1. Introduction

In the last two decades, the internal material processing by using a femtosecond (fs) laser is a burgeoning field, and an emerging technology for fabricating versatile three dimensional (3D) photonic devices possessing outstanding precision for use in various fields like optical telecommunication, and bio-photonic applications [1–10]. Of particular interest, fs-laser allows various permanent modifications to be written in most transparent materials such as glasses, polymer, and crystals. In most glasses (apart from pure silica), the first permanent “damage” threshold is usually defined a laser-induced negative volume changes (i.e., a permanent expansion) surrounded by a stress field [5,6,8,10]. The second threshold is defined by the appearance of a strong birefringence, which can be controlled by using the writing laser polarization including repetition rate, pulse energy, and scanning speed [4–6,8]. These 3D localized permanent changes in linear optical properties have several advantages, and are used in numerous applications [1–10] including waveguide lasers, 3D-optical data storage, optofluidics, and for polarization controlled birefringent devices, which is a today rapidly expanding field.

Fs-laser-induced birefringence can be engineered either through a control of the stress-field in response to permanent volume changes or in response to a controllable form of birefringence that is related to the formation of self-organized nanostructures [2,3,5] known as Nanogratings. These nanostructures, and related birefringence were initially observed in a handful of materials namely fused silica [4], and slightly doped silica glasses [2,4], but they now appear in many materials including bulk crystalline oxide materials namely TeO₂ [11,12], and sapphire [13]. One major drawback is related to a strong scattering losses, and light depolarization that occur due to the intrinsic nature of these nanogratings, which are based on a nanoscale phase separation [14,15].

On the other hand, stress field provides a laser-free area that can be exploited not only for writing low-loss birefringent devices [16] to create so-called Type-II optical waveguides [1,17–19], but also a positive index changes zone [1,5] usually by writing the optical cladding. For example, the “double line” technique especially offers a guiding zone that is located between two or more parallel lines for forming surrounding index depressed claddings based on type II modifications, and also preserved well-bulk material properties of the core like materials with specific luminescent properties, materials with metallic or oxide nanoparticles, and also laser-materials, as it was already reported within last year’s [1,17–19].

Despite Nd³⁺ doped yttrium aluminum garnet (Y₃Al₅O₁₂, YAG) was considered as a commercial laser material, it possesses low solubility of rare earth (RE) ions, low effective distribution coefficient of Nd³⁺ ions, and high phonon energy (approximately 941 cm^-1) [20]. However, CaF₂ is an optimal optical raw material, and has high RE solubility, low phonon energy (approximately 322 cm^-1), wide transparency range of 0.15 - 9.0 µm, high thermal conductivity, and ability to impede optical damage due to near-infrared laser pumping [19,20–27]. Recently, it has been found that an excellent laser action with outstanding slope efficiency can be achieved by using different bulk RE ions (RE = Er³⁺, Tm³⁺, and Yb³⁺) with single doped CaF₂ crystal under appropriate pumping conditions [22–24]. Whereas, no laser action was occurred in Nd³⁺ single doped CaF₂ crystals [21]. However, the co-doping of Y³⁺ into Nd³⁺: CaF₂ single crystals provided better bulk laser characteristics, as proven by Jiang et al. [21]. Recently, cladding type waveguides engineering have been demonstrated [27] in Nd³⁺ doped CaF₂ crystal using fs laser at 1kHz. In this paper, we decide to investigate the fs-laser permanent changes in Nd³⁺, Y³⁺ codoped CaF₂ crystals. We have thus studied the laser writing kinetics of quantitative refractive index (Δn) changes in and around the laser-written lines, and also the stress-induced birefringence according to the pulse energy, and repetition rate. To this end, we provided a spectroscopic data in the form of Raman spectral changes, and micro-photoluminescence properties at typical conditions to get some “double-line” waveguides.

