Open Access
Add to BibSonomyAbstract
We report a study of ultrafast laser waveguide inscription in two magneto-optical glasses. Two types of femtosecond laser systems operating in kHz and MHz repetition rate regimes are used for waveguide fabrication. Single mode waveguides in the visible are obtained in both writing regimes and exhibit distinct optical properties depending on laser writing conditions and the nature of glasses. Photodarkening, produced as a fabrication byproduct and associated with waveguide propagation loss, is shown to be reversible via annealing. Photodarkening behaves differently in the magneto-optical glasses studied, most likely due to large differences in the concentration of lanthanum and/or gallium in the materials.
© 2013 Optical Society of America
1. Introduction
In 1996, it was demonstrated that focused IR femtosecond laser pulses can induce refractive index changes in glasses [1, 2]. Since then, a variety of optical devices, such as waveguides, gratings, filters, resonators and microfluidic devices, have been fabricated using the femtosecond laser direct write (FLDW) technique [3]. Depending on glass properties and parameters of laser inscription, mainly the repetition rate, pulse energy and writing speed of laser, waveguide fabrication using the FLDW technique occurs in athermal (heat diffusion) or thermal (heat accumulation) regimes [4]. Generally, waveguides fabricated using kHz repetition rate FLDW lasers are produced in the athermal regime, while thermal effect typically occurs in waveguide inscription on MHz repetition rate laser systems. The mechanisms of laser writing in these regimes are distinct, resulting in different types of modifications in glasses. For example, in the case of borosilicate glasses, waveguide writing in the athermal regime creates localized changes to the glass chemistry via breaking of R-O bonds (where R= Si, B etc) and formation of non-bridging oxygen centers, while in the thermal regime, the modification is due to expansion and contraction of bond lengths and changes in angle resulting in changes in density [5].
With the rapid development of the FLDW technique, more complex optical systems consisting of multiple 3D components will be integrated into small glass chips. Reflections from device junctions, which can affect system stability or cause signal degradation, will become a significant problem. However, no optical isolator, which is the key component to mitigate stray light in an optical system, has been developed in the FLDW platform so far. A natural approach to incorporate optical isolators for the integrated platform is to apply FLDW itself to magneto-optical glasses to construct one-way structures based on the Faraday effect.
In a previous study [6], waveguides exhibiting Faraday rotation were fabricated in a single magneto-optical glass using a 25 MHz 800 nm oscillator laser (firmly in the thermal regime). However, there has been no subsequent systematic study on the variation of waveguide properties in the thermal regime. There have been no investigations at all of FLDW waveguides in Faraday glasses fabricated in the athermal regime, let alone observations of Faraday effect. Based on the knowledge of different mechanisms behind kHz and MHz repetition rate regimes of laser writing, it is important to verify the possibility of FLDW waveguides in the kHz regime, and to identify and understand differences between waveguides created in the two regimes.
In this paper, we explore the potential for waveguide fabrication in both kHz and MHz femtosecond laser repetition rate regimes for two magneto-optical glasses. Usable guiding structures exhibiting different optical properties are obtained in the two repetition rate regimes. An associated photodarkening phenomenon, which results in high waveguide propagation loss and behaves differently in the two glasses, is examined comprehensively. The mechanism of photodarkening is analyzed in view of the glass chemical composition, and the absorption and Raman spectra of both unmodified and photodarkened glasses. Based on the study, we attempt to mitigate photodarkening via thermal annealing. The results have implications for the design of future magneto-optical glasses directed towards laser writing applications.
2. Glass properties
Materials containing Tb3+ ions frequently exhibit a strong Faraday effect [7, 8], characterized by the Verdet constant of the material. The two types of commercial terbium-doped borosilicate glasses investigated in this work, known as Faraday glasses TG20 [9] and MR3-2 [10], are commonly used as core components in commercial bulk optical isolators. Some essential properties of the magneto-optical glasses are displayed in Table 1. The refractive index of TG20 was measured on an Abbe refractometer. Since the Abbe refractometer has an upper limit of 1.71, the refractive index of MR3-2 at different wavelengths was calculated using the Hartmann dispersion formula based on information from the manufacturer’s datasheet. The Verdet constant of both glasses was measured on a bulk glass Faraday rotation setup, composed of a solenoid and two polarizers, with a red HeNe laser (λ =632.8 nm) as the light source. The concentration of Tb in glasses was measured by an electron microprobe (CAMECA SX-100). The high Verdet constants in Table 1 are the result of the high concentration of Tb in the glasses.
Table 1. Basic properties of magneto-optical glasses.
3. Waveguide preparation and analysis
Two FLDW setups were used for waveguide fabrication. For the athermal regime, a regeneratively amplified Ti:Sapphire laser system (λ =800 nm, repetition rate 1 kHz, pulse duration 120 fs) was used to fabricate waveguides into both types of glasses. Since the Faraday effect is stronger at shorter wavelengths (closer to the 485 nm terbium absorption band), we sought to create single mode waveguides at 632 nm. Thus a 40× (NA 0.6) objective was used for laser focusing, combined with a 520 μm slit to promote symmetric waveguides [11]. A circularly polarized laser beam was focused into the glasses at 170 μm depth, and the samples were perpendicularly translated through the laser focus at a speed of 1.5 mm/min. A range of pulse energies from 225 nJ to 325 nJ, in increments of 25 nJ, were tested. Six structures with 1, 2, 4, 8, 12, 16 overscans were written at each energy level.
