Open Access
Add to BibSonomy
Abstract
This study reports the fabrication and characterization of various configurations of mid-infrared waveguides and beamsplitters within the chalcogenide glass IG2 using ultrafast laser inscription (ULI). Our investigation reveals two distinct regimes of ULI modification: weak and strong. The strong regime, marked by higher pulse energies, presents darker and prominent waveguide morphology, enabling efficient light guiding at 4.55 µm, but with higher scattering losses at shorter wavelengths. In the weak regime, we observed a significant enhancement in the mode confinement and a reduction in the propagation loss within the multilayer structures. We have investigated key geometric and inscription parameters such as inscription pulse energy and number of layers, as well as arm separation and splitting angles for beamsplitters. We have successfully fabricated beamsplitters with configurations ranging from 1 × 2 to 1 × 8, achieving a uniform splitting ratio over 96% and a splitting loss as low as 0.4 dB at 4.55 µm. These findings highlight the significant potential of ULI-based IG2 waveguides and beamsplitters for mid-infrared photonics.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Mid-infrared (MIR) waveguides offer a compact and efficient alternative to free-space optics, making them crucial for a range of applications, including spectroscopy [1], chemical sensing [2], and remote sensing [3]. Notably, spectroscopy in the 3-5 µm spectral window is promising due to its ability to characterize different molecules using distinctive ro-vibrational bands, often referred to as ‘fingerprint’ bands, due to minimal interference from water vapor [4,5].
Chalcogenides are amorphous glasses consisting of chalcogens, such as sulfur, selenium, or tellurium, bonded to network formers, such as arsenic, germanium, antimony, gallium, silicon, or phosphorus. Consequently, these elements confer unique optical properties that are beneficial for infrared, nonlinear, and waveguide optics. Chalcogenides possess a red-shifted bandgap to the visible or near-infrared (NIR) region, a result of their covalently bonded heavy elements leading to weak inter-atomic bonds and low vibrational and phonon energies, making them transparent in the MIR region at wavelengths up to 15 µm [6,7]. The chalcogenide glass IG2 (Ge33As12Se55), which has a broad transparency window and low absorption [8], spans two atmospheric windows at 3-5 and 8-12 µm, making it well-suited for remote sensing and astrophotonics [4].
Ultrafast laser inscription (ULI), used to fabricate photonic waveguide structures in transparent materials, is a versatile technique that enables the design of three-dimensional structures. By employing tightly focused femtosecond laser pulses, this method generates localized micro or submicrometric modifications, leading to permanent or highly stable refractive-index changes in the material [9–13]. Although the majority of MIR integrated astrophotonics research has centered on planar waveguide devices, recent advances in ULI technologies have enabled greater flexibility in designing three-dimensional structures through out-of-plane inscription [4]. Despite its potential, application of ULI for MIR device development is still nascent, with current research focused on minimizing device losses through 3D inscription techniques.
ULI-based waveguide beamsplitters, with their fiber-matching efficiency and 3D adaptability, play a crucial role in applications such as couplers [14] and modulators [15]. They have previously been fabricated inside a range of materials including crystals [16–19], glasses [20,21], and polymers [22]. For instance, BGO crystals have been key to fabricating 3D waveguide beam splitters suitable for MIR wavelengths [16].
Previous investigations showed the superiority of IG2 over Gallium Lanthanum Sulphide (GLS) glass for ULI-based MIR waveguides, with lower propagation loss (1 dB/cm versus 6.4 dB/cm at 7.8 µm) and higher peak refractive index contrast (0.015 versus 0.006) [23,24]. On the other hand, the performance of GLS improved with multiscanned ULI, reducing the propagation loss to 0.22 dB/cm at 4 µm. Moreover, while IG2 waveguides could be inscribed to the chip edges without any issues, this was unachievable with GLS owing to facet material damage, necessitating post-inscription facet polishing. ZBLAN glass displayed promising single-mode waveguides with 0.29 dB/cm loss at the same wavelength [25–27]. ZnSe crystals have also become an area of interest. However, challenges arose owing to increased damage and insertion losses, but a solution emerged by inscribing a cladding around an unmodified core, which reduced the propagation loss to 1.9 dB/cm [28,29].
Our team's recent work focused on single-scan ULI to inscribe type-I (propagation within the region where the optical index has been modified by ULI) and type-II (propagation between two tracks written by ULI) waveguides in IG2 glass, exploring key factors such as pulse energy, duration, repetition rate, and polarization state that impact the track morphology [30,31]. These studies showed that type-II waveguides, exhibiting periodic grating-like morphology, had better performance compared to waveguides with grain-like morphology, achieving a propagation loss of 2.3 dB/cm at 1.064 µm. However, in the MIR region, specifically at 4.55 µm, type-I waveguides exhibited a predictable propagation loss trend with a minimum at 2.1 dB/cm, thereby simplifying optimization compared to type-II waveguides. Although type-II waveguides achieved a lower propagation loss of 1.2 dB/cm at 4.55 µm, they require more precise parameter adjustments owing to their less predictable loss trends.
Type-II waveguides, which rely on overlapping refractive index modifications, can lead to increased scattering and higher loss. The use of two tracks that confine light between them is not suitable for realizing complex waveguide structures. In contrast, type-I waveguides, which confine light within the core of the track inscribed at a lower peak intensity, allow for direct fabrication of complex 3D geometries using ULI. Their simpler parameter tuning and adaptability for 3D structures position type-I waveguides as promising avenues for developing photonics applications.
