Effects of corundum-structured seeds on the low-temperature growth and hardness of the alumina-based films

The corundum-structured seeds have been introduced into alumina films using Al + α-Al2O3 composite and Al100−xCrx (x = 10, 20, 30) alloy as targets by reactive magnetron sputtering. The effects of the seeds on the elemental composition, phase composition, structure characteristic, surface morphology and nano-hardness of the alumina-based films have been investigated. The Al target has also been sputtered to prepare the alumina film as reference. The film is composed of single-phase α-Al2O3 by sputtering the Al + α-Al2O3 composite target at 550 °C, while the solid solution α-type (Al0.7Cr0.3)2O3 was detected in the film prepared with the Al70Cr30 target at the same temperature. The nano-hardness of the pure α-Al2O3 film and the α-(Al0.7Cr0.3)2O3 film were measured as ∼23.5 GPa and ∼28.5 GPa, respectively, which are much higher than that of the film deposited from Al target (∼16.3 GPa).


Introduction
Corundum-structured alumina (α-Al 2 O 3 ) films exhibit great physical and chemical properties including high hot hardness, good oxidation resistance, high physical and chemical stability. These properties make them one of the most attractive candidates for diffusion barrier, wear resistance and anti-oxidization coatings [1][2][3]. However, most of the polymorphs such as γ, κ or amorphous phases can hardly meet the requirements for these high-temperature applications [4][5][6][7][8], and only the corundum-structured α phase is considered thermodynamically stable. In commercial scale α-Al 2 O 3 films have been synthesized by chemical vapor deposition (CVD) at a high substrate temperature of over 1000°C [9], where the high cost cemented carbide is one of the few suitable substrate materials and some unwanted reactions between the film and substrate material might occur. Unfortunately, the metastable phases have a great change of being present in the film deposited at relatively low temperature, which will transform into the α-Al 2 O 3 irreversibly at high temperature and lead to a relatively huge volume decrease associated to the cracking and failure [4]. Therefore, decreasing the deposition temperature of the high purity α-Al 2 O 3 film is the crucial objective for expanding its practical applications.
Various methods have been performed to realize the low temperature growth of the α-Al 2 O 3 films and the physical vapor deposition (PVD) is the priority for many researchers. According to the progressive work reported in last years, two types of experimental approaches have been performed [10]: one approach benefits from the high ionization proportion and energetic ion bombardment by using new pulsed plasma technologies or highly ionized PVD processes. Zywitzki et al [11,12] reported that films exhibited single α phase when deposited at a temperature higher than 760°C in a pulse magnetron sputtering system. Selinder et al [13,14] successfully reduced the deposition temperature to 650°C by using reactive high-power impulse magnetron sputtering (HIPIMS). The same low temperature was also achieved by Brill et al's research [15], in which α-Al 2 O 3 films were obtained in a filtered arc device with high substrate bias voltage of −300 V. A lower temperature of 560°C for the α-Al 2 O 3 films was reached by Jiang et al [16] in a high-power plasma-assisted Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. chemical vapor deposition (PACVD). Unfortunately, the temperature reduction in this type of approach is still considered insufficient.
Considering the surface energies and thermodynamic phase stability in nanocrystalline alumina [17], another approach to synthesize and stabilize α-Al 2 O 3 at lower temperature suggests the introduction of the isostructural nanocrystals. The surface energy during the nucleation and growth could be decreased to a relatively low level due to the additional nucleation sites or center induced by the introduced materials. Especially, the nucleation of α-Al 2 O 3 can be facilitated by using Cr 2 O 3 interlayers as template which is isostructural with α-Al 2 O 3 and easily obtained at low temperature [18][19][20][21]. Diechle et al [10] reported that the deposition temperature was reduced to 500°C by reactively sputtering Cr and Al targets and the film was detected as predominant α-(Al, Cr) 2 O 3 . Andersson et al [22] reached the same temperature by deposited α-Al 2 O 3 films onto Cr 2 O 3 interlayer. In addition, the α-Al 2 O 3 seeds are also supposed to promote the lowtemperature growth of the stable single-phase α-Al 2 O 3 since the same structure to the Cr 2 O 3 template [23,24]. Lin et al [25] prepared Al films with α-Al 2 O 3 seeds by glow plasma technique followed by in situ oxidation at 580°C and the outer layer was tested as pure α-Al 2 O 3 film. However, the pure α-Al 2 O 3 films can still hardly be synthesized or confirmed in reactive magnetron sputtering deposition and the difference between the effects of these corundum-structured seeds have been rarely investigated.