For the thermal regime, waveguides were fabricated on a high-power oscillator laser system (λ =800 nm, repetition rate 5.1 MHz, pulse duration 50 fs, circular polarized). A stronger 100× (NA 1.40) oil immersion objective was used to compensate for the lower energy provided by the MHz system. Structures were also written into both glass samples 170 μm under the surface. Pulse energies from 30 nJ to 40 nJ, in increments of 2.5 nJ, were tested and the effect of writing at different speeds from 100 to 2500 mm/min was examined at each energy level.
After fabrication, a large number of tests were performed. The waveguides were examined under a transmission differential interference contrast (TDIC) microscope. Light was coupled into the waveguides on a precision 6-axis stage for loss measurement; their mode field diameters (MFD) were measured by imaging the waveguide output onto a camera interfaced to a laser beam analyser. The refractive index profiles of induced waveguides were measured on a refractive near-field profilometer (RINCK electronik). Based on experience with other glasses, we commonly find that the RINCK profilometer gives an accurate index profile of waveguide structures, but the absolute value of index contrast is slightly high. Therefore, we quantitatively estimated the waveguide index changes by solving for the modal field in a commercial beam propagation tool [12] using the measured index profile but a varying peak index contrast in order to match the measured MFD.
3.1. Properties of waveguides created in the kHz regime
In the kHz pulse rate writing regime, waveguides were obtained in both glasses. However, the properties of the waveguides, in both glasses under investigation, differed significantly as we now show.
3.1.1. TG20 waveguide properties
For the TG20 glass, single mode waveguides at 632 nm were obtained for 275 nJ, 300 nJ and 325 nJ pulse energies. However, waveguides created at 325 nJ pulse energy level with more than 4 overscans were multimode in the visible. There was no evidence for guiding structures created by pulse energies below 275 nJ. Compared to other borosilicate glasses, this window for obtaining waveguides in TG20 is narrow. Overall, as the writing laser energy increased, the diameter of the waveguides measured under the microscope increased from approximately d = 0.5 μm to 3 μm (Fig. 1(a)) while their cross-sections changed from quasi-circular to a triangular form (Fig. 2), a feature that is commonly observed in the kHz regime using the slit beam shaping method. In Fig. 2(b), the red spot in the middle of cross section indicates physical damage within the waveguide. For the single mode waveguides, the MFD decreased from w = 6 μm to 3 μm (Fig. 1(b)) even though the diameter slightly expanded. This phenomenon results from the increasing waveguide refractive index change as the writing laser energy is elevated. The refractive index distribution of the single mode waveguides all exhibited positive Gaussian profiles and both the measured data from the refractive index profilometer and simulation yielded relatively high peak index changes ranging from 6 × 10−3 to 1.3 × 10−2. The propagation loss (PL) of the single mode waveguides ranged from 3 to 7 dB/cm (Fig. 1(c)). We attribute this high loss to the physical damage and also to photodarkening, which is discussed below in detail. In summary, as the writing laser dose increased, the diameter and the refractive index change of the resultant waveguides increased, resulting in more compact modes (a lower MFD). The propagation loss of the waveguides also increased with elevated laser dose.
Fig. 1 Properties of waveguides in TG20 created in kHz regime. Zero values mean no guiding was observed. Waveguides created by 325 nJ laser with 8, 12, 16 exhibited multi-mode at 632 nm. a) waveguide diameter d; b) MFD w of single-mode waveguides; c) waveguide propagation losses.
Fig. 2 Bright field microscope images of cross-sections of waveguides in TG20 fabricated using 2 different pulse energies. a) for writing laser 300 nJ, 1 pass; b) for writing laser 325 nJ, 1 pass.
3.1.2. MR3-2 waveguide properties
In MR3-2, guidance was observed for writing laser pulse energies of 225 nJ and higher. The properties of the guiding structures, such as the waveguide diameters (see Fig. 3(a)) and the cross-section of the guiding structures, were similar to those in TG20. However, the MFDs of single mode waveguides in MR3-2 were larger, ranging from 4 μm to 14 μm (Fig. 3(b)). Even though guidance and near-field modes were observed for structures created by pulse energies of 275 nJ and smaller, under the TDIC microscope, the waveguides appeared as brownish lines with vague borders. As a result, the diameters could not be measured reliably. Under the same laser writing conditions, no brownish lines were observed in TG20. On the other hand, waveguides in MR3-2 created by a pulse energy above 275 nJ presented clear edges, but their MFDs could not be obtained due to strong light scattering. The propagation loss of these waveguides exceeded 9 dB/cm (Fig. 3(c)). The propagation loss of the waveguides in MR3-2 was higher than those waveguides in TG20 produced under the same writing conditions. Based on the investigation of photodarkening, described in detail below, photodarkening in MR3-2 is stronger than that in TG20. The considerable loss of waveguides partially results from physical damage, but is mainly due to photodarkening as we discuss in section 4.