2. Experimental details

The Nd³⁺ (0.5 mol. %), Y³⁺ (10 mol. %) codoped CaF₂ single crystal was grown by the Bridgman-Stockbarger method and was labelled as a Nd,Y:CaF₂ in this paper. The selected transparent Nd,Y:CaF₂ crystals were cutted into rectangular plates with dimension of 5 (x-axis) × 5 (y-axis) × 1 (z-axis) mm³, and then polished to reach an optical grade. For laser direct writing, a pulsed Yb³⁺ fiber laser with an operating wavelength of 1030 nm (Satsuma, Amplitude Systems Ltd.) is used. The delivering pulses of 250 fs were focused inside the CaF₂ plates through a 0.6 numerical aperture (NA) aspheric lens. Finally, these CaF₂ crystals were translated into XYZ motorized stage using a computer-controlled software with a constant writing speed of 0.2 mm s^-1. Here, the laser linear polarization $(\mathop {E)}\limits^ \to $ was perpendicular to the scanning direction $(\mathop v\limits^ \to )$. Many laser lines were thus written using various pulse energies in the range of 0.1-5.0 µJ at two different repetition rates of 10 kHz (no heat accumulation from pulse to pulse) and 500 kHz, where heat accumulation regime can occur to provide enough energy that has been deposited. For the following measurements, after laser material processing, end-facet of the written-lines was polished.

After fs-laser inscription, quantitative phase microscopy (QPM) at λ = 550 nm was used to determine the average refractive index changes $\overline {\Delta n} = \frac{\lambda }{{2\pi L}}\Delta \varphi$ ($\Delta \varphi$ is the phase shift (in radians) of the non-polarized light at 550 nm, and L is defined as the thickness of the laser track in the laser propagation direction) within the lines, including within the stress-affected zone surrounding the lines. Additional quantitative birefringence (B) measurements were carried out. Here, a polarized Olympus BX51 optical microscope was used to detect the neutral axis, for defining the birefringence slow axis orientation, and for estimating the linear birefringence using the Sénarmont method [6,28]. This method was based on a Sénarmont compensator (i.e., a quarter wave plate) integrated with a highly precise rotating analyzer to perform retardance R = B × t measurements (here, t is defined as the thickness of the birefringent zone). The precision of the retardance measurements typically tends to λ/100.

Then, based on the refractive index measurements, we wrote a few prototype waveguides using the double line technique for illustrating our purpose. The double laser tracks with separations of 5 µm, 8 µm, 11 µm, 15 µm, and 20 µm were inscribed 150 µm below the crystal surface for the two chosen pulse energies of 0.5 µJ, and 0.8 µJ in order to avoid too high damage within the irradiated area, which would lead to cracks.

From the spectroscopic side, the absorption spectrum of pristine crystal was recorded using a double-beam spectrophotometer (TU-1900, PG Instruments Co., Ltd.) with a typical spectral resolution of 1.0 nm in the spectral range of 300-900 nm. Micro-Raman spectra were performed using micro-Raman spectroscopy (Renishaw in Via Raman spectrometer) under 532 nm laser light excitation. The laser beam was focused on a selected irradiated line cross-section using a 50 × magnification microscopic objective (0.55 NA). The spatial resolution is defined by the waist size that is typically around 0.6 µm. In addition, micro-photoluminescence (µ-PL) was monitored by a confocal optical microscopy (Nanofinder FLEX2, Tokyo Instruments, Inc) under 532 nm laser used as an excitation source. The laser beam was focused on the Nd,Y:CaF₂ crystal waveguides core, and cladding through a 50 × microscopic objective, and then subsequent emitted light was collected, and transmitted to a spectrometer (MS 350i, SOLAR TII) equipped with Andor, IDUS DU420A-OE detector. The PL instrument was calibrated with a standard poly(3-octylthiophene-2,5-diyl material. All optical measurements were carried out at room temperature.

3. Results

Figure 1 portrayed the optical absorption spectrum of the pristine Nd,Y:CaF₂ crystal from 300 nm to 900 nm. In particular, optical absorption bands in the region of 750-900 nm were reproduced the trend of the dominant absorption band identified at approximately 796 nm, and a shoulder at around 802 nm mainly caused by doping of YF₃ ions into Nd³⁺:CaF₂ crystals [29]. Our experimental result is in good compatible with a similar work performed in the reported literature [29]. Thus, the prevailing 796 nm absorption band is likely originated from the new formation of Nd³⁺-Y³⁺ clusters [29], where the yttrium can play as a buffer ion. Note that, no significant absorption was reported at the laser pump wavelength (λ_p=1030 nm) for similar crystals in the literature [24,30], whereas they observed.

