1. Introduction:

There is considerable interest in advancing laser systems to higher peak power and average energy for a variety of applications including fusion research [1,2], materials science under extreme conditions [3,4] and astrophysics [5,6]. For a number of laser systems, including one that is pivotal to the performance of the National Ignition Facility [7], the dielectric multilayer-coated reflective optics play a significant role in the determination of the maximum allowable power and energy. Hafnia/silica (HfO₂/SiO₂) is perhaps the most promising and widely used multilayer dielectric film design for ultraviolet (UV) applications [8] and it is fluence limited due to nanoscale features that interact with lasers leading to localized energy deposition and subsequent laser-induced damage [9–12]. The damage typically occurs in the high index material, so for this multilayer system hafnia is primarily where damage is first observed; therefore, there is a strong interest in studying laser-induced damage in hafnia films [13–17].

The incorporation of inert gas during deposition in sputtered films is a well-known phenomenon [17–21]. These earlier studies have shown that the incorporation probability of sputtering species tends to scale with the ratio of the mass of the target atom to the impinging ion mass [18]. For example, we have observed significant argon (Ar) incorporation, approximately 6% (atomic %), in the hafnia layers in ion beam sputtered hafnia films, whereas in the silica layers it was much less, closer to ∼2% [16,17,22]. This contrast in inert gas incorporation is attributed to the large difference between the average atomic mass of the targets (Hf: 178 amu, SiO₂: 20 amu) compared to the inert gas (Ar: 40 amu). Xenon (Xe) has an atomic mass (131 amu) that is much closer to that of hafnia than Ar, suggesting that the probability of incorporation should be significantly lower if Xe is used as the working gas; this hypothesis is supported indirectly by studies on other film systems [18,19,23].

The impact of the entrapped gas on laser performance of dielectric films remains elusive; however, there is also some indication in the literature that trapped Ar could induce film ablation when exposed to intense excimer laser irradiation [17,24]. In a recent work, we observed the formation of nanobubbles which consisted predominantly of Ar and identified a pathway where the nanobubbles could serve as laser damage precursors [17,22]. The reduction of noble gas entrapment in dielectric films offers the potential means to suppress these damage prone precursors and improve their laser performance. Here we propose a suppression mechanism by substituting the Ar working gas with Xe to significantly reduce the gas entrapment.

In this study, Xe is substituted for Ar as the working gas in the ion beam sputtered hafnia thin films. While the optical properties of the hafnia films produced by Xe working gas are similar to that of the Ar-deposited coatings, the laser damage resistance of the Xe produced hafnia films to UV ns-laser exposure is significantly higher than those produced by Ar sputtering gas. Chemical, structural, and morphological analysis revealed that the laser performance enhancement is directly related to the reduction of inert gas entrapment and suppression of nanoscale defects within the thin films. Our results provide a fundamental basis for novel strategies of fabricating laser damage resistant thin films for next generation of high power and high energy laser applications.

2. Experimental method:

Hafnia single layer films with a half-wave optical thickness at 355 nm light and 45° angle of incidence are produced by ion beam sputtering (IBS) in a commercially available Veeco Spector-HT system. The IBS deposition was performed in a reactive oxygen environment with a polycrystalline hafnium target with either Ar or Xe as the working gas on both fused silica and silicon substrates. The gun settings were adjusted such that the beam voltage was the same for both cases but the current was optimized to enable gun stability and similar deposition rates, namely, 1.0 Å/s for Ar working gas and 1.2 Å/s for the Xe deposition, respectively [16,17]. The oxygen flow and partial pressure were the same between the two deposition processes. Finally, all depositions were conducted with ample neutralizing electrons via radio frequency neutralizers in the chamber [16].

An ample suite of techniques was employed to fully characterize the optical, structural and chemical properties of the deposited films, both pre- and post- laser exposure. The optical properties of the deposited hafnia films were characterized by a Cary 7000 Universal Measurement Spectrophotometer (UMS). Both the near normal transmission and reflection from the films were measured. The spectral scans were used to determine the dispersion relation using the OptiLayer commercial reverse engineering package (RE) for the index of refraction [25]. The optical bandgap of the films was estimated from the extrapolation of the Tauc plots which were derived from the transmission data assuming scattering is minimal over the measured spectral region.

