1. Introduction

Silver is an ideal metal for the front surface mirrors for optics in both visible and infrared wavelength ranges. In this range, silver possesses the highest reflectivity, the lowest emissivity, and lowest polarization splitting of all known metals [1]. Silver, however, can be oxidized very easily and requires reliable protection to be stable in the aggressive operating environment. The problem of protection of silver layers is a very important one for a number of practical applications, such as telescope (ground and space-born) mirrors, reflective IR imaging optics, photovoltaic concentrator mirrors, III-V laser back-reflectors, etc. [2]. The space environment especially results in severe damage arising from the exposure to the highly oxidizing atmosphere on low Earth orbit (LEO) [3].

Atomic oxygen, which is the most abundant corrosion precursor on LEO, is formed in the space in the process of photo dissociation of O₂ molecules by ultraviolet (UV) photons (< 243 nm) of solar radiation which has enough energy to break a 5.12 eV O₂ bond in an environment where the mean free path is sufficiently long and the probability of recombination or formation of ozone is quite small [3].

Protecting the mirrors with coatings, in this situation, is the only option to extend the lifetime span of the silver mirrors towards reasonable values. A number of different materials were reported in this respect, from silicon nitride and silicon oxide to ultra-thin Ni layers that allow not only slowing the tarnishing of silver but also lifting the reflection in the UV region, somewhat, though, affecting reflection in the infrared [4]. Low temperature PECVD layers of SiN_x produce multiple absorption bands in the IR region due to N-H and Si-H bonds. Various deposition techniques producing densely-packed layers can be used, such as Ion Assisted Deposition or PECVD. More common thermal evaporation and electron beam evaporation, unfortunately, produce a columnar layer structure that allows diffusion of foreign species from the environment to the buried interface.

Using atomic layer deposition (ALD) for the protection of silver-based front surface mirrors was first reported by [5–7]. While compatible with substrates of large sizes, ALD is especially well suited for the protection of small to medium-sized front surface mirrors. ALD technology can be purely thermal or plasma-assisted (PEALD). PEALD “as is” cannot be used for deposition of the entire thickness of the coating, because active oxygen will immediately oxidize the silver surface. So, before using PEALD, an additional protective layer was required [5]. This step is not needed in purely thermal ALD, consequently making it the best option. The most commonly ALD-produced material is Al₂O₃, with Trimethylaluminum (TMA) and water used as precursors. It has a very large transmission window with only some minor absorption at about 900 cm⁻¹, but at the typical thickness of several tens of nanometers, this absorption is completely negligible. High quality Al₂O₃ film can be deposited at temperatures as low as 100°C, though more hard and dense films are obtained at the temperatures above 150°C [8].

In this article, we report on the study of the resistance of radio-frequency (RF) magnetron sputtered front surface silver mirrors with ALD-deposited Al₂O₃ protective layers to the erosion in oxygen plasma, generated in high-density electron cyclotron resonance (ECR) plasma system. Accelerated aging tests in plasma systems (given the high active oxygen flux to the surface) can help to optimize protective coatings performance and are not uncommon [9]. The obvious advantage of using ECR plasma system instead of more common RF capacitively coupled plasma systems for simulation of the space environment is a very low sheath voltage, which limits the energy of the oxygen ions striking the surface to just several eVs. This is consistent with the maximal impact energy of atomic oxygen on front surface mirrors during space flights [3].

2. Experimental

Silver films with the thickness around 200 nm were sputtered onto 375 µm thick 100 mm in diameter (100) silicon wafers in Alliance Concept Dp650 RF magnetron sputtering system in the following conditions: RF-power 75 Watts (resulting in -275 V of DC bias) using 90 mm in diameter silver target, Ar pressure of 6.5 mTorr at 250 °C (substrate holder temperature). These conditions were deduced from the optimization study performed in order to improve the stability of the silver mirrors towards the ALD deposition. Before loading into the sputtering system, the silicon wafers were dipped in 5 percent hydrofluoric acid solution in water for 30 sec in order to remove native oxide. It was found that such procedure produced films of consistently better quality, denser and having considerably lower roughness. No additional adhesion layer was used. HF treatment alone was sufficient for mirrors to pass standard scotch tape adhesion test. The base vacuum before the depositions was always below 5·10⁻⁷ Torr. After sputtering, the mirrors were transferred into the Picosun R200 Advanced deposition system, where they stayed under constant N₂ flow for one hour in order to reach stable temperature before the deposition. Thermal ALD depositions were carried out at 150°C temperature using alternating 100 ms pulses of TMA and water at a chamber pressure of 9 Torr with constant nitrogen purge flux of 1 SLM.