Fig. 3 Properties of waveguides in MR3-2 created in kHz regime. Zero values indicate the edges were too faint to measure the structure diameter. a) waveguide diameter d; b) MFDs of single mode waveguide w; c) waveguide propagation losses.
3.2. Properties of structures created in MHz regime
In contrast to the kHz repetition rate regime, the structures created at the MHz pulse rate exhibited similar properties in both glasses. While only positive index changes were obtained on the kHz system, both positive and negative refractive index changes were created with the MHz system. Figure 4 shows a positive index change (top) and a structure with a central negative index change surrounding by a positive cladding (bottom) in TG20. Similar structures were also observed in MR3-2. Both types of structures offer the potential for yielding usable waveguides since one could either guide light directly in the positive index modifications, or use the negative index change to form a depressed cladding waveguide as has been shown in other glasses [13]. Among the positive index changes created on the MHz laser system, we obtained single mode waveguides at 632 nm in both glasses for 37.5 nJ pulse energy. The MFDs of these waveguides varied from 6 to 9 μm and their near field patterns were more elliptical. These large MFDs and elliptical modes may induce coupling issues and birefringence which would likely compete with Faraday rotation. As the diameter of these waveguides was only about 1.5 μm but their MFDs were larger, we deduced most of the transmitted light was guided in the unmodified area. Thus the propagation loss of these waveguides was lower than that of waveguides fabricated on the kHz laser system, ranging from 1 to 3 dB/cm. On the other hand, the bend loss of these positive single mode waveguide is likely to be poor.
Using larger pulse energies changed the sign of the refractive index change, creating modifications comprised of a negative index change core surrounded by a raised index cladding (lower structure in Fig. 4). This is related to the onset of cumulative heating, where the strong heat diffusion results in modifications with diameters that exceed, by several times, the focal spot size [14]. In our case, for pulse energies slightly above the threshold for cumulative heating modifications with 5 μm diameter as opposed to 1.5 μm were created. Therefore in order to induce the largest exclusively positive index change in both glasses, the pulse energy in the MHz repetition rate regime was kept as close to the cumulative heating threshold as possible but yet below.
No perceptible photodarkening was observed in either type of glass under the MHz writing conditions. This is likely to be a result of the difference in thermal conditions between the kHz and MHz writing processes.
3.3. Infra-red photodarkening
Attempting to avoid the physical damage within waveguides mentioned earlier, which is usually associated with high energy fabrication, we replaced the 40× objective on the kHz femtosecond laser system with a 20× (NA 0.45) objective combined with a 500 μm slit for focusing the writing laser beam. In this case, no guiding structure was induced in either Faraday glass though the writing laser pulse energy reached 3 μJ. However, undesirable photodarkening was observed under the microscope (Fig. 5), indicating that the energy of the writing laser was strongly absorbed by the glass samples before reaching the focus.
Fig. 5 Bright field microscope image of cross section (highest power end face) of IR photodarkening scan. Writing laser incident from left.
We examined the photodarkening produced by different writing laser pulse energies from 1 μJ to 3 μJ in increments of 0.5 μJ. The absorption spectra of the photodarkened and unmodified glasses were measured on a Cary UV-Vis-NIR spectrophotometer and are shown in Fig. 6. We see bands of increased absoprtion over the range 400–600 nm, decreasingly smoothly with increasing wavelength. Bands of increasing intensity centered at 420 nm have previously been observed in the absorption spectra of Tb-doped fibers [15] and crystals [16] after laser exposure. Therefore in the inset to Fig. 6 we characterise the relation between the absorption at 420 nm (in Fig. 6) and the writing conditions. The 420 nm absorption increases linearly with writing laser energy, indicating that the photodarkening was not saturated and would likely further increase with the writing laser energy.
Fig. 6 Absorption spectra of IR induced photodarkening in MR3-2. Inset: Energy dependence of IR photodarkening in both Faraday glasses examined.
The photodarkening in MR3-2 was 3 to 10 times stronger than that in TG20 at similar writing laser energies, which correlates with the higher measured propagation loss described in section 3.1.
4. Examination of kHz-induced photodarkening
As established in the previous section, photodarkening was observed within guiding structures fabricated on the kHz rate femtosecond laser system. This photodarkening increased glass absorption in the visible and leads to the rather high propagation loss of those waveguides. To understand the mechanism of photodarkening so as to ultimately avoid the high propagation loss of waveguides, we further applied absorption spectroscopy and Raman spectroscopy to the photodarkened materials.
The absorption spectra of both TG20 and MR3-2 have negligible absorption at the writing wavelength of 800 nm, but a previous study reported strong absorption of Tb3+-doped borosilicate glass at 267 nm and observed absorption at 800 nm for femtosecond laser writing that was attributed to a three-photon process [17]. Given the requirements in writing time and complexity, it is much more efficient to create a large area modification for absorption measurements using UV laser irradiation than femtosecond laser processing. Therefore, as another route to produce and characterise photodarkening, we exposed the glass samples to a UV laser. The Faraday glasses were exposed to a UV nanosecond laser that operated at 266 nm with a repetition rate of 15 kHz. In order to ensure uniform exposure conditions and produce large area modification, a spinning wedge was used to repetitively scan the beam in a circle at a frequency of 40 Hz and a diameter of 0.25 mm. These samples were then longitudinally translated through the beam at a speed of 1 mm/min (Fig. 7). Different laser powers and over writing passes were examined (Table 2). The range of processing conditions for TG20 was broader than that for MR3-2 due to its higher damage threshold. The darkened materials were further examined on the Cary spectrophotometer for absorption spectra and a RENISHAW Ramanscope with a 532 nm or a 442 nm excitation laser source.