The chemical composition of the films was analyzed by Rutherford backscatter spectroscopy (RBS), a technique well-suited for obtaining quantitative compositional information on hafnia films [16,17,26]. The 2 MeV He⁺ beam used for analysis is generated from a 4 MV Dynamitron ion accelerator and the ion scattering energies were calibrated using a gold film standard before data collection. The He⁺ is collected at a scattering angle of 175 degrees with a silicon (Si) surface barrier detector. Analysis of the RBS data is performed with the aid of the computer simulation program RUMP [27] to determine hafnia film compositions. All RBS measurements are performed on hafnia films deposited on Si wafers. Due to the proximity of the Xe and Hf peaks in the RBS spectrum we also employed an Energy Dispersive analysis of X-rays (EDX) technique in the environmental scanning electron microscope (ESEM); this was to provide qualitative compositional information to compliment the RBS results.

The crystallinity of the hafnia films is determined by glancing incidence x-ray diffraction (GI-XRD). A Bruker D8 Discover instrument utilizing a Cu source emitting x-rays of 1.5406 Å wavelength is used for diffraction measurements with a glancing angle of incidence of 0.45 degrees. A corundum internal standard and alignment to the zero points for 2-theta, z-height and theta is run prior to data collection. All GI-XRD data was collected by the Eurofins EAG Materials Science LLC team in Sunnyvale, CA.

Grazing incidence small angle x-ray scattering (GISAXS) was measured from hafnia films deposited on both Si and fused silica to resolve the nanoscale morphology of the films [17]. Each sample was secured to the sample holder using scotch tape on the side of the sample in such a way that ∼ 25 mm of the bare film was exposed to a 2 mm (h) and 0.5 mm (v) beam of Cu-α X-rays using a Xeuss 3.0 instrument under vacuum. In each case, the sample was tilted in the out-of-plane scattering angle, α, by 0, 0.2°, 0.4°, and 0.6°, which corresponds to an increase in penetration depth of 0 nm, 2 nm, 10 nm and 60 nm, respectively. The scattered X-rays were collected using a Pilatus3 300k in the “virtual-detector” mode whereby three horizontal frames were collected in the x-direction to expand the q-space; a total of six frames were collected to remove the dead zones on the detector. The GISAXS was also collected from bare Si to understand the background scattering from the instrument. The 2-D images were analyzed using FitGISAXS module for Igor Pro [28].

Cross-sectional transmission electron microscopy (XTEM) measurements were performed with a FEI Titan 80-200 TEM operating at 200 keV. Images were acquired in scanning TEM (STEM) mode using a high angle annular dark field (HAADF) detector. To measure the Xe content in the films, electron energy loss spectroscopy (EELS) in STEM mode was used. We conducted 15 second exposure point analyses with an electron beam convergence angle of 14 mrad and collection angle of 25 mrad.

Samples of hafnia on fused silica substrates were loaded into 5 mm quartz EPR/NMR tubes and placed in the resonant cavity of an X-band Bruker Elexsys E500 CW-EPR spectrometer with an operating frequency of 9.845 GHz. The modulation amplitude was set to 5 Gauss, and all samples were evaluated by EPR spectroscopy at room temperature.

The laser performance of the films was evaluated by conducting laser damage testing via a 1-on-1 test protocol [14,16,17,29,30] using the 355 nm, third harmonic of a Nd:YAG laser system which was operated in a single-longitudinal mode (Quanta-Ray Model PRO-350-10, Spectra-Physics, Inc, injection seeded) with a pulse duration of 8 ns (FWHM) and a near Gaussian temporal profile. All tests were performed at 45 degrees angle of incidence using P in-plane polarized light. The near Gaussian laser beam diameter is formatted to 650 µm at 1/e² maximum intensity. Two hundred damage sites in a rectangular test pattern were separated by 2 mm to ensure minimal shot to shot overlap for adjacent laser exposures. The laser testing was repeated on at least four samples to validate the results. A diagnostic reference arm is used to record the laser energy and spatial beam profile at an equivalent sample plane on every shot. In situ optical microscopy with an approximate resolution of 1 µm is used to collect images before and after laser irradiation to detect any laser-induced surface modifications. More details on the test protocol, beam diagnostics and data analysis can be found elsewhere [16,17]. The overall fluence error is approximately 10-15% and includes the energy meter and beam profile measurement uncertainties.

The surface morphology and damage density of the hafnia films after laser exposure were examined using ESEM and confocal microscopy. ESEM allows for the imaging of dielectric coatings surfaces without a conductive coating by flowing 0.35 Torr of water vapor over the sample, mitigating and dissipating the occurrence of charging on the surface due to the electron beam impinging on the surface. Chemical composition was also performed on this ESEM which has the capability for the Energy Dispersive x-ray Spectroscopy (EDS) technique; a voltage of 10 kV was used.

The simulation software package Transport and Range of Ions in Matter, Sputtering Edition (TRIM.SP) [31] was used to estimate the energy of reflected neutrals as a function of the radial position/location on the substrate holder; the program relies on Monte Carlo simulation techniques and we have successfully used it in past studies. All applicable TRIM.SP inputs have been discussed previously and are the same for this study. [16]

3. Results & discussion:

The hafnia films produced by the Ar and Xe working gas, respectively, exhibit similar optical properties such as the index of refraction and optical bandgap thus warranting their utilization in dielectric coatings for laser applications. The refractive index at 355 nm is 2.09 for the Ar- and 2.18 for the Xe-sputtered films, respectively. The spectrophotometric transmission measurements used to determine the indices in the ultraviolet-visible (UV-Vis) wavelength range are displayed in Fig. 1(a). Apparently, the index of refraction for the films deposited with Xe working gas is ∼4% larger. Since the refractive index generally increases with film density, this result suggests that the Xe produced films are denser than the films using Ar as the sputtering gas [32,33].

 

Fig. 1. (a) Transmission as a function of wavelength from ultraviolet-visible spectrophotometry measurements. This was performed for hafnia films deposited with an argon and xenon working gas, respectively, to determine the index of refraction of each film, which is shown. (b) Tauc plots derived from the UV-vis spectral data collected on the two hafnia films. The optical band gap is estimated from the plot for the hafnia films deposited with an argon and xenon working gas, respectively. The Ar deposited film is denoted by the black traces and the red traces denote the Xe deposited film.

Download Full Size | PDF

The optical bandgap, on the other hand, shows an opposite trend where the Xe produced hafnia film yields a slightly smaller value at ∼5.52 eV, compared to ∼5.58 eV for the Ar produced films. The optical bandgap of the materials was determined by the UV-Vis spectrophotometry measurements shown in Fig. 1(b). While the trend is consistent with previous reports [34–36] in that the index and bandgap are inversely proportional, the overall difference is only ∼1%, suggesting that there is no degradation in optical quality by substituting Xe for Ar as the working gas and the materials in the two hafnia films should have, to a first order, similar response to UV laser exposure (if produced defect-free).

We found the chemical composition of the hafnia films produced by the two different sputtering gases (Ar or Xe) to be largely the same except for the amount of entrapped inert gas. RBS spectra of the hafnia films are shown in Fig. 2(a). Fitting of the spectra yielded ∼1.8% Xe and ∼ 6% Ar for the films deposited with Xe and Ar working gas, respectively. We note that there is a complication associated with the Xe content estimate due to the proximity of the Xe peak to the large Hf peak shown in Fig. 2(a). However, the Hf signal does not extend to the spectral region where the Xe signal from the film bulk is located. This is evidenced by comparing the RBS spectra of the Ar- and Xe-sputtered sample. The partial overlap between Xe and Hf signals is not expected to significantly affect the content determined from RBS. On the other hand, the small bump at the onset region of the RBS spectrum of the Xe sputtered film between channel numbers 850 and 900 may be originated from the high surface roughness caused by the utilization of the Xe working gas. This would suggest that a few percent of the film surface area be covered with highly rough regions, showing a few large islands (e.g., 100×100 nm) or hundreds of small islands (e.g., 10×10 nm), all having a height greater than 100 nm, within the surface area of 1 µm² on average. However, SEM images along with confocal optical images indicated a smooth thin film surface and no trace of nanoscale islands was detected. For the Xe sputtered hafnia films, EDS measurements (data not shown) were also made at several different locations including one at different electron beam energy and no Xe was detectable. In addition to electron beam generated Bremsstrahlung radiation backgrounds that set a lower bound in detection sensitivity, the detection of Xe in the sample with a higher-Z element Hf could be further limited by matrix dependent effects, such as attenuation of Xe x-rays in hafnia and pulse pileups of strong matrix signals [37]. The ∼1.8% of entrapped Xe estimated from RBS may be an upper bound and the actual Xe concentration in the film could be lower, as will be discussed later when EELS results are presented. Regardless, the RBS and EDS results show that the entrapped Xe is significantly lower than the entrapped Ar in the hafnia films.

 

Fig. 2. (a) The chemical composition of the two hafnia films is characterized by the RBS analysis. (b) The structure of the hafnia films produced by Ar or Xe working gas is characterized by the glancing x-ray diffraction. The film deposited with Ar is denoted by the black trace while the Xe deposited film is denoted by the red trace.