This work presents the ULI fabrication of type-I multilayer-multiscan waveguides and beamsplitters inside IG2, unveiling innovative 3D configurations. Section 2 describes the methods for fabricating waveguides and beamsplitters in IG2 using ULI, the parameters for inscribing multilayer multiscan type-I waveguides, and performance assessment techniques. Section 3 explores the influence of the pulse energy and waveguide cross-section on ULI-based waveguides in IG2. Finally, Section 4 presents the ULI fabrication and characterization of beamsplitters in IG2, paving the way for advancements in the MIR photonic devices.
2. Material and methods
Waveguides were inscribed inside an IG2 glass substrate, with dimensions x × y × z = 10 × 5 × 2.5 mm3, and polished to optical quality on the relevant facets [ Fig. 1(a)]. The ULI setup is based on an ultrafast laser source with a pulse duration of 300 fs at a wavelength of 1030 nm and repetition rate ranging from a single shot to 2 MHz. The femtosecond pulses were focused approximately 200 µm beneath the substrate’s surface by a 20× microscope objective (NA = 0.5) to fabricate the type-I tracks. The glass sample was placed on a high-precision three-axis stage for optimal translation in three dimensions during the fabrication [Fig. 1(b)].
Fig. 1. (a) IG2 substrate [x × y × z = 10 × 5 × 2.5 mm3] (b) Illustration of a 3D view of the ultrafast laser inscription process for type-I multiscan waveguide and 1 × 4 and 1 × 8 beamsplitters. (c) & (d) Output cross-section of a multiscan layer (13 × 13 µm2), and multilayer-multiscan waveguide model with 3 vertical layers (28 × 28 µm2). (e) & (f) Output cross-section view of 1 × 4 and 1 × 8 beamsplitters. The scale bar denotes 10 µm.
To fabricate a square cross-section waveguide [Fig. 1(c)], a closely spaced series of tracks was inscribed along the x-axis. A multilayer approach was employed along the z-axis to enlarge the cross-sectional area of type-I waveguides, as shown in Fig. 1(d). Here, “multiscan” signifies multiple overlapping laser scans, while “multilayers” refers to vertically stacked multiscan layers along the y-axis and z-axis, respectively. Additionally, various beamsplitters configurations, including 1 × 4 and 1 × 8, were fabricated inside the IG2 substrate, as shown in Fig. 1(e) and 1(f).
Table 1 provides details of the fabrication parameters for inscribing straight multilayer multiscan type-I waveguides. The precise control of focal spot overlap, achieved through a 1-µm overstep along the y-axis and a 7.3-µm overstep along the z-axis, in conjunction with the 2D distribution of the focused femtosecond laser beam, yields spot overlaps of 50% and 27% in the y and z directions, respectively. This specific inscription strategy results in a square waveguide cross-section and is evaluated for configurations ranging from one to five vertical layers.
Table 1. Fabrication parameter ranges for the type-I multilayer-multiscan waveguides
Guided modes within the waveguides were characterized using a free-space coupling method (Fig. 2), in the NIR with a 1.064-µm Nd:YAG laser and in the MIR with a 4.55-µm quantum cascade laser (QCL). A linearly polarized laser beam was upcollimated to the desired dimension using a pair of lenses (L1 and L2) arranged in a telescope configuration. The upcollimated laser beam is then focused onto the input facet of the waveguides using a 40-mm aspheric lens (MO1). Its polarization was controlled by a λ/2-plate (HWP), and by default, the polarization was set as TM polarization (parallel to the tracks). Measurements performed in both orthogonal TM- and TE-polarizations revealed that the waveguiding properties do not significantly depend on the polarization state. The output from the waveguides was collected using an aspheric lens or microscope objective (MO2), and subsequently directed towards a camera or power detector for visualization or analysis.
The propagation loss (LP) was calculated by subtracting the inherent transmission loss of IG2 due to absorption (LT), coupling loss (LC) and Fresnel loss (LF) from the measured insertion loss (LI) of the waveguide, using the formula LP = LI – LT – LC – 2LF. LT, the transmission loss due to IG2 absorption, is 0.37 dB at 1.064-µm and 0.11 dB at 4.55-µm, respectively. LC, representing the mismatch in mode fields, was computed using the mode field overlap integral between the input beam and waveguide modes. LF, which corresponds to light reflection at an interface between air and IG2, was calculated using the corresponding refractive indices. To express the propagation loss (PL) in dB/cm, the calculated LP was divided by the length of the waveguide (1 cm).
The reported propagation loss values in this study have an associated uncertainty range. This ranges from approximately 0.4 dB for losses around 1 dB to 0.1 dB for higher losses of about 2.2 dB. This uncertainty mainly arises from wavelength-dependent Fabry-Perot resonances caused by internal reflections. We have detailed the calculation and experimental validation of these uncertainties, as well as their implications on our measurements, in our previous publications [30] and [31].
3. Fabrication and characterization of straight waveguides
3.1 Effects of pulse energy on single-layer straight waveguides
The pulse energy influences both the magnitude of the induced refractive index change and the morphology of the ultrafast laser-inscribed waveguides in IG2 glass, thereby revealing two distinct regimes. The threshold for Type I modification was at 3-nJ pulse energy. Upon increasing the pulse energy, the modified tracks underwent a morphological evolution, transitioning from a smooth to a significantly darker structure. Figure 3 presents microscope images depicting waveguide cross-sections inscribed using 6-nJ and 8-nJ pulse energies, along with their respective guided mode profiles at 1.064 µm and 4.55 µm. These images demonstrate notable differences in waveguide morphology and mode confinement between the two pulse energies.
