1. Introduction

Broadly tunable lasers have numerous applications such as swept wavelength interferometry [1], swept source optical coherence tomography [2,3] as well as in common laboratory use. Diffraction gratings [4,5], volume-Bragg gratings [6], large free-spectral range Fabry-Perot etalons [7], acousto-optical tunable filters [8,9], multimode interference filters [10] are typically used as wavelength selective elements.

The diffraction of light by devices exhibiting a periodic modulation of surface relief or material permitivity on a spatial scale similar to wavelength has been the subject of study since the eighteenth century. Practical applications of diffraction gratings include spectroscopy [11], wavelength selection in lasers [6], beam splitting [12], pulse stretching and compression [13] and many more.

Metallic diffraction gratings are usually used with TM polarization of incident light which allows to achieve remarkable bandwidth and good diffraction efficiency at shallower grooves compared to TE polarization [14]. Mixed metallo-dielectric gratings are used to decrease the number of dielectric layers and relax stress in high diffraction efficiency gratings [15].

Resonant metallo-dielectric gratings, also called leaky-mode resonant mirror-based gratings showed 99% efficiency in the -1 st reflected order [16]. These gratings consist of a dielectric waveguiding multilayer on top of metallic film and a corrugation of surface dielectric layer. They are based on the leaky-mode resonance, when the incident light is resonantly coupled into the dielectric waveguide by the grating and the guided mode is diffracted again giving birth to interference effects [17,18]. Large diffraction efficiency of leaky-mode resonant gratings with a metal layer included in bottom reflective structure was predicted by Tishchenko et al. [19].

Wideband operation leaky-mode resonance filters at normal incidence have been reported by several groups [20,21].

Here we present a different type of grating which we call metallic resonant leaky-mode diffraction grating. The name of the component reflects the fact that it can be theoretically reduced to an infinitely thin perfect metal grating strips standing in front of a perfect metal plane at an appropriate distance. Dielectric layers are not necessary for proper functioning of the component and serve as support for grating strips modifying in the same time the resonance condition. The grating period should be selected in such a way that light is diffracted into 0$^{th}$ and -1$^{st}$ reflected and transmitted orders only. The bottom metallic layer provides reflection for nearly any angle of incidence. Waves reflected from the bottom plate are subject to diffraction again and outcoupled beams interfere with beams reflectively diffracted from the grating. By properly selecting distance of the grating from the metallic plate, $0^{th}$, the reflected diffraction order can be suppressed by destructive interference and the -1$^{st}$ order maximized by constructive interference. The top dielectric layer can be made of the same or different dielectrics and protects the grating from the environment. Both dielectric layers modify the resonance condition and their thickness should be appropriately selected in order to get desired spectral dependence of the $-1^{st}$ order diffraction efficiency. Very narrow band diffraction resonances can be theoretically achieved for strip width $w$ close to the grating period $\Lambda$, while broadband diffraction can be achieved for strongly leaky structures with $w/\Lambda \rightarrow 1/2$. In this paper we will present a broadband grating with $w/\Lambda =1/2$.

In Section 2 we will present details of the grating design, in Section 3 the grating fabrication by lift-off technique will be described, in Section 4 results of the diffraction efficiency measurement will be shown and in Section 5 the application of a grating in a broadly tunable laser will be demonstrated. Detail description of the leaky-mode metallic grating theory and dependence of its diffraction efficiency on various parameters is out of scope of this paper and will be published elsewhere.

2. Design and optimization of metallic resonant leaky-mode diffraction grating

We used a new type of wavelength selective component which we call metallic resonant leaky-mode diffraction grating to distinguish it from other types of leaky-mode resonant gratings and from metallo-dielectric diffraction gratings. The component structure is seen in Fig. 1.

 

Fig. 1. Leaky-waveguide metallic diffraction grating showing the silica glass substrate, reflective metallic layer of thickness $\tau$, dielectric layer of thickness $h$, metallic strips of thickness $t$ and width $w$ periodically arranged with a period $\Lambda$, and protective dielectric layer of thickness $d$.

