1. Introduction

There is currently strong interest in developing 3D integrated optics (IO) technologies for a wide range of applications including telecommunications [1], biophotonics [2], miniaturized lasers [3], optofluidics [4] or astrophotonics [5]. Unlike currently well-established IOs fabrication techniques such as annealed proton-exchange [6] or ion-indiffusion [7], direct laser writing (DLW) has proven so far to be a technique capable of producing embedded high quality 3D waveguide circuits inside transparent glasses [8–10], but it is still not so mature for processing crystals. This technique relies on the tight focusing of ultrashort laser pulses to produce a localized index change only at the point where the threshold for modification is exceeded, i.e. at the focal volume. There are generally two possible approaches for creating DLW waveguides in crystals: one in which a refractive-index increase is produced at the laser focus and therefore a guided mode is directly obtained (Type I waveguides) [10], and another in which the refractive-index is directly reduced (Type II), as a result of causing some kind of lattice damage or modification. In the case of Type II reduced-index modifications, waveguiding can be produced either by conforming a cladding and therefore obtaining a leaky mode [11–13], or by additionally making use of the surrounding stress-fields and the associated piezo-optic refractive index increment, thus obtaining a properly confined guided mode [14–16].

The DLW of increased refractive-index regions (Type I waveguides) in crystals has been proven to be very elusive, as crystals already have an optimized lattice structure which can hardly be modified without inducing disorder and damage which generally lowers the index. As a result, the fabrication route of Type II waveguides is the most widely spread for processing crystals with ultrashort laser pulses. This approach simply relies on the fact that all materials have a certain damage threshold, so that theoretically a low-index cladding region can always be produced both in glasses and crystals [11,12]. In addition to this damage cladding, and in order to have a strictly guided non-leaky mode, an increased refractive index region is also required [13]. This guiding (refractive index increased) region can be provided by the strain which typically forms around damage tracks [14–16]. Due to this simplicity, the DLW fabrication of Type II stress-induced waveguides have already been demonstrated in a very wide variety of crystals, such as quartz [14], LiNbO₃ [15,17], sapphire [18,19], YAG [12,16,20,21], KGW [22], YVO₄ [23], GdVO₄ [24], YAB [25], and KTP [26], amongst others.

Regardless of this success, Type II stress-induced waveguides have a number of very important limitations which need to be overcome for the development of future 3D IO circuits: (1) the cladding damage tracks which fasten the stress-fields inevitably introduce scattering losses, and (2) the smooth refractive index profiles induced by stress-fields are not easy to control in terms of contrast, and always decay gradually rather than in a controlled way. Consequently, a high modal confinement is extremely difficult to obtain, unless extra damage cladding regions are introduced and these would increase the propagation losses even more [20]. Due to these constraints, many future potential implementations of DLW IO devices in crystals are likely to be severely compromised. So far, the longest guided wavelength reported in a Type II waveguide is 1.55 µm [27], and the typically reported refractive index contrasts (C = ∆n/n₀, with Δn = n_core-n₀) of low-loss Type II waveguides (i.e. with propagation losses lower than 1 dBcm⁻¹) are no higher than 0.04% [16].

To overcome these limitations one needs to fabricate Type I waveguides, in which the focused laser pulses produce a localized index increase in the focal region without much damage, allowing for low-loss 3D waveguide circuits with tailored properties (such as step-index profiles, precise core sizes and index contrasts) to be easily designed and implemented. However, Type I waveguides in crystals have proven to be very difficult to achieve, and until now they have only been demonstrated in LiNbO₃ [17,28,29], and recently in ZnSe crystals [30], with maximum guiding wavelengths of 1.55 µm and estimated index-contrasts typically lower than 0.14% [29]. And yet, many of the previously mentioned IO applications require the use of longer wavelengths in the near-IR and mid-IR regions, and minimum core index-contrasts of ~0.5%, for which the bend radiation losses start to be acceptable for bend radii ~50 mm [31,32].

In this work we report the ultrafast fabrication of step-index Type I waveguides in borate crystals with estimated index-contrasts in excess of 0.6% and propagation losses of 1.1 dBcm⁻¹ at 1.55 µm. We further demonstrate guiding at the longer wavelengths of 1.95 µm and 3.39 µm, for the first time to our knowledge in a laser written crystal. By performing µ-Raman mapping we have analyzed the waveguide profiles, enabling the visualization of the modified index regions and confirming that the waveguides have very well defined step-index profiles.