Fig. 3. Microscope images of output cross-sections (13 × 13 µm2) of single-layer multiscan waveguide inscribed with two different laser pulse energies; (a) 6 nJ and (c) 8 nJ. The corresponding measured guided mode profiles at 1.064 µm (MFD = 16 µm) and 4.55 µm (MFD = 37 µm) are shown in (b) and (d), respectively. MFD represents the mode field diameter. The red dashed lines overlaid on the microscope image indicate the relative position of the waveguide cross-section. The scale bars represent 10 µm.
A single-layer waveguide, inscribed at a 6-nJ pulse energy within a weak modification regime displays a smooth and uniform cross-sectional morphology [Fig. 3 (a)]. This regime enables efficient guiding at 1.064 µm, but not at 4.55 µm. This limitation can be attributed to the insufficiently induced refractive index change, which inhibits the confinement of modes at longer wavelengths. The guided mode at 1.064 µm exhibits a mode field diameter (MFD) of 16 µm and a propagation loss of 1.5 dB/cm [Fig. 3(b)]. In contrast, a single-layer waveguide inscribed at an 8-nJ pulse energy reveals a morphology that can be interpreted as a strong modification regime [Fig. 3 (c)]. These waveguides efficiently guided light at 4.55 µm, albeit not at 1.064 µm. The darker appearance in the microscopy image suggests a higher induced refractive index change, ensuring mode confinement at longer wavelengths. However, potential structural modifications may amplify scattering losses at shorter wavelengths [20,32]. As elaborated later in this section, this phenomenon is attributed to the proliferation of microstructures and a denser waveguide core, which emerges predominantly at higher pulse energies. For this configuration, the guided mode at 4.55 µm exhibits an MFD of 37 µm and a propagation loss of 3.8 dB/cm [Fig. 3 (d)].
Furthermore, Figure 4 presents the impact of different pulse energies (8, 10, and 12 nJ) on the guiding properties of the single-layer multiscan MIR waveguides. The microscope images of the output cross-sections are displayed in the top row [Fig. 4(a)] while the middle row features the corresponding beam profiles at 4.55 µm [Fig. 4(b)]. The bottom row displays the top views of the tracks, highlighting the increasing morphological irregularities and scatterers as the pulse energy increased [Fig. 4(c)]. The MFD for these waveguides remains relatively constant at 37 µm, 36 µm, and 37 µm at 8-nJ, 10-nJ and 12-nJ pulse energies respectively. On the other hand, the corresponding propagation losses trend upwards with increasing pulse energy (3.8 dB/cm, 6.8 dB/cm, and 9.1 dB/cm for 8 nJ, 10 nJ, and 12 nJ, respectively).
Fig. 4. Left to right column: single-layer, multiscan MIR waveguides inscribed at 8-nJ (left column), 10-nJ (middle column) and 12-nJ (right column) pulse energies. (a) Microscope images of output cross-sections of the waveguides. (b) Corresponding guided mode profiles at 4.55 µm. (c) Top view of the tracks, showcasing the increase in morphological irregularities and scatterers with rising pulse energy. The scale bars represent 10 µm.
The increase in scattering loss and saturation of mode confinement suggests a saturation in densification along the waveguide morphology, due to the formation of grain-like structures. The prevalence of these structures, as reported in [30], increases at higher pulse energies. Therefore, achieving the desired performance necessitates a fine balance between sufficient waveguide confinement at specific wavelengths and the minimization of scattering losses. The above observation indicates that the optimal pulse energy for a single-layer MIR waveguide is slightly above 6 nJ, marginally exceeding the upper limit of the weak modification regime.
In general, the narrow modification window for the weak modification regime and the propensity for darkening in the strong modification regime are attributed to densification and the presence of numerous scatterers. These phenomena arise from a combination of factors, including both the intrinsic physical and thermodynamic properties of the material as well as variations in the ULI parameters. At lower pulse energies, ultrafast laser pulses induce minor, controlled refractive index changes, resulting in smooth waveguide structures [33]. However, the limitations of chalcogenide glasses, characterized by their compact atomic structure, restrict further densification [34].
Local densification or rarefaction, which play a critical role in the observed refractive index changes, depend on the initial structural flexibility or connectivity of glass [35]. When the pulse energy increases, various effects occur, such as heat accumulation, which can induce localized phase transformations, stress-induced birefringence, and the creation of light-absorbing color centers, leading to darker and more irregular features [36]. In general, darker tracks inscribed at higher pulse energies reflect the intricate interplay between the refractive index changes and scattering properties, highlighting the role of pulse energy in shaping the guiding properties of ultrafast laser-inscribed waveguides.
3.2 Effects of the waveguide cross-section on straight waveguides
The previous section detailed how, in a weak modification regime, a single-layer waveguide, inscribed with a 6-nJ pulse energy, fails to confine light at 4.55 µm. To overcome this problem, it is essential to increase the cross-sectional area of the waveguide. Figure 5 presents multilayer waveguides with one to five layers, which correspond to cross-sections ranging from 13 × 13 µm2 to 43 × 43 µm2. The MFD decreased as the number of layers increased from two to four but broadened again when the number reached five. Despite this broadening, the lowest propagation loss of 0.9 dB/cm is achieved at the five-layer configuration. In the weak-modification regime, multilayer geometries effectively minimize propagation loss and enhance mode confinement. These improvements were attributed to the increased optical power confined within the ULI-modified region. The MFDs for waveguides with 2, 3, 4, and 5 layers were 59, 48, 49, and 52 µm, respectively, whereas the corresponding PLs are 2.0 dB/cm, 1.6 dB/cm, 1.0 dB/cm, and 0.9 dB/cm.
