1. Introduction

Compact solid state laser sources emitting in the visible range are nowadays devices of special attention, which can be used in many different lightning applications. Red Green Blue sources (RGB) are necessary to cover videoprojection applications. Furthermore, orange laser sources emitting around 606 nm are required for quantum information processing applications [1]. This current context promotes investigations on rare earth (RE) doped fluoride materials, in particular Pr³⁺-doped fluorozirconate glasses. Praseodymium enables the emission in the visible and the fluoride matrix provides a window of transparency (0.2 – 7 µm [2]) perfectly suitable for the targeted applications and relatively low phonon energies preventing non radiative processes detrimental to the emission of this RE ion in the visible [3]. Therefore, it can easily be pumped directly in the blue range, leading to a population inversion (³H₄ → ³P₀) and an efficient emission in the blue, green, orange and red spectral range [4]. Laser emission around these wavelengths (488 nm, 521 nm, 635 nm), was recently demonstrated in a Zr–Ba–La–Al–Na (ZBLAN) fluoride fiber by Okamoto [5, 6], as well as in fluoride crystals [7–9].

By a chemical exchange process, channeled ZBLA waveguide can be fabricated offering the opportunity to develop integrated compact laser sources. Moreover, ZBLA glasses [10] which are among the most stable of heavy metal fluoride glasses can accept a high amount of rare earth without showing any crystallization phenomena [11]. In this paper, we report on the experimental demonstration of amplification at 639 nm on Pr³⁺-doped ZBLA channel waveguides. The first part deals with the synthesis of ZBLA substrate, fabrication of planar and channeled waveguide and the characterization of the waveguide. Spectroscopic calculations are detailed in a second part while the gain measurements and gain modeling are presented in the last part.

2. Experimental part

The first step of the channel waveguide fabrication is the glass synthesis used as substrates. The composition prepared is the following: 57 ZrF₄ – 34 BaF₂ – 4 AlF₃ – 4.5LaF₃ – 0.5 PrF₃. 7 g batches were prepared from reagent grade (>99.9%) fluorides compounds. All chemicals reagents were stored and handled in a dry box to prevent hydroxide and oxide contamination (H₂O ≤ 2 ppm, O₂ ≤ 0.5 ppm). Fluorides crystals are crushed and powders are mixed in covered carbon crucibles. Reactants are melted at 850°C for 1 hour in a flowing dry nitrogen atmosphere. The melts were then quenched in air during 20 seconds and annealed during 16 h at 290°C. Finally, the glass is slowly cooled to room temperature. The main physicochemical characteristics are sum up in Table 1 . After synthesis, samples were polished for optical measurement and photolithography process.

Table 1. Physicochemical and optical properties of ZBLA glasses

View Table

Thin film of SiO₂ (200 nm thick) was deposited by radio-frequency (RF) magnetron sputtering system (MRC 822 Sputtering System) on ZBLA substrates [13].The quality of the silicon dioxide thin film is characterized on a regular basis, by capacitance measurements of aluminum/RF sputtered SiO₂/Single crystalline silicon wafer MIS (Metal/Insulator/Semiconductor) structures. After deposition of the SiO₂ layer, the guides are made by techniques commonly used in microelectronics technology. The first step is to spin-coat a thin film of photoresist onto the wafer. We choose the UV210-0.6 product from Rohm and Haas Electronic Material Company which is a multipurpose DUV photoresist that can be used for contact holes or trench applications in the sub-half micrometer range. The photoresist coating parameters are: spin-speed 1500 rpm, acceleration 5000 rpm.s⁻¹, duration 30 s, leading to a 800 nm-thick film, followed by a soft-bake step at 140°C for 3 min. By means of DUV-lithography (30 mJ.cm⁻² at λ_DUV = 248 nm on a BA6 Süss Microtec Mask Aligner) the patterns defined on a quartz/chromium mask specially designed for this purpose, are transferred in the UV210 layer. A Post-Exposed-Bake for 1 min at 120°C is necessary to cross-link the polymer. Then the recommended developer, a TMAH-based product, Microposit MF CD-26, is used for 30s to obtain the photoresist patterns. No subsequent annealing is needed to finish the photolithography process. The SiO₂ layer is etched (Fig. 1 ) by the method of Reactive Ion Etching with a CF₄ plasma (Microsys 400 RIE from Roth&Rau AG), with pure CF₄ plasma at 50 W of RF power, 10 sccm of CF₄ flow rate and 1 mT of partial pressure leading to a highly anisotropic profile and negligible contamination of the ZBLA material.

