In this contribution, we investigate the properties of antireflective coatings on iodine-filled absorption cell windows. These coatings are subject to high temperatures during the cell production process and are in direct contact with the absorption medium, which influences their optical performance. We tested the thermal resistance of TiO2- and Ta2O5- based coatings produced using conventional electron beam evaporation (e-beam) and ion-assisted deposition (PIAD). We prepared a set of iodine-filled absorption cells that were used to test the coatings’ resistance to iodine vapors. We show that the choice of coating materials, coating methods, and a well-chosen bakeout procedure can mitigate any unwanted effects, such as temperature-induced spectral shifts and optical losses inhomogeneities or settling of the absorption medium in the coating.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
There are multiple deposition techniques of dielectric optical thin films. The most common are electron beam evaporation, plasma ion assisted deposition, magnetron sputtering or ion beam sputtering . In this article we will deal with electron beam evaporation both conventional and plasma ion assisted. Conventional electron beam (e-beam) evaporation uses an electron beam to melt and evaporate the source material in a vacuum chamber. During deposition, the chamber pressure is low enough for the source material molecules to reach the substrate.
Plasma ion assisted deposition (PIAD) uses a plasma ion source in the vacuum chamber which directs a cone of plasma towards the substrates. The main difference from e-beam is a higher kinetic energy of the condensing material particles. Coatings prepared using this method have a higher refractive index and packing density than coatings prepared using e-beam only .
One of many applications of thin film coatings deposited on optical elements is in the area of absorption cells used as optical frequency references for frequency locking of lasers. The physical properties of the absorption medium and the optical properties of the cell itself determine the achievable parameters of the standard. The absorption cells are commonly developed in the form of bulky glass tubes, where the laser beam enters the inner volume through an optical window, interacts with the absorption medium inside the cell and leaves through an output optical window to a photodetector [3,4]. The most common absorbing media for stabilization of laser standards in present days are molecular iodine (mainly for visible spectral range) and acetylene (telecom wavelengths), because they offer the best spectral properties for laser locking [5–10]. Several absorption media (including iodine) are highly corrosive (especially for metallic materials) and sensitive to the presence of impurities and contaminants [3,4,11]. Moreover, especially in term of alkali vapors filled cells, problem of relaxation of excited atoms to ground state due to collisions with the cell walls caused reduction of achievable frequency stabilities. Also in these cases, a special coatings of the cell walls can reduce these unwanted effects [12,13].
Modern approach to the design of optical frequency references includes utilization of hollow-core photonic crystal fibers, where the mass and volume of the reference is drastically reduced and the interaction lengths of the light beam and the medium can be extended to units/tens of meters [14–17]. The main factor limiting broader application of these fiber-based references can be seen in lower achievable frequency stability due to transit-time and collision broadening of the absorption lines.
Antireflection coatings deposited on the cell windows and optical fiber ends can significantly reduce optical losses, laser light back-reflections and unwanted resonator effects . Especially in compact laser setups, the cells are often used in multi-pass arrangements [10,19]. In this case the inner surface of the windows can be coated with a combination of antireflection and high-reflective coatings to produce the multiple reflections of the beam through the medium. As the quality and spectral properties of the coatings affect the optical parameters of the reference, it has also a direct impact on the performance of the whole laser standard.
The optical windows can be attached to the cell body/tube by several techniques. One possibility is using special vacuum-compatible glue or solder. The windows can be also connected by optical contacting technique, where no additional material/chemical is needed. This technique needs demanding equipment for high precision polishing of optical surfaces, and it is more suitable for connecting optical elements with larger surfaces than the 1-2 mm wide joints between the cell tube and windows . The third approach for connecting the optical windows to the cell body is the quartz welding method. The potential risk of thermal stress and degradation of the coatings caused by welding flame can be minimized by proper mechanical design of the windows . The windows are in this case typically equipped with facets, so the thermal energy needed for melting of windows boundaries is reduced and transit of the heat to the coatings is minimized (Fig. 1).
Depositing the coating on every active surface means that some of the coatings are in a direct contact with the medium inside the cell. Due to this fact, the material of the coating itself must also ensure non-reactivity and inertness to the absorption medium. Furthermore, the coatings have to be prepared in a way that avoids settling of the medium particles in the coatings structure.
