As introduced in 1996 by Davis et al. (1996), tightly-focused near-infrared femtosecond lasers have shown to be a highly versatile tool to address nonlinear multi-photon 3D-localized energy deposition in dielectric materials. In 2006, Marshall et al reported the first femtosecond direct laser writing (DLW) of integrated waveguide-Bragg gratings (WBGs) in bulk fused silica (Marshall et al., 2006). Similarly to fiber-Bragg gratings (Hill and Meltz, 1997; Haque et al., 2014), a WBG is a waveguide bearing a periodically-modulated refractive index structure along the wave-guided light propagation, allowing for the backward wave-guided reflection of some specific so-called Bragg wavelengths whereas the other ones are transmitted and remain guided in the forward direction. Such distributed Bragg reflectors (DBRs) are therefore a key-element for integrated guided optics, allowing for many applications such as wavelength selective devices as band-rejection filters, dispersion compensation, Bragg-grating sensors, Fabry-Perot resonators based on the combination of two DBRs, and even photonic schemes for slow light (Hill and Meltz, 1997; Bernier et al., 2001; Bernier et al., 2009a; Bernier et al., 2009b; Wen et al., 2013; Haque et al., 2014; Ams et al., 2017).

The laser inscription of efficient WBGs in dielectrics requires two aspects. First, the grating periodicity needs to be as perfectly controlled as possible, which can be achieved via a point-by-point approach, by modulated irradiation or even better by the combination of laser inscription with a periodic phase mask (Bernier et al., 2001; Bernier et al., 2009a; Bernier et al., 2009b; Thomas et al., 2012; Wen et al., 2013). Second, the index modulation should be as high as possible, typically a few 10⁻³ or even more in transparent dielectrics, to enhance the reflectivity R and thus the associated backward coupling coefficient κ [R = −1—T = tanh (κL)²] that typically spans in the 0.1 to 1.5 mm⁻¹ range after femtosecond laser inscription (Ams et al., 2017). The overall WBG efficiency is adjusted by adapting the physical length L of the DBR, classically requiring structures in the centimeters’ length (Bernier et al., 2001; Bernier et al., 2009a; Bernier et al., 2009b; Thomas et al., 2012; Wen et al., 2013): such required WBG lengths may thus appear as a limitation in the downsizing process of integrated waveguide functionalities. Moreover, the WBG production mostly relies on smooth type I index modifications being related to homogeneous density changes or to moderate structural modifications, these modifications being diffraction-limited (or larger) because the modified material’s voxel is similar to the multi-photon energy deposition voxel. Such limitation restricts the experimental access to small WBG periodicities compatible with first-order Bragg diffraction in the visible range. Nevertheless, to access periodicities in the 200-nm range, one can consider the self-organized formation of periodic nanogratings, associated to the so-called type II refractive index modifications (Smelser et al., 2005; Li et al., 2012; Haque et al., 2014). However, the range of parameters for laser inscription of type II nanogratings may be very limited in some materials, and such structures are likely to introduce strong light scattering and losses. Moreover, such nanogratings can also show limited long-range stability of the periodic structures, and finally, due to their intrinsic formation mechanism, it is very difficult to tune the associated periodicities.

As a consequence, it is highly challenging to address the production of WBGs with smooth periodic structures beyond the diffraction limit (down to the 200-nm range) and high intrinsic backward coupling efficiencies. Such requirements are needed for downsizing integrated DBRs to achieve really compact efficient first-order WBGs even down in the visible range.

Up to now, many laser parameters and irradiation regimes have been investigated in standard materials (Hill and Meltz, 1997; Bernier et al., 2001; Smelser et al., 2005; Marshall et al., 2006; Bernier et al., 2009a; Bernier et al., 2009b; Li et al., 2012; Thomas et al., 2012; Wen et al., 2013; Haque et al., 2014; Ams et al., 2017), but specialty materials showing tailored photosensitivity for the DLW process should also be considered (Royon et al., 2011). We have been working in such a framework for more than a decade with silver-containing oxide glasses, leading to remarkable photosensitivity at the origin and high-contrast linear/nonlinear optical properties (Petit et al., 2018). More specifically, some inner features of femtosecond laser induced spatial distributions of silver clusters have been measured down to 80 nm by means of high resolution scanning electron microscopy (Bellec et al., 2009). Moreover, we have recently demonstrated a new type of positive refractive index change sustained by the femtosecond laser induced photochemistry of silver elements, labeled type A for Argentum, with ∆n values up to 6x10^-3 compatible with the creation of innovative waveguides (Abou Khalil et al., 2017; Khalil et al., 2019). This led to remarkable refractive index sensing abilities of the environment refractive properties by means of evanescent coupling. Thus, laser-induced Type A refractive modifications appear attractive to achieve sharp sub-micron scale periodic structures, to create efficient first-order WBGs.