This work aimed to identify the distinguishable effects of the corundum-structured seeds on the lowtemperature growth and the mechanical property of the alumina-based films deposited by reactive RF magnetron sputtering. Two types of seeds were introduced into the films by employing the the Al+15 wt% α-Al 2 O 3 composite and Al 100−x Cr x (x=10, 20, 30) alloy as targets. Single-phase α-Al 2 O 3 and α-type (Al 0.7 Cr 0.3 ) 2 O 3 solid solution were detected in the films deposited at 550°C. The preferentially formed corundum-structured species could serve as seeds and induce the epitaxial growth of α-Al 2 O 3 . The phase evolution leaded to significant improvement of the nano-hardness and the difference of these two seeds were finally investigated.

Methods
The depositions of the alumina-based films were performed in a radio frequency magnetron sputtering (RFMS) system. The Al (99.99% purity) and Al 100−x Cr x (99.99 purity, x=10, 20, 30) targets were commercially ordered and the Al+α-Al 2 O 3 composite target was manufactured by powder metallurgy method using a mixture consisted of 15 wt% α-Al 2 O 3 (99.999% purity) and 85 wt% Al (99.995% purity) powders. The dimensions of all the targets are Φ 60 mm * 3 mm and a 15 min pre-sputtering was performed to remove the contaminants on the surface in each deposition. In order to make most of the oxidation reactions of the sputtered species occur at the substrate surface, the sputtering and reaction gases were separately injected into the chamber. The inlet of the argon was in the vicinity of the sputtered target while that of oxygen was near the substrates. The Si (100) substrates (10×10 mm 2 ) were cleaned in an ultrasonic device using acetone then ethanol. The distance between the target and the substrate was set as ∼85 mm. The residual pressure of the vacuum chamber was evacuated to less than 5×10 −4 Pa. The Ar gas flow was kept at 15 sccm and the O 2 flow rate was 1.5 sccm. The work pressure was modified to 1.0 Pa and kept constant during the deposition. The power of the RF generator was held constant at 200 W and all the depositions were carried out at 550°C. The deposition time for all alumina-based films was 3 h and the thickness of the films was in the range of 350∼370 nm.
Electron probe microanalysis (EPMA) was performed to analyze the elemental compositions of the aluminabased films in an EPMA-1600 instrument. Grazing incident x-ray diffraction (GIXRD) was employed to characterize the phase composition of the films in a PANalytical diffractometer (Cu-Kα x-ray). Transmission electron microscopy (TEM) was carried out in a JEOL JEM-2100F system to identify the information on the microstructure of the as-deposited films. Scanning electron microscopy (SEM) was implemented in a NOVA Nanosem 430 system to observe the morphology of the alumina-based film. A thin platinum layer was sprayed on the samples to ensure that the surface conductivity is sufficient for the SEM measurement. Depth-sensing indentation (DSI) technique was performed to measure the hardness of the films in an AntonPaar NHT3 equipment. After the pre-measurement, the maximum load was set as 0.5 mN and 5 indentations were made for each sample. The Oliver and Pharr method [26] was used to calculate the values of the hardness from the obtained data. Table 1 shows the elemental compositions of the alumina-based films deposited from Al, Al+α-Al 2 O 3 composite and Al 100−x Cr x (x=10, 20, 30) alloy targets at 550°C. The alumina films (sample 1 and 2) both consist of ∼60 at% O and ∼40 at% Al, which implies that the O/Al atomic ratio is ∼1.50. The O:(Al+Cr) ratios are also calculated and the similar value is obtained for the Al-Cr-O films (sample 3, 4 and 5) deposited from the Al-Cr alloy targets. Thus, it can be assumed that the stoichiometric alumina-based films were successfully synthesized at 550°C with all the different targets. In addition, the calculated value of Cr:(Al+Cr) ratios are 0.31, 0.19 and 0.12 for the Al-Cr-O films deposited from Al 70 Cr 30 , Al 80 Cr 20 and Al 90 Cr 10 alloy targets, respectively, which suggests the great consistency between the Cr contents in Al-Cr-O films and those in alloy targets. Figure 1 demonstrates the GIXRD patterns of alumina-based films prepared at 550°C using Al, Al+α-Al 2 O 3 and Al 70 Cr 30 targets. The alumina film synthesized by using the Al target (sample 1) presents a composite phase consisted of