2. Experimental

2.1 Ultrafast 3D laser fabrication setup

For DLW a IMRA µJewel D400 fiber laser system emitting at 1047 nm wavelength was used in combination with precision air bearing Aerotech 3D translation stages. Pulse energies were studied in the range 0.22-1.33 µJ at the focus position within the sample. Laser powers were calibrated after the focusing lens and taking into account sample reflectivity. Laser polarization was in all cases linear and perpendicular to the laser scan direction. Two different lenses were used for focusing: a low numerical aperture (NA) aspheric 0.6 NA lens, and a high 1.4NA oil immersion Olympus UPlanSAPO microscope objective, as used in previously reported work [33]. Focusing was typically performed 100 µm below sample surface.

The low NA lens was intended for fabricating Type I multiscan waveguides [34]. The laser writing speeds were 1.7 cms⁻¹ and 6 cms⁻¹, this being a 10-fold increase in scan speeds over those used for previous multiscan Type I waveguides in crystals [29]. The transverse scan separations studied were 0.21, 0.42, 0.63, 0.85 and 1.06 µm. These spacings correspond to 1/e² spot multiscan overlaps of 90%, 80%, 70%, 60% and 50%, respectively. The pulse duration and repetition rate were 350 fs and 200 KHz, respectively. The high NA focusing lens was intended to be used for fabricating Type II damage cladding regions were light confinement could be obtained due to the higher laser fluence at the focal spot. For this, writing was performed at a 100-fold slower speed of 0.1 mms⁻¹, with longer pulse duration of 1.2 ps, and a lower repetition rate of 100 KHz. Figure 1 shows the sketch diagram of both waveguide fabrication approaches with low and high NA focusing lenses.

 

Fig. 1 Fabrication schemes studied in this work: Laser focusing is always performed along the positioning stage vertical z-axis. Waveguide light propagation is along the horizontal y-axis. A low NA focusing lens is used to fabricate square cross-section waveguides. Multiple scans are performed along the y-axis and overlap along the x-axis to build up a homogenous step-index core profile. A high NA lens is also used to fabricate waveguides with smaller cross-section sizes and closely-packed 2D arrays.

Download Full Size | PDF

The borate Nd³⁺:YCa₄O(BO₃)₃ crystals (Nd:YCOB) were Czochralski grown with a Nd³⁺ ion concentration of 5 at.% and sizes of 3x3x5 mm³. Used samples were originally cut for phase matching 1.06 µm and 0.53 µm wavelengths along the directions (66.5°, 35.5°) and (33°, 9°) and therefore waveguides were written to propagate along those directions. All sample facets were always polished to optical quality before and after laser microfabrication.

2.2 Characterization setup

All the characterization data presented in this work was obtained from 1 year old samples. Waveguides were fabricated in May 2010 and the final data presented here was obtained in May 2011. Importantly, this ensured that the waveguides were stable with time at room temperature. Routine sample examination was always undertaken using visible light transmission optical microscopy. Waveguide propagation in the near-IR was tested with a 0.98 µm laser diode, a 1.55 µm tunable laser, and a 1.94 µm laser diode. In-coupling was carried out with a SMF28 fiber and out-coupling with a 0.85NA microscope objective. Results at 0.98 µm wavelength are not shown in this paper for the sake of brevity. At the mid-IR a 3.392 µm HeNe laser was used, coupling was performed by using a pair of 0.25NA ZnSe objectives and the power and polarization were controlled by using a MgF₂ λ/2 plate and a BaF₂ wire grid polarizer. All near-field profiles were imaged using a FLIR SC700 camera. Propagation modes were also numerically simulated using COMSOL Multiphysics® software. Propagation losses were measured by using the tunable source at 1.55 µm and by recording the Fabry-Perot fringes at the output of waveguides using the same setup as reported in Ref. [30].