In this article we investigate the influence of high temperatures and the presence of corrosive iodine vapors on the antireflective coatings properties and its dependence on coating materials and deposition technologies.
In the following investigation we especially concentrate on performance of antireflection coatings for 532 and 633 nm wavelengths, where molecular iodine-stabilized laser standards commonly operate. However, the general results are applicable for different coating designs and desired wavelengths as well.
2. AR coatings for 532 and 633 nm wavelengths
For our experiments we prepared four types of coatings deposited on four sets of fused silica substrates (FS). The aim of these experiments was to investigate the differences in coating materials and deposition techniques and their influence on sensitivity to thermal stress, and on the optical performance of the iodine absorption cells.
The fused silica substrates are 1” in diameter and 5 mm thick. One surface is polished to optical quality of λ/10, the other surface is fine ground to minimize back surface reflections.
When designing an antireflective coating there are several coating materials to choose from. In our case we wanted to investigate the materials we commonly use as they are relatively cheap, easy to work with and have a good optical performance. We chose TiO2 and Ta2O5 as the high refractive index material and SiO2 as the low refractive index material. The optical designs of the coatings are summarized in Table 1.
The coatings were produced using a SYRUSpro 710 coating system made by Bühler (formerly Leybold Optics). All coatings of one particular type were prepared in a single batch to ensure the same conditions for all samples. The deposition parameters can be found in Table 2. The substrate temperature was 300 °C in case of e-beam and approx. 260 °C in case of PIAD. The plasma source used was APSpro by Bühler.
The residual spectral reflectance characteristics measured by Varian Cary 5 spectrophotometer are in Fig. 2. All of the samples exhibit residual reflectance equal to or lower than 0.1% at the desired wavelengths (532/633 nm).
3. Thermal resistance of dielectric coatings based on TiO2 and Ta2O5 materials
The temperature influence on the coatings’ optical properties (degradation, spectral shifts) were investigated by exposing the samples to various temperatures (350-1000°C) in a PID regulated electrical furnace. The applied temperature influence control loop steps include: 1/ heating up from room temperature to the desired temperature (heating rate 1000°C/1 hour), 2/ keeping the desired temperature for 2 hours, 3/ spontaneous cooling back to room temperature. For each bakeout procedure temperature level, new AR coated FS sample from the set was used. The residual reflection of tested substrates was evaluated by the spectrophotometer (Varian Cary 5). The thermally induced spectral shifts of all samples (e-beam TiO2, e-beam Ta2O5, PIAD TiO2, PIAD Ta2O5) are summarized in Fig. 3, detailed results from desired AR wavelengths (532 and 633 nm) are pointed out in Fig. 4.
The results show that PIAD coatings have a better resistance to high temperatures. They exhibit very low spectral shifts and perform reasonably well at both 532 nm and 633nm up to 1000 °C in case of TiO2 and 900 °C in case of Ta2O5. The e-beam prepared coatings spectral performance fails at lower temperatures. The plasma ion assist transfers its kinetic energy to material particles, which are condensing on substrate . E-beam coatings form a columnar structure, while PIAD coatings form a dense structure on the surface, which is less susceptible to thermal treatment . The measurement also confirms that comparing to Ta2O5/SiO2 coatings, where the thermally induced spectral shifts play a role in case of bakeout temperatures over ~600°C, TiO2/SiO2 coatings show better properties. The same ion energy was used for both TiO2 and Ta2O5. As there is a difference in mass of these materials, they might be affected differently by the ion assist, which may result in a different microstructure and susceptibility to thermal treatment. The surface morphology of the samples was inspected by AFM (atomic force microscopy) and we did not observe any changes of the surface with the temperature.
It must be pointed out that the presented results of bakeout procedures were done for a relatively short exposure times (~2 hours) and slow thermal shock rate (1000°C/hour). The mid-term and long-term impacts are planned for investigation in future work together with dramatically faster cooling of the heated coatings. However, we did not observe any differences in the results of two hours and ten hours long bakeout procedures. Moreover, repeated measurement of the spectra after one month of the initial bakeout procedure did not show any spectral changes during this period. We also expect that the thermal expansion coefficient of used substrate material (fused silica) is small enough to prevent the coatings against thermal shock induced mechanical cracking (discussed below in section 4).