In this paper, we present the pioneer demonstration of first-order WBGs sustained by laser-induced photochemistry of silver elements, allowing the access to Bragg wavelengths in the visible spectral range. Section 2 details the proposed method, describes the mechanisms at play such as the management of the silver reservoir and reports on sub-wavelength index modulation. Section 3 reports on the spectral characterizations of the produced WBGs, including the 2-step approach with the line-by-line inscription of the DBR followed by the waveguide inscription as well as another method based on a true 3D-DLW approach.

Refractive index modifications show typical values of a few 10^-3, as measured in Figure 3B, in agreement with previous measurements (Abou Khalil et al., 2017; de Castro et al., 2019; Khalil et al., 2019). The effective refractive index derived from the guided mode solution is thus reasonably assumed to be close to the value of the pristine bulky glass. Refractive index measurements of the pristine glass have independently been measured by ellipsometry over the 270 nm—2 µm spectral range, with an uncertainty on ± 0.002 after fine adjustment with a reference silica. For the inscribed periods Λ = 240, 260 and 280 nm, the experimentally measured spectral positions of the first-order Bragg resonances λ B are peaking at 736.4, 796.3, and 856.1 nm, respectively. This led to the associated effective modal indices n e f f (defined as λ B = 2 n e f f Λ / p with p = 1) to be estimated to 1.5342, 1.5313 and 1.5288, respectively, while the refractive index of the pristine glass was measured to 1.5376, 1.5355, and 1.5337 at these three wavelengths of 736.4, 796.3 and 856.1 nm, respectively. Unexpectedly, the refractive index values of the pristine glass are slightly higher than that of the effective guided modes: this is only to be attributed to an error in the spectral calibration of our spectrometer. Still, these indices are in rather good agreement, and it does not affect the demonstration of Type A-based WBGs.

The relative transmission spectra of these WBGs (length L = 500 µm) are presented in Figure 4A. At each expected spectral position, a narrow but significant transmission drop is recorded, demonstrating the first observation of WBG behavior where both the waveguide and the Bragg grating are refractive structures being sustained by Type-A refractive index structures. It should be noted that periods of 240 and 260 nm allow for reaching the visible range with first-order Bragg grating resonances. Transmission drop due to WBG resonances shows a decreasing efficiency for decreasing WBG periods. Relative transmission measurements along the 500-µm long WBGs led to narrow-bandwidth transmission drops corresponding to attenuations (sub-nm spectral FWHM) from 2.29 dB to 6.25 dB for periods from 240 nm to 280 nm, respectively. Under the hypothesis of a transmission attenuation being purely associated to reflection (absence of light decoupling and/or of additional absorption in the WBG structure), the transmission drop has been converted, by means of standard coupled mode theory formula, in order to provide an estimation of the backward coupling coefficient (Kogelnik, 1969; Ams et al., 2017), as shown in Figure 4B. The obtained backward coefficient values, spanning from 1.66 to 2.69 mm⁻¹, show remarkably high values compared to the commonly achieved values with direct femtosecond laser inscription without phase mask approach, especially in the visible spectral range (Ams et al., 2017). The associated spectral widths (FWHM) of the measured Bragg resonances were about 0.5–0.7 nm, as shown in Figure 4B; Figure 4C reports on the top-view image of the WBG thanks to the high spatial resolution confocal fluorescence imaging, showing both the Bragg grating (period Λ = 250 nm) and the waveguide, similarly to the description in Figure 2B. Figure 4D shows the associated near-field single-mode profile of the guided mode at the exit facet of the sample, demonstrating the overall wave-guiding behavior of such a two-step inscribed WBG structure. In order to produce as efficient WBGs as possible, further investigations will have to address the optimal management of the silver reservoir and distribution of induced silver clusters, by considering the inscription of the waveguide first and then that of the Bragg grating, contrarily to the present approach.