 

Fig. 1. Absorption spectrum of pristine Nd,Y:CaF₂ crystal in the spectral range of 300-900 nm.

Download Full Size | PDF

After laser writing, the quantitative phase microscopy (QPM) is directly employed to determine the mean value of the permanent refractive index changes in the plane perpendicular to the direction of the laser beam propagation [6,28,31]. Typically, a three dimensional QPM image, and the corresponding two-dimensional integrated profile image across the line are shown in Fig. 2(a). The inset of Fig. 2(a) presents a typical 3D-QPM image. This corresponds to phase changes for a laser line (e.g., a waveguide cladding) written in Nd,Y:CaF₂ crystal at low repetition rate of 10 kHz with a pulse energy of 2.0 µJ. The black line in the resultant QPM image represents negative refractive index changes, whereas the surrounding white line edge corresponds to the positive index changes [6,28,31]. The high negative phase shift (typ. - 1.7 π radians at 2 μJ) is likely due to a permanent volume expansion. The refractive index changes $\overline {\Delta n}$ can reach up to −1.5 × 10⁻² on the basis of the considered modified length (typ. around L = 30 µm). As a result, an elastic strain field develops in and out of the irradiated area leading to positive index changes that are estimated to be approximately a few + 10⁻³ (e.g., up to 6×10⁻³ at 2μJ) for laser tracks written at 10 kHz.

 

Fig. 2. Detailed analyses of the laser track (a) phase change profile, inset shows QPM image of the Nd,Y:CaF₂ crystal for 2.0 µJ at 10 kHz, and (b) phase change in radians according to the pulse energy for two different repetition rates (10 and 500 kHz).

Download Full Size | PDF

Corresponding quantified laser-induced phase shift (in radians) versus the pulse energies of a Nd,Y:CaF₂ crystal for different repetition rates are depicted in Fig. 2(b). As the pulse energy increases up to 2.0 µJ, the refractive index decreases exponentially at center of the tracks, and then remains rather invariant beyond 2.0 µJ (up to 5.0 µJ) for lines fabricated at a repetition rate of 10 kHz. On the contrary, for laser lines made at a repetition rate of 500 kHz, the refractive index decreases with the pulse energy until 0.8 µJ, and then suddenly increases for higher energy together with increasing line width. The first part of the curves is related to the increase of the permanent volume expansion, whereas the increase (> 0.8 µJ) corresponds to the transition towards the heat accumulation regime (in agreement with a typical heat diffusion time that is estimated to be around a 2 μs in CaF₂). Figure 2(b) reveals that the writing kinetics of refractive index changes are quite similar for both 10 kHz, and 500 kHz repetition rates at low energy. However, at saturation, the net negative phase change (i.e., the decrease of refractive index) is higher for low repetition rate (10 kHz) than for high repetition rate (500 kHz) waveguides.

Then, we studied the laser-induced birefringence B (B is defined as the difference between the ordinary refractive index (n_o), and an extra-ordinary refractive index (n_e)) by estimating the optical retardance as a function of the pulse energies (0.01 µJ - 5.0 µJ), and for two different repetition rates (10 and 500 kHz) (Fig. 3). The retardance (R) is defined as the product of the birefringence (B) and thickness (t) of the birefringent region. For the waveguides fabricated at a repetition rate of 10 kHz, there is no significant variations in optical retardance typically below 0.4-0.5 µJ. Then, the retardance increases monotonously up to 2.5 µJ, and remains constant until 5 µJ to reach around 15 nm that corresponds a birefringence of 5 × 10⁻⁴. At 500 kHz higher repetition rate waveguides, the optical retardance increases from 0.1 µJ up to a 1.0 µJ. Then, there is a dramatic decrement, and no optical retardance was present beyond a pulse energy of 2.0 µJ. The disappearance of the photo-induced birefringence can be attributed to the heat accumulation that reaches a temperature on the order of melting temperature and that temperature is maintained for a long duration (typ. at least few 100’s µs) to allow structural relaxation, and an apparent complete stress release. The maximum stress-induced optical retardance is around 45 nm for irradiated lines at 1 μJ and 500 kHz. The corresponding linear birefringence B is approximately 1.5 × 10⁻³ for the measured laser track length. To compute the temperatures of a Nd,Y:CaF₂ single crystal in the modified region at two different repetition rates for the experimental laser exposure conditions, a finite-difference thermal diffusion model can be employed [32]. The typical heat diffusion time $\tau$ is characterized by d²/D_th where d is the focal spot diameter and D_th the thermal diffusivity, which is around 3.57 × 10⁻⁶ m²s^-1 in CaF₂. So, we estimated τ to 2 μs for a 2.5 μm spot diameter, which is four times faster than in SiO₂ glass. At low repetition rate, the calculated temperature can be as high as the melting temperature for a short time (a few hundreds of nanoseconds) even for low pulse energy of 1 µJ. In contrast, in the heat accumulation regime (e.g., energy higher than 1 µJ at 500 kHz), temperature can be higher than the melting temperature for a few hundreds of microseconds.

