1. Introduction

Laser-induced damage of optical coatings has been one of the limiting factors in developing high-power laser systems in recent years [1–4]. It is a complex subject because a great number of factors may lead to catastrophic failure of thin film devices, especially in the UV wavelength range. Lots of experiments reveal that laser-induced damage in optical coatings is associated with the presence of micron and sub-micron scale defects [5–8]. A number of techniques have been introduced to minimize these defects on the LIDT of optical coatings, such as the reactive evaporation technique [9], laser conditioning [10], ion post-treatment [11] and so on. Due to its attractive optical properties, such as low absorption of light, high resistance, high transmittance over a wide spectral range even in the near-ultraviolet (UV) region, HfO₂ is widely used as high refraction index material in UV multilayer coatings [12, 13], such as mirrors and optical filters in high power laser systems. The improvement of the laser resistance of HfO₂ films shows great importance for the further development of high power performance.

In our previous work, influence of oxygen plasma treatment on LIDT of ZrO₂ thin films at 1064 nm wavelength was performed, and satisfying results were obtained [14]. In this work, HfO₂ thin films were prepared by e-beam evaporation technique, and the samples were treated with plasma with different flow rate ratio of oxygen to argon gases. The dependence of the plasma treatment on LIDT of the films is investigated.

2. Experimental details

HfO₂ thin films were prepared on BK7 glass substrates at 200 °C by e-beam evaporation in box coater with molecular bump system. The high purity HfO₂ (99.99%) in tablet form was served as coating material. During deposition the vacuum chamber pressure was 2.5×10^-3Pa. Film thickness was controlled by an optical monitor, and all sample thickness was about 1000nm. The deposition rate was about 55nm/min. The assistant ion gun in Veeco spector deposition system was employed to treat the prepared samples with different Ar/O₂ flow rate ratio. The ion beam is irradiated on the sample surface in near-normal direction. The treatment parameters in the experiment are shown in Table 1 and Table 2, respectively. The treatment time is 10 minutes.

Table 1. Ion gun parameters of plasma treatment

View Table | View all tables in this article

Table 2. Flow rate ratios (Ar/O₂) for different samples

View Table | View all tables in this article

The surface defect density of the samples was measured in dark field mode by microscope under 100 × magnifications, and the visible smallest size of the defects is about 3μm in this case. 10 sites were selected randomly on each sample to measure, and the mean value of test results will be regarded as the defect density of the sample. The weak absorption of the samples was measured with the surface thermal lensing (STL) technique. The technique can test the absorption as low as 1ppm, and basic principle of the technique have been described in elsewhere [15, 16]. A single-mode, continuous wave YAG 1064 nm wavelength laser was selected to serve as the pump source. The beam was split so that the power could be measured, modulated at 540Hz with a mechanical chopper wheel synchronized to a lock-in amplifier, and focused on the sample at a near-normal angle of incidence with a beam diameter of 60μm. A 20 mW He-Ne laser was used as the probe beam. The beam was focused onto the sample surface coincidentally with the pump laser beam. The detector was located about 50 cm away from the sample to detect the STL signal; a slit and an interference filter used to reduce scatter noise were placed in front of the photocell. In this way, the detector monitored the intensity change at the center of the probe beam. The intensity is proportional to the optical absorbance of the sample. The calibration coefficient was obtained by measuring the STL signal of a calibrated thin film sample with known absorptance. 10 spots were selected randomly on one sample surface to measure, and the mean value of these test results will be regarded as the absorption of the sample.

Damage testing was performed in the “1-on-1” regime according to ISO11254-1: 2000 [17]. A Q-switched Nd: YAG single-mode laser with a wavelength of 355nm and 8 ns pulse length was used. The laser was focused to provide a far-field circular Gaussian beam with a diameter of 0.467 mm at 1/e ² of the maximum intensity. The sample was set on a three-dimensional precision stage driven by a stepping motor. The angle of incidence was 2°–3° off normal to avoid interference effects due to reflection from the substrate front surface. The laser energy used to damage the sample was obtained by adjusting the attenuator, and the pulse energy from a split-off portion of the beam was measured with an energy meter. A He–Ne laser was used to help in monitoring. Damage onsets were detected through a microscope with a charge coupled device camera. The laser-induced damage threshold was defined as the incident pulse’s energy density when the damage occurs at 0% damage possibility (J/cm²).

3. Results and analysis

 

Fig.1. Surface defect density of the as-grown and treated samples

Download Full Size | PDF

Figure 1 is the surface defect density of the as-grown sample and treated samples. The error bars represent the standard deviation of defect density of 10 sites. The surface defect density of the as-grown sample is about 23 /mm², and the defect density decreased to about 13 /mm² after plasma treatment. The defect reducing effect is very obvious. At the same time, no notable distinguishment was observed for the plasma treatment in reducing surface defect density under different Ar/O₂ flow rate ratios. The reducing of surface defect density is due to the impact of Ar atoms with the defects.