Fig. 5. Microscope image (top) and guided mode profiles at 4.55 µm (bottom) of various cross-section waveguides inscribed with a 6-nJ pulse energy. The number of layers of the waveguide and the corresponding MFD of each waveguide are provided at the top of the microscope image and the mode profiles, respectively. The red dashed lines overlaid on the mode profiles indicate the relative position of the waveguide cross-section. The scale bars indicate a length of 10 µm.
Figure 6(a) presents the decreasing trends of modal diffusion and propagation loss of the waveguides as the cross-section increases from 21 × 21 µm2 to 43 × 43 µm2, representing an increase from two to five layers. Modal diffusion, an inverse measure of modal confinement, is defined as the relative magnitude of the MFD to twice the characteristic radius (r) based on a step-index circular index approximation of the given geometry, MFD/2r. These waveguides can be approximated by a square buried channel waveguide profile with a cross-section of d × d (µm2). The equivalent characteristic radius for the step-index circular index profile can be calculated and substituted with r (µm) = d (µm)/$\sqrt {\pi} $. These results suggest that enlarging the cross-section decreases modal diffusion, enhances light confinement and subsequently leads to a lower propagation loss. This balance between confinement and propagation loss is the key to understanding the behavior of waveguide modes, as discussed further.
Fig. 6. (a) Modal diffusion (MFD/2r) and propagation loss (PL) trends for waveguides inscribed with a 6-nJ pulse energy guiding at 4.55 µm. The diamond black markers represent modal diffusion, and the square red markers represent propagation loss. (b) MFD as a function of waveguide cross-section for different numbers of layers. (c) Modal diffusion (MFD/2r) as a function of V-number. The markers represent experimentally measured data, and the adjacent numbers indicate the corresponding number of layers in the waveguides. The dashed lines indicate theoretical calculations based on Δn = 1.2 × 10−3.
Generally, the number and distributions of guided modes in waveguides depend on factors such as wavelength, refractive index profile, and core radius. Mathematically, the V-number (or normalized spatial frequency), defined in Eq. (1), is used to determine the modes that propagate in a waveguide. The V-number is defined as
where λ is the vacuum wavelength and NA is the numerical aperture. Similarly, the ratio of the MFD to the characteristic radius, r, can be estimated using Marcuse's equation, which is expressed as [37]Assuming a constant induced refractive index change (Δn = 1.2 × 10−3) across all layers, Figure 6(b) presents MFD as a function of waveguide cross-section, determined using equations (1) and (2). The smallest MFD is measured for the 28 × 28 µm2 waveguide cross-section, and the MFD increases for smaller and larger cross-sections. This trend results from the wave nature of light, where a much larger core radius relative to the wavelength of light leads to multiple modes and light primarily confined within the core [38]. Conversely, as the core radius decreased from 5 to 3 layers, a reduction in the MFD was observed. This is consistent with the principle that increased light confinement, a result of smaller core radii, causes a decrease in the MFD. However, when the core radius becomes exceptionally small, the waveguide operates in a single mode, and light starts to extend into the cladding, leading to an increase in the MFD [39,40]. In this scenario, the boundary conditions of the wave equation result in a significant extension of the light into the cladding. Similarly, Figure 6(c) shows the modal diffusion (MFD/2r) as a function of V-number calculated using Marcuse's equation [Eq. (2)]. This further supports the observed trends in Fig. 6(b), where smaller core radii result in larger higher-mode diffusion, while larger core radii lead to a high V-number potentially leading to higher order modes along with enlarged fundamental mode. Hence, careful tuning of the core radius is crucial for balancing the confinement and propagation for optimal fundamental-mode waveguide performance in the weak-modification regime.
After exploring the weak modification regime, we expanded the scope of this study to examine the strong modification regime. In this case, we employed the same multilayer geometry ranging from one to five layers. Figure 7 displays the guided mode profiles at 4.55 µm across varying pulse energy levels and number of layers. Each row corresponds to a specific energy (8, 10, and 12 nJ from bottom to top), and each column represents the number of layers (from 1 to 5). As shown in this figure, when the pulse energy exceeded 8 nJ, the MFD remained stable at approximately 37 ± 4 µm. This observation indicates that the fundamental mode can be adequately confined within a single layer, without the need for additional layers.
Fig. 7. Guided mode profiles at 4.55 µm for different number of layers (1 to 5) and energies (8 nJ, 10 nJ, and 12 nJ from bottom to top). Each row corresponds to a specific energy level, while each column represents the number of layers. The scale bars indicate a length of 10 µm.
Figure 8 further illustrates the experimental map depicting the dependence of the PL [Fig. 8(a)] and MFD [Fig. 8(b)] on the ULI pulse energy and the number of layers. A particularly notable trend is the increasing propagation losses (PLs), ranging from a minimum of 3.8 dB/cm to a maximum of 12.2 dB/cm, as the energy levels increase from 8 nJ to 12 nJ and the number of layers from 1 to 5.
Fig. 8. Interpolated experimental map of (a) Propagation loss (PL), and (b) Mode Field Diameter (MFD) as a function of ULI pulse energy (vertical axis) across different number of layers, numbered from 1 to 5 (horizontal axis).
These observations can be rationalized by considering the material dynamics of the strong modification regime. At the focal point of the ULI, multiple cycles of melting and re-solidification lead to a localized refractive index reduction owing to material rarefaction at the center [41]. As the pulse energy increases, this rarefaction effect becomes more pronounced, forming small anti-waveguides characterized by a sharp decrease in the refractive index at the core of grain-like structures. The sequence of scanning, particularly at higher energies, leads to disruptions in periodicity and density of the grain-like structures formed in the material, as shown in Fig. 4(c), contributing to the asymmetry of the guided modes. These effects counterbalance any potential increase in the refractive index change induced by a higher pulse energy or larger cross-section. Consequently, type-II (stress-induced) waveguiding was observed at even higher pulse energies, a phenomenon discussed in our previous studies on IG2 [30,31].