 

Fig. 3. Optical retardance R (proportional to the linear birefringence) of Nd,Y:CaF₂ crystal as a function of pulse energies for two different repetition rates (10 and 500 kHz).

Download Full Size | PDF

Waveguides writing, and spectroscopic investigations: Then, based on the above refractive index change measurements, we wrote a few prototype waveguides using “double line” technique for illustrating our purpose before to perform some spectroscopic investigations. The double laser tracks were inscribed 150 µm below the crystal surface for two chosen pulse energies of 0.5 µJ, and 0.8 µJ. The double line distance was varied with a separation of 5 µm, 8 µm, 11 µm, 15 µm, and 20 µm. Figure 4 (left side) exemplifies a typical cross-sectional optical image of a Nd,Y:CaF₂ crystal waveguides end-facet for a pulse energy of 0.5 µJ. The represented circular broken line indicates the guiding zone between the written claddings. The width of the laser tracks increase from 3.0 to 5.0 µm, when the pulse energy is increasing from 0.5 to 0.8 µJ. The end-face coupling technique was conducted at visible wavelength of 632.8 nm (He-Ne laser) for TM polarization to probe the waveguide mode profiles written in a Nd,Y:CaF₂ crystal [18,19,31,33], as displayed in Fig. 4 for a pulse energy of 0.8 µJ at 10 kHz. Note that, for TE polarization, the producing waveguide modes are not circular, and also achieved higher propagation losses, which are not exploitable.

 

Fig. 4. Examples of waveguide end face cross-section images written using double line for a typical pulse energy of 0.5 µJ and a repetition rate of 10 kHz. Dotted line represents light guiding and mode profile pictures of Nd,Y:CaF₂ crystal with double line separations of (a) 5 µm, (b) 8 µm, (c) 12 µm, (d) 15 µm and (e) 20 µm for TM polarization with a pulse energy of 0.8 µJ at 10 kHz.

Download Full Size | PDF

Here, we only write a few prototype waveguides for illustrating our purpose. For high quality waveguides, the reader could refer Ref. [23] about cladding type waveguides engineering. The insertion losses (ILs) are computed using the following formula $ILs(dB) = 10 \times \log (\frac{{{P_{out}}}}{{{P_{in}}}})$, where P_out is the power of output beam coming out of the waveguide end facet and P_in is the power of the input beam passing into the waveguide [17,25]. The coupling losses (CLs) are roughly estimated using the formula $CLs = {\left( {\frac{{2{w_1}{w_2}}}{{w_1^2 + w_2^2}}} \right)^2}dB$ [30], taking into consideration the overlap of the incident light beam (w₁) and the waveguide mode (w₂) profile. Finally, the propagation losses (PLs) are computed from (IL-CL)/L (expressed in dB/cm), where L is the length of the waveguides (typ. a few cm). The estimated PLs decreased to 2.81 dBcm^-1 from 4.16 dBcm^-1 for TM polarization, when increasing double line separation from 5 µm up to 20 µm, as shown in Fig. 5. This might be due to smaller acceptance angle, and some higher order modes are cutoff for small diameter of waveguides in contrast to a higher separation [27]. In addition, it appears that higher pulse energy implies lower propagation loss, which could be attributed to the higher index contrast in agreement with QPM results in Fig. 2(b).

 

Fig. 5. Propagation loss (dB cm^-1) of prototypes waveguides inside Nd,Y:CaF₂ crystal as a function of double lines separation at two different pulse energies. Losses were measured for TM polarization.

