Recent highly sensitive absorption measurements of lithium niobate (LiNbO3) show that within the spectral range between 500 and 2900 nm the transparency is limited by impurities such as iron or hydrogen. In order to reduce the residual absorption, 5-mol.-%-MgO-doped and undoped congruent as well as undoped stoichiometric LiNbO3 crystals are annealed in dry oxygen atmosphere. The extinction coefficient of the treated crystals is measured using whispering-gallery-resonator-based absorption spectroscopy. The conducted measurements show that the treatment of stoichiometric crystals leads to scattering centers. For the congruent material residual metallic ions like iron and copper dominate the absorption in the spectral region from 400–2000 nm, and oxidization only shifts the center of the absorption in our case from that of iron to that of copper, thus inhibiting reaching the theoretical loss limit. Nevertheless, starting from 2000 nm, where absorption caused by hydrogen dominates, annealing leads to a significant drop in absorption, narrowing here the gap to the theoretical loss limit.
© 2015 Optical Society of America
The fundamental limit of transparency in optical materials is usually given by their band edge towards the UV, phonon scattering towards the IR, and Rayleigh scattering in between . Additionally, there are also residual impurities present which originate from the manufacturing process of the optical material. In silica, e.g., OH− represents such an impurity, separating the telecommunication windows in fibers.
Lithium niobate (LiNbO3) is one of the driving optical materials. It provides large nonlinear-optical coefficients and it can be periodically poled allowing for quasi-phase-matching (QPM) . Furthermore, it’s wide transmission range from the UV to the MIR makes LiNbO3 interesting for many applications. Consequently, LiNbO3 is utilized, e.g., for optical parametric oscillation (OPO)  and second harmonic generation (SHG) . It is used as well for optical modulators and filters . But also in LiNbO3 reaching of the fundamental transparency limit is hindered by residual impurities, such as Fe2+/3+, Cu+/2+, Cr3+/4+, or OH− .
These impurities lead to absorption, causing heating and also the photorefractive effect, limiting the light powers that can be handled with LiNbO3. Several efforts, such as heat treatments in different atmospheres, have been made to diminish the absorption in LiNbO3 and to overcome the photorefractive effect [7–10]. So far the effects of such treatments on bulk material have been studied only at the edges of the window of low absorption in nominally undoped LiNbO3 or for the entire spectral range in intentionally doped crystals. The effect of heat treatments on LiNbO3 in its window of lowest absorption (700 – 2900 nm) has not been determined yet. This is due to the lack of highly sensitive absorption measurement techniques.
Here we present a study of the effect of heat treatments in a dry oxygen atmosphere on 5-mol.-%-MgO-doped and undoped, congruently grown LiNbO3 as well as on undoped, stoichiometric LiNbO3 crystals by measuring the absorption coefficient of the treated samples within the spectral window of lowest absorption by using the whispering-gallery-resonator-based absorption spectroscopy .
1.1. Sample preparation
The first step is the sample preparation. A commercially available LiNbO3 waver (z-cut, diameter: 3”, thickness: 0.5 mm) of each material investigated is cut into pieces with the dimensions of 10 × 9 × 0.5 mm3 (x, y, z). For each annealing recipe (see section 1.2) from about the center of the respective wafers one piece of LiNbO3 is taken, as well as one piece which not is treated, serving as a reference. The samples are annealed, and a whispering gallery resonator (WGR) with 0.5 mm height (z) and a diameter of 2 mm (x, y) is manufactured out of each sample (see Fig. 1, left). Then, the resonators are polished first with a slurry of diamond particles of diameters down to 1 μm and afterwards with isotropic wet etching. By these means, a sufficiently low surface roughness is obtained to neglect scattering losses at the surface of the resonator in order to get access to the extinction coefficient of the bulk material .
1.2. Annealing recipes
Thermal treatments are performed by placing the samples into a crucible made of platinum, and the actual temperature is measured by a sensor placed in the crucible. With the goal to oxidize the residual Fe2+ impurities into the Fe3+ valence state, the samples are annealed together in a dry, nominally pure oxygen atmosphere, as proposed by Falk and Buse , at a temperature of 1060 °C. This high temperature is also proposed for driving hydrogen out of the samples. The heating ramp is chosen to be 7 °C/min (see Fig. 1, right). Once 1060 °C are reached, the temperature is kept at this value for 13 hours. To avoid reducing effects of Fe3+ to Fe2+ during the cooling process , the cooling ramp is chosen quite high: a moderate cooling ramp of −6 °C/min (recipe Tslow) and a fast ramp of −44 °C/min (recipe Tfast) are applied. To achieve the moderate cooling ramp, the oven is turned off, and after reaching a temperature of 850 °C, the oven is opened slowly. The faster cooling is obtained by opening the oven at 1060 °C and removing the crucible as quickly as possible. At 500 °C the samples are removed from the crucible.