α-Al 2 O 3 and γ phase, which can be clarified by the dynamics of film growth. Due to the particular setting for the reaction gas inlet, most of the sputtered Al atoms or ions were oxidized immediately near the substrate surface and formed numerous alumina species. Meanwhile, the exothermal oxidation reactions emitted additional energy which could boost the surface migration of the species. Part of the species tended to stack and crystallize into α-Al 2 O 3 while the others could only transform into metastable phases (i.e. γ or amorphous). Then, the large energy barrier suppressed the phase transition from γto α-Al 2 O 3 during the deposition and the metastable phases were retained in the film. Furthermore, the widen α(012) peak with low density suggests the poor crystallinity of the alumina film, which makes the assumption on the presence of amorphous phase reasonable. However, the pattern of the film prepared with the composite target (sample 2) exhibits a single phase of α-Al 2 O 3 with high intensity peaks, while the (220) peak from γ-Al 2 O 3 or those from other metastable phases cannot be detected. Inspired by the research reported by Lin et al [25], it can be assumed that the particles sputtered from the Al+α-Al 2 O 3 target consist of Al atoms or ions and α-type alumina molecules. The α-type alumina molecules developing from the α-Al 2 O 3 in the composite target would crystallize into α-Al 2 O 3 nuclei. Meanwhile, the Al species developing from the Al in the composite were oxidized at the substrate surface and transformed into new alumina molecules accompanied with the generative heat. The heat could promote the surface migration of the alumina molecules so that these molecules could transfer to the α-type nuclei. As a result, the α-Al 2 O 3 could grow homoepitaxially around the seeds with the formation of the metastable phases restrained.  The films deposited from the Al 80 Cr 20 and Al 90 Cr 10 targets were introduced to investigate the structure evolution and confirm the formation of the α-type solid solution. Figure 2(a) presents the phase structures of the films prepared with the Al 100−x Cr x (x=10, 20, 30) target at 550°C. Compared to the film with 30% α-Cr 2 O 3 (x=30), the pattern of the film prepared with the Al 90 Cr 10 target exhibits distinguishable (012) peaks from α-Cr 2 O 3 and α-Al 2 O 3 , and the (220) peak from γ-Al 2 O 3 can also be observed. The film deposited from Al 80 Cr 20 demonstrates the similar phase composition composed of α-Cr 2 O 3 , αand γ-Al 2 O 3 with an additional weak peak of α(104). As reported in previous work [27], the Cr content in the Al-Cr-O films has a significant impact on the phase evolution. When synthesized at relatively low temperature of 550°C, increasing the Cr content in the film could provide more α-Cr 2 O 3 as α-type nuclei, which implies increasing the distribution density of α-Cr 2 O 3 could promote the heteroepitaxial growth of α-Al 2 O 3 . When the α-Cr 2 O 3 proportion reaches to 30%, the formation of γ-Al 2 O 3 can be completely suppressed. The enlarged patterns of the Al-Cr-O films focusing on the position of (012) peaks are given in figure 2(b). The standard positions of (012) peaks from α-Cr 2 O 3 (PDF No.38-1479) and α-Al 2 O 3 (PDF No.10-0173) are marked as two vertical dot lines, respectively. The (012) peaks of the measured patterns locate between the reference positions. There is a trend for the α-Cr 2 O 3 peaks to shift to higher diffraction angle as the Cr content (i.e. the α-Cr 2 O 3 content) in the films increase. On the contrary, the α-Al 2 O 3 peaks tend to move to the opposite direction at the same time. This phenomenon relates to the intersubstitution of the Al 3+ and Cr 3+ ions which results in expansion or shrink of the lattice parameters. It is well known that the Cr 3+ has larger ionic radius (∼0.615 Å) than Al 3+ (∼0.535 Å). The lattice parameters of α-Cr 2 O 3 shrank when part of the Cr 3+ were replaced by Al 3+ . In the same way, the lattice parameters of the α-Al 2 O 3 expanded as some Al 3+ were substituted by Cr 3+ . With the α-Cr 2 O 3 content of 30% in the film deposited from the Al 70 Cr 30 target, the (012) peaks from α-Cr 2 O 3 and α-Al 2 O 3 overlap into one peak, which is mostly attributed to the full inter-substitution between the metallic ions of Al 3+ and Cr 3+ , indicating the formation of singlephase solid solution α-type (Al 0.7 Cr 0.3 ) 2 O 3 film.