Despite the relatively short length of the gratings, the observed attenuations are very strong for a line-by-line approach (Ams et al., 2017). These attenuations have been observed by transmission only, and we have not conducted reflection measurements up to now. However, thanks to the localized spectral positions of the WBGs, the observed attenuations do well correspond to the spectral signature of first-order WBGs, so that the expected reflectivities should follow the expected spectral and efficiency behavior. With this assumption, the observed WBG spectral attenuations are considered to correspond to reflectivities R = 46.2%, 66.7% and 76.3% for periods Λ = 240, 260 and 280 nm, respectively. With a length L = 500 µm, and considering the classical formula R = tanh (κL)² (Ams et al., 2017), the backward coupling coefficients κ are respectively estimated to 1.66, 2.29 and 2.69 mm⁻¹, such values being significantly higher than those reported elsewhere, especially in the visible/near-IR spectral range (with coupling coefficients usually spanning from a few 0.01 s mm^-1 to a few 0.1 s mm^-1, and possibly up to 0.85 mm^-1 for Bragg resonances between 643 nm and 1,030 nm) while considering DLW without phase masks (Ams et al., 2017). The remarkably large κ values obtained with our approach take benefit of both the high contrast of the refractive index modulation of a few 10^-3 even for the involved sub-diffraction periodicities, and of the significant overlap of the guided mode with the Bragg grating.

As introduced in Section 2.3, the 100x100 μm² gratings have been inscribed with the same periodicities as those of the WBG depicted in Figure 4. With the combination of these values ∆n with that of the coupling coefficients κ, one could estimate the modal spatial overlap Γ of each WBG, thanks to the formula κ = πΔn Γ/λ (Ams et al., 2017). The three WBGs led to the same overlap Γ = 28 ± 1%, which supports the consistency of our measurements and analyses. Indeed, in this case, the three WBGs had been inscribed with respect to the same unique waveguide. All the measured and calculated specifications of these three WBGs are reported in Table 1.

The difference between the expected and measured spectral positions of the first-order Bragg grating resonances (difference inferior to 0.5%) can be explained by the uncertainty of the estimated refractive index dispersion, but also by the miscalibration of our spectrometer. The precision of the stages for such short periods may also affect the effective inscribed period. The narrow bandwidth of the Bragg resonances (FWHM = 0.5, 0.6 and 0.7 nm) appears nevertheless to be larger than what may be expected with such coupling coefficients κ and WBG length (Ams et al., 2017), corroborating imperfections in the periodic inscription of the WBG. Still, these results correspond to the first demonstration of WBGs being sustained by localized silver cluster distributions, remarkably highlighting the reliability of such a new process and allowing for the achievement of first-order Bragg grating resonance in the visible range.

The decrease of the observed transmission drops in Figures 4A,B, and correlatively the estimated coupling coefficients κ, with respect to the period Λ can be interpreted as following. One certainly needs to consider the physical thickness of the individual refractive index walls. Indeed, while reducing the periods getting closer to the 200-nm range, the refractive index walls may start to partially overlap laterally, leading to a smoothing of the amplitude of the refractive index modulation ∆n_AC (as shown in Table 1) and thus to a lower coupling coefficient κ.

To go further in the interpretation of the Type-A based WBGs, we investigated the achieved quality of the WBG behavior with respect to the length L of such Bragg gratings. As shown in Figure 5A, we observed that the effective backward coupling coefficient κ was increasing from 1.23 mm^-1, 1.56 mm^-1 up to 2.41 mm^-1 while decreasing the WBG length L from 800 µm, 600 µm down to 400 µm, whereas a constant coupling coefficient κ is to be expected if the hypothesis of negligible absorption relies valid. This behavior is directly due to the stability and repositioning repeatability of the stages, which currently restricts the overall efficiency of our process. Indeed, as the grating length increases, the accumulation of positioning error increases, artificially limiting the coherence length of the process and thus lowering the effective value of the backward coupling coefficient. Oppositely, it suggests that WBGs with shorter lengths may correspond to even larger coupling coefficients κ than those estimated experimentally, as suggested by Figure 5A. These considerations tend to suggest that our approach could lead to much larger coupling coefficients κ if the overall periodicity was improved. Moreover, as illustrated by Figure 2A and Figure 3A, improving the periodic pattern should also include the improvement of the stability of the thickness and amplitude of the periodic refractive index modulations.