2.3 Confocal µ-Raman setup

After the guiding characterizations, confocal μ-Raman high-resolution surface mapping of the waveguide facets was performed using a Renishaw inVia Reflex microscope attached to a 514 nm argon laser. The Raman signal was obtained in a back-scattering configuration by using a filled 1.4NA oil immersion Olympus UPlanSAPO microscope objective with an estimated focal diameter spot of ~450 nm at a 1/e² intensity (d_spot = 1.22λ₀/NA), at 514 nm wavelength. All measurements were performed whilst keeping the laser power at 1 mW to avoid sample heating or modification as a result of the high fluence visible illumination. Spectral data analysis was performed using WiRE® 3.2 software, by deconvoluting each measured spectra to obtain the different phonon mode components: energy, linewidth, and intensity. Analysis of the spectral changes in the ⁴F_3/2-⁴I_9/2 neodymium ions emission band at around 900 nm wavelength was also performed by using the same microscope setup.

3. Results and discussion

3.1 Low NA written multiscan waveguides

Initial experiments were performed by inscribing two equal parallel tracks separated by 30 µm at a speed of 1.7 cm·s⁻¹, and observing if there was any Type II stress-induced mode guiding between each of the two inscribed parallel tracks at 1.55 µm wavelength (see §1 for more detailed definition of Type II index changes). These writing conditions were exactly the same as those recently reported by the authors for writing Type II waveguide lasers in Nd:GdVO₄ crystals, where two damage tracks were fabricated and the characteristic stress-field induced light confinement between them was obtained [24]. Surprisingly, in the case of Nd:YCOB crystals, no Type II waveguides could be observed between any of the double track structures even when using the maximum available irradiation pulse energy of 1.33 µJ. On the contrary, light guiding could be observed through most of each single tracks (rather than between them), and with a very high level of light confinement. This suggests therefore that the written track modifications correspond to Type I increased refractive index regions (see §1 for more detailed definition of Type I index changes).

Figure 2(a) shows the different double track structures which were fabricated, as seen under the optical transmission microscope. Visible light transmission allows three different regimes to be distinguished: a very localized low-index change regime appears for pulse energies in the range 0.2-0.3 µJ (#1), for energies in the range 0.36-0.82 µJ a second regime appears where the index has clearly increased (#2), and for higher energies of 0.96-1.33 µJ the tracks appear to be darker and more complex (#3), with pre-focal regions were the index modification is more similar to the first low-index regime due to the lower laser fluence at these points. Only waveguide tracks corresponding to pulse energy regimes #2 and #3 were observed to guide 1.55 µm near-IR light, whereas tracks corresponding to the regime #1 where not observed to guide. All guided modes were always observed to be strongly polarized along the vertical direction (laser writing direction), so that no horizontally polarized light was transmitted.

 

Fig. 2 Scheme of laser fabrication parameters studied and resulting waveguides when using a 0.6NA focusing lens, 200 KHz and 350 fs pulse duration. (a) Single scans where studied inscribing 2 parallel lines for 12 different energies. Speed was fixed at 1.7 cms⁻¹. (b) Multiscan waveguide study, performed with an increased speed of 6 cms⁻¹ and different combinations of pulse energy and scan overlaps. Pulses arrive from the top of the images.

Download Full Size | PDF

In order to obtain well defined core structures for near-IR guiding, the multiscan procedure was implemented which implies overlapping several single line tracks, in order to “paint” a rectangular cross-section refractive index increased core [34]. However, for this operation we required the use of even higher writing speeds. As it can be observed from Fig. 2(a), all waveguide lines corresponding to the energy regimes #2 and #3 have a very long vertical size (~20-70 µm), but if small waveguide cores are required, their sizes should be limited to no more than 10 µm in cross section. For this reason, the writing speed was increased and set to 6 cm·s⁻¹, in order to fabricate smaller size cores, and to allow further tailoring of the waveguide size by simply controlling the pulse energy. Figure 2(b) shows the results of producing such ultrafast written multiscan waveguides. As it can be seen, the waveguide vertical cross-section was clearly reduced to about 10 µm on average for the 3 pulse energies of 0.67, 0.79, and 0.93 µJ studied. As described in §2.1, for each of these 3 energies, 5 different multiscan overlaps of 50, 60, 70, 80 and 90%, were used. In order to easily perform the µ-Raman mapping study of these different structures, waveguides with multiscan overlaps from 50% to 80% were placed together, so that a single mapping experiment could be carried out over the whole set of parameters (see §3.4). As labeled in Fig. 2(b), these structures are here referred to as MS1, MS3 and MS5. For scan overlaps of 90% only, a single independent waveguide was made for characterization, each of these corresponding to waveguides MS2, MS4 and MS6.