Absorption is one of the important factors of LIDT of thin films. Fig. 2 shows the weak absorption measurement results of HfO₂ thin films after plasma treatment with different Ar/O₂ flow rate ratio. Absorption of the as-grown sample is about 40ppm, and the absorption decreases to as low as 18ppm after treatment when Ar/O₂ flow ratio is 5:25. With the increasing of the Ar/O₂ flow rate ratio, the absorption is increasing. When the flow ratio of Ar/O₂ is 5:15, the absorption value increases to a value as large as that of the as-grown sample and the absorption value is even larger than that of the as-grown sample when the Ar/O₂ ratio is 5:10. There are two different types of plasma in our treatment process: oxygen plasma and argon plasma. Their effect on reducing absorption of the HfO₂ thin film samples is almost converse. As a high-valence metallic oxide, the suboxide composition of HfO₂ is easily formed in the film deposition process. Generally, this suboxide composition has a higher absorption compared to the oxides [18]. Due to the high oxidation of oxygen plasma, the oxygen plasma treatment will repair the oxygen vacancies and modify the sub-stoichiometry defects to some degree. So this process will lead to a corresponding absorption reduction. On the other hand, for the argon plasma treatment process, no chemical reaction occurred to reduce the sample absorption of HfO₂ thin films because argon is an inert gas; Besides this, preferential sputtering effects maybe exist in coating plasma treatment process due to the mass difference of hafnium and oxygen atoms. So HfO₂ molecules could be decomposed to some degree when impacted by argon atoms in the treatment process and some new substoichiometric compositions could be formed. In our experiment, the flow rate of Ar is constant, so the increase of absorption with the decrease of O2 flow rate is reasonable. Due to the fact that the effect of argon and oxygen plasma on reducing absorption of HfO₂ thin films is mutually exclusive, it is reasonable to deduce that an optimal Ar/O₂ flow rate ratio must exist to obtain low absorption HfO₂ thin film samples.

 

Fig. 2. Weak absorption of the samples

Download Full Size | PDF

The LIDT of the as-grown and treated samples at 355 nm wavelength is shown in Fig. 3. The error bars in the figure represent errors induced by the fluence uncertainty of the laser system. We found that the LIDT varied from 1.2J/cm² to 2.7J/cm² for different samples. No obvious increase of LIDT is found after plasma treatment compared to the results of previous work. In our previous work [14], the LIDT of the ZrO₂ thin films at 1064nm increases obviously after treatment with oxygen plasma. The drastic difference of the two results may be related to the damage mechanisms at different laser wavelengths. At 1064nm wavelength of YAG, micrometer scale nodule defects are the dominant cause for the film damage. Oxygen plasma treatment reduces not only the nodule defect density, but also the weak absorption in some degree. Therefore, the LIDT improvement of the sample is reasonable. While at 355nm wavelength of YAG laser, the scale of the harmful defects related to LIDT is as little as subwavelength, and the native absorption of the sample increases rapidly. In this case, the electron-avalanche-ionization could occur easily. So the above two factors maybe the real reasons why the LIDT at 355nm of the samples does not increase accordingly with the decrease of the surface defect density.

 

Fig. 3. Laser-induced damage threshold of the sample at 355nm wavelength

Download Full Size | PDF

Figure 4 shows the typical laser-induced damage morphologies of thin film samples at 355 nm wavelength, and Fig. 4(b) is the magnified image of the box area in Fig. 4(a). The damage morphology shows the different characteristics compared with the same type of thin films at 1064nm wavelength. It is clear that damage area consists of many circular damage scars with a size ranging between 1 and 4 μm. At the same time, we found that most damage scars initiated from a nanometer pit center (the arrow pointed sites in Fig. 4(b)). It is natural to conclude that such nanopits are initiated by subwavelength particulate absorbers in the subsurface layer. The nanometer scale absorber defect is the key point in laser damage at 355nm wavelength.

 

Fig. 4. Laser-induced damage morphology of the sample at 355nm wavelength. (a) Microscopic laser damage morphology near LIDT. (b) Magnified image of the box area in Fig. 4(a).

Download Full Size | PDF

From the damage morphologies, we can deduce that there are two failure stages in the damage process: melting and mechanical cracking. In the first stage, the nanoscale impurity defect absorbs the laser energy, and the temperature will increase rapidly and reach the impurity inclusion melting point. During this process, a tiny cavity around the center will be melted and the nanopits will form. In the second stage, the heat-induced expansion of the impurity inclusion will lead to a rise in stresses. When these stresses reach a critical value, this thermo-mechanical effect will break down the host materials, and the circular damage scars will form.

Table 3. Parameters in simulation

View Table | View all tables in this article

The model, which is most frequently employed in the study of impurity-dominated damage, assumes that the damage sites start with a spherical, absorbing particle embedded in host layers. In order to simplify the problem, it is also assumed that the thermal properties are independent of temperature. The temperature rise (T) as a function of position (r) from the absorber center could be determined by the heat conduction equation [19]. The breakdown criterion used in this model is the melting temperature of the inclusion or the host material. The numerical simulation solution of the heat conduction equation was performed for our case. The simulation parameters are assumed in Table 3. Figure 5 shows the temperature variation with the radial distance from inclusion center. It reveals that the temperature in inclusion center is about 6500 K. When the radius of the melting zone is about 200 nm, the temperature is more than 2500 K. It is easy to deduce that the dimension of the damage pit formed due to the inclusion melting is about 400nm from the above estimation. This agrees well with the result shown in Fig. 4(b).

 

Fig. 5. The relations between the temperature rise of inclusion and host layer and the radial distance from the center of laser irradiation.

Download Full Size | PDF

4. Conclusions

Based on the above experiment results and analysis, some conclusions could be obtained as following: