1. Introduction

Downscaling of semiconductor devices necessitates a thin dielectric gate oxide layer (typically < 2 nm) [1,2]. Traditionally used SiO₂-based gate oxides have reached their operational limit due to a high gate leakage current of ultrathin layers [3]. HfO₂ is a promising alternative to SiO₂ in semiconductor technology, e.g., in metal-oxide-semiconductor field-effect transistors (MOSFET) due to its high dielectric constant and high density [4–6]. Due to the shrinking size of memory devices, HfO₂-based resistive random access memory (RRAM) devices are promising candidates for the new generation of non-volatile memory units because of complementary metal-oxide-semiconductor (CMOS) compatibility, high switching speed, and increased endurance [7]. HfO₂ is also widely used in optical coatings, e.g., in antireflection coatings, mirrors, and optical filters due to its high refractive index in the ultraviolet, visible, and infrared spectral range [8–11]. HfO₂ has a high melting point, as well as high thermal and chemical stability; hence it is widely studied experimentally and theoretically [12–14].

A low deposition temperature is often desired for optical applications to obtain relatively thick, amorphous, and smooth layers. High-power laser applications mandate thin films with low impurities to guarantee low optical losses and a high laser-induced damage threshold (LIDT) [15]. The current development of quantum technologies relies on optical components that put even more stringent requirements on thin film properties and optical performance. Traditionally, HfO₂ thin films are deposited by physical vapor deposition [16–20] or chemical vapor deposition techniques [17,18]. However, for optimal performance of nanostructured components or strongly curved substrates, conformal and uniform coatings are demanded. Since atomic layer deposition (ALD) is based on self-limiting surface reactions and cyclic layer-by-layer growth with alternately pulsed precursors into the reaction chamber, it provides precise thickness control, a uniform coating on large-area substrates [21–25] and conformal growth on 3D and high aspect-ratio surfaces [22,26–29].

In conventional thermal ALD processes, the surface reactions are activated by heating the substrate [29–31]. Various metal-organic precursors with functional groups like halides [32–36], alkyls [36–38], alkoxides [36,39–41], amides [36–38,42–45], β-diketonates [36–38], and cyclopentadienyls [36–38,46–48] have been used to develop thermal ALD HfO₂ thin films. High deposition temperatures are required to break the relatively strong metal-halide and metal-oxide bond in HfO₂ precursors containing halide and alkoxide functional groups [38]. Additionally, the by-product resulting from the metal halide reaction with surface functional groups are highly corrosive and might damage the substrate or contaminate the film [38]. The precursors containing β-diketonates and cyclopentadienyl functional groups are relatively unstable at low temperatures. Moreover, they are bulky and cause steric hindrance [38]. In contrast, amide precursors are volatile and relatively reactive at low deposition temperatures [38]. Hence, they have found broad interest in the growth of HfO₂ thin films. Thermal ALD processes have been reported for various alkyl amide precursors and oxidizing agents such as H₂O [42–45,49–54] and D₂O [55,56], H₂O₂ [57,58], molecular O₂ [59,60], or ozone [61–64].

High carbon and nitrogen impurities have been observed in HfO₂ thin films deposited by thermal ALD at high deposition temperatures (>300°C) due to decomposition when using amide precursors [43,45,65,66] and at low deposition temperatures (<200°C) due to insufficient activation energy of precursor reactions [60,64,67]. At high temperatures, the chemical bonds break as the energy of surface species exceeds the activation energy. On the other hand, a high hydrogen content has been observed in the films deposited using an alkyl amide precursor and water at low deposition temperatures due to incomplete surface reactions [45,66,68,69]. The hydrogen impurities can be reduced by introducing longer purging durations after the alkyl amide precursor pulse or/and by post-deposition annealing [49,68,70]. In this context, plasma enhanced atomic layer deposition (PEALD) can provide dielectric films with high purity, high density, and high refractive index at lower deposition temperature compared to thermal ALD [30,69,71,72]. In PEALD, highly energetic plasma reactants are used as an oxidizing agent, enabling a low-temperature deposition with good thin film quality. Plasma enhanced ALD processes of HfO₂ using metal-amides, metal-cyclopentadienyls, metal-alkoxides, and metal β-diketonates have previously been reported [9,13,24,26,28,36,50,64,73–79].

In this study, we present a thorough approach in the development of HfO₂ thin films grown by PEALD using tetrakis-dimethylamino-hafnium (TDMAH) and oxygen plasma. The influence of the deposition temperature and plasma parameters were studied. So far, structural, optical, and mechanical properties of HfO₂ layers deposited by PEALD have not been systematically investigated in the literature. Most ALD process development articles report on optimizing the pulse/purge durations of the precursors towards a constant growth rate per cycle (GPC) [64,74,76,78]. Detailed optimization procedures to achieve an optimum thickness distribution, i.e., minimum non-uniformity (NU), are rarely considered. Due to the strong influence of the reactor geometry on the NU, specific gas flow rates and pulse durations are required for each reactor design [80]. Thin film properties strongly depend on deposition parameters like temperature, plasma conditions, reactants, gas flow rates, and pulse and purge durations. To meet the optical coatings requirements with respect to optical constants, surface roughness, impurities, and optical losses, a reactor-specific process development and an in-depth investigation of the thin film properties are required. Furthermore, we discuss the impact of different plasma configurations on these properties to prepare an optimal basis for the incorporation of hafnium oxide thin films in optical interference coatings.

High power laser systems warrant high-resistance coatings on optical elements. Conventionally these coatings are prepared using e-beam evaporation [81–87], ion assisted deposition [82,84,87], ion beam sputtering [88,89], magnetron sputtering [84,88,90], RF sputtering [91], and sol-gel methods [84,92,93]. In a thin film laser damage competition held at the Boulder Damage Symposium 2010, antireflection coatings manufactured with six different techniques were submitted and tested at 355 nm laser wavelength with a 7.5 nm pulse length [84]. It was shown that the coatings exhibited a high laser induced damage threshold (LIDT), but these values strongly depend on the selected materials and the deposition method.

Recently, ALD has raised interest for high power laser applications due to its ability to coat 3D substrates with precise thickness control, good uniformity and conformality [88,94–96]. Although the LIDT of ns-pulsed lasers is mainly defect driven (particles, defects, etc.), the laser beam parameters like wavelength, beam diameter, or the absorption in the film also play a significant role. The absorption in the film can cause thermal effects where heat diffusion resulting from the high energy of lasers increases laser damage probability and lowers LIDT. We demonstrate highly efficient and low absorption antireflection coatings consisting of hafnium oxide, silicon oxide, and aluminum oxide prepared by PEALD. The antireflection coatings were designed for the fundamental wavelength of a Nd:YAG laser at 1064 nm and the higher harmonics at 532 nm, 355 nm, and 266 nm.

