. In this context Ge-rich SiGe graded index waveguides have been recently demonstrated as a promising platform benefiting from the wide transparency window of Ge to achieve deep-MIR operation, beyond 8 µm [17–20]. Furthermore, the strong 3rd order nonlinearity of SiGe shows a huge potential in the field of non-linear active devices for efficient optical frequency generation and conversion [11,21,22].

The integrated resonator is a key building block for the enhancement of on-chip sensing, spectroscopy and nonlinear optics. Si-based on-chip MIR resonators have been reported previously [23–27], however on-chip integrated resonant structures are still missing for wavelengths beyond 5.6 µm. We have thus chosen to use broadband Ge-rich SiGe graded index waveguides to develop on-chip resonators working at large wavelength in the MIR. In comparison with ring resonators, Bragg grating Fabry-Perot resonators present enhanced flexibility in terms of achievable free spectral range, as it only depends on the cavity length. Because of the low index contrast between graded waveguide core and Si substrate, a careful design of the Bragg grating is needed to achieve a reasonable grating coupling coefficient keeping low leakage towards to the substrate.

In this paper, we experimentally demonstrate Bragg grating waveguides up to 8.4 µm wavelength. Fabry-Perot resonators with a maximum quality factor Q = 2200 are demonstrated at 7.95 µm. This work paves the way for future investigation of cavity-enhanced sensing circuits and non-linear active devices in the MIR region.

2. Bragg grating waveguides

The Bragg grating waveguides are implemented on Ge-rich SiGe graded platform. Here, the Ge concentration increases linearly from the Si substrate to the pure Ge, across the 6-µm-thick waveguide core. The waveguide etching depth (D_WG) is 4 µm and the width (W_WG) is 5 µm, which provide high optical confinement of the guided mode in the Ge-rich region over a wide range of MIR wavelengths [18]. A cross-section schematic view (y-z plane) of the SiGe waveguide is shown in Fig. 1(a), jointly with the guided transverse magnetic (TM) mode profile at a wavelength of 7.3 µm. The vertical refractive index profile of the SiGe graded platform is shown at the right hand side.

 

Fig. 1 (a)Schematic of the waveguide cross-section with the simulated transverse magnetic (TM) mode profile in the grating region at 7.3 µm wavelength. Right side: The vertical refractive index profile of the graded SiGe waveguide platform. (b) Calculated coupling efficiency of the Bragg grating waveguide as a function of the ratio r between the shallow-etch grating width (W_Etch) and the nominal waveguide width (W_Wg) at a wavelength of 7.3 µm. Inset: Three-dimensional (3-D) schematic of a Bragg grating waveguide.

Download Full Size | PDF

The Bragg grating waveguide is defined using a surface-corrugated approach, by a shallow etching of the top surface of the waveguide. As shown in the inset of Fig. 1(b), the etched region is narrower than the waveguide (W_Etch<W_WG) to improve the alignment tolerance between the two mask layers used in the device fabrication. Numerical simulations have been performed to optimize the dimensions of the Bragg grating waveguides. First, an etching depth (D_Etch) of 0.4 µm is chosen as it provides a reasonable interaction between the top-surface waveguide corrugation and the optical mode, while keeping a good optical confinement of the propagating mode within the corrugated waveguide. The coupling efficiency describes the interaction of the waveguide mode and the Bragg grating [28], and can be calculated as follows:

The SiGe graded-index waveguides were firstly grown using a low energy plasma enhanced chemical vapor deposition (LEPECVD) technique, which allows a tight control of the alloy composition in the growth direction [30]. The Bragg grating waveguides were fabricated using an electron beam lithography, followed by an inductive coupled plasma (ICP) etching. Firstly, a partial shallow–etch level of 0.4 µm was performed to define the Bragg grating, followed by a second etching step of 4 µm to define the waveguides. A scanning electronic microscope (SEM) image of the fabricated Bragg grating waveguide is shown in Fig. 2(a).

 

Fig. 2 (a) The scanning electron microscopy (SEM) image of a fabricated Bragg grating waveguide. (b) The simulated (blue curve) and the measured (orange curve) transmittance spectra of the Bragg grating waveguides with a period of 0.74, 0.88, 1.02 µm, leading to an operation at 5.4, 6.4 and 7.3 µm wavelengths, respectively. (c) Measured transmittance of Bragg grating waveguides as a function of wavelength for various Bragg grating waveguides, featuring a different number of grating periods in a range from 210 to 280. (d) Minimum transmittance of the Bragg grating waveguides in the rejection band (simulation and measurement) as a function of the number of grating periods, with a Bragg grating period of 0.88 µm.

Download Full Size | PDF

Fabricated devices were characterized using a free-space configuration with a tunable quantum cascade laser (QCL). Additional details about the testing set-up can be found in Ref [18]. Figure 2(b) shows the simulated (blue curve) and the measured (orange curve) spectral responses of Bragg grating waveguides with 250 grating periods and various grating period lengths (Λ = 0.74, 0.88, 1.02 µm), thereby shifting the operating wavelength of the rejection band from 5.4 to 7.3 µm. The experimental results are in good agreement with the numerical simulations. The slight difference in wavelength is due to the different grating width between simulation (W_Etch = W_WG), and experiment (W_Etch = 0.7 × W_WG) due to the restriction of 2D simulations. A reduction of the effective index of the Bloch mode is thus responsible for a blue-shift of the rejection band in the simulations in comparison with the experiments. Figure 2(c) shows the measured transmission of Bragg grating waveguide, comprising different lengths, i,e. different number of Bragg grating periods, while maintaining a constant grating period of 0.88 µm, providing rejection centered at λ = 6.4 µm. It can be seen that the minimum transmittance (i.e. rejection of the Bragg grating) decreases as the number of periods increases, which corresponds to an increase of the Bragg grating reflection. A maximum rejection of 21.6 dB was measured. The minimum transmittance has also been calculated as a function of the number of periods and compared to experiments in Fig. 2(d), obtaining comparable trends.