4. Fabrication and characterization of waveguide beamsplitters
4.1 Beamsplitters based on a 1 × 8 geometry with 80-µm arm separation
To develop optimized functional devices, a variety of beamsplitters utilizing different numbers of arms were implemented for guiding at 4.55 µm. Figure 9(a) provides a three-dimensional schematic of a 1 × 8 beamsplitter, which features a conical configuration. In this design, all eight arms diverge from a straight waveguide starting at splitting point P.
Fig. 9. (a) Schematic of a 1 × 8 beamsplitter with conical geometry. LS and LD define the lengths for the straight and splitting beamsplitter sections. darm represents the arm separation at the sample's output. The guided beam's direction is highlighted by the green dashed arrow. (b) and (c) show the output patterns for the 1 × 8 beamsplitter with conical geometry and its associated divergence triangle, respectively.
The design foundation for each inscribed beamsplitter is based on a 1 × 8 configuration, as shown in Fig. 9(b). The key parameters include a circumradius (R) of 104.5 µm, output arm separation (darm) of 80 µm, straight length (LS) of 1 mm and splitting length (LD) of 9 mm. These specifications form the geometric backbone of the designs and are crucial for determining the splitting behavior. This design approach ensures a constant splitting angle of 0.7° across various beamsplitter configurations from 1 × 2 to 1 × 8. The aim was to achieve uniform beam splitting by maintaining consistent splitting angles across all arms, regardless of the number of arms.
All beamsplitters were inscribed using a 7-nJ pulse energy, which was determined to be the optimal level for balancing mode confinement while minimizing both insertion and propagation losses and mitigating the effects of spherical aberration. For a more in-depth discussion of the choice of pulse energy, Appendix A presents a comprehensive study examining the impact of spherical aberration and pulse energy on waveguide losses and mode confinement. During the inscription of the arms located at equal depths, the laser’s scanning motion in the multiscan technique was alternated directionally, moving sequentially to the right and then to the left. This approach ensures uniform energy deposition and consistent material modifications across the corresponding arms.
Figure 10 displays the cross-section of the straight waveguide and beamsplitters (1 × 2, 1 × 4, 1 × 6, and 1 × 8), observed with a microscope and the output mode profiles at 4.55 µm. The straight waveguide (first column) exhibits insertion and propagation losses of 4.5 dB and 2.2 dB/cm, respectively. Adjacent to this are beamsplitters with two, four, six, and eight arms, all of which are inscribed using single-layer multiscan geometry. This approach employs a 68% focal spot overlap, equivalent to a 0.6-µm overstep, between horizontal scans to achieve a 13 × 13 µm2 square cross-section. This increased overlap, in contrast to the previously inscribed 50% overlap straight waveguides in Sections 3.1 and 3.2, aims to induce a higher refractive index change, potentially improving confinement and reducing splitting loss.
Fig. 10. Microscope images (top) and guided mode profiles at 4.55-µm (bottom) for straight, 1 × 2 v (vertical), 1 × 2 h (horizontal), 1 × 4, 1 × 6, and 1 × 8 beamsplitters. The red dashed lines overlaid on the tracks demonstrate the relative position of the circle with a 104.5-µm circumradius. The scale bars represent 30 µm.
To evaluate the performance of the beamsplitters, we focused on three primary metrics: splitting loss (SL), splitting ratio (SR), and SR uniformity. SL is calculated by assessing the difference in insertion loss between the combined loss from all beamsplitter arms and a straight waveguide inscribed with identical parameters. SR is defined as the ratio of the output power of each arm to the highest observed output power. SR uniformity evaluates the consistency of the power distribution among the arms. It is defined using the root mean square variability as
Table 2. Performance of beamsplitters with different configurations, showing splitting loss (SL) in dB, splitting ratio (SR) in %, and SR uniformity in %.
Overall, the 1 × 2 horizontal configuration exhibited the lowest SL of 0.4 dB and ensured a nearly balanced power distribution between the two arms. However, as the number of arms increases, both SL and SR uniformity worsened. This deterioration is likely due to overwriting in the splitting region, leading to the generation of more grain-like structures that increase the scattering loss and reduce the availability of material for further structural development in multiple arms. Despite this, a uniform power distribution between the left and right arms was achieved, thereby validating the efficacy of the applied ULI writing sequence. Attempts to inscribe with other configurations, such as a smaller 40-µm arm separation or eight arms in a staggered 2 × 4 configuration (comprising two rows of four output waveguides at the same depth), led to poor results in terms of SL and SR uniformity. For example, for the staggered 2 × 4 configuration, the ‘two-stage splitting process’ involves the beam first splitting into two arms and then each of those arms further dividing into four. However, this configuration failed to achieve MIR confinement, which could be attributed to either a larger splitting angle of 1° or substantial splitting losses.
4.2 Effect of splitting angle on 1 × 4 beamsplitters
To understand the effect of the splitting angle on beamsplitter performance, several 1 × 4 beamsplitters (multiscan, single-layer) were inscribed using varying pulse energies and splitting angles. The distances between the arms at the output, which correspond to splitting angles of 0.3°, 0.4°, and 0.5° at the splitting location, were set at 60, 80, and 100 µm, respectively. Waveguides inscribed in the weak modification regime, with pulse energies below 6 nJ, guided the NIR laser at 1.064 µm, whereas those in the strong modification regime, with pulse energies above 6 nJ, guided the MIR laser at 4.55 µm.