Table 2. Processing conditions for the UV exposure study.
Photodarkening was observed in both Tb3+-doped borosilicate glasses after the UV exposure and exhibited similar properties to the IR photodarkening. Furthermore, the UV and IR-induced photodarkening had very similar absorption spectra, hence we considered the UV and IR-induced photodarkening to be comparable. In the interests of practicality, we performed most of our remaining characterisation on the UV induced photodarkening in order to investigate the photochromism.
4.1. Absorption spectra
Figure 8 shows absorption spectra for the UV-darkened glass: the degree of darkening and its associated absorption increased with writing laser power and number of over-writing passes. In fact, a large area of photodarkening in MR3-2 could be seen unaided whereas that in TG20 was subtle and could only be identified under a microscope. The absorption spectra of the glasses darkened by the IR and UV lasers had very similar forms (compare Fig. 6 and Fig. 8): the spectra show a rising absorption in the visible due to photodarkening, with a tail extending to the near-IR. The sensitivities, however, were different: the photodarkening created by UV exposure in MR3-2 was about 2 to 5 times stronger than that in TG20 as the writing laser power increased, while the contrast of the IR photodarkening in different glasses was greater (being about 3 to 10 times stronger in MR3-2 than in TG20), most likely due to the higher laser energy density and the three-photon mechanism.
Fig. 8 Absorption spectra of UV induced photodarkening in MR3-2. Inset: Absorption spectra corrected by subtraction of the spectrum of the unirradiated glass.
4.1.1. Interpretation
The increasing absorption in MR3-2 (Fig. 8 inset) corresponds to a broad band centered at approximately 370 nm (27,000 cm−1 in wavenumber) with a range extending to 650 nm. The increased absorption in TG20 has the same centre and range but is lower in intensity. These bands are consistent with the picture that photodarkening relates to charge transfer (CT) transitions of Tb4+ ions [18], which was also reported in other oxygen host materials [15, 19]. In these reports, the position of the Tb4+ CT bands shifted slightly in the range from 350 to 440 nm, and their bandwidth varied depending on the material structure and oxygen coordination. Atkins et al. [15] and Hosono et al. [19] attributed the photodarkening process to charge transfer between Tb3+ and Tb4+, followed by electron trapping at color centers. Previously, Shih et al. observed an increase in Tb4+ concentration and a small decrease in the Verdet constant after femtosecond laser (λ =800 nm) writing [6]. A reduction of the Verdet constant was also detected in a crystalline sample due to formation of Tb4+ after thermal annealing in an oxidizing atmosphere [20]. Based on the discussion above, we suggest the photodarkening generated by UV and IR radiation relies on an analogous mechanism and studies of the UV photodarkening can be applied to IR photodarkening. This would also suggest that the photo-darkening not only increases waveguide losses, but may weaken the Faraday effect due to Tb3+ reduction as well.
4.2. Raman spectroscopy
We applied Raman spectroscopy to both the untreated materials and photodarkened samples. Generally we expect Raman spectroscopy to provide information about changes in the covalent network of the material. Here the Raman spectra are largely dominated by Tb3+ fluoresence which obscures the vibrational response in most parts of the spectra, but this proves useful in its own right. Figure 9 shows the spectra for pumping at 532 nm; Fig. 10 displays the spectra for pumping at 442 nm. Due to Tb3+ fluorescence around 490 nm, 545 nm, 580 nm and 620 nm, many regions of the Raman spectra were obscured. The peaks corresponding to Tb3+ fluorescence are marked in these figures. Tb3+ is more easily excited by 442 nm radiation than by 532 nm, which explains the larger noise in Fig. 9. The sharp peaks around 2400 cm−1 and 500 cm−1 in Fig. 9 can be attributed to residual room light contamination from fluorescent lighting. The strong peaks at 2400 cm−1 correspond to 611 nm emission from Eu:Y2O3 fluorophor and the narrow peaks seen at 500 cm−1 correspond to 546 nm emission from mercury. In contrast, the Tb3+ emission peaks around 500 cm−1, 1800 cm−1 and 2700 cm−1 differ significantly in intensity, especially for the MR3-2 data in Fig. 9(b). For the 442 nm pumping in Fig. 10(a), the strong peaks around 2300 cm−1 correspond to Tb3+ emission at 490 nm. Again, the intensities of Tb3+ emission peaks differ significantly, while other peaks changed slightly. Here, the contrast in the Tb3+ emission peaks of the raw and photodarkened material is much more pronounced for MR3-2 (blue and orange curves) than for TG20 (red and black).
Fig. 10 Raman spectra of Faraday glasses pump at 442 nm. a) Experimental data; b) Normalized data (for the range unaffected by Tb3+ emission).