3. Fabry-Perot resonators

From the demonstration of MIR Bragg grating waveguides, it has been possible to design and fabricate Fabry-Perot resonators. This was implemented by integrating two identical Bragg gratings as reflecting mirrors, with a straight waveguide in between, thereby acting as a cavity. It is worth mentioning that despite being able to sweep the central wavelength of the Bragg grating at will over the studied wavelength range in the MIR (as seen in Fig. 2(b)), we decided to focus the demonstration of FP resonators especially around of λ ≈8 µm, where on-chip resonant structures are crucially missing.

The schematic view of the fabricated FP resonator are shown in Fig. 3(a). Figure 3(b) shows a reference simulated transmittance of a FP resonator with following parameters: Λ = 1.1 µm, N = 280 and a cavity length L_cav of 70 µm. The resonant peak is situated at a wavelength of 7.906 µm. Figure 3(c) shows the experimental results achieved for two FP resonators using Bragg mirrors with N = 280, L_cav = 70 µm and different periods of 1.1 and 1.16 µm, operating at wavelengths of 7.95µm and 8.35 µm, respectively. The −3 dB resonance bandwidths are measured to be 5.3 and 6.5 nm, respectively. This corresponds to loaded Q-factors of 1514 and 1272, respectively. Bragg grating reflectivity values of around 88% and 85% are thus estimated. Both FP resonators exhibit comparable performance in terms of maximum transmission at the resonance peak of about −1.3 dB. Based on this value and taking into account propagation loss of 3 dB/cm ± 0.5 dB/cm [18], we can estimate that mirror losses are between 1.2 and 1.8%.

 

Fig. 3 (a) Schematic view of a FP resonator. (b) Simulated transmittance of a FP resonator with Bragg grating period of 1.1 µm, number of period is 280, L_cav = 70 µm, showing a resonance at around of 7.9 µm (c) Measured transmittance spectrum of two different FP resonators with different design parameters: Λ = 1.1 µm (left), Λ = 1.16 µm (right), N = 280, L_cav = 70 µm. Orange lines correspond to the Lorentz fit of the central resonance. (d) Measured transmittance spectrum of a FP resonator with Λ = 1.1 µm, N = 500, L_cav = 70 µm. The used Lorentzian function is: $y=y_0+\frac{1}{\pi }\frac{2\Gamma }{\left[4{\left(x-x_0\right)}^2+{\Gamma }^2\right]}$. The R-squared value of Lorentzian fit is 0.97.

Download Full Size | PDF

The cavity Q-factor can be improved by an increase of the cavity length at the expense of a smaller FSR or by increasing the Bragg grating reflectivity. We fabricated a FP resonator using longer Bragg gratings, keeping a cavity length of 70 µm. The measurement is shown in Fig. 3(d), where the number of grating periods in the resonator is equal to 500. The −3 dB bandwidth at a wavelength of 7.95 µm is reduced down to 3.6 nm, which in turn, yields a loaded Q-factor of 2200. It can be noted that the linewidth of the external cavity based mid-IR laser in pulsed regime used in the experiment is specified to be below 1 cm⁻¹ (i.e. below 6.5 nm at 8 µm wavelength). Furthermore, the wavelength scan has been performed in steps of 1 nm, attaining the maximum resolution of the laser. This fact limits the maximum value of the measured Q-factor, which is conditioned by the experimental set-up. Furthermore Bragg grating reflectivity and loss can be further optimized through design. It is thus interesting to evaluate the intrinsic Q factor of the cavity which is limited only by the waveguide loss and given by the following equation [31]:

4. Conclusion

In conclusion, we demonstrated Bragg gratings waveguides and Bragg-grating-based Fabry-Perot resonators operating in the long-wave MIR region. Benefiting from the wideband waveguide design, and following the Bragg condition, the Bragg gratings and Fabry-Perot resonators were investigated over a wavelength range from 5.4 µm up to 8.4 µm. The Bragg grating structure is based on a top-surface waveguide corrugation that provides a rejection higher than 20 dB. We also implemented Fabry-Perot resonators by facing two Bragg grating mirrors one in front of each other within a certain distance to control the cavity length. The resonators are demonstrated up to 8.4 µm wavelength with a Q-factor higher than 1000 in all cases. A maximum Q-factor reaching 2200 is demonstrated at 7.95 µm wavelength by increasing the grating length. This first demonstration of resonators in such deep-MIR region paves the way to further investigation of new MIR resonance-enhanced sensing circuits in the molecular fingerprint region. Moreover, benefiting from the strong nonlinearity, the enhancement of non-linear effects in SiGe alloys is also anticipated using these new MIR cavities.