4. Spectral and optical properties of baked out AR coatings for iodine absorption cells windows
Considering the results from comparison between TiO2 and Ta2O5 coatings material in section 3, we decided to use TiO2/SiO2 material combination and PIAD deposition method for testing its suitability for AR coatings for iodine absorption cells windows. The design of the coatings was the same as in bakeout procedure test described in section 3. The cells set consisted of 6 cells filled with ultra-pure molecular iodine 127I2. The whole cells (including optical windows) were made of fused silica material, the windows were equipped with safety facets and were joined to the cell body midsections by quartz welding. The active length was 200 mm and active diameter of the windows (AR coatings diameter) was 22 mm.
4.1. AR coatings spectral properties in context of iodine absorption cells technology steps
The cell windows inner surfaces were AR coated, then the windows were baked out. After baking out, the windows were welded onto the midsection and then their outside surfaces were AR coated. The evaluation of the coatings spectral properties after each individual cell manufacturing technology step consists of: 1/ residual reflectance measurement of the window surfaces after the welding of the window to midsections of the cell body, 2/ transmittance measurement through the cell after welding all of its parts together (midsections with windows welded to central part), 3/ transmittance measurement after the filling of the cell with the iodine media.
Considering the results of thermal resistance presented in section 3, we decided to apply the bakeout procedures as follows: PIAD-no bakeout, 800°C, 900°C and e-beam-no bakeout, 600°C, 800°C. The baked windows with connected midsection parts were then coated from outer sides and residual reflectance of both sides of the windows were evaluated by the optical spectrophotometer Agilent Cary 7000. The diagrams in Fig. 5 present means of averaged residual reflectance measured from pairs of windows tested from inner and also outer sides.
Investigation of the results found that the baking procedure did not cause significant spectral profiles changes, the residual reflectance at desired wavelengths still stays below 0.25% for all baking temperature cases. The main difference can be observed in the graph PIAD / 900°C with spectral red shift of ~10 nm.
The finished windows with midsections were than welded to the central parts of the cells, the cells were completed and prepared for filling with iodine. The transmittance of the completed cells (before filling) is shown in Fig. 6. The size of the spectrophotometer sample compartment allows for placing of the whole cell to the path of the focused light beam, with the optical waist in the middle of the cell length. Because of the length of the cells (~200 mm) and the divergence of the beam, the cross section of the beam at the windows positions is ~15x15 mm and covers almost the whole surface of the coatings.
The investigation of the final transmittance through the cells was done after the filling of the cells with iodine media using the same optical arrangement in the spectrophotometer as in the previous step. To avoid the optical absorption caused by the iodine vapor, the cold finger of the cell was placed in liquid nitrogen (LN2) trap where the iodine was stored in a solid state. After several tens of minutes of LN2 cooling, the transmittance spectra were recorded.
The transmission measurement results in Fig. 6 confirm the thermally induced effects in the spectral properties of the coatings. In case of the PIAD method, the bakeout procedure of up to 800°C level causes negligible changes (losses ~0.1% per coating) of optical transmittance at desired wavelengths. With the thermal stress induced by bakeout temperature of 900°C, the losses increase to levels of 0.6% (532 nm) and 0.3% (633 nm) per coating respectively. In comparison to these results, the coatings deposited by conventional e-beam method exhibit much higher sensitivity to applied bakeout temperatures, optical losses reach 0.5% (532 nm) and 0.3% (633 nm) per coating for 600°C and 0.65% (532 nm) and 0.25% (633 nm) per coating for 800°C bakeout temperature levels.
Moreover, in comparison to conventional e-beam technology, the use of the PIAD method in optical coatings deposition dramatically reduces the sensitivity of the coatings to the settling of the iodine molecules in the coating structure (Fig. 6). The results also confirm that this unwanted effect in e-beam deposition can be mostly suppressed by proper coatings bakeout before filling the cell with the iodine media (but finding an acceptable compromise against deformation of the spectra caused by baking out is necessary).