Overall, the intensity of fluorescence in the Raman spectra of photodarkening was consistently weaker than those of the unirradiated materials. This phenomenon can be explained by the formation of Tb4+, since Tb4+ does not fluoresce under these conditions [21]. In addition, this is consistent with the lower contrast in the TG20 photodarkening spectra as compared to the MR3-2, since the photodarkening in TG20 is comparatively smaller.
Only a small range (200–1,600 cm−1) of the 442 nm-pumped Raman spectra escaped from the overshadowing effect of Tb3+ emission. Outside this range, no Raman peak was observed though the spectrum range extends to 5,500 cm−1. In this range, the system background was subtracted and the noise was reduced by Fourier transform and a low-pass filter. The original Raman data is enhanced at low frequencies because of the Bose-Einstein thermal phonon population n(ν) = 1/[exp(hν/kT) − 1], where ν is the Raman shift, h and k are the Planck and Boltzmann constants, T is the temperature [22, 23]. This temperature factor was removed by dividing our experimental data by [n(ν) + 1]. The processed spectra were normalized to the strongest peak (Fig. 10(b)) and were fitted by pseudo-Voigt functions. The ratio between Raman peaks indicates the relative amount of certain kinds of glass structures.
In Fig. 10(b), the Raman spectra with different colours correspond to photodarkening regions created by different laser writing conditions including the untreated material. The spectra of the darkened areas are essentially unchanged from the spectra of the unmodified material. This suggests that the photodarkened process is an ionic transformation rather than a covalent conversion, since Raman spectroscopy is insensitive to charge transfer between ions. The position of the Raman peaks for the two glasses is roughly similar. The strongest peaks around 950 cm−1 were assigned to Q2 that represent a SiO4 tetrahedral with two nonbridging oxygen atoms [24, 25]. Peaks around 400 cm−1 and 500 cm−1 were assigned to the Si-O-Si linkage; their precise position relates to the average intertetrahedral Si-O-Si angle [24–26]. Peaks in the range of 600–800 cm−1 arise due to various metastructural units related to the BO4 tetrahedron along with the vibration of Si-O-Si bridges [23, 27]. Raman peaks around 1400 cm−1 are associated with short-range structures in borate groups [28–30]. The MR3-2 spectra show ratios of the intensities of other peaks relative to that of the strongest peak that are noticeably smaller than for TG20. Based on the assignment above, we note that the Si-O-Si linkage, the metastructure of the borate groups (such as B-O− and BO2O−) and the vibration between Si-O-Si and borate groups were weaker in MR3-2. As shown in the result of mass spectrometry (Fig. 14) in section 6, the concentration of B and Si is similar in both glasses. We conclude that the smaller ratio between the peaks in the MR3-2 Raman spectra does not result from different concentrations of B and Si, but from distinct structures in the two glasses.
5. Thermal annealing
We attempted to reduce the high loss resulting from photodarkening via thermal annealing, which has recently been found to markedly improve FLDW waveguides in other glasses [31]. Photodarkened test samples and glasses containing waveguides were placed into a program-controllable oven for thermal annealing at 100 °C to 300 °C in increments of 50 °C. The absorption spectra of the thermal-annealed samples (not shown) were analogous to those of the photodarkened glass, but the broad Tb4+ charge transfer bands had reduced in strength. The photodarkening in TG20 vanished completely after 200 °C annealing, while the stronger photodarkening in MR3-2 practically disappeared after 300 °C annealing (Fig. 11(a)). At the same time, the propagation waveguide losses and the absorption of photodarkening in both types of glasses decreased rapidly between 100 °C to 200 °C, before flattening out (Fig. 11(b)). The waveguide losses in the annealed MR3-2 only fell to a level comparable to that of TG20 before thermal annealing. We infer that Tb4+ ions, which affect waveguide transmission and Faraday rotation, are reduced back to Tb3+ during annealing.
Fig. 11 Photodarkening reduction after thermal annealing. a) absorption (at 420 nm) of photodarkening; b) waveguide propagation loss (waveguide created by 300 nJ pulse energy with 1 over-writing pass).
The waveguides in TG20 were relatively robust to thermal annealing. The MFDs of these waveguides did not change until a temperature of 250 °C was reached. After that, some multi-mode waveguides became single mode waveguides, and the near-field mode pattern of waveguides written at 275 nJ pulse energy spread out (see Fig. 12(a)), which is to be compared to Fig. 1(b)). This phenomenon implies that the refractive index change had decreased after thermal annealing.
Unlike the robust waveguides in TG20, even though guiding structures can be created in MR3-2 by lower energy laser pulses (below 300 nJ), the near-field patterns of these waveguides spread out quickly once the temperature exceeded 100 °C and vanished fully after heating to 200 °C. This change is consistent with the trend that the photodarkening started to erase when the temperature reached 100 °C and disappeared after annealing at 200 °C (Fig. 12(a)). Since the waveguide propagation loss decreased as the photodarkening was annealed, the near-field patterns of waveguides in MR3-2 created at 300 nJ and 325 nJ energy levels appeared gradually (Fig. 12(b)) on the near-field camera. Only a small increase was observed for the MFDs of these waveguides, even as the temperature reached 300 °C.