1.3. Absorption spectroscopy
The samples are characterized regarding their extinction coefficient κ at different wavelengths employing the whispering-gallery-based extinction spectroscopy . In order to quantify the change of the OH− concentration in the samples, κ is measured in the peaks of the OH− absorption bands , i.e. at 2257 nm, 2311 nm, 2361 nm, 2484 nm and 2870 nm for undoped, 2830 nm for doped and 2885 nm for stoichiometric LiNbO3. In order to check for changes of the absorption caused by iron, κ is measured also from 500–1250 nm.
To cover this broad wavelength range, a tunable Ti:sapphire laser without and with frequency doubling as well as an optical parametrical oscillator (OPO) pumped by the Ti:sapphire laser are used. Thus, the wavelength regions from 700 to 1000 nm (Ti:sapphire laser), around 500 nm (frequency-doubled Ti:sapphire laser), from 1000 to 1500 nm for the signal wave, and from 1700 to 2600 nm for the idler wave of the OPO are available. Furthermore, the strong OH− peak beyond 2800 nm is detected with a grating spectrometer.
Except for near-stoichiometric crystals, where bulk scattering centers are observed after the annealing (see section 2), the extinction values are supposed to be identical with the absorption coefficients α.
Figure 2 shows the absorption coefficient α for the samples treated with different recipes (Tslow and Tfast) for undoped (a) and 5-mol.-%-MgO-doped (b) congruent LiNbO3 as well as the absorption coefficient for the untreated samples (Tref). The theoretical lower limit for the absorption coefficient in the highly transparent spectral range of LiNbO3 is shown as well . Depicted are the results for ordinarily polarized light. The results for extraordinarily polarized light are similar.
At all measured wavelengths below 2000 nm, the absorption coefficient increases, which is caused by a new absorption band which is centered around 1040 nm, extending into the region of lowest absorption around 1800 nm. Hereby recipe Tslow leads for the doped material to a stronger increase in absorption than recipe Tfast. For the undoped material, no significant difference in α is observed for the samples treated according to the two recipes.
Beyond 2000 nm, all recipes lead to a reduction of the absorption coefficient. For the undoped congruent material Tslow leads to a reduction of the absorption by a factor of 5 in the OH− peaks. For the 5-mol.-%-MgO-doped congruent material, the improvement factor is 3. For the undoped material the reduction factor for recipe Tfast is smaller (factor 2) than for recipe Tslow (factor 5), while for the doped material both annealing procedures lead to similar results.
Annealing of the stoichiometric material leads to an increase of the extinction by at least a factor of 10 compared to that of the untreated sample, for both light polarizations. Solely the strong OH− absorption peak at 2830 nm drops by a factor of 3. Optical microscopy (see Fig. 3) reveals micrometer-sized structures in the bulk material for all annealed stoichiometric LiNbO3 samples, while for the untreated, as well as for the congruent samples, such structures are not observed.
Regarding the first goal, an oxidization of Fe2+ to Fe3+, the results in Fig. 2 indicate that the absorption caused by Fe2+ with its center around 1200 nm is replaced by another broad absorption centered around 1040 nm extending also to 1800 nm. Earlier work with doped LiNbO3 crystals shows that Cu2+ can cause such an absorption originating from crystal field splitting . As mentioned above, copper is one residual impurity that results from the growing process of the crystals. Copper is incorporated in the crystal lattice only in the valence states Cu+/2+ . So, the oxidization process may affect not only the residual Fe2+ but also Cu+. From the measured absorption coefficient and the absorption cross section calculated from published values , one can calculate the concentration of Cu2+ in the oxidized samples, leading for our case to a concentration of cCu2+ ≈ 0.2 – 0.4 wt. ppm for the different samples which is a realistic number. Determination of the Fe2+ concentration from its absorption coefficient at 1250 nm wavelength with the cross section given by Kurz et al.  leads for the untreated samples to a value of cFe2+ ≈ 0.04 wt. ppm for both congruent crystals.
Furthermore it is found, that for undoped LiNbO3 the absorption that is attributed to the oxidized copper is the same for moderate and strong cooling ramps, showing that reducing effects upon cooling are unlikely. For the doped sample, however, they could be responsible for the weaker absorption due to Cu2+ in the case of moderate cooling.