Results and discussion
To estimate the lattice size of the α-type (Al 0.7 Cr 0.3 ) 2 O 3 solid solution, the parameters can be calculated as Vegard's law:  However, the existence of the amorphous phase in the films can hardly be detected by the GIXRD technique. Thus, the TEM analysis was employed to investigate more details on the microstructure of the alumina-based films. The cross-section TEM image for the film prepared with composite target (sample 2) is displayed in figure 3(a), which indicates that the film consists of nano-particles. Figure 3(b) demonstrates the selected area electron diffraction (SAED) patterns of sample 2 and all the diffraction rings can be assigned to α-Al 2 O 3 , which is consistent with the phase composition detected by GIXRD (see figure 1). Figure 3(c) presents the highresolution transmission electron microscopy (HRTEM) images of sample 2, which also confirms the distribution of the nano-crystalline α-Al 2 O 3 . The enlarged view of the area marked with white box and the corresponding fast Fourier transform (FFT) patterns in figure 3(c) can be identified as the (2 12) and (2 04) planes of α-Al 2 O 3 . Furthermore, no pattern or plane corresponding to amorphous or other metastable phases can be discovered in these TEM images. Hence, these TEM results confirm that single-phase α-Al 2 O 3 film was obtained by reactively sputtering the Al+α-Al 2 O 3 composite target at 550°C and the formation of the metastable phases was completely suppressed. Figure 4 shows the microstructure of the film prepared at 550°C using the Al 70 Cr 30 target (sample 3). The SAED patterns of the film in figure 4(a) exhibits α phase only (see figure 4(b)). According to the GIXRD patterns shown in figures 1 and 2, these diffraction rings correspond to the solid solution α-(Al 0.7 Cr 0.3 ) 2 O 3 . The HRTEM image (see figure 4(c)) presents a microstructure with no amorphous phase around. The enlarged view of the area marked with white box and the corresponding FFT patterns also demonstrate the planes of corundum structure. Therefore, the TEM analysis is considered consist with the GIXRD results where the single-phase solid solution α-(Al 0.7 Cr 0.3 ) 2 O 3 film was synthesized by using Al 70 Cr 30 alloy target.
The surface morphologies of the films prepared with Al, Al+α-Al 2 O 3 composite and Al 100−x Cr x (x=10, 20, 30) alloy targets by RFMS at 550°C are shown in figure 5. The typical granular surface is observed in each film and the uniformity of grain size may lead to the good surface quality. The grains in the film deposited from Al+α-Al 2 O 3 target are approximately twice the size of those in the film synthesized by using Al target (see figures 5(a) and (b)), which is probably attributed to the promotion of α-Al 2 O 3 nuclei on the homoepitaxial growth and crystallization. As shown in figure 5(c), the grain size of the film deposited from the Al 70 Cr 30 target increases observably and is much larger than that of the alumina films. This might benefit from the oxidation of the Cr species, which could not only provide sufficient α-type nuclei but also release the generating heat, resulting in the enhanced heteroepitaxial growth and grain growth. Moreover, there is a tendency for the grain size of the films prepared with Al-Cr alloy targets to expand as the increasing α-Cr 2 O 3 content in the films (see figures 5(d), (e) and (c)). More α-Cr 2 O 3 in the film can provide more nucleation sites or centers to promote the  nucleation during the deposition progress, which implies that the required energy for nucleation was reduced and more energy could be expended for grain growth. However, the distinguishable grain size of the films is in the range of a dozen to dozens of nanometers, which illustrates the formation of the nanocrystal alumina-based films. The cross-section view of these alumina-based films are similar to each other and the thickness are measured as 350∼370 nm (not shown here). Based on the deposition time of 3 h, the deposition rate of these films is in the range of 117∼123 nm/h, which could be considered at a relatively low level. However, this low deposition rate may also have a positive effect on the formation of corundum-structured phase at low temperature since the species have sufficient time for surface migration. On the contrary, the kinetics of nucleation and growth of the α phase at larger rate may be unfavorable and higher deposition temperature will be required.
