Study on reactive sputtering of yttrium oxide: Process and thin film properties

Surface & Coatings Technology 276 (2015) 39–46

Contents lists available at ScienceDirect

Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat

Study on reactive sputtering of yttrium oxide: Process and thin film properties Pei Lei a, Wouter Leroy b, Bing Dai a, Jiaqi Zhu a,⁎, Xiaoting Chen c, Jiecai Han a, Diederik Depla b a

Center for Composite Materials, Harbin Institute of Technology, P.O. Box 3010, Yikuang Street 2, Harbin 150080, PR China Research Group DRAFT, Department of Solid State Sciences, Ghent University, Krijgslaan 281(S1) 9000 Gent, Belgium Key Laboratory for Liquid–solid Structural Evolution and Processing of Materials (Ministry of Education), School of Materials Science and Engineering, Shandong University, Jingshi Road 17923, Jinan 250061, PR China

b c

a r t i c l e

i n f o

Article history: Received 21 November 2014 Revised 7 April 2015 Accepted in revised form 23 June 2015 Available online 27 June 2015 Keywords: DC magnetron sputtering Yttrium oxide thin films Microstructure Optical properties Mechanical properties

a b s t r a c t This paper investigates the influence of deposition conditions on the properties of yttrium oxide thin films. The paper focuses on the texture, optical and mechanical properties. With this objective, a series of yttrium oxide thin films with different thicknesses were deposited by direct current (DC) unbalanced reactive magnetron sputtering at high and low pumping speed. By changing the oxygen flow, depositions were performed in the three characteristic deposition modes for reactive magnetron sputtering, i.e., metallic, transition and poisoned mode. By using an oxygen flow directed to the substrate, full oxidation of the samples, as shown by X-ray photoelectron spectroscopy (XPS), in the three modes is obtained. Crystallographic characterization by X-ray diffraction (XRD) shows that films crystallize in the cubic phase with a strong (222) out-of-plane orientation at low oxygen flow. As the oxygen flow increases a mixture of cubic and monoclinic phase is obtained. In poisoned mode, the films consist of the cubic phase with preferred (420) orientation. Scanning electron microscopy (SEM) cross sections show, with increasing oxygen flow, a loss of the columnar structure. As the oxygen flow rates increase through the metallic, transition, and the poisoned mode, the grain size becomes gradually smaller. An overview diagram of all experimental results uncovers that the textural changes are closely linked to the oxygen partial pressure rather than the oxygen flow. The optical properties of films were investigated by spectroscopic ellipsometry (SE). The films with a columnar structure demonstrate superior hardness and modulus as well as the high plasticity. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Yttrium oxide, a rare-earth oxide, has attracted considerable attention over the past several years due to its variably crystallographic characteristics and unique properties, which makes it an important technological material. Briefly, owing to high chemical and thermal stability (melting point is up to ~ 2349 °C) [1,2], and its mechanical properties (high strength and fracture toughness) [3], yttrium oxide films and particles have been used in thermal or reaction barrier coatings [4] and oxide dispersion strengthened steels [5,6]. Particularly, due to the excellent optical and electric properties, including a wide transmittance range, high refractive index (~2), low absorption, large band gap (~ 5.4 eV), and high permittivity (~ 14–18) accompanied with a lattice match with Si and GaAs (for the cubic phase) and graphene (for the hexagonal phase), yttrium oxide thin films become one of the most interesting materials widely used in optical waveguides [7–9], and as an antireflective layer [10], or as a high ⁎ Corresponding author. E-mail address: [email protected] (J. Zhu).

http://dx.doi.org/10.1016/j.surfcoat.2015.06.052 0257-8972/© 2015 Elsevier B.V. All rights reserved.

efficiency phosphor by doping with other rare-earth elements [11,12], as well as one component of high-quality metal-oxide-semiconductor (MOS) based devices [13–18]. Structurally, yttrium oxide has several polymorphic phases, such as A-hexagonal (P32m), B-monoclinic (C2/m), C-cubic (Ia3), H-hexagonal (P63/mmc), and X-fluorite (Fm3m) [19,20]. At ambient pressure and room temperature, the C-cubic phase is the most stable phase, which transforms to H-hexagonal phase at about 2600 K or to B-monoclinic phase at ~10 GPa [17,21]. As the material properties are structurally dependent, it is important to control the thin film deposition conditions which influence the structure. Compared to other methods, DC reactive magnetron sputtering is an important physical vapor deposition technique in engineering and scientific fields due to its inherent advantages, which include low temperature deposition with ion assistance, high deposition rate, good film quality [22], and its scalability. However, the addition of the reactive gas (such as oxygen or nitrogen) increases its complexity. Indeed, in DC reactive magnetron sputtering a hysteresis of the deposition parameters as a function of the oxygen flow is often reported. The transition from the high deposition rate regime (metallic mode) to the low deposition rate regime (poisoned mode) often occurs