From Fig. 2 it is furthermore obvious, that a significant decrease in absorption can be achieved in the region where hydrogen, incorporated into the crystals lattice, dominates the absorption. Especially for the undoped material, the treatment with a slower cooling curve (Tslow) gives better results, i.e. lower absorption. This is probably due to the fact that, in order to get the strong cooling ramp for the annealing procedure Tfast, the oven is opened at 1060 °C. Thus, air from the surrounding can stream into the oven, exposing the samples to water vapor. This can result in a reincorporation of hydrogen from the air water content. For the moderate cooling ramp, the oven was opened at lower temperatures, where the mobility of hydrogen in LiNbO3 drops. Although the mobility is smaller, it is not zero, and some hydrogen might still be built in also for the moderate cooling speed. We attribute this to be the reason why the decrease by a factor of 10 in absorption  could not be reached for the undoped, congruent LiNbO3. Here an improvement just by a factor 5 is observed. For the 5-mol.-%-MgO-doped material the improvement is weaker (factor 3). This might be due to the lower OH− concentration in commercially available crystals of this material in general. An optimization of the recipe towards a treatment only of the OH− would be to increase the annealing time and to cool smoother in a dry, oxygen free atmosphere.
The micrometer-sized structures seen in the stoichiometric material may stem from local phase variations, i.e., of the lithium content, as the melting point for stoichiometric LiNbO3 is lower than that for congruently grown LiNbO3 . Nevertheless, regardless of the increased scattering, it can still be recognized that the OH−-induced absorption is greatly reduced by the treatments.
Between 500 and 2900 nm impurities such as iron, copper, and hydrogen are responsible for residual absorption in lithium niobate. Annealing in dry oxygen atmosphere decreases the absorption for wavelengths above 2000 nm drastically, which is attributed to the removal of hydrogen. For shorter wavelengths, however, the absorption increases, which is explained by oxidizing Cu+ to Cu2+ causing absorption because of a crystal-field-splitting transition.
Financial support by the German Research Foundation (DFG) is gratefully acknowledged. The article processing charge was funded by the German Research Foundation (DFG) and the Albert Ludwigs University Freiburg in the funding program Open Access Publishing.
References and links
1. M. Lines, “Ultralow-loss glasses,” Annu. Rev. Mater. Sci. 16, 113–135 (1986). [CrossRef]
2. M. M. Fejer, G. A. Magel, D. H. Jundt, and R. L. Byer, “Quasi-phase-matched second harmonic generation: tuning and tolerances,” IEEE J. Quantum Electron. 28, 2631–2654 (1992). [CrossRef]
4. P. A. Franken, A. E. Hill, C. W. Peters, and G. Weinreich, “Generation of optical harmonics,” Phys. Rev. Lett. 7, 118–119 (1961). [CrossRef]
5. L. Arizmendi, “Photonic applications of lithium niobate crystals,” phys. stat. sol. (a) 201, 253–283 (2004). [CrossRef]
6. M. Leidinger, S. Fieberg, N. Waasem, F. Kühnemann, K. Buse, and I. Breunig, “Comparative study on three highly sensitive absorption measurement techniques characterizing lithium niobate over its entire transparent spectral range,” Opt. Express 23, 21690–21705 (2015). [CrossRef] [PubMed]
7. G. E. Peterson, A. M. Glass, and T. J. Negran, “Control of the susceptibility of lithium niobate to laser-induced refractive index changes,” Appl. Phys. Lett. 19, 130–132 (1971). [CrossRef]
8. J. R. Schwesyg, C. R. Phillips, K. Ioakeimidi, M. C. C. Kajiyama, M. Falk, D. H. Jundt, K. Buse, and M. M. Fejer, “Suppression of mid-infrared light absorption in undoped congruent lithium niobate crystals,” Opt. Lett. 35, 1070–1072 (2010). [CrossRef] [PubMed]
9. E. M. de Miguel-Sanz, M. Carrascosa, and L. Arizmendi, “Effect of the oxidation state and hydrogen concentration on the lifetime of thermally fixed holograms in LiNbO3:Fe,” Phys. Rev. B 65, 165101 (2002). [CrossRef]
10. M. Falk and K. Buse, “Thermo-electric method for nearly complete oxidization of highly iron-doped lithium niobate crystals,” Appl. Phys. B 81, 853–855 (2005). [CrossRef]
11. A. Dhar and A. Mansingh, “Optical properties of reduced lithium niobate single crystals,” J. Appl. Phys. 68, 5804–5809 (1990). [CrossRef]
12. K. Peithmann, J. Hukriede, K. Buse, and E. Krätzig, “Photorefractive properties of LiNbO3 crystals doped by copper diffusion,” Phys. Rev. B 61, 4615–4620 (2000). [CrossRef]
13. H. Kurz, E. Krätzig, W. Keune, H. Engelmann, U. Gonser, B. Dischler, and A. Räuber, “Photorefractive centers in LiNbO3, studied by optical-, Mössbauer- and EPR-methods,” Appl. Phys. 12, 355–368 (1977). [CrossRef]
14. J. R. Carruthers, “Nonstoichiometry and crystal growth of lithium niobate,” J. Appl. Phys. 42, 1846–1851 (1971). [CrossRef]