Download Full Size | PDF

The two types of the hafnia films exhibit relatively similar microstructural characteristics as suggested by the GI-XRD spectra collected from the hafnia films deposited with Ar or Xe as the working gas (Fig. 2(b)). The GI-XRD spectral data from both films exhibit similar broad peaks at 30 and 51 degrees, respectively, which are consistent with previous observations from amorphous hafnia films [16]. However, the relatively narrower peak at 31 degrees suggests there may be small grains within the film with an average grain size of 1.4 nm by the Scherrer equation [38]. The similarity in GI-XRD measurement indicates that the structure of the hafnia films is not affected by the chemical nature of the sputtering gas under the deposition conditions utilized.

The 2D GISAXS data reveal randomly distributed phases within the film deposited with Ar, as evidenced by the diffuse halo in Fig. 3(a). By comparison, very little diffuse scattering is observed in the film deposited with Xe (Fig. 1(b)). As this halo is isotropic, a horizontal line profile, from the Yoneda peak, [39] was extracted from the images to be fit to a simple scattering model. The q-dependent shape of these profiles does not change significantly with the sample angle (penetration depth). Rather, only an increase in scale is observed (see supporting information, Fig. S1). Therefore, the GISAXS data collected at a sample angle of 0.6 degrees (60 nm depth) is analyzed and shown here, as these data have the highest signal to noise and largest penetration depth. In all cases, the horizontal line profile located between the horizon and 0.03 Å⁻¹ above was extracted using FitGISAXS. The apparent broad peak in the line profile in Fig. 1(a) indicates that a preferred distance between phases is present and therefore a polydisperse system of quasi-crystalline spheres was fit to the data by assuming a log-normal distribution and local-monodisperse approximation for the hard sphere structure factor; a flat background was also included to account for the incoherent scattering from the hafnia [28]. While the line profile collected from the Xe film yields a significantly lower signal, a faint Guinier knee can be observed, which is an order of magnitude higher than the background scattering from the Yoneda wing obtained from blank Si substrate below and above the critical angle. The resulting model fits shown for each line profile are shown in Fig. 3(a) and Fig. 3(b) fit well to the data and the best fit parameters are shown in Table 1. Using relative scaling values and bubble sizes obtained from the fitting, the relative volume fraction of bubbles can be estimated. Details of this calculation can be found in the supporting information, Section S2. Briefly, the volume and contrast contributions to the scaling factors are accounted for using the mean bubble size and the molar volumes of Ar and Xe obtained by RBS. The ratio of the bubble volume fractions in the Xe to Ar films is found to be 1/3. Therefore, far fewer bubbles are found in the film formed with Xe as the working gas.

 

Fig. 3. GISAXS images obtained from the hafnia films deposited with Ar (a) and Xe (b) along with the 1-D profiles obtained from each, which are compared and used in model fitting.

Download Full Size | PDF

Table 1. Table 1 contains the results from the model fitting of the GISAXS data along with the fit error estimated from the least-squares fitting routine in Igor Pro. The “*” superscript denotes cases where the fit error is large due to low signal.

View Table

The local structure of the hafnia films produced by the two different working gases is starkly different at the nanometer scales. A close-up structural examination of the films using XTEM is shown in Fig. 4. For the hafnia films deposited by Ar sputtering, nanometric bubbles are consistently found throughout the film as indicated by the black regions in Fig. 4(a). As determined in a previous report [17,22], these nanobubbles are Ar-rich and appear regularly distributed throughout the hafnia matrix with an average estimated diameter of 1-2 nm. The insert in Fig. 4(a) displays the distribution of these nanobubbles at a different length scale. However, when Xe was substituted for Ar as the working gas a striking decrease in detectable nanobubbles is apparent. At the similar STEM resolution as that used in Fig. 4(a), no visible nanobubbles are observed (Fig. 4(b)). Examination at finer scales do show, however, that minute amounts of nanobubbles exist in the film stochastically (Fig. 4(b) insert). To obtain a more precise estimate of the Xe content in the films, STEM/EELS point analyses were used. It was revealed that the ratio between Xe and O in hafnia is 1 to 99. This indicates that the amount of entrapped Xe is ∼0.7 at% in the hafnia film if an ideal stoichiometric ratio is assumed. The 0.7 at% Xe estimate is in between the values presented earlier using EDS (0%) and RBS (1.8%). This result implies that the 1.8 at% Xe value could be considered an upper bound for Xe entrapment in the hafnia films. It is noted that the different Xe contents measured with RBS and TEM/EELS would reflect the difference in Xe distributions between the macroscopic and nanometer scales. In addition to the Xe entrapped within nanobubbles, individual Xe atoms or small Xe clusters should be included in the Xe content measured with RBS. TEM/EELS could probe a different Xe concentration at the nanoscale structural inhomogeneity.