2. Experimental techniques

Plasma enhanced ALD was employed to deposit hafnium oxide thin films. Tetrakis-dimethylamino hafnium (TDMAH) [Strem Chemicals Inc., Bischheim, France] was used as a metal-organic precursor and oxygen as a plasma gas using a Silayo ICP330 equipment [Sentech Instruments GmbH, Berlin, Germany]. A detailed description of the deposition tool is given elsewhere [97,98]. Various substrates were used to satisfy the requirements for the experimental investigations. For the initial optimization of the non-uniformity and control of the GPC, 9 single-side polished, p-doped and (100) oriented Si substrates were arranged in a concentric circular pattern on the heating plate in the reactor: 1 in the center, 4 at 100 mm diameter, and the remaining 4 at 200 mm diameter around the center of the heating plate. The UV/VIS spectroscopy, LIDT, and the absorption loss measurements were carried out on HfO₂ thin films deposited on vertically (double-side coating) and horizontally (single-side coating) placed fused silica substrates. For the atomic force microscopy (AFM), X-ray diffraction (XRD), X-ray reflectometry (XRR), Fourier transform infrared (FTIR) spectroscopy, spectroscopic ellipsometry (SE), and Auger electron spectroscopy (AES) analysis, silicon substrates with a native SiO₂ layer were used. Furthermore, 75 mm Si (100) double-side polished wafers were employed for determining the mechanical stress.

The process optimization was carried out applying 200 ALD cycles. Further, the effect of temperature and plasma parameters on the properties of HfO₂ were investigated on thicker films deposited with 1482 ALD cycles, whereas the density of the films was investigated from thin HfO₂ films grown with 184 ALD cycles (∼20 nm thickness). The film thicknesses and optical constants were determined by spectroscopic ellipsometry using the 850 spectroscopic ellipsometer and the SpectraRay 3 software [Sentech Instruments GmbH, Berlin, Germany]. The measurements were performed in the spectral range from 200 to 980 nm wavelength. The Kramers-Kronig consistent Tauc-Lorentz dispersion model using three oscillators was used to fit the refractive index and the extinction coefficient. For film thicknesses over ≈100 nm, an additional effective medium approximation (EMA) layer using the Bruggemann model was added to improve the fit by taking into account the surface roughness into the model. The fraction of void inclusion was assumed to be 0.5. During the process optimization, the thickness non-uniformity (NU) across 200 mm diameter area was estimated by (d_max-d_min)/2d_average, whereby d_max, d_min and d_average are the maximum, minimum and average thickness of HfO₂ films, respectively. Additionally, optical constants of HfO₂ films deposited on fused silica substrates have been estimated via spectrophotometry by modeling the transmittance and reflectance spectra using the LCalc software with a Lorentzian multi-oscillator dispersion model [99]. The model consists of fused silica substrate, HfO₂ thin film, and a roughness layer with 50% void inclusion. The hafnia dispersion curves were modelled by using 7–9 Lorentzian oscillators.

The reflectance (R) and transmittance (T) spectra were recorded with a Lambda 950 spectrophotometer [PerkinElmer Inc., Waltham, MA, USA] in a wavelength range from 200 to 1200 nm and at an angle of incidence at 6°. The optical losses (OL), the sum of the scattering and the absorption, were calculated from the R/T data using OL = 100 - R - T. For validation of the film thickness, x-ray reflectometry (XRR) was performed using a Bruker D8 Discover system [Bruker AXS, Karlsruhe, Germany]. The mechanical stress was determined by measuring the change in curvature of a double-side polished Si (100) wafer (400 µm thickness, 75 mm diameter) before and after deposition [Model FLX-2320, KLA-Tencor, San Jose, CA, USA]. The stress was calculated using the Stoney equation [100]:

The laser-induced damage threshold (LIDT) was measured at Layertec GmbH under normal incidence using a Nd:YAG, frequency-tripled, Q-switched LITRON NanoTRL-650-10 laser [Litron Lasers, Rugby, UK] at 355 nm with 7 ns pulse duration (FWHM), 10 Hz repetition rate and approximately 240 µm beam diameter (1/e²). Because of the limited test area available, the R-on-1 test method was used. This testing method is useful for production monitoring as well as comparison of manufacturing processes for vendors. In this investigation, each test site was irradiated with 10 pulses of 5 J/cm² initial fluence. The pulse burst was reapplied with fluence increments of 5 J/cm² until damage occurred. Damage onset was detected in-situ using the scattering method. For each test sample, roughly 40 test sites were irradiated to ensure statistical reliability. From these results, damage probability over fluence was calculated. Each pulse was interpretated as an individual test. This way, the number of available datapoints, i.e., pulse fluence and the information whether or not damage occurred, greatly increases and the statistical error is further reduced. The damage probability P for a given fluence F is then determined using equation: P(F) = n₁(F)/(n₁(F) + n₀(F)), where n₁(F) is the number of damaged test sites and n₀(F) is the number of undamaged ones at that fluence. In accordance with ISO 21254, a linear regression of P(F) in the transition region from 0 to 100% damage probability was performed and LIDT was determined by the intersection of the fit function and the fluence axis. However, due to aging and conditioning effects, the R-on-1 regime results may differ from those in the S-on-1 regime.

Absorption losses were measured at 355, 532, and 1030 nm wavelengths via photothermal common path interferometry (PCI), using a custom-built set-up of Layertec GmbH. In PCI, optical layers are irradiated by a pump beam, and part of the incident radiation is absorbed by the layers. Due to thermal conduction, the absorbed energy propagates from the layers into the underlying substrate, creating a thermal lens. The phase-front deviations of a probe beam resulting from the thermal lens were interferometrically measured. By employing a modulated pump beam and a lock-in amplifier, resolution in the sub-ppm range can be achieved.

3. Results and discussion

3.1 Process development

The initial development of the hafnium oxide processes aimed at finding adequate deposition conditions for the precursor pulse and purge durations, whereby a low thickness non-uniformity (NU) and a constant growth per cycle (GPC) were desired. Therefore, a series of processes varying different parameters like TDMAH pulse duration, carrier gas (Ar) flow rate, purge duration after TDMAH pulse, and O₂ plasma pulse duration were carried out at 100°C deposition temperature and 200 ALD cycles. Additionally, the focus of the work was to keep the total cycle time as short as possible without compromising the film quality. The saturated growth rate and thickness uniformity of PEALD films depends on several factors such as reactor type, precursor exposure time, reaction mechanisms, and sticking probabilities [22]. The GPC and NU values obtained in the initial development processes are summarized in Figs. 1(a)–1(d).

 

Fig. 1. The effect of process parameters on the non-uniformity (NU) HfO₂ thin films on a 200 mm diameter surface and growth per cycle (GPC) deposited varying a) the precursor (TDMAH) pulse with a carrier gas (Ar) flow rate of 80 sccm b) the carrier gas (Ar) flow rate with TDMAH pulse of 3.12 s c) the precursor (TDMAH) pulse with a carrier gas (Ar) flow rate of 160 sccm, and d) the precursor and plasma purge with precursor (TDMAH) pulse of 3.12 s and Ar flow rate of 160 sccm. All thin films were deposited at 100°C with 200 ALD cycles. The sequence of the PEALD cycle is given where x denotes the varied parameters.