4.2. Optical losses homogeneity of the cells’ windows AR coatings
Considering the limitations of spectrophotometer spot size of ~15 x 15 mm, we decided to map the AR coatings’ optical losses homogeneity at λ = 532 nm using laser beam. The filled cells were put into simple optical system consisted of frequency doubled Nd:YAG laser and a photodetector unit. The frequency of the laser was detuned far away from any visible iodine absorption line in the cell and further the iodine in the cell under the test was trapped in liquid nitrogen (LN2) trap. In this arrangement we were able to monitor optical losses on the cell windows (4 interfaces in total) by the optical power meter. Thanks to small diameter size of the laser beam (~1 mm) we were able to monitor these losses in different areas of the windows. The correctness of this method has been verified by comparison of the transmittance measurement of the iodine filled cell (solid iodine trapped by LN2 trap) and measurement of reopened and washed cell. The washing procedure included slight heating of the cell windows with the hot-air gun (~60°C) with simultaneous rinse of the argon gas for helping the iodine to get off the inner volume of the cell. The washing was applied for 30 minutes. The resulting transmittance, which was ~0.2% lower after the washing than in case of the iodine filled cell (due to absorption caused by residual iodine settling on AR coatings) confirms that using of LN2 trap in our measurements is a reliable method for keeping the iodine out of the coatings.
The results of total optical losses measured through both cell windows (1 beam pass) at 9 different positions (1 in center, 4 between center and periphery and 4 in the coating periphery) for each of the cells are plotted in Fig. 7.
Considering the estimated accuracy of the laser measurement ( ± 0.1%), the results show the same trends as with previously obtained data from the spectrophotometry measurement. The use of the ion assist (PIAD) in the coating formation significantly contributes to better thermal resistance of the coating. According to the results of the tested coatings, there is a minimal impact of spectral shifts or degradation of the coatings to resulting optical losses for bakeout temperatures for up to at least 800 °C. The standard deviations of the homogeneity of the optical losses at different window locations are well below 0.5% (averaged per 4 coatings interfaces).
Regarding the coatings produced without ion assist, there is an evident impact of the iodine absorption medium to the resulting optical losses. The iodine molecules probably settle and are trapped in the porous structure of the coatings and cause additional light absorption. The worst case (the windows deposited with coating without ion assist and without bakeout procedure) exhibit very high losses at levels up to 6% per pass through the cell. When we consider that this unwanted iodine settling effect acts only on the inner surfaces of the cell and the outer losses per coating is at or below 0.2% level, we obtain the losses at least at 2.8% per one coating-iodine interface.
However, the bakeout procedure can significantly reduce this effect, as shown by results from “e-beam / 600°C” and ”e-beam / 800°C” diagrams in Fig. 7. Considering the coating design under the test and minding the results from above presented measurements, one can find a suitable compromise between bakeout temperature level (spectral shift and degradation of the coating) and the level of iodine settling effect between 600 and 800°C. Furthermore, thermal impact on the coating structure is confirmed by the inhomogeneity of optical losses of the “e-beam – no bakeout” cell. The standard deviation of the losses at different location of the windows is at 1.5% level, at least three times higher than in case of the PIAD deposited cells. Moreover, in general this cell exhibits lower level of losses close to the coating edge, caused by the glasswork burner induced thermal stress around the edges during the welding of the windows. This result confirms heat influence on the coatings-iodine reactivity/settling effect.
4.3. Evaluation of the iodine purity
One of the potential reasons for degradation of the coatings optical properties can be found in contamination with the iodine and formation of unwanted iodine based compounds settling down/reacting with the windows coatings. The evaluation of the chemical purity and non-reactivity of the iodine media against materials of the coatings was realized through the laser induced fluorescence method (LIF), well described in [3,4,22]. This technique is based on measurement of dependency between the relative level of induced fluorescence on iodine pressure, expressed by Stern-Volmer formula  and resulted in Stern-Volmer coefficient (K [Pa]) describing the purity level. The cells are rated as: excellent (K<1 Pa), good (K between 1 and 1.5 Pa) and contaminated (K>1.5 Pa). All of the iodine cells were tested by the LIF system and they show excellent purity with the Stern-Volmer coeficient (K) at 0.76-0.93 Pa level (Table 3).