Figure 11 displays the output cross-section images and guided light modes of the straight waveguides and 1 × 4 beamsplitters for both the weak and strong modification regimes. Figure 11(a) presents waveguides and beamsplitters fabricated using a 5-nJ ULI pulse energy and characterized at 1.064 μm. This straight waveguide exhibited a propagation loss of 1.3 dB/cm.
Fig. 11. Output cross-sections (left) and mode profiles (right) (a) at 1.064 µm for a straight waveguide and 1 × 4 beamsplitters with arm separations of 60 µm, 80 µm, and 100 µm (from top to bottom row), in the weak modification regime (5-nJ ULI pulse energy). (b) at 4.55 µm for the same arm separations, but written in the strong modification regime (7-nJ ULI pulse energy). The scale bars represent 20 µm.
Table 3 summarizes the performance of these NIR 1 × 4 beamsplitters, showing lowest SL of 0.8 dB at 60-µm arm separation while the SR uniformities across different arm separations did not show significant variation.
Table 3. Performance of NIR 1 × 4 beamsplitters with different arms separation, showing splitting loss (SL) in dB, splitting ratio (SR) in %, and SR uniformity in %.
Figure 11(b) presents waveguides and beamsplitters fabricated using a 7-nJ ULI pulse energy and characterized at 4.55 μm. The straight waveguide exhibited a propagation loss of 3.3 dB/cm. Table 4 shows that both the SL and SR uniformity, of these MIR 1 × 4 beamsplitters improved at the 100-µm arm separation to the values of 1 dB and 98.2% respectively.
Table 4. Performance of MIR 1 × 4 beamsplitters with different arm separations, showing splitting loss (SL) in dB, splitting ratio (SR) in %, and SR uniformity in %.
These findings suggest that the splitting angle alone is not a determining factor in beamsplitter performance. Interestingly, the performance at a 100-µm arm separation for the MIR beamsplitter appears to improve, potentially because of the reduced interaction between the arms. Factors such as evanescent field coupling [15], cross-talk between the arms, sequential ULI writing scheme and spherical aberrations introduce additional layers of intricacy that complicates any straightforward correlation between splitting angle and the beamsplitters performance. Therefore, a more detailed study of the ULI parameters that considers all these variables incorporating computational simulations along with experimental validation, could help reduce splitting loss and improve the splitting ratio between the arms.
5. Conclusion
This study presents a comprehensive investigation of the inscription of waveguides in IG2 glass using ultrafast lasers with multilayer and multiscan approaches, leading to multilayer-multiscan type-I waveguides at 1.064 µm and 4.55 µm. The weak type I modification starts at 3-nJ pulse energy, transitioning to a strong regime above 6 nJ, causing darker waveguides. In the weak modification regime, a 6-nJ pulse energy led to efficient light confinement at 1.064 µm. However, single-layer waveguides at this energy level did not confine light at 4.55 µm. Higher pulse energies applied to single-layer waveguides resulted in irregular and darker formations; however, they were efficient for light guidance at 4.55 µm, indicating a shift to a strong modification regime. The weak modification regime revealed an improvement in confinement at this wavelength upon the introduction of the multilayer configurations, leading to the lowest propagation loss of 0.9 dB/cm in a five-layer configuration. Beamsplitters with multiple configurations, ranging from 1 × 2 to 1 × 8, were fabricated for the first time using various pulse energies and splitting angles. Relatively uniform splitting ratios were achieved, with splitting loss at 1.064 μm and 4.55 μm ranging from 0.4 dB to 2.9 dB.
The ability to control mode confinement in the MIR and the successful implementation of beamsplitters have significant implications in various fields, including environmental monitoring, chemical and biological sensing, medical diagnostics, and nonlinear propagation of ultrashort optical pulses, e.g., supercontinuum generation. The use of multilayer-multiscan techniques can pave the way for the realization of advanced MIR photonic lanterns that distribute MIR light more efficiently and precisely.
Appendix A
This appendix aims to clarify the impact of spherical aberration on straight waveguides, providing supplemental information to explain the fabrication parameters for the beamsplitter in the strong-modification regime [Sections 4.1 and 4.2].
Figure 12 shows the output cross-section images of straight waveguides inscribed with pulse energies ranging from 6 to 8 nJ, and at depths ranging between 50 µm and 450 µm from the top surface of the IG2 sample, and the corresponding beam profiles guided at 4.55 µm. By utilizing a multiscan approach with 22 scans and 0.6-µm overlaps, an increment over the 13 scans and 1-µm overlap waveguides previously presented in Sections 3.1 and 3.2, these waveguides demonstrated improved mode confinement, enabling guidance even at pulse energies as low as 6 nJ.
Fig. 12. (a) Output cross-section images and (b) corresponding guided mode profiles at 4.55 μm for straight waveguides inscribed with pulse energies between 6 and 8 nJ at depths ranging from 50 to 450 µm from the top surface of the IG2 sample. The scale bars represent 30 µm.
A total of 45 waveguides were fabricated with energies between 6 and 8 nJ at depths between 50 and 450 µm. As the focal location of the laser descends deeper, the spherical aberration reduces the fluence, which leads to progressively weaker modifications. For instance, at a 50-µm depth, only waveguides inscribed with pulse energies of 6.0 nJ and 6.5 nJ showed smooth modification. Conversely, at a depth of 450 µm, smooth modifications were observed for almost all pulse energy levels, except at 8 nJ, as shown in Fig. 12(a). For clarity, a red line delineation is added between the smooth and strong modification regimes. For the beam profiles shown in Fig. 12(b), the zones of smooth modification are visually distinct from the areas with more intense modification. A significant shift to profound mode confinement was evident when transitioning from smooth to strong modification regimes.