Download Full Size | PDF

The process optimization started with a constant argon gas flow rate of 80 sccm while the precursor pulse duration was varied from 0.62 to 4.12 s (Fig. 1(a)). The durations of the plasma pulse, precursor purge, and plasma purge were kept constant at 5 s. At a TDMAH pulse of 0.62 s, a relatively high NU (18.4%) and a GPC of 1.88 Å/cycle were observed. This indicates the absence of the self-limiting surface reaction regime and poor gas distribution in the reactor. The lowest NU of 3.4% was observed at 3.12 s TDMAH pulse. Additionally, at this TDMAH pulse duration, the GPC reached the highest value of 2.55 Å/cycle. Although the NU decreased significantly at 3.12 s, the high GPC may be due to a lower film density and the formation of voids resulting in higher incorporation of impurities. The gas flow rate of 80 sccm (TDMAH pulse < 3 s) resulted in a thickness gradient across the surface area (see Supplement 1 Fig. S1). A higher thickness was observed close to the precursor inlet and lower thickness at the maximum distance from the inlet. It is important to note that the same gas flow rate is also used for purging reactant by-products after the TDMAH pulse. Hence, a carrier (and purging) gas flow rate of 80 sccm and 3.12 s of TDMAH pulse may be sufficient to cover the surface but inadequate to efficiently remove reactant by-products for this reactor design [45].

In the second optimization phase (step 2), the TDMAH carrier gas flow was varied (Fig. 1(b)) while the durations of the precursor pulse, the plasma pulse, the precursor purge, and the plasma purge were 3.12 s, 5 s, 5 s and 5 s, respectively. Due to the optimal distribution of the precursor and effective flow rate during purging, the lowest NU of 0.5% was achieved for 160 sccm argon gas flow and 3.12 s precursor pulse duration. The average GPC decreased from 2.58 Å/cycle to 1.92 Å/cycle. While a relatively constant GPC was obtained in the first optimization step with the precursor pulse variation, a significant influence of the gas flow rate on the GPC and NU is observed in the second optimization phase. These results are critical towards ALD process optimization since gas flow optimization is rarely considered in the literature. Further details on the film properties were not discussed here, since the NU is already a main process selection criterion in the development of optical and semiconductor devices.

In step 3, considering the optimized GPC and NU obtained from step 1 and step 2, the precursor pulse duration was varied again from 0.62 s to 4.12 s, using the higher Ar flow rate of 160 sccm (Fig. 1(c)). The durations of the plasma pulse, precursor purge, and plasma purge were 5 s. The NU decreased significantly with increasing TDMAH pulse duration and attained the lowest value of 0.5% at 3.12 s of TDMAH pulse. This indicates a uniform distribution of the precursor molecules in the chamber and hence uniform surface coverage of the substrate. The GPC increased significantly from 1.51 Å/cycle to 1.88 Å/cycle on increasing the duration of TDMAH pulse from 0.62 s to 4.12 s. The low GPC at lower precursor pulse duration may be due to the unsaturated surface as the result of an insufficient TDMAH dose. The surface saturation occurs when the duration of the TDMAH pulse increased beyond 1.12 s, and consequently, a higher GPC of 1.88 Å/cycle is observed. A similar trend was reported by Kim et al. for HfO₂ films deposited at 80°C [64]. The GPC of 1.88 Å/cycle at 3.12 s TDMAH pulse is lower compared to the GPC obtained using 80 sccm Ar gas flow rate. This may be the result of a more efficient removal of residual impurities after the TDMAH pulse or loosely bound precursor residuals.

Finally, the durations of the precursor and plasma purge times were decreased from 5 s to 2 s to investigate the shortest purge duration required for the complete removal of reaction by-products (Fig. 1(d)). Shorter purge durations are desirable in ALD processes for the sake of cost and time efficiency. It is important to note that, in this investigation, the duration of the plasma pulse was reduced to 3 s as compared to 5 s used in the optimization steps 1–3, as it has a negligible impact on the NU and GPC. However, the detailed investigation of the effect of plasma pulse duration on material properties of HfO₂ will be discussed in a later section.

Initially, the plasma purge duration was kept at 5 s, and the precursor purge duration was varied. With decreasing precursor purge duration from 5 s to 4 s, the GPC remained reasonably constant at 1.89 Å/cycle, whereas the NU increased slightly from 0.5% to 0.7%. For a precursor purge duration below 4 s, another minor increase of the NU to 0.8% was observed, whereas the GPC increased significantly, indicating an increase of impurities and inefficient removal of precursor residuals. Hence, the precursor purge of 5 s was considered as an optimum parameter, and the effect of the plasma purge duration was investigated.

For the plasma purge duration of 5 s and 4 s, the GPC and NU remained reasonably constant at 1.90 Å/cycle and 0.5%, respectively. On further decreasing plasma purge duration below 4 s, there was a slight increase in GPC, whereas a significant increase in NU was observed. However, NU remained below 1%. The higher GPC below 4 s of precursor purge and plasma purge indicate incomplete removal of residual precursor molecules and reaction by-products.

Consequently, for the optimized process for PEALD hafnia using TDMAH and oxygen plasma in the Silayo Sentech ICPEALD tool, the following process parameters were considered: 3.12 s precursor pulse, 160 sccm Ar carrier gas flow, 5 s precursor purge, and 5 s plasma purge. Using the above-mentioned optimized parameters and 5 s plasma pulse, the GPC of HfO₂ thin films was found to be 1.88 Å/cycle, and thickness NU was 0.5% over a 200 mm diameter surface.

The HfO₂ thin films deposited using various precursors, oxygen sources, and deposition parameters have been summarized by Blaschke et al. and Sharma et al. [28,45]. The GPC value observed in our study was significantly higher than 1.50 Å/cycle, as reported by Kim et al. using the same precursor and oxygen plasma at 100°C [71]. This may be due to the higher reactivity of plasma species generated by the uniquely designed planar triple spiral antenna (PTSA) source [97,98,101]. Additionally, there was a constant supply of Ar and oxygen during the entire duration of deposition. The generated plasma consists of O and Ar radicals and ions. Increased GPC of HfO₂ films in the presence of Ar during the O₂ plasma pulse was reported earlier [76]. It was shown that HfO₂ thin films deposited using tris(dimethylamido) cyclopentadienylhafnium (TDMACpHf) and Ar/O₂ plasma had a 1.6 times higher GPC compared to pure oxygen plasma. A fraction of argon in the mixed plasma is excited to metastable Ar* or ionized Ar⁺ by electron collisions. The non-radiative momentum transfer after collisions of Ar* atoms with electrons enhances the kinetic energy of low energy electrons of the O₂ plasma. Hence, the addition of Ar increases the amount of high-energy electrons in the plasma, which further increases the amount of oxygen radicals and thereby the GPC [76,102]. It is important to emphasize that the GPC and NU depend on the reactor configurations, even when the same chemicals and similar deposition temperatures are employed. The pressure conditions, which are also influenced by the purge gas flow rates, pumping configurations, etc., will affect the surface functional groups and consequently the chemical reactions.