 

Fig. 4. STEM/HAADF analyses of the hafnia layer after IBS using a) argon and b) xenon as working gas. Inserts are zoomed in views of the two regions at larger magnification.

Download Full Size | PDF

EPR spectra of Ar- and Xe- sputtered hafnia are shown in Fig. 5 by black and red lines, respectively. The red spectrum (dotted line) indicates that a Xe environment substantially decreases the concentration of paramagnetic centers in the material compared to the Ar sputtered hafnia (black solid line), as depicted by the diminution of EPR signal in the red spectrum. Fitting the experimental line shapes in Fig. 5 using the Bruker Anisospinfit module in XEPR software indicates that two paramagnetic centers exist in the material. We assign the two centers to superoxide radical (O₂⁻) and an oxygen vacancy center (O_V) based on the simulated g-factors and previous assignments in similar materials from the literature [40,41]. The EPR spectra indicate that the paramagnetic dipoles of both superoxide and oxygen vacancies have rhombic symmetry and overlap one another.

 

Fig. 5. EPR spectrum collected from hafnia film sputtered by Ar and Xe, respectively.

Download Full Size | PDF

We found the laser-induced damage onset for the Xe sputtered hafnia films to be much higher than that of Ar sputtered films. The probability of damage as a function of the 355-nm laser fluence for both hafnia films is displayed in Fig. 6(a). The laser damage onset fluence is ∼2.2 ± 0.2 J/cm² for films deposited with Ar and is ∼3.1 ± 0.2 J/cm² for films deposited with Xe. This constitutes a nearly 1.0 J/cm² fluence increase which is significant for nanosecond laser applications at UV wavelengths.

 

Fig. 6. (a) Probability of damage as a function of the laser fluence for hafnia films deposited with either argon or xenon as the working gas. (b) Damage density and (c) damage diameter as a function of 1-on-1, p-pol fluence. The performance of films deposited with argon are denoted by black, square data points while the performance of xenon deposited films is denoted by the red, triangle data points.

Download Full Size | PDF

The observed damage densities (Fig. 6(b) along with the size of damage sites (Fig. 6(c)) for the same fluence range above the damage onset show a similar dependence on fluence for both hafnia films. The density, however, is higher for the Xe sputtered hafnia films and the difference becomes apparent at fluences greater than 3.5 J/cm². Similarly, the average size of the individual damage sites is also larger for the Xe produced films at the same exposure laser fluence. While the damage density increases exponentially with fluence, the diameter grows somewhat linearly with respect to the laser fluence.

Figure 7 shows regions in the deposited hafnia films after laser irradiation at higher fluences (5 J/cm²) where damage is known to occur for both films. Figures 7(a)–7(b) are secondary electron images taken by SEM showing the damage sites in a top-down view from hafnia films deposited with Ar and Xe working gases, respectively. The films deposited with Xe show a consistent and significantly different morphology in the damaged regions compared to the films deposited with the Ar working gas. The Ar produced films show a primary circular-shaped damage area with a foamy texture as seen in Fig. 7(a). TEM cross-section images (Fig. 7(c) and 7(e)) confirm this texture originates from the formation and segregation of bubbles at the damage surface. These bubbles were identified as Ar gas inclusions [14,16]. Based on the TEM images, it seems the damage mechanism is consistent with Ar bubbles formation upon temperature rise, damage nucleation, and local hafnia layer failure or vaporization resulting in full removal of material from the damage region. On the other hand, the primary circular-shaped damage sites in the Xe-based hafnia films are surrounded by a bulging ring (Fig. 7(b)). This morphology feature in the Xe produced film is most likely formed by the flow of molten material during damage propagation as seen in the TEM cross sections (Fig. 7(d) and 7(f)). Moreover, this flow results in numerous, larger bubbles trapped in the ring upon solidification. These observations provide evidence of a different damage mechanism in the Xe sputtered hafnia layer where the temperature may not have been high enough to locally vaporize the hafnia but rather melt it.