Figure 13 presents the interpolated maps of the insertion loss (IL), propagation loss (PL), mode field diameter (MFD), and change in refractive index (Δn). Figures 13(a) and 13(b) show a clear trend that both IL and PL initially improve as the waveguide moves from the weak to the smooth modification regime. However, the losses begin to increase as the waveguide moves from smooth to strong modification regimes. This can be attributed to two main factors: initially, the increase in confinement as the modification regime transitions from weak to smooth and then to strong modification.
Fig. 13. Interpolated experimental map of (a) Insertion loss (IL), (b) propagation loss (PL), (c) mode field diameter (MFD) and (d) refractive index change (Δn) with varying ULI pulse energy (horizontal axis) and the waveguide (WG) location (vertical axis).
Furthermore, the confinement continued to increase with increasing pulse energy up to the strong modification regime, beyond which it plateaued with only minor increases, as shown in Fig. 13(c) and 13(d). As the ULI pulse energy continued to increase within the strong modification regime, additional scatterers appeared, leading to an increase in losses.
Funding
National Aeronautics and Space Administration (80NSSC20C0027, 80NSSC21C0638, 80NSSC22PA936).
Acknowledgments
Portions of this work were presented at the Conference on Lasers and Electro Optics in 2023, Paper AM2R.5. The authors thank Dr Anthony Yu (NASA Goddard Space Flight Center) for fruitful technical discussions and the hardware loan that supported this study. Special appreciation is extended to Dr. Wei Hu and Dr. Muhammed Kilinc for developing the QCL-based characterization setup.
Disclosures
The authors declare that there is no conflict of interest.
Data availability
The data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. D. Sorak, L. Herberholz, S. Iwascek, et al., “New Developments and Applications of Handheld Raman, Mid-Infrared, and Near-Infrared Spectrometers,” Appl. Spectrosc. Rev. 47(2), 83–115 (2012). [CrossRef]
2. Y.-C. Chang, P. Wägli, V. Paeder, et al., “Cocaine detection by a mid-infrared waveguide integrated with a microfluidic chip,” Lab Chip 12(17), 3020–3023 (2012). [CrossRef]
3. J. Hodgkinson and R. P. Tatam, “Optical gas sensing: a review,” Meas. Sci. Technol. 24(1), 012004 (2013). [CrossRef]
4. A. Arriola, D. Choudhury, and R. R. Thomson, “New generation of photonic lanterns for mid-IR astronomy,” Advances in Optical and Mechanical Technologies for Telescopes and Instrumentation 9151, 915116 (2014). [CrossRef]
5. H. Lin, L. Li, Y. Zou, et al., “Chalcogenide glass planar photonics: from mid-IR sensing to 3-D flexible substrate integration,” in Laser Resonators, Microresonators, and Beam Control XV (2013), p. 86000 K.
6. B. J. Eggleton, B. Luther-Davies, and K. Richardson, “Chalcogenide photonics,” Nat. Photonics 5(3), 141–148 (2011). [CrossRef]
7. V. Singh, “Chalcogenide Glass Materials for Integrated Infrared Photonics,” Doctoral dissertation, Massachusetts Institute of Technology (2015).
8. J. Hu, J. Meyer, K. Richardson, et al., “Feature issue introduction: mid-IR photonic materials,” Opt. Mater. Express 3(9), 1571–1575 (2013). [CrossRef]
9. D. Choudhury, J. R. Macdonald, and A. K. Kar, “Ultrafast laser inscription: Perspectives on future integrated applications,” Laser Photonics Rev. 8(6), 827–846 (2014). [CrossRef]
10. S. M. Eaton, G. Cerullo, and R. Osellame, “Fundamentals of Femtosecond Laser Modification of Bulk Dielectrics,” Top. Appl. Phys. 123, 3–18 (2012). [CrossRef]
11. R. R. Gattass and E. Mazur, “Femtosecond laser micromachining in transparent materials,” Nat. Photonics 2(4), 219–225 (2008). [CrossRef]
12. S. Gross and M. J. Withford, “Ultrafast-laser-inscribed 3D integrated photonics: challenges and emerging applications,” Nanophotonics 4(3), 332–352 (2015). [CrossRef]
13. F. Chen and J. R. V. de Aldana, “Optical waveguides in crystalline dielectric materials produced by femtosecond-laser micromachining,” Laser Photonics Rev. 8(2), 251–275 (2014). [CrossRef]
14. J. Tepper, L. Labadie, S. Gross, et al., “Ultrafast laser inscription in ZBLAN integrated optics chips for mid-IR beam combination in astronomical interferometry,” Opt. Express 25(17), 20642–20653 (2017). [CrossRef]
15. K. Liu, C. R. Ye, S. Khan, et al., “Review and perspective on ultrafast wavelength-size electro-optic modulators,” Laser Photonics Rev. 9(2), 172–194 (2015). [CrossRef]
16. R. He, I. Hernández-Palmero, C. Romero, et al., “Three-dimensional dielectric crystalline waveguide beam splitters in mid-infrared band by direct femtosecond laser writing,” Opt. Express 22(25), 31293–31298 (2014). [CrossRef]
17. J. Lv, Y. Cheng, W. Yuan, et al., “Three-dimensional femtosecond laser fabrication of waveguide beam splitters in LiNbO3 crystal,” Opt. Mater. Express 5(6), 1274–1280 (2015). [CrossRef]
18. C. Cheng, C. Romero, J. R. V. de Aldana, et al., “Superficial waveguide splitters fabricated by femtosecond laser writing of LiTaO3 crystal,” Opt. Eng. 54(6), 067113 (2015). [CrossRef]
19. Y. Ren, L. Zhang, H. Xing, et al., “Cladding waveguide splitters fabricated by femtosecond laser inscription in Ti:Sapphire crystal,” Opt. Laser Technol. 103, 82–88 (2018). [CrossRef]
20. V. A. Amorim, J. M. Maia, D. Viveiros, et al., “Loss Mechanisms of Optical Waveguides Inscribed in Fused Silica by Femtosecond Laser Direct Writing,” J. Lightwave Technol. 37(10), 2240–2245 (2019). [CrossRef]
21. J. Tang, Q. Hu, J. Wang, et al., “Ni2+:KZnF3 glass-ceramics waveguide beam splitters inscribed by femtosecond laser,” in Laser Congress 2018 (ASSL) (OSA, 2018), p. ATu2A.2.