3.2 Influence of the deposition temperature

It is widely known that thin films of several high refractive index oxides may undergo significant morphological changes with increasing film thickness. The deposition temperature also critically influences the film morphology. Hence, the properties of a thin (≈30 nm/184 ALD cycles) and a thick (≈250 nm/1482 ALD cycles) HfO₂ film were investigated. HfO₂ samples were deposited at a temperature ranging from 100 to 250°C using 160 sccm carrier (Ar) gas flow. Figure 2 shows the variation of the NU and GPC of the HfO₂ films deposited using 1482 cycles and an ALD cycle sequence of 3.12s/5s/3s/5s as function of the deposition temperature. The HfO₂ film deposited at 100°C shows a GPC of 1.80 Å/cycle and a NU of 1.3%. Noteworthy, the NU deteriorated slightly for the thick film compared to the thin film. There is also a slight variation of the GPC, although all process conditions are kept constant. This could be related to the influence of the substrate and reactor conditioning, or to the precursor conditioning. The GPC significantly decreases to 1.71 Å/cycle, and the NU slightly increases to 1.6% at a deposition temperature of 150°C. On further increasing deposition temperature to 200°C, the GPC increases slightly to 1.78 Å/cycle, whereas the NU increases to 1.8%. At the highest deposition temperature of 250°C, the GPC increases further and reaches a maximum value of 2.05 Å/cycle, whereas the NU decreases slightly to 1.7%. The NU for thick films is in the range of 1.3% to 1.8% and has not been further optimized since for antireflection coatings rather thin films (d<50 nm) are required (see section 3.4).

 

Fig. 2. The effect of deposition temperature on NU and GPC. The thin films were deposited using 1482 ALD cycles, 160 sccm Ar flow and an ALD sequence of 3.12s/5s/3s/5s.

Download Full Size | PDF

A similar result of decreasing GPC with increasing deposition temperature from 230°C to 350°C was reported by Heil et al. for PEALD HfO₂ films deposited using tetrakis(ethylmethylamido)hafnium (TEMAH) precursor and O₂ plasma [78]. The slightly higher GPC at low deposition temperature was attributed to a low film density and a high H content. Hence, the films deposited at 100°C in this study may also possess higher hydrogen content originating from the incorporation of residual OH groups. Due to a higher film density, the H content could have decreased when the temperature was increased to 150°C (Table 1). A similar parabolic dependence of the GPC on the deposition temperature of HfO₂ thin films deposited by thermal ALD (TALD) using TDMAH and H₂O has been observed [49]. The minimum GPC of 1.0 Å/cycle was found at 250°C for HfO₂ deposited by thermal ALD for TDMAH + H₂O and a GPC of 1.2 Å/cycle using TEMAH + H₂O at 275°C. Hence, it is important to mention that the ALD window for TALD HfO₂ films using TDMAH precursor is at lower temperatures compared to TEMAH precursor, which is critical for precursor selection for low-temperature deposition. Additionally, it is well known that the temperature window of thermal ALD can be shifted to lower deposition temperatures by using PEALD [71]. The high GPC of HfO₂ at the deposition temperature of 250°C may occur either due to an increased surface reactivity and/or precursor decomposition [46]. This increase in the surface reaction rate may also result in an altered surface morphology as observed in AFM and SEM images (Figs. 3(a)–3(d)).

 

Fig. 3. AFM and SEM images of the film deposited at 100°C and 250°C with 1482 ALD cycles; a) AFM image of a hafnium oxide thin film, d ≈ 273 nm, deposited at 100°C with a roughness RMS = 0.5 nm; b) AFM image of hafnium oxide deposited at 250°C with d = 309 nm and RMS = 7.3 nm; SEM images of hafnium oxide at c) 100°C and d) 250°C, respectively; crystalline grains are observed at 250°C while the films grow rather smooth at 100°C. The thin films were deposited using an ALD sequence of 3.12s/5s/3s/5s.

Download Full Size | PDF

Table 1. Thin film properties of the HfO₂ deposited at 100°C - 250°C.^a

View Table | View all tables in this article

The HfO₂ film deposited at 100°C (Figs. 3(a) and 3(c)) and 150°C shows a smooth surface with nearly constant root-mean-square (RMS) surface roughness (Table 1). It increases linearly from 0.5 to 10.1 nm on increasing deposition temperature from 100°C to 200°C (Table 1) due to the onset of crystallization, as evident from the XRD pattern shown in Fig. 4. However, at the deposition temperature of 250°C, the roughness decreases to 7.25 nm due to the coalescence of islands [45] (Fig. 3(d)) and formation of laterally oriented grains of different orientations [28,33,53] confirmed in the XRD pattern shown in Fig. 4. HfO₂ is well known to be prone to crystallization when deposited at elevated temperatures [28,45,49,53,61,64,68,74,96]. The SEM image in Fig. 3(d) shows polycrystalline HfO₂ with the grain size of 22 to 25 nm (calculated using the Scherrer equation from XRD data).

 

Fig. 4. XRD diffractogram of HfO₂ films (1482 ALD cycles) deposited at deposition temperature from 100°C to 250°C. The peak positions corresponding to tetragonal (ICDD: PDF card No. 08–0342), orthorhombic (ICDD: PDF card No. 21-0904) and monoclinic (ICDD: PDF card No. 43-1017) phase are given as reference. The thin films were deposited using an ALD sequence of 3.12s/5s/3s/5s.

Download Full Size | PDF

The effect of deposition temperature on the surface roughness due to crystallization is further supported by the XRD measurements shown in Fig. 4. The HfO₂ films deposited at 100°C and 150°C show a broad feature around 2θ = 32°, indicating amorphous films [28,45,64]. At a deposition temperature of 200°C, the onset of crystallization occurs. At the deposition temperature of 250°C, intense peaks with different phase orientations are visible, indicating an increase in the degree of crystallization. Contrary to previous studies, the microstructure transition of HfO₂ thin films from amorphous to polycrystalline at deposition temperatures below 250°C in this study may be due to increased surface energy via momentum transfer of high energetic plasma species in PEALD [27,103]. Hence, a balance between the thermal energy and the energy of the plasma species must be found to grow smooth but pure and dense thin films. The increase in crystallization with an increase in deposition temperature also influences the density of HfO₂ films.

The density of HfO₂ films deposited at 100°C is 7.7 g/cm³ as determined by XRR. A similar value was reported by Provine et al. for HfO₂ films deposited at 200°C [100] by TDMAH precursor and O₂ plasma [59]. The density obtained here at 100°C is slightly higher than the reported value of 7.3 g/cm³ for HfO₂ film deposited at 230°C using TEMAH and O₂ plasma [78]. Kim et al. have reported a density of 8.1 g/cm³ for HfO₂ film deposited at 80°C using TEMAH and oxygen plasma. On increasing deposition temperature from 100 to 250°C, the density of the HfO₂ films increased and approached the bulk value of 9.68 g/cm³, which is comparable to the densities observed in HfO₂ films deposited by IBS, IAD, PIAD with xenon as working gas and ion plating [104,105]. A similar effect of an increase in density with increasing deposition temperature has been reported earlier [28,74,78]. The increase in density has been attributed to decreased C, N, and H impurities in the film. The hydrogen impurities in the films are mainly present in the form of hydroxyl groups. Therefore, on increasing substrate temperature, the peaks corresponding to isolated OH groups decreased significantly, as shown in the FTIR spectra (Fig. 5). Additionally, the impurities, such as carbon and nitrogen in the HfO₂ films, act as defects and result in deteriorated film properties [64]. The decrease in the impurity content with increasing deposition temperature (Table 1) is confirmed by AES measurements. The carbon content in the films decreased from 12.1% to 5.2%, and the nitrogen content decreased from 3.6% to 1.2% on increasing deposition temperature from 100°C to 250°C. Although the impurities content decreases in HfO₂ films deposited at a higher temperature, the absorption (Table 2) increases significantly, possibly due to defects in the film in the form of oxygen vacancies [94]. Noteworthy, the density could be increased at 100°C by applying a longer plasma pulse.

 

Fig. 5. The region of FTIR spectra corresponding to -OH stretching for the HfO₂ films (1482 ALD cycles) deposited at a temperature from 100°C to 250°C.

Download Full Size | PDF

Table 2. Thin film properties of HfO₂ films (1482 cycles) deposited at 100°C to 250°C.

View Table | View all tables in this article

The residual tensile stress of HfO₂ films increases from 588 MPa to 916 MPa with increasing deposition temperature from 100°C to 250°C. The residual intrinsic stress in thin films can be influenced by several factors like thickness, density, grain growth, grain boundary generation, surface roughness, voids, and impurities [81,106,107]. As the deposition temperature increases, crystallization occurs, as confirmed by larger grains observed in Fig. 3(d). The change in the structure and morphology of ALD HfO₂ with the increase in deposition temperature is similar to the effect of annealing temperature on HfO₂ films reported by Shen et al. [108]. As the temperature increases, the crystallite grain size increases, and the grain boundary area decreases. Since the density of the film at the grain boundaries is low, the reduction in grain boundaries results in the lateral film contraction, which leads to an increase in tensile stress [109,110]. This can be further explained in terms of the total energy change of the film. The two main contributions leading to the change in the total energy of the film are the free surface energy of the crystallites and grain boundary free energy. As the grain growth in the amorphous matrix proceeds, crystallites snap together to form grain boundaries; therefore, the energy gained by grain boundary formation between the crystallites replaces their free surface energy; thereby leading to high tensile stress [111,112].

Additionally, a decrease in micro defects and pores containing OH (loosely bound hydroxyl groups or H₂O, which are not entirely removed at low temperature) can lead to an increase in tensile stress. This is confirmed by the reduction in OH content with an increase in deposition temperature, as shown in Fig. 5. A similar observation of decreasing OH content with increasing deposition temperature for PEALD HfO₂ was previously reported [9]. Thin films with incorporated water mainly possess compressive stress due to the repulsion of the permanent dipole moments of the adsorbed water molecules on the walls of the pores [113–115]. On increasing the deposition temperature or a post-deposition annealing, the films densify, and porosity decreases. As a result, the adsorbed water and OH amount decreases with increasing temperature to 250°C, and consequently, tensile stress increases from 588 MPa to 916 MPa [9,49,116]. The change in density on varying deposition temperature is also reflected in the refractive index of HfO₂ films (see Table 1).

A relatively low refractive index of 1.92 (at 633 nm wavelength) was observed for the HfO₂ films deposited at 100°C (Table 1 and Fig. 6(a)). At a deposition temperature of 250°C, the refractive index reaches the maximum value of 2.07 as the consequence of increased density in accordance with previous studies [68,96,117]. The absorption edge of HfO₂ thin films deposited at higher temperatures shifts towards lower wavelengths as seen by the shift in the extinction coefficient k towards lower values (Fig. 6(b)). It indicates an increase in optical band gap with increase in deposition temperature. The Tauc-Lorentz analysis of the ellipsometry data in contrast, indicates increasing extinction coefficient with increasing temperature. It is important to note that the increase in extinction coefficient already starts below 400 nm wavelength for HfO₂ films deposited at 200 and 250 °C which can result in high absorption and high scattering losses (data not shown). Defects like oxygen vacancies in HfO₂ thin films can increase absorption losses, which significantly affects the laser-induced damage threshold (LIDT) [96,118]. Absorption losses determined with high accuracy at selected wavelengths are summarized in Table 2. The absorption for all HfO₂ thin films at 1030 nm before annealing was 2 ppm, which is comparable to 3 ppm reported by Liu et al. at 1064 nm for the HfO₂ films deposited using TEMAH and H₂O at 200°C [95]. The actual absorption was reported to be between 1 and 2 ppm for the thermal process as the slightly higher absorption of 3 ppm was contributed by the fused silica substrate. The lowest absorption is observed for PEALD HfO₂ films deposited at 150°C at all three wavelengths (Table 2). The absorption in the HfO₂ films increases with increasing deposition temperature and annealing, which may be attributed to the increase in crystallization. As the crystallization increases, the absorbing defects at grain boundaries and the substrate interface also increase, thereby increasing the absorption [81].

 

Fig. 6. Dispersion of HfO₂ thin films (1482 cycles) deposited at different deposition temperatures determined from a) ellipsometry and b) transmission-reflection measurement using UV-VIS spectroscopy. The thin films were deposited using an ALD sequence of 3.12s/5s/3s/5s. Details of the analysis are provided in Supplement 1.

Download Full Size | PDF

The LIDT measured for HfO₂ deposited at 100°C is not shown because the film was entirely removed at minimal fluence values. For HfO₂ films deposited at 150°C and 200°C with a post process annealing for 4 h at 400–450°C, an LIDT value of 8 J/cm² was measured. Jena et al. showed that the LIDT at 532 nm and 1064 nm laser wavelength of annealed HfO₂ thin films increased with increasing density [81]. It was shown that increased density leads to low thermal barriers and superior thermal conductivity and, thereby, improvement of LIDT. However, LIDT decreases due to an increase in absorbing defect concentration with increase in deposition temperature. In this study, the LIDT decreases to 2 J/cm² for HfO₂ deposited at 250°C and annealed for 12 h at 400°C. This may be due to an increase in crystallinity at elevated temperatures and will be investigated in detail in the future. Wei et al. have shown that the reflection and refraction of the incident laser can occur at the grain boundaries where a higher concentration of defects is present, which increases the absorption due to the increased optical path [119].