The nano-hardness of the alumina-based films were evalutated by the nano-indentation measurement. The load versus displacement (P-h) curves of the films prepared with Al, Al+α-Al 2 O 3 composite and Al 70 Cr 30 alloy targets are show in figure 6(a) and those curves of the films despoited from the Al-Cr alloy targets are demonstrated in figure 6(b). The value of the calculated hardness (H) are given in figure 6(c). The hardness of the alumina-based films are mostly ascribed to the type and amount of the phases in the films. As a reference, the hardness of the amorphous, γand α-Al 2 O 3 films deposited by magnetron sputtering has been reported as ∼10 GPa, ∼19 GPa and ∼22 GPa, respectivley [12]. According to the phase composition analysis (see figure 1),  According to the GIXRD results shown in figure 2, it can be assumed that the film prepared with Al 90 Cr 10 alloy shows relatively low hardness of ∼16.7 GPa due to the existence of γ-Al 2 O 3 and amorphous phase. The low α-Cr 2 O 3 content (10%) in the film can barely provide sufficient α-type nuclei and the heteroepitaxial growth of the α-Al 2 O 3 has been greatly limited, which results in a similar growth pattern to the film deposited from Al target. When the α-Cr 2 O 3 content increases to 20%, the hardness of the film increases to 20.2 GPa with the enhanced growth of the α-Al 2 O 3 . However, the formation of γ-Al 2 O 3 or amorphous phase can hardly be completely suppressed and the value of the hardness is still slighly lower than that of the film composed of singlephase α-Al 2 O 3 . Owing to the combined contribution of the single α phase and the solution strengthening, the hardness of the film deposited from Al 70 Cr 30 target reaches a highest value of 28.5 GPa. In a word, the α-Al 2 O 3 seeds in the film could suppress the formation of metastable phases and lead to the pure α-Al 2 O 3 films with high hardness. The α-Cr 2 O 3 seeds with certain distribution density could not only play the same role in the synthesis of single-phase α-type films but also induce the solution strengthening for the full inter-substitution between the metal ions.

Conclusions
Two types of corundum-structured seeds were introduced into alumina-based films by using Al+α-Al 2 O 3 (15 wt%) composite and Al 100−x Cr x (x=10, 20, 30) alloy as targets via radio frequency magnetron sputtering. The epitaxial growth of the α-Al 2 O 3 is enhanced by the presence of these seeds and the nucleation of γ or other metastable phases is suppressed in the meantime. The nanocrystalline single-phase α-Al 2 O 3 film was synthesized by using the Al+α-Al 2 O 3 composite target at 550°C. The homoepitaxial growth of α-Al 2 O 3 at low temperature was promoted by the existence of α-Al 2 O 3 nuclei, which leads to a hardness of ∼23.5 GPa. The single α-type (Al 0.7 Cr 0.3 ) 2 O 3 solid solution film with hardness of ∼28.5 GPa was synthesized at 550°C using Al 70 Cr 30 alloy as target. The high hardness of the α-(Al 0.7 Cr 0.3 ) 2 O 3 film is ascribed to the combined effects of sufficient distribution density of corundum-structure nuclei (i.e. α-Cr 2 O 3 ) and solution strengthening.