 

Fig. 1 SEM images of RIE etched SiO₂ mask on ZBLA substrate.

Download Full Size | PDF

Waveguides are obtained by an ionic exchange (F^- →Cl^-) [14] performed on ZBLA substrates, leading to an refractive index increase of the exchanged layer located close to the sample surface. During ion exchange process, samples were set in an alumina tube in a furnace which is exposed to an Argon diluted HCl gas flow. This step can be followed by an annealing of the sample under Ar gas flow (2 l.h⁻¹). Five main parameters are adjustable: time of treatment (5h – 9h), temperature of treatment (230°C – 280°C),total gas flow (2l.h⁻¹ – 8l.h⁻¹), ratio between HCl gas flow and Ar gas flow (0.5 – 2), annealing time (4h – 24h) and annealing temperature (200°C – 280°C). Finally, a clad layer is created at the waveguide surface by reverse exchange (Cl^-→F^-) during 2h at the treatment temperature. Indeed, the chloride layer is highly hygroscopic and an exposure at ambient atmosphere leads to a degradation of the guiding structure which is avoid thanks to the superficial cladding obtained by a reverse exchange restoring a chemical barrier against the atmosphere.

Specific investigations on refractive index profile of the exchanged layer have been made, using M-lines measurements (Metricon Model 2010) at 403.8, 532, 633, 825, 1311 and 1551 nm. Absorption spectra of Pr³⁺:ZBLA glasses were measured using a spectrophotometer (Perkin Elmer Lambda 1050). Optical losses were measured by studying the scattered light from the surface of the waveguides [15]. The laser light was single-mode fiber-coupled into the waveguide. The intensity of the scattered light was recorded with a digital camera placed above the sample and the attenuation values were the average of several measurements.

To assess the laser potential of the Pr³⁺:ZBLA channel waveguide, specific gain measurements were performed using a pump-probe technique (Fig. 2 ). The set-up consisted of two laser sources: a GaN laser diode emitting at 444nm for the optical pumping of the channel waveguide and a homemade Pr³⁺:YLF laser. The beam of the GaN laser diode was separated in two parts; one for pumping the Pr³⁺:YLF laser and the other for the waveguide optical pumping. The homemade laser is composed of a Pr³⁺:YLF crystal within a plano-concave cavity. It provides a 10mW output power at 639 nm. The pump beam and the probe beam were combined together with a beam splitter and injected into the channel waveguide with a x5 microscope objective. The output beam was collected by a x5 microscope objective and focused onto the input slit of a monochromator which was set at 639 nm. The pump beam was modulated at 133 Hz while the probe beam was modulated at 1.26 kHz. Two lock-in amplifiers were used to measure the signal at each frequency. The lock-in amplifier set at the probe frequency gave the mean probe signal with and without the pump. The other lock-in set at the pump frequency gave the variation of the probe signal induced by the pump.

 

Fig. 2 Experimental set-up for gain measurements.

Download Full Size | PDF

3. Results and discussion

Exchanged layers between 1 µm and 10 µm can be obtained, depending on experimental parameters set, with a refractive index contrast (Δn) between the layer and the glass substrate varying from 0.02 to 0.09. Thus, opto-geometric properties of the layer can easily be modified by changing the exchange parameters (Fig. 3a ). Annealing appears to be a very important factor allowing us to play on an additional parameter to refine and fine-tune the optogeometric parameters required for the final optical component. An annealing of the sample under Ar flux (2l.h⁻¹) after the ion exchange increases the Cl^- diffusion, leading to an increase of the layer thickness and a decrease of the Δn.

 

Fig. 3 a) Variation of layer thickness and refractive index with temperature of exchanged ZBLA: Pr3 + planar waveguide, b) Refractive index profile of an exchanged ZBLA: Pr³⁺ planar waveguides.