Few articles report on the LIDT values of HfO₂ at 355 nm wavelength. For HfO₂ films deposited using e-beam evaporation, Zhang et al. have shown that nanometer-scale absorber defects are a crucial factor for the LIDT at 355 nm [83]. Also oxygen deficiency creates sub bandgap states which increases coupling of laser energy and thereby reduces LIDT [118,120]. Gallais et al. have shown a comparison of the LIDT at 355 nm for HfO₂ films deposited using different deposition techniques. The reported LIDT values at 355 nm were 2.1 J/cm² (electron beam deposition using HfO₂ source - EBD-HfO₂), 2.8 J/cm² (electron beam deposition using Hf source - EBD-Hf), 2.3 J/cm² (reactive low voltage ion plating - RLVIP), and 0.22 J/cm² (dual ion beam sputtering - DIBS) [82]. The low LIDT for DIBS films was attributed to a high extinction coefficient and damage morphology due to the high intrinsic absorption level as a result of high oxygen deficiency. This is in agreement with the low LIDT and higher absorption in HfO₂ deposited at 100°C in this study. Additionally, the absorption of HfO₂ deposited at 100°C using a higher plasma pulse duration of 7 s shows a significant decrease compared to HfO₂ deposited at the same temperature using 3 s plasma pulse (Table 2), and consequently an improved LIDT.

However, these LIDT values cannot be properly compared with our PEALD films because of different test regimes (1-on-1 vs. R-on-1) and different laser parameters (beam diameter, pulse length and repetition rate). Therefore, Layertec GmbH has produced single layer HfO₂ layers using magnetron sputtering (MS) and ion assisted deposition (IAD). These single layers have LIDT values of 20 J/cm² and 11 J/cm², respectively, as measured with the same system as the ALD coatings. The corresponding reference value for the uncoated substrate is 85 J/cm². The LIDT values of the PEALD films depend on the deposition conditions, with the highest value of 16 J/cm², when a longer plasma pulse was applied. Thus, the effect of plasma parameters was investigated and is discussed in detail in the next section.

3.3 Influence of the plasma parameters

In the influence of temperature section 3.2, the duration of the TDMAH pulse was 3.12 s, the precursor and plasma purge were 5 s, the carrier (Ar) gas flow rate was 160 sccm, the ICP power was 100 W, and the oxygen gas flow was 200 sccm. The effect of plasma parameters (ICP power, oxygen flow rate and plasma pulse) on HfO₂ properties were investigated on films deposited at 100°C using 1482 ALD cycles, which leads to films with a thickness above 250 nm. The variation of parameters is marked in Table 3. The thickness non-uniformity (NU) over 200 mm diameter area surface was 1.3% except the HfO₂ films deposited using 20 sccm oxygen flow. The NU of HfO₂ films deposited using 20 sccm oxygen flow was 2.4%, probably due to insufficient oxidation of surface functional groups. Various plasma parameters were varied here, but the study is punctual in order to limit the number of samples and experiments. Tables 3 and 4 summarize the film properties discussed in this section as function of these parameters.

Table 3. Thin film properties of the HfO₂ deposited at different plasma conditions.^a

View Table | View all tables in this article

Table 4. The effect of plasma parameters on refractive index and absorption in HfO₂ thin films (1482 ALD cycles).

View Table | View all tables in this article

The plasma pulse duration was varied from 3 s to 7 s. While the duration of the precursor and the purge pulses as discussed in the section 3.1 had a significant influence on the NU, no considerable effects of the variation of plasma pulse durations were observed. The NU of HfO₂ remained under 1% on varying plasma pulse duration from 3 s to 7 s due to the uniform O₂ gas distribution in the reactor. However, the thin film refractive index increased from 1.90 to 1.96 (Fig. 7) with the increase in plasma pulse duration from 3 s to 7 s. The prolonged exposure of plasma enables the energetic reaction of plasma species with intermediate surface functional groups leading to a densification of the thin films. The increase in density and the refractive index can be correlated to a decrease in GPC from 1.90 to 1.80 Å/cycle with increasing plasma pulse duration. A similar effect of decreasing GPC with increasing density was observed with increasing deposition temperature, as mentioned in the section 3.2. In contrast, an increase in GPC with an increase in plasma exposure time was reported earlier [28,64]. The low GPC at low plasma pulse durations in those studies was associated with unsaturated growth. Hence, the effects of variation of plasma pulse duration, O₂ gas flow, and ICP power on thin film growth, mechanical properties, optical properties, and chemical composition of HfO₂ films were investigated thoroughly in this study.

 

Fig. 7. Variation of the refractive index at 632.8 nm and growth per cycle (GPC) with plasma pulse time. The ALD cycles of 200 were used in these depositions.

Download Full Size | PDF

At the initial plasma conditions (3 s plasma pulse), HfO₂ films show a smooth surface (See Supplement 1 Fig. S4a) with a roughness of ≈2 nm estimated from EMA using spectroscopic ellipsometry fits and an RMS roughness of 0.51 nm obtained from AFM measurements (See Supplement 1 Fig. S5a). The RMS roughness increases slightly from 0.51 to 0.68 nm on increasing plasma pulse duration from 3 s to 7 s (Table 3), which is visible as the slightly rougher texture of the film deposited at 7 s plasma (See Supplement 1 Fig. S4). A significant decrease in the GPC from 1.80 to 1.70 Å/cycle is observed. This may be due to the densification of the film, as confirmed by XRR measurements (Table 3). The density of the HfO₂ films increases from 7.7 to 8.2 g/cm³ on increasing plasma pulse duration. The higher plasma pulse duration provides sufficient time for the oxygen to react with intermediate functional groups, thereby removing impurities associated with carbon and nitrogen. The effect of increasing density is reflected in an increase in the refractive index. For thicker HfO₂ films (> 260 nm), the refractive index at 632.8 nm increases from 1.92 to 1.96 on increasing plasma pulse duration from 3 s to 7 s. The longer plasma pulse duration of 7 s (Table 4) leads to higher quality films considering the film composition and the optical losses. The tensile stress increases from 588 MPa to 653 MPa. This can be explained by the onset of crystallization (See Supplement 1 Fig. S6), which should be carefully monitored in a process development.