22. P. Gaso, D. Pudis, D. Seyringer, et al., “3D Polymer Based 1 × 4 Beam Splitter,” J. Lightwave Technol. 39(1), 154–161 (2021). [CrossRef]
23. H. L. Butcher, D. G. MacLachlan, D. Lee, et al., “Demonstration and characterization of ultrafast laser-inscribed mid-infrared waveguides in chalcogenide glass IG2,” Opt. Express 26(8), 10930–10943 (2018). [CrossRef]
24. H. L. Butcher, D. G. MacLachlan, D. Lee, et al., “Ultrafast laser-inscribed mid-infrared evanescent field directional couplers in GeAsSe chalcogenide glass,” OSA Contin. 1(1), 221–228 (2018). [CrossRef]
25. T. Gretzinger, S. Gross, A. Arriola, et al., “Towards a photonic mid-infrared nulling interferometer in chalcogenide glass,” Opt. Express 27(6), 8626–8638 (2019). [CrossRef]
26. T. Gretzinger, S. Gross, M. Ams, et al., “Ultrafast laser inscription in chalcogenide glass: thermal versus athermal fabrication,” Opt. Mater. Express 5(12), 2862–2877 (2015). [CrossRef]
27. S. Gross, N. Jovanovic, A. Sharp, et al., “Low loss mid-infrared ZBLAN waveguides for future astronomical applications,” Opt. Express 23(6), 7946–7956 (2015). [CrossRef]
28. J. R. Macdonald, S. J. Beecher, P. A. Berry, et al., “Efficient mid-infrared Cr:ZnSe channel waveguide laser operating at 2486 nm,” Opt. Lett. 38(13), 2194–2196 (2013). [CrossRef]
29. J. R. Macdonald, P. A. Berry, K. L. Schepler, et al., “Directly Written Mid-Infrared Waveguides in Zinc Selenide,” in Lasers, Sources, and Related Photonic Devices (OSA, 2012), p. IF1A.3.
30. W. Hu, M. Kilinc, W. Gebremichael, et al., “Morphology and waveguiding properties of ultrafast-laser-inscribed type-II waveguides in IG2,” Opt. Mater. Express 12(1), 360–373 (2022). [CrossRef]
31. W. Hu, W. Gebremichael, J. Fernandes, et al., “Single-scan ultrafast laser inscription of waveguides in IG2 for type-I and type-II operation in the mid-infrared,” Opt. Express 31(12), 18949–18963 (2023). [CrossRef]
32. R. R. Thomson, S. Campbell, I. J. Blewett, et al., “Active waveguide fabrication in erbium-doped oxyfluoride silicate glass using femtosecond pulses,” Appl. Phys. Lett. 87(12), 121102 (2005). [CrossRef]
33. S. M. Eaton, H. Zhang, P. R. Herman, et al., “Heat accumulation effects in femtosecond laser-written waveguides with variable repetition rate,” Opt. Express 13(12), 4708–4716 (2005). [CrossRef]
34. M. Somayaji, C. D’Amico, Y. Wu, et al., “Influence of thermal annealing on ultrafast laser-induced local densification in bulk sulfur-based chalcogenide glasses,” Phys. Status Solidi A 216(3), 1800568 (2019). [CrossRef]
35. Y. Shimotsuma, P. G. Kazansky, J. Qiu, et al., “Self-organized nanogratings in glass irradiated by ultrashort light pulses,” Phys. Rev. Lett. 91(24), 247405 (2003). [CrossRef]
36. V. R. Bhardwaj, E. Simova, P. P. Rajeev, et al., “Optically Produced Arrays of Planar Nanostructures inside Fused Silica,” Phys. Rev. Lett. 96(5), 057404 (2006). [CrossRef]
37. D. Marcuse, “Loss analysis of single-mode fiber splices,” Bell Syst. Tech. J. 56(5), 703–718 (1977). [CrossRef]
38. K. Okamoto, “Planar optical waveguides,” in Fundamentals of Optical Waveguides (Elsevier, 2006), pp. 13–55. [CrossRef]
39. G. P. Agrawal, “Optical fibers,” in Fiber-Optic Communication Systems (Wiley, 2021), pp. 24–74.
40. F. Kong, C. Dunn, J. Parsons, et al., “Large-mode-area fibers operating near single-mode regime,” Opt. Express 24(10), 10295–10301 (2016). [CrossRef]
41. A. Mermillod-Blondin, I. M. Burakov, Y. P. Meshcheryakov, et al., “Flipping the sign of refractive index changes in ultrafast and temporally shaped laser-irradiated borosilicate crown optical glass at high repetition rates,” Phys. Rev. B 77(10), 104205 (2008). [CrossRef]