Before and after depositing the protective ALD layers of different thicknesses, all mirrors were measured by spectroscopic ellipsometry and reflection spectroscopy. Additionally, some samples were analyzed with pulsed glow-discharge optical emission spectroscopy (GD-OES) to establish the compositional profile of the coatings [10]. Then the mirrors were exposed to a high flux of oxygen ions and radicals at room temperature in a high-density ECR plasma system for 1 minute. After oxygen exposure, we again studied the mirrors with reflection spectroscopy, spectroscopic ellipsometry (when possible) and pulsed GD-OES in order to find out the changes in reflectivity and composition and evaluate the efficiency of a protective layer.

Spectroscopic ellipsometer Uvisel-1 by Horiba was used for ellipsometric data acquisition in a spectral energy range 1.0 - 4.6 eV and DeltaPsi 2 software, also by Horiba, was used for data analysis. Reflection measurements on samples were performed using Perkin-Elmer Lambda 950 system equipped with a specular reflectance attachment in the wavelength range 200-2500 nm at 8 degrees incidence. For pulsed GD-OES analysis, we have used GD Profiler 2 by Horiba, while SEM photos of the silver surface were made in Hitachi S4800 scanning electron microscope (SEM). Oxygen plasma treatment was performed in the designed in-house MDECR plasma enhanced chemical vapor deposition (PECVD) system [11] at 2 mTorr pressure using 40 sccm O₂ flow and 1000 Watts microwave power for the typical duration of 1 minute (sometimes longer). Those conditions produce constant oxygen ions flux of about 6·10¹⁵ ions/cm²·sec onto the grounded wafer (around 1 mA/cm² when measured with flat Langmuir probe). Estimated oxygen radicals flux should be close to 7·10¹⁶ particles/cm²·sec.

3. Results and discussion

We have found that the conditions of sputtering are very important for mirrors to withstand subsequent ALD coating deposition. Our standard conditions for silver sputtering, consequently, had to be modified in order for Ag mirror to survive Al₂O₃ deposition at 150 °C. Initially tried conditions of the deposition – no heating of the substrate, 50 Watts of RF power (instead of 75 Watts after optimization), and use of silicon wafers without removing native oxide resulted in silver layers, that degraded in the process of ALD deposition. Step-by-step optimization consisted of (1) introducing the HF treatment of silicon wafer for native oxide removal, (2) finding the values of temperature, RF power, and Ar flow, providing compact, low roughness silver layers that were able to withstand ALD deposition process. Results of silver sputtering optimization are presented in Figs. 1 and 2, which compare electron microscopy images before the depositions and visual appearance of silver layers after ALD depositions.

 

Fig. 1. Scanning electron microscopy images of sputtered silver films: standard (left) and optimal (right) conditions.

Download Full Size | PDF

 

Fig. 2. Impact of ALD deposition on reflection of silver sputtered in two different conditions (note inserts for silver morphology before ALD deposition). The surface of silver, deposited at non-optimal conditions (full wafer), becomes rough and highly-scattering after the ALD process, while square cut from the wafer deposited in optimal conditions and placed onto substrate holder for the same deposition run remains intact and highly reflective.

Download Full Size | PDF

Optical properties of ALD-deposited Al₂O₃ protective layers are broadly in line with those reported for the films deposited in similar conditions (Table 1). The optical properties do not change over the thickness, as demonstrated by perfect ellipsometric fits, as seen with Fig. 3 for a film with a thickness of 63 nm. No measurable absorption is detectable by spectroscopic ellipsometry, whether in the film deposited over clean silicon wafer or over silver mirror layer.

 

Fig. 3. Ellipsometric spectra and fit for 600 cycles Al₂O₃ film on silver (thickness is 63 nm, material properties are given in Table 1). Points are experimental data, solid lines are the result of the fit.

Download Full Size | PDF

Table 1. Lorentz dispersion formula parameters for dielectric function of ALD-deposited Al₂O₃

View Table

As evidenced from Fig. 4, the deposition rate versus cycle number is linear, with minor variations at a low number of cycles, what could possibly be due to the fact, that silver mirrors were exposed to air before the growth for a somewhat different amount of time. It may, according to literature, affect the absorption of precursors within the first several alternating pulses. Those fluctuations will be studied more carefully in future work.

 

Fig. 4. Dependence of Al₂O₃ layer thickness on the number of ALD cycles. Solid points are experimental data. The solid line is only guide for the eye.