The impurities in the form of N decrease from 3.6% to 2.0% and C from 12.1% to 8.4% on increasing plasma pulse duration to 7 s (Figs. 8(a), 8(d)). Additionally, the O:Hf ratio remains relatively constant at around 2.1. The high impurity content in the HfO₂ films deposited at 3 s compared to 7 s plasma exposure correlates very well with the higher absorption losses at 355 nm wavelength, as shown in Table 4. Moreover, the optical losses in the UV range decrease (Fig. 9) due to a reduction of the impurity concentration. The FTIR measurement confirms the decrease in residual OH groups with increasing plasma exposure to 7 s (See Supplement 1 Fig. S7). As explained in the previous section, an increase in crystallinity results in a decrease in grain boundaries. Hence, the tensile stress increases with an increase in crystallization (See Supplement 1 for Figs. S4 and S6).

 

Fig. 8. AES depth profiles of HfO₂ thin films (1482 ALD cycles) deposited a) using an initial condition with 3 s plasma pulse, 200 sccm O₂ gas flow, 100 W ICP power b) 3 s plasma pulse, 20 sccm O₂ gas flow, 100 W ICP power c) 3 s plasma pulse, 200 sccm O₂ gas flow, 500 W ICP power d) 7 s plasma pulse, 200 sccm O₂ gas flow, 100 W ICP power.

Download Full Size | PDF

 

Fig. 9. Comparison of optical losses (OL = 100% - R – T) of hafnium oxide (thickness 260 - 275 nm/1482 ALD cycles) thin films deposited at various plasma conditions. The varied parameters are marked in bold.

Download Full Size | PDF

We also investigated the effect of the oxygen flow rate on HfO₂ properties. The HfO₂ films deposited using 20 sccm of oxygen flow show smoother surface with RMS roughness of 0.38 nm compared to 0.51 nm (See Supplement 1 for Fig. S5a,b) for the films deposited using higher oxygen flow (200 sccm). For HfO₂ deposited using 3 s plasma pulse, the GPC decreases from 1.80 to 1.72 Å/cycle on decreasing O₂ gas flow from 200 sccm to 20 sccm. The lower GPC can be attributed to the low oxidation of the surface ligands. Hence, less dense films with a low refractive index contain a higher amount of carbon and nitrogen impurities, as summarized in Table 3. As a result of the high impurity level, the absorption at 355 nm wavelength increases drastically from 463 to 3920 ppm (Table 4). Additionally, under stoichiometric films may also be responsible for the higher absorption. Correspondingly, high optical losses in the UV range are observed in Fig. 9. The mechanical stress in the film is lower compared to HfO₂ deposited using 200 sccm O₂ gas flow due to the low density and a porous microstructure. The incomplete oxidation of surface ligands due to insufficient oxygen flow (20 sccm) leads to higher residual OH groups (See Supplement 1 Fig. S7). Hence, a higher O₂ flow (200 sccm) is necessary for the plasma species to sufficiently react with the surface ligands, thereby achieving lower impurity content and absorption losses.

The impact of the ICP power on HfO₂ deposited at 100°C using a 3 s plasma pulse and 200 sccm O₂ gas flow were also investigated in this study. On increasing ICP power from 100 W to 500 W, the GPC increases slightly from 1.80 to 1.84 Å/cycle, and the density increases from 7.7 g/cm³ to 8.9 g/cm³. The increase in density correlates with an increase in the refractive index of HfO₂ deposited at 500 W ICP power. For example, the refractive index at 632.8 nm increases from 1.92 to 2.02 on increasing ICP power. Also, the surface roughness estimated from the EMA layer increases from 2 to 14 nm. Similarly, the RMS surface roughness measured using AFM also shows an increase from 0.51 to 9.55 nm (See Supplement 1 for Figs. S5a,d). These effects of increasing roughness, density, and refractive index may be attributed to an increase in crystallization, as evident in the surface morphology (See Supplement 1 for Figs. S4a,c) and XRD pattern (See Supplement 1 Fig. S6). The increase in crystallization can be due to higher ad-atom mobility facilitated by increased ion energy with increasing ICP power. The increased surface energy through highly energetic oxygen ions reacting at the surface assists the removal of C% and N% impurities, as shown in Table 3. The HfO₂ deposited using 500 W ICP power also showed lowest incorporation of residual -OH groups (See Supplement 1 Fig. S7). The lower amount of OH groups may be attributed to higher density of the HfO₂ films. The absorption at 355 nm increases significantly from 463 ppm to 1570 ppm and at 532 nm from 12 to 110 ppm with increasing ICP power from 100 W to 500 W (Table 4). The impurities in the thin films, as well as defects at grain boundaries, can be the source of absorption [121,122]. Sub-bandgap absorption was observed in polycrystalline films [122]. These defects can exist in the form of oxygen vacancies or interstitials. However, AES measurements show a nearly constant O:Hf ratio of 2.2 for HfO₂ deposited at 500 W and 100 W ICP power. Theoretical calculation by Lyons et al. has shown that oxygen interstitials are more stable in oxygen-rich conditions [123]. The tensile stress in the film deposited at 500 W ICP power is low (460 MPa) compared to the film deposited at 100 W power (588 MPa), possibly due to stress-release because of crack formation.

It is important to note that the standard deviations of the N and C impurities in the HfO₂ films deposited by 500 W plasma power (3 s plasma pulse, 200 sccm O₂) and 7 s plasma pulse (100W ICP power, 200 sccm O₂) are lower compared to the other two plasma conditions (Table 3). In-depth insight into this phenomenon is given by the AES depth profiles shown in Fig. 8. At the initial conditions of 3 s plasma pulse duration, a significant increase of the carbon content and a slight increase in the nitrogen content towards the substrate were observed (Fig. 8(a)). This effect is reduced when the plasma pulse duration was increased to 7 s. However, a minor increase in the carbon content towards the substrate is seen in Fig. 8(d). In contrast, the atomic impurity content in the depth profile of the HfO₂ deposited with 20 sccm oxygen flow was relatively uneven due to the poor gas flow while showing the same increasing tendency of the C and N content towards the substrate. As a result, the standard deviation is high. This observation shows the influence of the substrate and the available functional surface groups, as explained below.

For the HfO₂ deposited at a low oxygen flow rate (20 sccm) or short plasma pulse duration (3 s), more TDMAH ligands, i.e., more nitrogen and carbon, remain on the surface during the plasma step and do not participate in chemical reactions during the first cycles until the film reaches a critical thickness. At this point, the influence of the substrate is no longer prominent, and the film solely grows from the new hydroxyl groups formed on the growing film. The chemical environment and the availability of functional groups are different in the vicinity of the substrate, and the self-limiting reaction regime during the plasma step changes with thickness during the thin film growth. Consequently, an increase of the plasma power (to 500 W) or the plasma pulse duration (7 s) leads to more ligand oxidation during the first few hundreds of PEALD cycles and a more uniform atomic nitrogen and carbon content throughout the film. However, despite the broad variation of the plasma parameters (20 sccm O₂ flow vs. 200 sccm, 100 W vs. 500 W), the Hf:O ratio remained relatively constant.