These results confirm the high-level purity of iodine medium in the cells and there is no chemical reaction of iodine with the coatings materials. This evaluation also confirms that the high level of optical losses found in the measurement (especially the cell ”e-beam / no bakeout”) is caused by the settling of iodine molecules in the structure of the coating surface. This effect can be reduced by using of the PIAD and proper bakeout procedure of the coatings before the filling of the cell with the iodine.
This work was oriented towards the influence of thermal annealing to spectral properties of dielectric optical coatings on iodine absorption cells windows. Two different coating material combinations (TiO2/SiO2 and Ta2O5/SiO2) and two different deposition methods (conventional e-beam and plasma ion assisted deposition) were studied from the point of view of antireflective coatings’ resistance to applied bakeout temperatures and chemically aggressive environment.
TiO2 based PIAD coating is the most resistant against temperature induced spectral shifts. Temperatures up to 1000°C do not produce any significant shifts in the spectrum and the coating performs well at both 532 nm and 633 nm wavelengths. In case of both TiO2 and Ta2O5, PIAD coatings show better thermal resistance than conventional e-beam.
In terms of iodine-filled absorption cells, the proper bakeout procedure can significantly reduce optical losses of the AR coatings due to the iodine settling into the coating. In case of TiO2 based PIAD coating baked to 900°C we achieved optical losses at the inner surface of 0.3% level, while in case of e-beam based coating without bakeout procedure this losses dramatically increased up to 2.8% level per inner surface.
The resulting spectral characteristics confirm that by selection of proper deposition method and optimized bakeout temperature, the unwanted effect of settling of the iodine in the coating structure can be effectively suppressed and the performance of laser standards based on these iodine cells can be further improved.
Grant Agency of the Czech Republic (GA15-18430S); Ministry of Education, Youth and Sports of the Czech Republic (LO1212); Ministry of Education, Youth and Sports of the Czech Republic and European Commission (CZ.1.05, 2.1.00, 01.0017); Czech Academy of Sciences (RVO: 68081731, Strategy AV21).
The authors wish to express thanks to Tatiana Šarlejová and Stanislav Šlechtický for their development of tested iodine absorption cells and preparation of the optical elements used in this research work.
1. H. A. Macleod, “Recent developments in deposition techniques for optical thin films and coatings,” in Optical Thin Films and Coatings - From Materials to Applications, A. Piegari and F. Flory, eds. (Woodhead Publishing Limited, 2013), pp. 3–25.
2. G. Atanassov, J. Turlo, J. K. Fu, and Y. S. Dai, “Mechanical, optical and structural properties of TiO2 and MgF2 thin films deposited by plasma ion assisted deposition,” Thin Solid Films 342(1-2), 83–92 (1999). [CrossRef]
3. M. Zucco, L. Robertsson, and J. P. Wallerand, “Laser-induced fluorescence as a tool to verify the reproducibility of iodine-based laser standards: a study of 96 iodine cells,” Metrologia 50(4), 402–408 (2013). [CrossRef]
5. J. Lazar, J. Hrabina, P. Jedlicka, and O. Cip, “Absolute frequency shifts of iodine cells for laser stabilization,” Metrologia 46(5), 450–456 (2009). [CrossRef]
6. G. D. Rovera, F. Ducos, J. J. Zondy, O. Acef, J. P. Wallerand, J. C. Knight, and P. S. Russell, “Absolute frequency measurement of an I-2 stabilized Nd: YAG optical frequency standard,” Meas. Sci. Technol. 13(6), 918–922 (2002). [CrossRef]
7. R. Holzwarth, A. Y. Nevsky, M. Zimmermann, T. Udem, T. W. Hansch, J. Von Zanthier, H. Walther, J. C. Knight, W. J. Wadsworth, P. S. J. Russell, M. N. Skvortsov, and S. N. Bagayev, “Absolute frequency measurement of iodine lines with a femtosecond optical synthesizer,” Appl. Phys. B 73(3), 269–271 (2001). [CrossRef]
8. P. Balling, M. Fischer, P. Kubina, and R. Holzwarth, “Absolute frequency measurement of wavelength standard at 1542nm: acetylene stabilized DFB laser,” Opt. Express 13(23), 9196–9201 (2005). [CrossRef] [PubMed]
10. C. Philippe, R. Le Targat, D. Holleville, M. Lours, T. Minh-Pham, J. Hrabina, F. Du Burck, P. Wolf, and O. Acef, “Frequency tripled 1.5 µm telecom laser diode stabilized to iodine hyperfine line in the 10-15 range,” Eur. Freq. Time Forum 1, 7477827 (2016).