As discussed here above, improving the effective WBG strength requires the improvement of the WBG periodicity. Another improvement can also rely on improving the modal overlap Γ of the guided mode with the WBG structure (previous estimation of Γ ∼ 28% in the line-by-line approach). This has been investigated by taking benefit of the true 3D DLW ability. As a demonstration, using the very same sample and writing conditions as well as the two-step scheme (inscription of the Bragg grating first and then of the waveguide), a squeezed helix with an elliptical basis (period Λ = 480 nm, large diameter leading to the horizontal width W = 40 µm, small diameter leading to the vertical height H = 6 µm, length L = 500 µm) and a waveguide (located at the center of such a helix) have been inscribed, as illustrated in Figure 5D. The transmission spectrum of such a squeezed helix structure is reported in Figure 5C, showing a very strong attenuation down to 9.4% transmission. This corresponds to an equivalent coupling coefficient κ_helix—3.7 mm^-1, which is even stronger than the best results shown in Figure 4 with the line-by-line approach. The relative enhancement of the helix approach versus the line-by-line one leads to a typical yield κ_helix/κ_line-by-line = 3.7/1.67–2.2. This enhancement may suggest thus a typical factor 2.2 enhancement of the modal overlap up to Γ_helix—61%, which is compatible with the typical doubling of the height of the helix-based Bragg grating compared to that of the line-by-line grating. Moreover, successive laser passes related to the upper and lower parts of the helix structure are at different depths, namely, shifted within the helix height of 6 µm, which may lead to a better refractive index modulation contrast with a larger ∆n_AC parameter. Such an improvement of the helix-based Bragg grating attenuation can also result from a more reliable process. Indeed, the inscription of a helix-shaped WBG based on a single trajectory may reduce the overall uncertainties of stage repositioning compared to the line-by-line approach.

Beyond the question of WBG efficiency and strength of the achieved gratings, one shall note that the obtained spectral profiles show secondary bumps. This may come from non-uniform Bragg gratings showing multiples domains with distinct periods, as possibly shown in Figure 4A. This may alternatively come from backward coupling in a non-perturbative regime where the modulation of the effective index ∆n_AC is of the order of magnitude of the refractive index modification that supports the effective index of the guided mode, as possibly observed in Figure 5B. Further modeling with the coupled-mode theory is further required to better detail the origin of such secondary bumps.

Within the last years, we had provided the pioneer demonstration of laser inscription of integrated Type-A waveguides, as well as the ability to perform their direct inscription at the surface of photosensitive glasses (Abou Khalil et al., 2017; Khalil et al., 2019). We report here the first demonstration of Type-A based WGBs (namely, WBGs sustained by laser-induced photochemistry of silver species), the direct access to periodicities down to 240 nm allowing for first-order Bragg grating behavior in the visible range. We report on remarkably strong attenuation over short WBG lengths, corresponding to high values of coupling coefficients up to 2.4 mm^-1 at a wavelength of 765 nm (period Λ = 250 nm and length L = 400 µm) and even to 2.7 mm^-1 at near-IR wavelength of 855 nm (period Λ = 280 nm and length L = 500 µm. Furthermore, an attenuation corresponding to an effective coupling coefficient up to 3.7 mm^-1 has been obtained by taking benefit of a true 3D inscription of a squeezed helix-shaped WBG in order to achieve a better modal overlap by the Bragg grating: such an architecture may also possibly take benefit of a better grating periodicity thanks to a 3D structure being inscribed with a continuous single trajectory (contrarily to the successive inscriptions for the line-by-line approach). These results extend the high potential of laser-inscribed photonics architectures based on Type-A refractive index modifications in silver-containing ortho-phosphate glasses (Kunwar and Soman, 2020). Further investigations shall address the design of apodized WBGs (Halir et al., 2015), the combination of Bragg gratings to form active/passive optical resonators (Ams et al., 2017) but also the direct inscription of WBG at the glass interface for enhanced sensing applications (Khalil et al., 2019).