3.2 Near-IR (1.94 µm and 3.39µm) waveguide propagation characterization

For propagation characterization the intermediate energy waveguide MS4 was chosen, as it was found to give the best mode confinement and photon flux on the FLIR camera, with respect to MS2 and MS6. All waveguides showed guiding only for vertical polarizations. Even if the MS4 waveguide was multimode at 1.55 µm, the number of high order modes which could be observed was no higher than 2 or 3 and therefore the propagation losses could be measured using the Fabry-Perot technique. This is because these higher order propagation modes do not make a significant contribution to the measurement as they extend further into the cladding and therefore suffer higher propagation losses. A value of 1.1 dBcm⁻¹ was obtained for the propagation loss at 1.55 μm wavelength.

The MS4 waveguide was found to be strictly single mode at 3.39 µm and quasi-single mode at 1.95 µm. Only one LP₁₁ mode was found in the vertical direction at 1.95 μm, and to excite it input light from the SMF fiber had to be launched at a misaligned vertical position in between core and cladding. The modal detailed characterization of the MS4 waveguide could be made using the well-defined fundamental LP₀₁ modes at 1.95 µm and 3.39 µm wavelengths. Figure 3 shows the guiding results overview. Figure 3(a) shows the waveguide facet microscope transmission image, where it can be seen that strong index change modification has only occurred in a highlighted squared section. From Fig. 3(a) the cross-sectional core size of the waveguide could be estimated to be 7.4 µm x 9.6 µm. Figure 3(b) shows the normalized horizontal intensity profiles of the near-field modes at 1.92 µm and 3.39 µm. From these waveguide core size and intensity profiles, a modal analysis was performed using COMSOL Multiphysics® software, by calculating the possible fundamental modes of a step-index waveguide with a constant refractive index core and a rectangular shape, as observed from Fig. 3(a). The Sellmeier equation for the crystal refractive index at each wavelength was used (Ref. [35]), and different values for the core refractive index were iteratively computed in ±10⁻⁴ index variation steps until the calculated mode profiles matched the experimental one. In this way, the dispersion of the refractive index modification could be estimated at each different wavelength. Both the final calculated and measured mode profiles show a very satisfactory agreement. The numerically calculated mode profiles are also shown together with the experimental ones in Fig. 3(b), so that good matching between experimental and numerically calculated modes can be seen. Figures 3(c) and (d) show the measured near-field mode images at 1.94 µm and 3.39 µm wavelengths and Figs. 3(e) and (f) their simulated counterparts.

 

Fig. 3 Optical near-IR characterization of the MW4 waveguide. (a) Visible light transmission image obtained with an oil-immersion 100X 1.4NA UPlanSApo Olympus microscope objective. From this image the estimated waveguide size is of 7.4 µm x 9.6 μm. (b) Horizontal intensity cross-sections of the near-field profiles of fundamental modes at 1.94 μm and 3.39 μm wavelengths. Both measured and calculated modes are shown. (c) and (d) Measured near-field images of single-modes. The waveguide core is also shown. (e) and (f) Corresponding calculated modes and waveguide core.

Download Full Size | PDF

From the numerical simulation of fundamental modes, the set of waveguide parameters can be estimated for each test wavelength. The core refractive index changes (index-contrasts) are 0.005 (0.29%) and 0.010 (0.59%) at 3.39 µm and 1.94 µm wavelengths, respectively. Approximating the waveguides as cylindrical step-index cores, the normalized frequency, V and the normalized propagation constant, b of the waveguides can also be calculated for each wavelength [36]. The theoretical circular core radius which gives an equivalent cross-sectional area to that of the MS4 rectangular one is 4.78 µm. From that radius the normalized frequencies of V~1.15 and V~2.84 were obtained for the wavelengths of 3.39 µm and 1.94 µm. Similarly, normalized propagation constants of b~0.08 and b~0.60 were obtained for the wavelengths of 3.39 µm and 1.94 µm, respectively. The low V value at 3.39 μm indicates that the fundamental mode has more than half of its modal power propagating in the cladding [36]. Conversely, at 1.94 μm the higher V value indicates that the fundamental mode is very well confined. These observations are in very good agreement with the experimental mode profiles as shown in Figs. 3(c) and (d).