Download Full Size | PDF

As can be noticed, the reflectance of the silver film in the wavelength range above 320 nm is barely affected by the presence of the Al₂O₃ layer, while for shorter wavelength reflectance differs considerably (see Fig. 5). This spectral region (below 400 nm), however, is not of interest for using silver mirrors due to the strong silver absorption feature arising from interband transition at around 320 nm [12].

 

Fig. 5. Reflection spectra of silver mirrors, unprotected and protected with Al₂O₃ layers produced with 50, 100, and 150 ALD cycles, respectively, measured at 8 degree incidence angle before plasma oxidation.

Download Full Size | PDF

After the oxidation step, the reflection of a non-protected mirror and mirrors with both 5 and 10 nanometer-thick coatings suffered significant deterioration (see Fig. 6). The non-protected silver layer was oxidized all the way to the silicon wafer, as evidenced by the feature in the reflection spectrum arising due to additional contribution from the polished backside of silicon wafer around the bandgap of crystalline silicon (approximately 1100 nm). While 5 nm and 10 nm layers somewhat protected the bulk of silver layer, the surface still degraded significantly, leading to a big drop in reflectivity in the visible wavelength range. Apparently, only the deposition of 15 nm-thick layer of Al₂O₃ was enough to provide full protection against erosion of silver surface by atomic oxygen. For this sample, an additional plasma exposure time (5 more minutes) also did not lead to the deterioration of reflectance. For mirrors to withstand atomic oxygen attack for sufficiently long working time on LEO, such coating is judged to be adequate. If other constraints are present, like mechanical damage, thermal cycling, etc. additional tests may be required.

 

Fig. 6. Reflection spectra of silver mirrors, unprotected and protected with Al₂O₃ layers produced with 50, 100, and 150 ALD cycles measured at 8 degree incidence angle after plasma oxidation.

Download Full Size | PDF

Analysis by GD-OES corroborates the data obtained by reflection spectroscopy. Depth profiles for the following elements were recorded: H, O, N, Ag (shown divided by 10 on the graphs due to very high intensity of emission line), C, S, Al, and Si, while Fi represents the total emission intensity. Unprotected silver film is fully oxidized through with more or less constant concentration of oxygen all the way to the point where the signal from silicon wafer appears (Figs. 7(a) and (b)). 5 nm Al₂O₃ film apparently slows down the oxidation, but about a quarter of silver layer is seriously compromised (Figs. 7(c) and (d)). 15 nm layer of Al₂O₃ film completely protects the underlying silver film from oxidation by active oxygen (Figs. 7(e) and (f)). Doubling the plasma exposure time for a silver film protected with 15 nm of alumina did not produce any difference. Silver protected by thicker Al₂O₃ films (>15nm) also did not show deterioration during oxygen exposure. Interestingly, for the films with thickness of 15 nm and more, a silver emission line was not appearing at the beginning of sputtering, while for thinner films, it appears simultaneously with carbon line, as carbon always present on the surface, and that is true even for non-oxidized films. We verified with ellipsometry 2D mapping with a step of 1 cm that the thickness of ALD deposited films was very uniform along the surface of the mirrors. Moreover, corrosion character did not manifest itself with a spot-like erosion. We have to note here that ellipsometry gives an average thickness across the beam spot (around one square mm) and is not sensitive to a small number of defects, such as tiny pinholes, but they are typically absent in ALD films.

 

Fig. 7. GD-OES spectra of silver mirrors before and after the exposure to O₂ plasma: a) 0 cycles, non-oxidized; b) 0 cycles, oxidized; c) 50 cycles, non-oxidized; d) 50 cycles, oxidized; e) 150 cycles, non-oxidized; f) 150 cycles, oxidized;

Download Full Size | PDF

4. Conclusions

Sputtered front surface silver mirrors can be quite sensitive to the ALD process, depending on the way they are deposited. In order to make a silver layer be compatible with 150 °C ALD deposition, one may need to optimize silver magnetron sputtering steps in order to produce low roughness dense layers. Protective films of Al₂O₃ deposited onto the optimized silver mirrors by ALD provide an effective barrier against active oxygen erosion even at 15 nm of protective film thickness and can be deposited at the temperature of 150 °C, provided silver film withstands the conditions of the ALD process. It is not yet clear why the thicknesses below 15 nm could not prevent the deterioration of a mirror surface in oxygen plasma, and further work is needed to clarify the details of the degradation process. Pulsed GD-OES technique provides a fast and accurate way to investigate the degradation of mirrors after different sorts of corrosion tests.