11. J. Hrabina, M. Šarbort, O. Acef, F. D. Burck, N. Chiodo, M. Holá, O. Číp, and J. Lazar, “Spectral properties of molecular iodine in absorption cells filled to specified saturation pressure,” Appl. Opt. 53(31), 7435–7441 (2014). [CrossRef] [PubMed]
12. R. Straessle, M. Pellaton, C. Affolderbach, Y. Petremand, D. Briand, G. Mileti, and N. F. de Rooij, “Microfabricated alkali vapor cell with anti-relaxation wall coating,” Appl. Phys. Lett. 105(4), 043502 (2014). [CrossRef]
13. W. H. Li, M. Balabas, X. Peng, S. Pustelny, A. Wickenbrock, H. Guo, and D. Budker, “Characterization of high-temperature performance of cesium vapor cells with anti-relaxation coating,” J. Appl. Phys. 121(6), 063104 (2017). [CrossRef]
14. A. Lurie, P. S. Light, J. Anstie, T. M. Stace, P. C. Abbott, F. Benabid, and A. N. Luiten, “Saturation spectroscopy of iodine in hollow-core optical fiber,” Opt. Express 20(11), 11906–11917 (2012). [CrossRef] [PubMed]
15. P. T. Marty, J. Morel, and T. Feurer, “All-fiber multi-purpose gas cells and their applications in spectroscopy,” J. Lightwave Technol. 28(8), 1236–1240 (2010). [CrossRef]
16. M. Triches, M. Michieletto, J. Hald, J. K. Lyngsø, J. Lægsgaard, and O. Bang, “Optical frequency standard using acetylene-filled hollow-core photonic crystal fibers,” Opt. Express 23(9), 11227–11241 (2015). [CrossRef] [PubMed]
17. T. Talvard, P. G. Westergaard, M. V. DePalatis, N. F. Mortensen, M. Drewsen, B. Gøth, and J. Hald, “Enhancement of the performance of a fiber-based frequency comb by referencing to an acetylene-stabilized fiber laser,” Opt. Express 25(3), 2259–2269 (2017). [CrossRef] [PubMed]
18. J. Seppa, M. Merimaa, A. Manninen, M. Triches, J. Hald, and A. Lassila, “Interference cancellation for hollow-core fiber reference cells,” IEEE Trans. Instrum. Meas. 64, 1595–1599 (2015). [CrossRef]
19. T. Schuldt, K. Doringshoff, A. Milke, J. Sanjuan, M. Gohlke, E. V. Kovalchuk, N. Gurlebeck, A. Peters, and C. Braxmaier, “High-performance optical frequency references for space,” J. Phys.: Conf. Ser. 723, 012047 (2016).
20. M. S. Farhan, E. Zalnezhad, and A. R. Bushroa, “Properties of Ta2O5 thin films prepared by ion-assisted deposition,” Mater. Res. Bull. 48(10), 4206–4209 (2013). [CrossRef]
21. S. H. Woo, C. K. Hwangbo, Y. B. Son, I. C. Moon, and G. M. Kang, “Optical properties of Ta2O5 thin films deposited by plasma ion-assisted deposition,” J. Korean Phys. Soc. 46, S187–S191 (2005).
22. S. Fredin-Picard, “A study of contamination in I-127(2) cells using laser-induced fluorescence,” Metrologia 26(4), 235–244 (1989). [CrossRef]