Download Full Size | PDF

A reflective metallic layer of thickness $\tau$ is deposited on top of the substrate, followed by a waveguiding layer made of transparent oxide of thickness $h$ and another metallic layer of thickness $t$. The surface layer is etched to form a periodic pattern of metallic strips, each with a width of $w=f \Lambda$, where $f$ is a fill factor. The surface of the grating is protected from the environment by another oxide layer, which can be formed of the same material as the first dielectric layer or of some hard oxide like $\mathrm {Al}_{2}\mathrm {O}_{3}$. In our metallic resonant leaky-mode diffraction grating, gold was used as metal and fused silica was used in place of dielectric layers. Diffraction efficiencies were numerically calculated using the rigorous coupled wave analysis, implemented in commercial computer software (RSoft diffractmod). 2D RCWA with 100 spatial harmonics over one grating period was used. The permitivity of evaporated gold layers was interpolated for thickness of $t = \tau = 100$ nm using a Yakubovsky model [22]. Refractive index of silica layers was taken to be the same, $n_{1}=n_{2}$, using the internal model of RSoft [23]. The grating was optimized for a wavelength of 1515 nm. Figure 2(a) shows mapping of a grating period $\Lambda$ and waveguide layer thickness $h$ to the $-1^{th}$ reflected order diffraction efficiency. Gold layers’ thickness was fixed to $t=100$ nm and the width of metal stripes was set to $w=0.5\Lambda$. Additionally, the top dielectric thickness was fixed at a value of $d=130$ nm. For each period, the angle of incidence was adjusted according to the Littrow condition,

 

Fig. 2. (a) Dependence of the $-1^{th}$-order diffraction efficiency on the waveguide layer thickness $h$ and period $\Lambda$. The black point denotes the bottom dielectric layer thickness and optimum period. (b) Dependence $-1^{th}$-order diffraction efficiency on the protective layer thickness $d$.

Download Full Size | PDF

For the experiment, we selected the grating period $\Lambda =960$ nm as shown by dashed line in Fig. 2(a). We can see, that the $-1^{th}$ reflected diffraction order can be effectively suppressed or enhanced by proper selection of the dielectric layer thickness $h$. To maximize grating bandwidth, the first resonance should be used, and corresponding dielectric thickness is $h=130$ nm. Further, we optimized thickness of the top dielectric layer (Fig. 2(b)). In order to achieve a flat diffraction efficiency in a spectral range 1300 nm to 1700 nm we selected a thickness od protective layer to be 380 nm.

3. Leaky-mode metallic resonant diffraction grating fabrication

The grating was fabricated on a high quality silica substrate using the e-beam lithography combined with lift-off technique. The technological process is depicted in Fig. 3. The gold layer of thickness 100 nm followed by the $\mathrm {SiO}_{2}$ layer of thickness 130 nm was deposited by the electron-beam evaporation method in the PLS 570 deposition system (Pfeiffer vacuum). To promote adhesion, we deposited 5 nm of titanium between the substrate and the first gold layer and 1.5 nm on all subsequent silica-gold interfaces. The substrate was rinsed subsequently with ethanol and purified water (Milli-Q, EMD Milipore), dried, and subjected to ozone for 10 min (UVO-Cleaner Model 30, Jelight Company Inc.). This cleaning and hydrophobication sequence was repeated two times. Then a layer of poly(methyl methacrylate) (PMMA, 950 kDa, 4 % solution in anisole, MicroChem Corp.) of thickness approximately 300 nm was spin-coated (Laurell WS-650, 2000 rpm, 60 s) and soft baked on a hot plate (120 °C for 120 s). A conductive polymer layer (AquaSAVE, Mitsubishi Rayon) was spin-coated on the top (2500 rpm, 30 s) and soft-baked (120 °C, 60 s). A pattern with period of 960 nm was exposed by the e-beam lithography (Raith e-LiNE Plus, 10 kV). The overall size of the grating area was 3 mm×1.5 mm. After exposure, the AquaSAVE layer was washed off by purified water and the sample was developed for 120 s in a 3:1 mixture of isopropyl alcohol (IPA) and methyl isobutyl ketone (MIBK) and immersed in a stopper (IPA) for 30 s. After development, the grating layer of gold (100 nm) was deposited in the vacuum deposition system, and the sacrificial layer of PMMA was dissolved in acetone (50 °C) in an ultrasonic bath (Elmasonic P30H). Finally, a protective $\mathrm {SiO}_{2}$ layer of thickness 130 nm was deposited in the PLS 570 deposition system (Pfeiffer vacuum).

 

Fig. 3. Process of leaky-wave metallic grating fabrication.

Download Full Size | PDF

The photographs of the fabricated grating from the scanning electron microscope (Tescan Lyra 3) and AFM (JPK NanoWizzard) are shown in Fig. 4.

 

Fig. 4. (a) SEM and (b) AFM image of the edge of leaky-wave diffraction grating.

Download Full Size | PDF

4. Metallic resonant leaky-mode diffraction grating testing

The fabricated leaky-mode metallic diffraction grating was tested in the setup shown in Fig. 5. Such setup gives good estimation of the grating efficiency in practical fibre laser configurations.