The experimental mode field diameters (MFD) at 1/e² intensity values are of 20.2 µm and 8.0 µm, at 3.39 µm and 1.94 µm wavelengths, respectively. As the waveguide modes are almost circular, the NA of the waveguides is also expected to be axially symmetric, and the standard definition of a diffraction-limited focal spot diameter at 1/e² intensity can also be used to estimate an upper value for the modes NA (NA~1.22λ/MFD). By using the measured MFDs, the NA_max are calculated to be 0.2NA and 0.3NA at 3.39 µm and 1.94 µm wavelengths, respectively. Additionally, by using the standard definition for the NA of a fiber (NA = (n_core ²-n_crystal ²)^1/2), the corresponding NA at 3.39 µm and 1.94 µm are 0.13NA and 0.18NA, which are clearly lower than the NA_max.

Finally, considering that the value of the crystalline core index probably follows a slowly varying function with wavelength, the core refractive index at 1.55 µm can also be estimated to a first order of approximation from the values at 3.39 µm and 1.94 µm. Accordingly, the core index change (index contrast) of the MS4 waveguide at 1.55 μm wavelength is found to be of ~0.011 (~0.67%). A summary of all measured and estimated parameters is included in Table 1 .

Table 1. Summary of the MS4 waveguide parameters. Unshaded, experimental data; grey shaded, estimated data.

View Table

3.3 High NA single scan waveguides

As explained in §3.2, none of the Type I waveguides showed guiding for horizontal polarization, nor stress-induced guiding. In order to test the possibility of inducing damage claddings as well as generating stress-fields capable of confining light, a much tighter focusing setup was used in addition with longer temporal pulse durations [15]. As explained in §2.1, an oil immersion 1.4NA objective was used at maximum pulse energies of 1.1 μJ at the focal volume ; at scanning speeds of 0.1 mms⁻¹, a pulse repetition rate of 100 KHz and a duration of 1.2 ps. A 2D hexagonal waveguide array was written as shown in Fig. 1. If these tracks were composed of damaged crystal, a highly confined region would be expected in the center crystalline channel inside the hexagonal array. Similar geometries have already been successfully fabricated in laser glasses [11].

In order to compare in a normalized manner the irradiation conditions between the previous low NA written waveguides and these high NA written ones, the mean laser fluence at the focus, F = E^pulse/a_spot, and the pulse overlap, O = 1-(s/f·d_spot), can be compared; where E^pulse is the energy pulse, a_spot is the spot area at a normalized intensity of 1/e², s is the scan speed, and f the pulse frequency. In the case of Type I multiscan waveguides, a maximum fluence of 34 J/cm² and a low pulse spatial overlap of 86% were used. In contrast to this, in order to see whether we could induce damage and optical breakdown in the NdYCOB crystals using the high 1.4NA lens and lower speed,, we used an almost 5-fold higher laser fluence of 154 J/cm² and a much increased overlap of 99.89%.

Even under these conditions, no stress-field modal confinement could be obtained inside the 2D array at 1.55 μm wavelength. Instead, big positive index regions surrounded by a more complex modification morphology were observed to have formed, as evident from Fig. 4(a) . These Type I positive index waveguides were therefore configured as a 2D array of evanescently coupled single mode waveguides, as can be seen in Fig. 4(b). The waveguides were observed to be strictly single mode at 1.55 μm.

 

Fig. 4 (a) Microscope image of the two-dimensional hexagonal array of single scans waveguides fabricated with 1.2 ps pulses, slow speeds of 0.1 mms⁻¹, high NA1.4 and 1.1 µJ pulse energy. Laser writing was performed from the top of the image. (b) C-band 1.55 µm wavelength near field image of the guided modes. Mode intensity color scale is the same as in Fig. 3. Guiding was only observed to occur for vertical polarization, and almost no stress-induced guiding could be observed inside or outside the hexagonal array.