The HfO₂ film using 7 s plasma pulse duration, 100 W ICP power, and 200 sccm O₂ gas flow showed low impurity and low absorption losses. Hence, these conditions were considered further for antireflection coatings, as discussed in the following section.

3.4 Antireflection coatings

Antireflection (AR) coatings were realized for the wavelengths of a Nd:YAG laser and its higher harmonics (up to the 4^th order) consisting of hafnium oxide in combination with silicon oxide and aluminum oxide. The individual layers were deposited using plasma enhanced ALD (PEALD) at 100°C temperature. The Al₂O₃ thin films were deposited using trimethyl aluminum (TMA) precursor and oxygen plasma with an ALD sequence of 0.08s/2s/3s/2s, 80 sccm carrier gas (Ar) flow, 100 sccm oxygen flow and 100 W ICP power. The SiO₂ thin films were deposited using bis(diethylamino)silane (BDEAS) precursor and oxygen plasma with an ALD sequence of 0.32s/5s/3s/2s, 30 sccm carrier gas (Ar) flow, 200 sccm oxygen flow and 100 W ICP power. The ALD cycle for the HfO₂ thin films consists of 3.12 s of precursor pulse, 5 s of precursor purge, 7 s of plasma pulse, and 5 s of plasma purge. The O₂ and Ar gas flow rates of 200 sccm and 160 sccm, respectively, were maintained for the entire HfO₂ deposition. The depositions were carried out on fused silica (FS) substrates placed in different orientations. Most FS substrates were clipped in a vertical orientation on the substrate holder for a double-side coating, while one FS substrate was placed directly on a 200 mm Si wafer (to prevent backside deposition) for single-side deposition. The AR designs and the total physical thicknesses are summarized in Table 5. Figure 10(a) shows the reflection spectra of the AR1 sample with an antireflection (AR) coating for 1ω and 2ω of the Nd:YAG laser. The measured reflectance spectrum matches very well with the design (data shown for single side coating, without backside reflection). The thickness gradient of the vertically aligned sample is small and the AR is successfully achieved for single and double-side deposition. All designs have been well achieved, and the total reflectance of the double side coatings are below 0.7% at all target wavelengths; however, the absorbance at 266 nm is larger than at 355 nm wavelength. The minima of the AR2 and AR3 samples is slightly shifted towards lower wavelengths than the target spectral positions. The ARs have been not further optimized. The mechanical stress of AR1 was 282 MPa, which was lower than 326 MPa and 380 MPa for AR2 and AR3 because of fewer HfO₂ layers in the AR stack.

 

Fig. 10. Reflectance spectra of an antireflection coating consisting of HfO₂, SiO₂, and Al₂O₃ on single and double-side coated fused silica substrates at a) 1064 nm and 532 nm b) 1064 nm, 532 nm, and 355 nm c) 1064 nm, 532 nm, 355 nm, and 266 nm. A comparison of the designed and deposited thin-film system is shown. The backside reflection of the single-side coated sample is subtracted.

Download Full Size | PDF

Table 5. The composition of Al₂O₃[HfO₂/Al₂O₃]_xSiO₂ thin films and design used in antireflection coatings.^a

View Table | View all tables in this article

The absorption in all AR coatings (single and double sides) at 1ω is 5 ± 2 ppm and at 2ω is 10 ± 4 ppm. Similar absorptions of 1 ppm (1ω) and 9.3 ppm (2ω) in AR coatings deposited on the single-side fused silica was reported by Liu et al. [95]. The authors also reported an absorption of 597.1 ppm at 3ω [95], which is significantly higher than the values obtained here 188 ± 14 ppm and 170 ± 14 ppm for single-side AR2 and AR3. Hence, a PEALD process might be more appropriate in terms of absorption losses than a thermal process.

The LIDT at 355 nm wavelength for all the three AR coatings is 27–28 J/cm² (see Table 5). Liu et al. reported LIDT of 4.9 J/cm² at 3ω wavelength (10 pulses, S-on-1, 10 ns, 100 Hz, beam focus Ø 300 µm) [95]. In a similar study, Yin et al. have shown that LIDT (1-on-1, 7.8 ns, beam focus 0.28 mm²) of 24.4 J/cm² at 3ω for AR coatings deposited by PEALD HfO₂/SiO₂ at 150°C was higher than AR coatings (20.6 J/cm²) deposited by electron beam deposition [94], which was attributed to the lower absorption of PEALD AR coating. Comparable AR systems consisting of HfO₂/SiO₂ multilayers have been deposited at Layertec GmbH by magnetron sputtering and ion assisted deposition on similar substrates. The LIDT values of these coatings reached values of 31 J/cm² and 28 J/cm², respectively. This indicates that the main limiting factor on the LIDT performance are the intrinsic properties of the oxides, specifically of the high refractive index component. Additionally, macroscopic defects such as nodules or particles are also low within the PEALD processes. The influence of different materials and designs for AR coatings at 355 nm made by PEALD is currently under investigation and will be presented in a sub-sequent article.

In this study, high-quality AR coatings with low absorption and high LIDT deposited using an optimized PEALD process at 100°C demonstrate a promising application in high power laser systems. The deposition processes must be carefully optimized to achieve the desired film properties to meet the performance of AR coatings.

4. Conclusion

The requirements of thin film coatings are a high density with pinhole-free morphology, amorphous materials with low defect density, low impurity and OH incorporation and a very high uniformity. This study provides a thorough approach towards the process development of HfO₂ coatings by plasma enhanced atomic layer deposition (PEALD). The process parameters, mainly gas flow rates, have been demonstrated to have a major influence on the non-uniformity of the coating. The deposition temperature has a high influence on the film density and refractive index, OH incorporation, and C and N impurities; the effect of crystallization was prominent by high absorption in the UV range, high surface roughness, and high tensile stress. Correspondingly, the films deposited above the crystallization threshold showed lower LIDT. The deposition process was further optimized by varying plasma conditions and keeping the deposition temperature at 100°C. The HfO₂ films deposited using 7 s of plasma exposure time, 100 W ICP power, and 200 sccm oxygen flow rate showed a high refractive index in the UV range, high density, low C and N impurities, and low absorption. The effects of crystallization on deteriorating thin film properties were also visible by depositing HfO₂ films at 500 W ICP power. Eventually, the shortcomings of HfO₂ films deposited using 3 s plasma exposure time were rectified using 7 s of plasma exposure. This also improved the LIDT value. Thus, oxygen plasma species with sufficient energy were successful in improving material properties. Subsequently, HfO₂ thin films deposited under optimized conditions were incorporated into antireflection coatings designed for 1ω, 2ω, 3ω, and 4ω wavelength of the Nd:YAG laser. The deposited films were in excellent agreement with the design, and a reflectance below 0.4% was observed for all the coatings. Besides, very good LIDT values were realized for these coatings. Consequently, it was shown that high-quality films at a deposition temperature of 100°C are achievable by using PEALD.