 

Fig. 5. A setup for estimating diffraction efficiency of the diffraction grating. (a) Measurement of diffracted signal in Littrow configuration, and (b) measurement of reference spectrum. (c) Measured diffraction efficiency (red) compared to designed dependence (blue) with $f=0.5$, $\Lambda =960$ nm, $h=130$ nm, $d=380$ nm and to the best fit with $f=0.53$, $\Lambda =950$ nm, $h=145$ nm, $d=360$ nm (green).

Download Full Size | PDF

We measured the grating’s diffraction efficiency using the incoherent light source obtained by combining amplified spontaneous emission (ASE) of three semiconductor optical amplifiers (Thorlabs BOA1080P-1550nm, Inphenix IPSAD1301, Innolume SOA1060). These sources together cover the spectral range 1000-1700 nm, and they are henceforth denoted as SOA 1550 nm, SOA 1300 nm, and SOA 1100 nm, respectively. Free standing FC/APC connector at the input and optical circulators at the output were efficient enough to suppress reflections and prevent any lasing of optical amplifiers. For diffraction efficiency measurement, the additional optical isolation was provided by a double stage polarizing optical isolator between the SOA and the circulator. Light at the second port of circulator was focused by the collimator (Thorlabs, PAFA-X-4-C) forming a Gaussian beam with the waist diameter of approximately 0.8 mm. The grating was fixed in a rotation stage. The peak power spectral density $S_{1}(\lambda )$ of light diffracted in -1$^{st}$ order was measured by an optical spectrum analyzer (ANDO AQ6317B) for set of angles as shown in in Fig. 5(a). Then the grating was replaced by a protected gold mirror and a reference power spectral density was measured for the whole range covered by given ASE source ($S_{2}(\lambda )$ in Fig. 5(b)).

Experimental diffraction efficiencies were found as

Measured diffraction efficiencies are compared to the values calculated numerically by the RCWA method in Fig. 5(c). The agreement between the measured and predicted diffraction efficiencies is rather good. Measured diffraction efficiency to the -1$^{st}$ order exceeds a value of 90 % in the interval 1417 nm to 1700 nm. Fluctuations of the measured diffraction efficiency originate from the drifting modulation of the ASE related to the facet reflectances of SOA chips. They manifest more severely for diffraction efficiency values close to 1 when we subtract two very close logarithmic spectral values and convert the result to a linear scale. The measured diffraction efficiency is slightly lower than the predicted efficiency which can be attributed to fabrication imperfections like slightly different layer thicknesses, strip width, and period. We mostly reproduced the wavelength dependence of diffraction efficiency with the following parameters: $f=0.53$, $\Lambda =950$ nm, $h=145$ nm, $d=360$ nm.

5. Application of the metallic resonant leaky-mode diffraction grating in the broadly tunable laser

We demonstrate application of the leaky-mode metallic diffraction grating in a broadly tunable laser operating in the Littrow configuration as shown in Fig. 6(a). The gain is provided by three semiconductor optical amplifiers (see Section 4). Their signals are combined in the polarization maintaining (PM) wavelength division multiplexers (WDM) and directed toward the grating by the aspheric achromatic collimator (Thorlabs, PAFA-X-4-C). The grating provided the wavelength selectivity and filtered ASE of those amplifiers that were out of the band. The filtered signal was used as a feedback for the amplifiers. A small part of filtered signal varying in the interval from 2 % to 20 % over the tuning range of the laser was coupled to the output using a PM coupler. We observed three lasing bands 60 nm, 139 nm, and 132 nm wide. All together the laser covered a bandwidth of 331 nm withing a spectral range 558 nm broad. Laser spectra recorded for different grating angles are shown in Fig. 6(b).

 

Fig. 6. (a) Scheme of the laser. (b) Measured output spectra.

Download Full Size | PDF

6. Conclusions

In conclusion, we designed a metallic resonant leaky-mode diffraction grating for application in low power, broadly tunable fibre laser. The grating was optimized for operation with TM polarization in Littrow configuration. Using the RCWA method, we predicted diffraction efficiency into $-1^{st}$ reflected order of more than 90 % over a spectral range from 1283 nm to 1716 nm. Experimentally we measured slightly lower diffraction efficiency which exceeded level of 90 % from a wavelength 1417 nm beyond the wavelength 1700 nm which was the limit of our measurement. We demonstrated a first application of the grating in the broadly tunable laser based on three fibre pigtailed semiconductor optical amplifiers. We achieved a lasing at wavelengths from 1058 nm to 1640 nm covering a spectral range of 331 nm within a tuning band of 582 nm without any adjustment of optics. The grating thus provided the feedback sufficient for lasing over a spectral band representing 43 % of the mean wavelength.