Download Full Size | PDF

The results showed the presence of two distinct layers: an upper layer, with a step index profile versus depth, and a lower one, with a profile presenting a refractive index gradient (Fig. 3b). Thickness of exchanged layers and effective refractive index of the waveguide are dependent of the exchange parameters. Nevertheless, the index profile shape is always comparable: a very uniform layer with a high index at the surface, then an abrupt drop of the index value and finally a thicker gradually decreasing index layer.

To validate the results obtained on refractive index profile, the SIMS (Secondary Ion Mass Spectroscopy) was used to monitor the concentration profile of atoms of Fluor, Chlore and Hydrogen using a cesium source. Adsorption and absorption of water at the surface of the samples can cause a problem of ionic charges. To remedy this, the electron gun was used to neutralize them. For all samples, this disturbance due to water at the surface is visible up to 500 nm.

The pattern observed (Fig. 4a and 4b) is consistent with observations made by M-lines (Metricon) highlighting the presence of two distinct layers: a first plateau of constant [Cl^-] concentration relatively narrow followed by a broader second layer of Cl^- atoms decreasing gradually. Thickness measurements obtained by M-lines method (with λ_min = 633 nm) for these two samples are 2.55 µm 4.35 µm respectively. The influence of an annealing step (16h,

 

Fig. 4 a) Cl^- concentration versus depth for an exchange of 7h at 270°C. b) Cl^- concentration versus depth for an exchange of 7h at 270°C followed by an annealing.

Download Full Size | PDF

270°C) is clearly observed by comparing the Fig. 4a to Fig. 4b. The annealing process mainly changed the total profile of chlore concentration. The thickness of the first layer appears larger to promote the diffusion of chlorine ions inside the glass. This diffusion modifies the concentration profile of the second layer making a smoother profile.

Furthermore, spectroscopic measurement has been performed to characterize the bulk material and calculate absorption and emission cross-section. The absorption cross-section has been calculated between 420 and 520 nm (Fig. 5a ). The three characteristic bands of the ion Pr³⁺ are observed. The band at 479.9 nm is assigned to the transition from the ground state ³H₄ to the ³P₀ level with a peak cross-section of about 0.66 10⁻²⁰cm². The band at 467.8 nm with a peak cross-section of 0.68 10⁻²⁰cm² is assigned to the transition from the ground state to ³P₁ and ¹I₆ manifolds. The band at 443.7 nm is the most intense one and is assigned to the transition to the ³P₂ level with a cross section of 1.28 10⁻²⁰cm². This band can be easily optically pumped by laser diodes based on GaN blue. Excitation at 469 nm (σ_a = 0.59 10⁻²⁰cm²) is also possible using a frequency double Nd:YAG special cavity [9].

 

Fig. 5 a) Absorption cross section of bulk glass, b) Emission of a channel waveguide of ZBLA: Pr³⁺.