Download Full Size | PDF

It is important to examine the complementary spectroscopic information related to the CaF₂ crystal structure. Thus, µ-Raman studies were conducted to probe the structural study for both non-modified (pristine), in the core and in the written cladding. Figure 6 presents the normalized µ-Raman spectra of Nd,Y:CaF₂ crystal for a pulse energy of 2.0 µJ. As presented in Fig. 6, a broad band was presented at approximately 287 cm^-1, which was attributed to the symmetric stretching of fluorine atoms in the vicinity of cations such as Ca [34]. Note that, the main Raman band position is shifted by 2 cm^-1 towards the higher wavenumbers in the cladding compared with an unmodified region. This “blue shift” and the increment in the Raman bandwidth is attributed to the lattice expansion, and an increase of the local disorder within the laser-written cladding of Nd,Y:CaF₂ crystal [35]. In the waveguide core region, we also observed a slight shift, which is attributed to the compressive stress field.

 

Fig. 6. Raman spectra of Nd,Y:CaF₂ crystal bulk and on the cladding (10 kHz).

Download Full Size | PDF

Following the same view, the photoluminescence (PL) studies were conducted, before and after fs laser irradiation, to investigate the influence of yttrium ions on the local environment of Nd³⁺ ions within the CaF₂ crystal. Figure 7 illustrates the PL spectra of Nd,Y:CaF₂ crystal irradiated with varying pulse energies (λ_ex=532 nm). A strong PL band located at about 895 nm associated with an electronic transition of Nd³⁺ (⁴F_3/2-⁴I_9/2) ions and two well-resolved peaks recognized in ∼1049, ∼1055 as well as a weak hump at approximately 1066 nm within the broadband of 1035-1085 nm spectral window [25,29,36]. The predominant emission band at approximately 867 nm is attributed to the R₁-Z₁ transition of the Nd³⁺ ions [36]. Moreover, ∼ 1049 nm, and ∼1066 nm bands are mainly originated from the Nd³⁺ ions in doped CaF₂ crystal. Whereas, the band at 1055 nm can be attributed to the addition of yttrium ions into the Nd³⁺:CaF₂ crystal [25]. The relative intensity of the band at about 1049 nm (site-II), 1066 nm (site-II) and 1055 nm (site-I) slightly enhanced after fs-laser irradiation at 10 kHz when compared to the pristine sample. Moreover, a marginal enhancement in the full-width at half-maximum is observed with increasing pulse energy. For high repetition rate waveguides (500 kHz), similar behavior of PL is notified, like for the low repetition rate irradiations, as shown in the inset of Fig. 7. This could be interpreted in turn an increasing multiple Nd-Y active centers, high density defect generation, change in the environment of Nd³⁺ ions, lattice defects, and disorder [35]. In addition, there is F^- + Y²⁺ defect centers formation as evidenced by the optical absorption in the reported literature [21]. For sake of comparison, we also add the PL spectra of the non-irradiated area and within the core waveguide region. No significant differences were observed, as shown in the inset of Fig. 7. Therefore, the Raman results demonstrated the fs-laser-irradiated Nd,Y:CaF₂ crystal retaining their crystalline nature, and PL features of Nd³⁺ ions, at all investigated pulse energies like reported for SrF₂ at 1μJ [35].

 

Fig. 7. PL spectra of Nd,Y:CaF₂ crystal waveguides as a function of pulse energies (λ_ex=532 nm) at 10 kHz with a power of 0.65 mW. The inset shows PL spectra of pristine, 10 kHz and 500 kHz at 0.3 µJ.

Download Full Size | PDF

4. Discussion

In the case of low repetition rate irradiations (e.g., 10 kHz in CaF₂), the deposited pulse energy can drive heat and developing more heat with increasing pulse energy that enables permanent changes within the laser focal spot i.e., it is local melting [1,37]. In the course of cooling down, the melted region recrystallizes again in a CaF₂ structure with rich lattice defects (additional disorder) and it causes an expansion of the lattice accompanied by a compressive stress, which led to the lower density of the CaF₂ crystal in the irradiated area [1]. The typical heat diffusion time (τ) has been estimated to be about 2 µs in CaF₂ material. Hence, this could lead to heat accumulation provide enough energy has been deposited, which is likely the case for energy higher than 1.0 μJ at 500kHz in agreement with results reports in Ref. [38]. This leads to a much slower re-solidification process on the millisecond (100’s μs up to ms) time scale leading to lower density changes, as we observed above 1.0 μJ for example.