Download Full Size | PDF

3.4 Raman analysis of the laser induced refractive index changes

Figure 5 shows the confocal µ-Raman spectra for the most intense phonon modes at around 930 cm⁻¹. This energy region of the Raman spectrum of oxoborate crystals has been reported to correspond to the stretching vibration of constituent B-O molecular rings within the crystalline lattice [37]. The normalized µ-Raman spectra of non-irradiated Nd:YCOB (blue) and refractive index increased waveguide cores (red) are shown for comparison. As can be clearly seen, the two phonon modes at 938.4 cm⁻¹ and 950.8 cm⁻¹ are blue-shifted at irradiated areas, as well as broadened. In addition to this, an extra phonon band is formed at around 918 cm⁻¹.

 

Fig. 5 Normalized Raman spectra inside the waveguide MW4, and at a non-irradiated point, showing the most active 938 cm⁻¹ and 950 cm⁻¹ phonon modes, and the defect mode associated with refractive index increased zones.

Download Full Size | PDF

The origin of the observed phonon band is not known, but due to its broad shape it can be tentatively associated with the creation of intrinsic defects within the Nd:YCOB lattice. In the case of LiNbO₃ crystals, similar phonon bands are known to appear in congruent LiNbO₃ crystals when compared to stoichiometric ones [38]. In that situation, the origin of the bands is due to an increase in the phonon density of states due to the slightly disordered congruent lattice with respect to the stoichiometric one, where the origin of the structural disorder is the slight Li deficiency. We believe that a possible phenomenological explanation for the appearance of this band in Nd:YCOB is the out diffusion of elements from the irradiated volumes, as it has been observed to occur when writing with 250 KHz fs pulse lasers in glasses [39]. However, in order to check this possibility, further experiments need to be done, but are beyond the scope of this study.

Confocal µ-Raman surface mapping was also performed in multiscan structures MS1 and MS3 (see §3.1). These structures have a rectangular shape, where the induced modifications (index change) are increasingly higher towards one direction. The composed modification was performed by increasing the multiscan overlaps in four steps. Figures 6 (a) and (b) show the microscope transmission image of structures MS3 and MS1, respectively.

 

Fig. 6 Micro-Raman analysis of multiscan structures MS3 and MS1, as shown in Fig. 2. Each structure is composed of 4 segments of equal horizontal size but different scan separations of 0.42/0.64/0.85/1.06 µm, from left to right. (a, b) Microscope images of the structures. (c, d) Normalized integrated area of lattice defects band at around 920 cm⁻¹ (see Fig. 5(b)). (e, f) 2D mapping of the 950.83 cm⁻¹ phonon mode energy. (g, h) 2D mapping of the phonon mode FWHM. All color scales are normalized to min. and max. values, respectively and are indicated in cm⁻¹ units. (i.e. Note that the max. shift values for the MS1 are approximately a half of those for the MS3).

Download Full Size | PDF

In these spectral maps the effects of gradually increasing the multiscan separation can be very clearly seen. The integrated area of the tentatively assigned phonon defect band at around 920 cm⁻¹ is first plotted in Figs. 6 (c,d). As can be seen, the creation of intrinsic defects is strictly limited to the rectangular fabricated core. This indicates that the refractive index increase is directly related to the creation of these defects and also that the core morphology and shape can be directly imaged by this technique. The energy and linewidth of the 950.83 cm⁻¹ phonon mode is also plotted in Figs. 6 (e)-(f). Results for the 938.4 cm⁻¹ phonon mode were essentially equal in features (relative energy, linewidth, and intensity changes with respect to laser parameters) to those of the other 950.83 cm⁻¹ band, and are therefore not shown here for the sake of brevity. Both the energy shifts and broadening of modes are also observed to be localized at the waveguide core and not present in its surroundings. This indicates that the modification of these modes is linked to defect generation. The increase in phonon energy (blue-shift) points to the distortion of B-O ring units caused by an increase in their stretching vibration energy, while the increase in linewidth indicates that a small increment in the lattice local disorder is also taking place. The reasons why these unknown lattice distortions produce an anisotropic index change, which implies that the waveguides only guide vertically polarized light, are however out of the scope of the present work, and are yet to be understood.