Download Full Size | PDF

Emission spectra of a channel waveguide (4µm x 4 µm x 1 cm) exchanged at 270°C using a total gas flow of 4 l.h⁻¹ and a HCl/Ar ratio gas flow equal to 1 during 7h, have been reported on Fig. 5b. The indicated thickness corresponds to the high refractive index plateau thickness. The excitation is done at 443 nm, pumping the ³P₂ level which relaxes trough the ³P₀ level. The green, orange and red band corresponds to the transitions from the ³P₀ and ³P₁, ¹I₆ levels through the lower levels. On the spectra, the band/shoulders (indicated by arrows in Fig. 5b) correspond to the transition from thermally populated level ³P₁ and ¹I₆. The blue band (not shown on the spectra) corresponds to the transition ³P₀ trough the ground state. The green emission trough the ³H₅ level is particular. Indeed, the ³P₀ → ³H₅ transition at 535 nm is weaker than the transition from ³P₁ and ¹I₆ to ³H₅ at 520 nm. The other strong visible emissions at 602 nm and 635 nm correspond to the transition to ³H₆ and ³F₂ levels, respectively. The profiles of the emission band keep the same shape in bulk and waveguides [10], but there is some difference in the intensity. For the 490 nm emission, we observed a reducing in the intensity, due to a classical reabsorption along the channel in comparison with bulk material. So, the ionic exchange with the Cl^-, used to create the index profile does not affect noticeably emission properties of the Pr³⁺: ZBLA. The main advantage of the absorption and emission profiles displayed in Fig. 5a and 5b is the broad width of absorption and emission bands, giving glassy samples a high tunability compared to crystals. Finally, gain measurements have been performed on 4μm wide and 4μm thick and 1 cm long waveguides. The measured gain versus launched pump power is reported in Fig. 6a , and shows a linear variation. A maximum signal amplification of 29% (1.1 dB.cm⁻¹) was measured on the Pr³⁺:ZBLA at 639 nm. It is worth noting that, at this wavelength, the emission cross-section value is three times lower than at the maximum (634.6 nm), so we could expect an amplification of 3.3 dB.cm⁻¹ at the maximum of the emission band. Furthermore, an on/off gain of 50% has been obtained as expected at 639 nm in a channel waveguide of 2.5 cm long and 4 µm large and deep. The amplification percentage value is obtained by doing the ratio of the variation of the 639 nm signal (with and without pump) by the signal itself, so it is not directly the net gain value. To have the net gain in the waveguide, one must take into account the losses. We simulated the amplification process in the channel waveguide.

 

Fig. 6 a) Signal amplification versus injected pump power at 639 nm in th chlorine anionic exchange waveguide of 1 cm long. b) Gain simulation including propagation losses versus waveguide length.

Download Full Size | PDF

The lifetimes, absorption section at 444 nm of the pump and emission cross-sections at 639 nm used in the calculations were respectively 37 µs, 1.27 10⁻²⁰ cm² and 2.01 10⁻²⁰ cm². These values were derived from the Füchtbauer-Ladenburg formula and a Judd Ofelt analysis which will be reported in a further paper. The simulation of the amplification in the 4 × 4 µm, 1 cm long channel waveguide is presented in the Fig. 6a. The input pump power was the only adjusted parameter to fit the experimental data which enabled us to estimate the coupling efficiency of the pump beam into the waveguide to about 25%. We also calculated the net gain (Fig. 6b) in the waveguide by taking into account propagation losses of 2 dB.cm⁻¹ measured on a channel waveguide by detecting the diffused light although this value could certainly be improved to 0.5dB.cm⁻¹ with optimized fabrication process. In the spectral range of amplification, no absorption band related to the presence of Praseodymium or impurity is present and no other additional optical losses are therefore to be considered. Comparison between gain values at 639 nm and at 634.6 nm are presented in Fig. 6b for different pump powers to estimate the potential of the waveguides. As expected, we observe that the length for a given maximum launched pump power can be optimized. Curve (1) shows the gain at 639 nm and corresponds exactly to the experimental situation depicted in Fig. 6a for 40 mW of launched pump power. We clearly see that the amplification cannot overcome the losses, so there is no net gain. By looking at the same case, but at the emission peak (634.6 nm), we have 30% net gain for a 1 cm long waveguide (curve (2)). On the curve (3) of the Fig. 6b, we observe that a high gain of 225% (5.1 dB) can be reached for 100 mW of injected pump power in a 16 mm long waveguide which is still a realistic length. So, by reshaping the pump beam size so as to couple more pump power into the waveguide and adjusting the waveguide length, laser threshold should be achieved in the near future.

4. Conclusion

We have reported signal amplification, in a praseodymium-doped ZBLA glass channel waveguide with dimension of about 4µm*4µm*1cm obtained by chlorine anionic exchange, with amplification up to 29% in the red region. According to simulation based on experimental data, we demonstrated that a high gain of 225% (5.1 dB) can be reached for 100 mW of injected pump power in a 16 mm long waveguide These results, along with simulations, show that laser threshold can be reached in ZBLA channel waveguides by increasing the launched pump power and the waveguide length. Works on the elaboration technique and photolithography process to obtain lower losses are also under progress to demonstrate the first integrated Pr: ZBLA visible waveguide laser.