It seems that the key process is a permanent volume change (i.e., specific volume increases and thus crystal density decreases) accompanied by an increase of the local disorder as indicated by the Raman studies. These permanent volume changes (i.e., permanent strain) led to the formation of an elastic strain field in and out of the laser-irradiated zone. This elastic response also gives rise to a significant stress-induced birefringence as we observed in our experiments. The measured negative refractive index changes within the irradiated lines are the result of superposition of a strong negative index changes related to the permanent expansion mainly, and a positive one associated to the elastic response of the irradiated zone. While the positive index changes around the laser tracks are solely due to the photo-elastic effect.

For a more quantitative approach, we could introduce a free of stress volume expansion $\varepsilon^p$, where p means permanent. As the surrounding matter resists, this expansion in the irradiated region should be compensated by a compressive stress, as it is described in Ref. [39]. This expansion of the irradiated region can damage the surrounding matter, releasing stress and fracturing the laser trace itself as it is observed in some of our experiments [39]. So for evaluating the related refractive index change, we must take into account the permanent change of refractive index and the contribution from stress. The first one is evaluated from Lorenz-Lorentz equation: $\Delta n_{ii}^p = - \frac{{({n^2} - 1)({n^2} + 2)}}{{6n}}(1 - \Omega )Tr({\varepsilon ^p})$, where Ω is the ratio between the relative changes in the molecular polarizability to the relative change in the specific volume. In Ref. [40], different values of Ω found for different compaction methods were compiled for comparison. Actually, the higher value of Ω, the lower index response for a given amount of compaction/expansion. For example, if we taken into account for Ω change with laser-induced compaction/expansion in fused silica [41], Ω=0.22, we can deduce the index changes and the stress-free contraction. Then by introducing an isotropic free of stress volume expansion ${\varepsilon ^p} = \left\{ {\begin{array}{{c}} {{\varepsilon _0}}\\ {{\varepsilon _0}}\\ {{\varepsilon _0}} \end{array}} \right.,\mbox{ }{\varepsilon _0}$ of 1.4×10⁻², we get a permanent index change contribution $\Delta n_{ii}^p$ around -1.6×10⁻². The contribution of elastic strain is then deduced from the photo-elasticity relationship:

Then, using p₁₁= 0.038, p₁₂= 0.226, and n = 1.435 for CaF₂, it yields a mean refractive index (Δn) change of −1.5×10⁻², and a linear birefringence of 4×10⁻⁴ that are in good agreement with our observations for irradiations performed at 10 kHz. We also deduce a positive index changes up to + 7×10⁻³ related to the stress-field-formation.

5. Conclusion

In this study, we investigated laser-induced refractive index changes kinetics in Nd³⁺, Y³⁺ co-doped CaF₂ crystal. Quantitative Phase measurements revealed a quite high refractive index contrast with a negative index changes down to −1.5×10⁻² that is accompanied that a positive index changes related to the formation of a stress field. The average index changes of the stress affected zone can reach up to + 7 × 10⁻³, a useful range to implement some optical waveguides. This also results a slight stress-induced birefringence of a few 10⁻⁴. Moreover, at 500 kHz, the stress-induced optical retardance reached 45 nm in the CaF₂ crystal i.e., a quarter wave plate in the VUV range. This can be further exploited to develop low loss VUV-UV-Vis waveplates based on femtosecond laser-induced stress birefringence engineering [12]. Indeed, the birefringent area is free of laser irradiation resulting in a high transparency, and the stress-induced birefringence exhibit a quite small spectral dependence allowing its exploitation at low wavelengths. From the spectroscopic point of view, we observed a slight blue-shift and an increase of the width of the main Raman band indicating perturbed fluorine environment, and the lattice expansion. The micro-photoluminescence study reveal only minor changes such as a slight increment in the NIR emission band and broadening with pulse energy by small local changes in the vicinity of the network. Thus, the commercial Nd³⁺, Y³⁺ codoped CaF₂ crystal appears as a promising alternative to existing materials for developing not only for compact waveguide lasers, but also for realizing UV-Vis birefringent devices, and for use in integrated photonics.