To gain further insight into the relation between lattice distortion and fabrication parameters, the horizontal cross section of maps presented in Figs. 6 (e) and (f) is analyzed. Figure 7(a) shows these cross-sections for structures MS1 and MS3. As can be seen, the cross-sections are modulated in steps which correspond to the spatial regions where different multiscan overlaps have been used. Measuring the mean value for each section, allows the phonon mode energy shift to be plotted as a function of overlapping. Figure 7(b) shows these plots for the two pulse energies of 0.67 µJ and 0.79 µJ, corresponding to structures MS1 and MS3, respectively. While for the low pulse energy of 0.67 µJ the energy shift increases gradually with increasing overlap (decreasing scan separation), for the higher pulse energy of 0.79 µJ a clear saturation is observed for multiscan separations shorter than 0.6 µm. This indicates a slight lattice re-accommodation could be taking place due to the laser overwriting, lowering the B-O ring distortion by a different extent, even if the density of intrinsic defects has continued to increase, as seen in Fig. 6 (a).

 

Fig. 7 Detailed analysis of the Raman energy-shifts of the 950.83 cm⁻¹ phonon mode in structures MS1 and MS3. (a) Centered horizontal cross sections showing the different dependences of the phonon mode energy on the combined values of pulse energy and laser scan separation. (b) Raman shift as a function of laser scan separations. A 3 fold increase in the energy shift is observed when the pulse energy is increased by 0.12 µJ, reaching a maximum of +1.92 cm⁻¹. A sudden decrease is observed for scan separations lower than 0.6 µm and for the higher energy of 0.79 µJ, indicating saturation and annealing of lattice defects.

Download Full Size | PDF

The spectral changes in the ⁴F_3/2-⁴I_9/2 neodymium ions emission band at the waveguide core were also studied. Due to the relevance of these waveguides to potential waveguide laser developments, an assessment of the luminescence properties of modified crystals is also required. In this case some peak broadening is observed, together with an increase in the measured intensity. This is shown in Fig. 8 , where the spectrum from bulk non-irradiated crystal is shown in blue and that from the center of the waveguide core is shown in red. Although some line broadening is always expected in directly irradiated zones due to the distortion of the lattice, the increment in the detected intensity is less common. Two hypotheses can be presented to explain this observed enhancement: either the observed behavior may be due to a purely optical effect attributable to the confocal microscope back-scattering setup involving coupling with the waveguide, or the concentration of Nd³⁺ ions has increased in the waveguide core as a result of the irradiation. Either way it is clear that the waveguide core retains the good luminescence properties of the original bulk crystal.

 

Fig. 8 Measured µ-luminescence spectra of the ⁴F_3/2-⁴I_9/2 neodymium ions emission band from bulk crystal (blue), and from the center of the waveguide core (red).

Download Full Size | PDF

4. Conclusions

We have demonstrated the ultrafast fabrication of high contrast, step-index channel waveguides in borate Nd:YCOB crystals for the first time. Very high processing speeds of 6 cms⁻¹ are demonstrated for the fabrication of waveguides capable of guiding up to mid-IR wavelengths around 3.4 μm. Modeling the measured fundamental modes at the wavelengths of 1.9 µm and 3.4 µm allowed us to estimate the laser-induced refractive index increments (index contrasts) to be 0.010 (0.59%), and 0.005 (0.29%), respectively. These contrast values are the highest so far reported for laser written waveguides in a crystal. Confocal µ-Raman spectral imaging of the sample cross-section allowed us to observe that the waveguides have well defined step-index cores. Furthermore, the increase in the refractive index was linked to the localized creation of intrinsic defects. The Nd ions luminescence band at around 900 nm was also measured to confirm the potential of the waveguides as waveguide laser candidates.

The intrinsic limitations that the fabricated waveguides have are: (1) they can only guide vertically polarized light, and (2) they may introduce changes in the electro-optic properties of the fabricated cores with respect to bulk Nd:YCOB due to the lattice distortion responsible for the refractive-index increase [40]. Limitation (1) is very fundamental and common to this type of multiscan waveguide, and it will therefore be difficult to overcome. Limitation (2) however, in the case it takes place, should be successfully overcome by properly tailoring the fabrication and waveguide parameters. As has been demonstrated here, the extent to which the crystalline lattice is distorted can be precisely controlled by changing the fabrication parameters, such as scan speed, pulse energy, multiscan overlapping, focusing optics and others. Furthermore, in addition to the 3D capability, the technique also allows the NA of the waveguide to be precisely varied to achieve good transitions for coupling into fibers or other IO components. We therefore believe that these waveguides are promising candidates for the near future development of integrated on-chip waveguide circuits for a great variety of optical applications.