 

Fig. 7. Scanning electron microscopy images of hafnia films deposited with (a) Ar and (b) Xe working gases after irradiation with 5 J/cm². TEM images of cross-sectional view of the laser damage site at different magnification: (c and e) Ar sputtered hafnia films and (d and f) Xe sputtered hafnia films. All are at the same laser fluence.

Download Full Size | PDF

The observed difference in the amount of entrapped inert gas in the hafnia films can be related to the energy of the reflected neutrals of the sputtering gas species. The radial distribution of the energy of the reflected neutrals estimated by the TRIM simulations for Ar and Xe is displayed in Fig. 8. As we have shown in a previous report [16], the high (∼6%) concentration of Ar entrapped in the hafnia film is in large part due to the impingement of the highly energetic (over 400 eV, black curve) Ar neutrals near the center of the rotational planet in the IBS system. The simulated energy distribution across the sun for Xe reflected neutrals is displayed in the red curve; the overall energy is much smaller than that of the Ar neutrals and is consistent across the substrate, thus across the growing film. The reduction in beam energy for the reflected Xe neutrals as opposed to reflected Ar is expected from the two-body collision model. As demanded by momentum and energy conservations, only neutrals colliding with Hf atoms rather than O atoms can be scattered back to the vacuum. The incident Ar or Xe ions (mass m₁) transfer to the target Hf atoms (mass m₂). The amount of energy transferred is proportional to $\frac{{4{m_1}\ast {m_2}}}{{{{({m_1} + {m_2})}^2}}}$, reaching a maximal at $\frac{{{m_1}}}{{{m_2}}} = 1$ and decreasing as the mass ratio deviates from 1. Compared to the case with Ar ions incident on Hf where $\frac{{{m_1}}}{{{m_2}}} = 0.22$, the mass ratio (0.74) between Xe and Hf is much closer to 1 and consequently, the Xe ions transfer a greater amount of incident energy to the target and would thus carry a much lower energy when reemerging into the vacuum [42]. This lower energy for the xenon is expected to result in less incorporation in the hafnia film which is consistent with what we have inferred above.

 

Fig. 8. The estimated energy of reflected Ar and Xe neutrals as a function of sun position using a Hf target and similar geometries to those used in the deposition. The Ar working gas condition is denoted by the black square data and the Xe working gas condition is denoted by the red triangular data points.

Download Full Size | PDF

We hypothesize the significant reduction of the Xe entrapment in the hafnia films is responsible for the higher laser damage onset and distinct laser damage morphology compared to those of the Ar sputtered hafnia films. Indeed, with much less entrapped inert gas atoms in the films, the overall defect density has been significantly reduced as suggested by the EPR results. Furthermore, the formation of nanobubbles is significantly suppressed or diminished. Since nanobubbles clustered in hafnia films have been shown to be precursors for the onset of UV ns-laser damage [16,17], the lessening of the nanobubbles in hafnia films by Xe sputtering effectively eliminates or drastically affects the lower fluence (near threshold) laser damage precursors and results in the elevated onset of laser damage. In addition, the lack of entrapped inert gas in the Xe produced films removes the pathway for plasma generation at fluences above the damage onset as it occurred in the Ar produced films, thus no abrupt morphological changes should be observed in the film, which is consistent with the results shown in Figs. 7(b) and 7(a), respectively. By the same token, the thermal conductivity should show much less local discontinuities in the Xe produced films as the structural interruption by the nanobubbles is essentially removed. This may be in part responsible for the observed material movement and building feature surrounding the laser damage sites. The fact that the Xe produced films damage at a higher fluence supports the existence of a new type of laser damage precursors. The identification and removal of such high fluence laser damage precursor are beyond the scope of this work and warrant future studies.

4. Conclusion:

We demonstrated that the onset of UV ns-laser induced damage of IBS produced hafnia films can be significantly increased by nearly 1 J/cm² through utilizing a heavier working gas, namely Xe. The replacement of the commonly used Ar working gas by Xe reduces the gas incorporation in the dielectric films from ∼ 6% for Ar to less than 1.8% for Xe. The drastic reduction of the entrapped gas lessens the formation of the nanobubble clusters which has been shown previously to be a class of damage precursors responsible for the observed lower onset for UV ns-laser induced damage in hafnia films. The reduced incorporation of Xe can be attributed to the much smaller energy of the reflected Xe neutrals. While the reduction of Xe incorporation may increase the index of refraction slightly comparing to that produced by an Ar-based process, the influence on the bandgap and structure of the hafnia films is minimal. Our results offer a practical strategy for producing improved laser damage resistance in dielectric films for high power and high energy laser applications.