In summary, guiding structures can be fabricated with lower energy laser pulses in MR3-2 but not in TG20, however, these structures in MR3-2 created by lower energies disintegrate with the photodarkened regions disappearing during thermal annealing. The MFD values of waveguides created by higher energy laser pulses did not change significantly during heating. We deduce that the photodarkening partially contributes to the refractive index modification. As the energy of the writing laser increases, structural changes, such as the change in molar refractivity [23], become the main component of the guiding structures.
In general, the increasing absorption resulting from photodarkening and the related propagation loss are stronger in MR3-2. After thermal annealing, the propagation loss of waveguides in MR3-2 only approached a similar level to that of TG20 (Fig. 13). Furthermore, waveguides in TG20 have more stable properties compared to those in MR3-2. Thus we identify TG20 as the more favourable candidate for FLDW devices.
6. Mechanisms for photodarkening in different glasses
We have clearly established that both UV and IR exposure can produce photodarkening in Tb-doped borosilicate glasses. The former is attributed to a single photon absorption process whereas the IR exposure is considered to induce a three-photon absorption process in magneto-optical glasses. In both cases the photodarkening process involves the photoionization of Tb3+ into Tb4+, associated with electron capture at color centers. This is supported firstly by the Raman spectroscopy results. Although these measurements were challenging due to the overshadowing influence of Tb3+ fluoresence, for the 442 nm pumped case there was a small region between 200 to 1600 cm−1 (see Fig. 10) that can be used for a comparative analysis of the two glasses in both their pristine and laser modified states. This region exhibited spectral features assigned as Si-O-Si linkages, BO4-Si-O-Si linkages, Q2 silica tetrahedral and borate groups. When comparing the modified, photo-darkened regions of TG20 with the unmodified regions we note that the Raman spectra have not undergone any significant change. The same is true for the MR3-2 glass. These observations are consistent with the inferences above that photodarkening is due to photoionization of Tb3+ and are not associated with a variation to the covalent bond structure.
A comparison of the Raman spectra for the TG20 and MR3-2 glasses reveal both similarities and marked differences in the covalent bond structure. The Si-O-Si, BO4-Si-O-Si and borate groups are more intense for the TG20 glass. In addition, our experimental observations have shown that femtosecond laser induced photodarkening occurs more readily in the MR3-2 glass than it does in the TG20 glass. To correlate the effect of glass chemical composition on photodarkening in the two types of magneto-optical glasses, we characterized the glasses by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). Electron microprobe analysis was used to obtain absolute fractions for a large number of species.
Figure 14 shows part of the LA-ICP-MS results: the first 9 items are elements present in high concentration in glasses; the last two items, Ce and Y, have been reported to improve photodarkening resistivity in rare-earth doped materials [32–34]. The results of LA-ICP-MS indicate that both glasses under study contain large amounts of Tb, Si, B and their concentration is similar in both glasses. We assume the differences in Tb, Si and B concentration are too small to either explain the stronger photodarkening in MR3-2 or the differences in the covalent bond structure between the two glasses. Al is also present in quite high concentration, and has previously been associated with a reduction in photodarkening in Yb-doped materials [35, 36], but the similarity in concentration again suggests it is not the primary cause here.
On the other hand, there is a distinct difference in the concentrations of La and Ga, both of which are more prevalent in TG20 by several orders of magnitude (Fig. 14). These elements may influence the packing density and geometry of the glass lattice, resulting in variations of the Raman signature. We have not found prior reports of La or Ga influencing photodarkening in glasses. However, the possibility is supported by work involving other rare-earth dopants. Dachraoui et al. [16] reported suppression of photodarkening by Ce dopants in (crystalline) terbium gallium garnet. They proposed a mechanism by which charge transfer reduces terbium to the Tb3+ while oxidising Ce. However, they also report that excess Ce, which is itself a rare-earth element, can induce photodarkening again via oxidation of Ce3+ into Ce4+. Other comparable investigations in fiber reported improvements in photodarkening resistivity in Yb-rich materials by codoping with trivalent ions, such as Ce3+ and Y3+ [32–34]. These trivalent ions enlarged the distance between Yb3+ ions in glass matrix reducing the degree photodarkening associated with Yb clustering. Again, the precise ratio between Yb and the codopant was important in these studies. In our case, the MR3-2 glass, which has higher concentrations of Ce and Y, was more prone to photodarkening. In any case, the concentrations of Ce and Y are low in both glasses (Fig. 14), making these elements unlikely to be related to the photodarkening difference.
Based on these observations however, we suspect that La3+ and Ga3+, which are 2,200 times and 90 times greater in TG20 respectively, may play a similar role to Ce3+ and Y3+ in these other materials. A higher concentration of Ga in TG20 gives a closer Tb-Ga spacing then a higher possibility of trapping electrons or holes before the formation of permanent color centers. La and Tb are both lanthanides and their trivalent ions have similar radius (106.1 for La3+, 92.3 for Tb3+), thus La3+ ions can easily substitute Tb3+, increasing the distance between Tb ions and suppressing photodarkening in TG20 from Tb clustering. Ga has multivalent states that exist as Ga+, Ga2+ and Ga3+ ions in glasses, and acts as an ideal candidate for charge compensation. Clearly, this is a complex issue and additional investigation, beyond the scope of this study, is necessary to resolve it.
7. Conclusions
Waveguide fabrication using femtosecond laser inscription methods occurs via different mechanisms depending on the repetition rate of the laser system. Using a MHz repetition rate laser system, we obtained positive and negative index changes in the magneto-optical glasses under investigation. Both types of structures might permit functioning isolators. No perceptible photodarkening was created by the MHz laser writing, but coupling issues and birefringence need to be considered. Single mode waveguides produced by the kHz pulse rate laser system have more compact and isotropic modes, but the competition between refractive index change and photodarkening is considerable. Photodarkening in MR3-2 is stronger than that in TG20 due to different chemical composition of glasses. La and Ga, which are more prevalent in TG20, likely play a major role in suppressing Tb-photodarkening. Photodarkening was also shown to be reversible via annealing in both magneto-optical glasses. The results may act as a guide for the design of future generations of magneto-optical glasses that might show superior performance under laser writing conditions. Following this study we can proceed to the development of complete isolators.
Acknowledgments
This research was supported by the Australian Research Council Centre of Excellence for Ultrahigh bandwidth Devices for Optical Systems (project number CE110001018) and performed in part at the OptoFab node of the Australian National Fabrication Facility.
References and links
1. K. Davis, K. Miura, N. Sugimoto, and K. Hirao, “Writing waveguides in glass with a femtosecond laser,” Opt. Lett. 21, 1729–1731 (1996). [CrossRef] [PubMed]
2. E. Glezer, M. Milosavljevic, L. Huang, R. Finlay, T. Her, J. Callan, and E. Mazur, “Three-dimensional optical storage inside transparent materials,” Opt. Lett. 21, 2023–2025 (1996). [CrossRef] [PubMed]
3. R. Gattass and E. Mazur, “Femtosecond laser micromachining in transparent materials,” Nature Photon. 2, 219–225 (2008). [CrossRef]
4. S. M. Eaton, H. Zhang, M. L. Ng, J. Li, W. Chen, S. Ho, and P. R. Herman, “Transition from thermal diffusion to heat accumulation in high repetition rate femtosecond laser writing of buried optical waveguides,” Opt. Express 16, 9443–9458 (2008). [CrossRef] [PubMed]
5. D. J. Little, M. Ams, S. Gross, P. Dekker, C. T. Miese, A. Fuerbach, and M. J. Withford, “Structural changes in BK7 glass upon exposure to femtosecond laser pulses,” J. Raman Spectrosc. 42, 715–718 (2011). [CrossRef]
6. T. Shih, R. Gattass, C. Mendonca, and E. Mazur, “Faraday rotation in femtosecond laser micromachined waveguides,” Opt. Express 15, 5809–5814 (2007). [CrossRef] [PubMed]
7. J. Qiu, K. Tanaka, N. Sugimoto, and K. Hirao, “Faraday effect in Tb3+-containing borate, fluoride and fluorophosphate glasses,” J. Non-Cryst. Solids 213, 193–198 (1997). [CrossRef]
8. A. Villaverde, D. Donatti, and D. Gozinis, “Terbium gallium garnet Verdet constant measurements with pulsed magnetic field,” J. Phys. C: Solid State Phys. 11, L495–L498 (1978). [CrossRef]
9. http://www.laserglass.com.cn/english-page/index/product/Faraday.htm.
10. http://www.xaot.com/sdp/173803/4/cp-4304059/0/Magneto-Optical_Glass.html.
11. M. Ams, G. Marshall, D. Spence, and M. Withford, “Slit beam shaping method for femtosecond laser direct-write fabrication of symmetric waveguides in bulk glasses.” Opt. Express 13, 5676–5681 (2005). [CrossRef] [PubMed]
12. “BeamPROP by RSoft Design Group, Inc.” http://www.rsoftdesign.com.
13. D. G. Lancaster, S. Gross, H. Ebendorff-Heidepriem, A. Fuerbach, M. J. Withford, and T. M. Monro, “2.1 μm waveguide laser fabricated by femtosecond laser direct-writing in Ho3+, Tm3+:ZBLAN glass,” Opt. Lett. 37, 996–998 (2012). [CrossRef] [PubMed]
14. C. Schaffer, J. García, and E. Mazur, “Bulk heating of transparent materials using a high-repetition-rate femtosecond laser,” Appl. Phys. A Mater. Sci. Process. 76, 351–354 (2003). [CrossRef]
15. G. R. Atkins and A. L. G. Carter, “Photodarkening in Tb3+-doped phosphosilicate and germanosilicate optical fibers,” Opt. Lett. 19, 874–876 (1994). [CrossRef] [PubMed]
16. H. Dachraoui, R. Rupp, K. Lengyel, M. Ellabban, M. Fally, G. Corradi, L. Kovacs, and L. Ackermann, “Photochromism of doped terbium gallium garnet,” Phys. Rev. B: Condens. Matter 74, 144104 (2006). [CrossRef]
17. S. Zhang, B. Zhu, S. Zhou, S. Xu, and J. Qiu, “Multi-photon absorption upconversion luminescence of a Tb3+-doped glass excited by an infrared femtosecond laser,” Opt. Express 15, 6883–6888 (2007). [CrossRef] [PubMed]
18. H. Ebendorff-Heidepriem and D. Ehrt, “Formation and UV absorption of cerium, europium and terbium ions in different valencies in glasses,” Opt. Mater. 15, 7–25 (2000). [CrossRef]
19. H. Hosono, T. Kinoshita, H. Kawazoe, M. Yamazaki, Y. Yamamoto, and N. Sawanobori, “Long lasting phosphorescence properties of Tb3+-activated reduced calcium aluminate glasses,” J. Phys.: Condens. Matter 10, 9541–9547 (1998). [CrossRef]
20. Y. Kagamitani, D. A. Pawlak, H. Sato, A. Yoshikawa, H. Machida, and T. Fukuda, “Annealing Effect in Terbium-Scandium-Aluminum Garnet Single Crystal,” Jpn. J. Appl. Phys. 41, 6020–6022 (2002). [CrossRef]
21. R. Verma, K. Kumar, and S. Rai, “Inter-conversion of Tb3+and Tb4+states and its fluorescence properties in MO-Al2O3: Tb (M = Mg, Ca, Sr, Ba) phosphor materials,” Solid State Sci. 12, 1146–1151 (2010). [CrossRef]
22. R. M. Almeida, “Vibrational Spectroscopy of Glasses,” J. Non-Cryst. Solids 106, 347–358 (1988). [CrossRef]
23. D. J. Little, M. Ams, P. Dekker, G. D. Marshall, and M. J. Withford, “Mechanism of femtosecond-laser induced refractive index change in phosphate glass under a low repetition-rate regime,” J. Appl. Phys. 108, 033110 (2010). [CrossRef]
24. P. Colomban, A. Tournie, and L. Bellot-Gurlet, “Raman identification of glassy silicates used in ceramics, glass and jewellery: a tentative differentiation guide,” J. Raman Spectrosc. 37, 841–852 (2006). [CrossRef]
25. P. McMillan, “Structural studies of silicate glasses and melts – applications and limitations of Raman spectroscopy,” Am. Mineral. 69, 622–644 (1984).
26. D. Manara, A. Grandjean, and D. Neuville, “Structure of borosilicate glasses and melts: A revision of the Yun, Bray and Dell model,” J. Non-Cryst. Solids 355, 2528–2531 (2009). [CrossRef]
27. T. Yano, N. Kunimine, S. Shibata, and M. Yamane, “Structural investigation of sodium borate glasses and melts by Raman spectroscopy. III. Relation between the rearrangement of super-structures and the properties of glass,” J. Non-Cryst. Solids 321, 157–168 (2003). [CrossRef]
28. D. Maniu, T. Iliescu, I. Ardelean, S. Cinta-Pinzaru, N. Tarcea, and W. Kiefer, “Raman study on B2O3-CaO glasses,” J. Mol. Struct. 651, 485–488 (2003). [CrossRef]
29. V. N. Bykov, T. N. Ivanova, and O. N. Koroleva, “Raman spectroscopy of borosilicate and germanate-silicate glasses and melts,” Russ. Metall. (Metally) 2011, 719–722 (2011). [CrossRef]
30. T. Yano, N. Kunimine, S. Shibata, and M. Yamane, “Structural investigation of sodium borate glasses and melts by Raman spectroscopy. I. Quantitative evaluation of structural units,” J. Non-Cryst. Solids 321, 137–146 (2003). [CrossRef]
31. A. Arriola, S. Gross, N. Jovanovic, N. Charles, P. Tuthill, S. Olaizola, A. Fuerbach, and M. Withford, “Low bend loss waveguides enable compact, efficient 3D photonic chips,” Opt. Express 21, 2978–2986 (2013). [CrossRef] [PubMed]
32. M. Engholm, P. Jelger, F. Laurell, and L. Norin, “Improved photodarkening resistivity in ytterbium-doped fiber lasers by cerium codoping,” Opt. Lett. 34, 1285–1287 (2009). [CrossRef] [PubMed]
33. M. C. Paul, A. V. Kir, Y. O. Barmenkov, S. Das, M. Pal, S. K. Bhadra, S. Yoo, A. J. Boyland, and J. K. Sahu, “Yb2O3 Doped Yttrium-Alumino-Silicate Nano-Particles Based LMA Optical Fibers for High-Power Fiber Lasers,” J. Lightwave Technol. 30, 2062–2068 (2012). [CrossRef]
34. S. Yoo, M. Kalita, A. Boyland, A. Webb, R. Standish, J. K. Sahu, M. C. Paul, S. Das, S. K. Bhadra, and M. Pal, “Ytterbium-doped Y2O3 nanoparticle silica optical fibers for high power fiber lasers with suppressed photodarkening,” Opt. Commun. 283, 3423–3427 (2010). [CrossRef]
35. M. Engholm and L. Norin, “Reduction of photodarkening in Yb/Al-doped fiber lasers,” Proc. of SPIE 6873, 68731E (2008). [CrossRef]
36. T. Deschamps, N. Ollier, H. Vezin, and C. Gonnet, “Clusters dissolution of Yb3+in codoped SiO2–Al2O3–P2O5 glass fiber and its relevance to photodarkening.” J. Chem. Phys. 136, 014503 (2012). [CrossRef]