40

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

abruptly resulting in process instability close to the transition points between both modes [23]. Although there have been many reports about thin film deposition of yttrium oxide by a large number of methods, including molecular beam deposition [17,24], electron beam evaporation deposition [25], pulsed laser deposition [26,27], radio frequency sputtering [7, 28], chemical vapor deposition [2,29], there are few reports systematically covering the controllable growth of yttrium oxide films by DC reactive magnetron sputtering [30]. The latter research paper still leaves an interesting gap to be investigated, especially, on understanding and controlling the growth of yttrium oxide films within the three different modes from metallic mode, transition zone to poisoned mode. This is a crucial issue to build the relationship of process–structure-properties. Based on the above background, in this paper, a series of yttrium films with three different thicknesses were deposited by DC reactive magnetron sputtering on glass and silicon substrates using local oxygen supply. In order to obtain stable process in the transition zone, a higher pumping speed was employed to remove the hysteresis loop [23,31]. Influence of different oxygen flow rates on the target state, oxygen pressure, deposition rate, film structure and the optical and mechanical properties was systematically investigated. 2. Experimental details 2.1. Thin film deposition Yttrium oxide thin films were prepared on glass and silicon (100) substrates by DC reactive unbalanced magnetron sputtering. The glass substrates were cleaned by distilled water and methanol in ultrasonic bath for 10 min. The silicon substrate was washed via RCA cleaning procedure. A metallic yttrium target (99.5% purity, 50.8 mm in diameter and 3 mm in thickness, Testbourne Ltd.) was mounted on a magnetron which is oriented perpendicularly to the center of substrate holder in a stainless steel vacuum chamber. The oxygen inlet was positioned close to the substrate in order to obtain fully oxidized films even in metallic mode. The vacuum chamber was evacuated to a base pressure lower than 10−4 Pa (10−6 mbar) by combination of a turbo-molecular and a rotary pump. The argon pressure was fixed at 0.5 Pa by the combination of a throttle valve and an argon flow controller. In this way, experiments were performed at low (120 L/s, 35 sccm Ar) and high pumping speed (500 L/s, 148 sccm Ar). Experiments were performed at constant current (I = 0.5 A, Hüttinger Elektronik 1500 DC power supply). Before deposition, pre-sputtering was performed for 10 min to clean the target surface. At the same time, the substrate was protected by a shutter. The distance between target and substrate was kept constant at 10 cm. The substrate was neither intentionally heated nor cooled. The films with thickness of about 250, 550 and 1000 nm were obtained by adjusting the deposition time from several minutes in metallic mode up to more than 10 h in poisoned mode. 2.2. Thin film characterization The film thickness was evaluated by contact profilometry (Taylor–Hobson Talystep) and checked again by cross section SEM images and spectroscopic ellipsometry (SE). The deposition rate can be calculated from the measured thickness and the deposition time. X-ray photoelectron spectroscopy (XPS) was employed to detect the chemical composition on the surface and subsurface of films after ion bombardment (4 keV argon). The XPS spectra were measured using monochromatized Al Kα (1486 eV). For survey measurements a pass energy of 152.55 eV was used, while the O1s and Y3d spectra were measured with a pass energy of 107.8 eV. Crystallographic properties were measured by X-ray diffraction (XRD) using Cu Kα radiation (0.154 nm) in Bragg–Brentano geometry with a LynxEye Silicon Strip detector. The samples were scanned in a 2θ range from 20 to 68° with

a step size of 0.04°. The cross-section morphology was studied on gold-coated samples by a FEI Quanta 200 scanning electron microscope using a high voltage of 20 KV. Ellipsometry (J. A. Woollam Spectroscopic Ellipsometer, SE) was used to evaluate the optical properties of films in the spectral range of 250–1690 nm at an incidence angle of 70°. The hardness and elastic modulus of the films were evaluated by Nanoindenter XP with continuous stiffness measurement mode. Four measurements were made on each sample and then the average value was used in this work.

3. Results and discussion 3.1. Selection of the deposition conditions To select the deposition conditions, a so-called hysteresis experiment is performed by a stepwise increase of the oxygen gas flow rate up to a maximum of 3.0 sccm for the low pumping experiment, and up to 8 sccm for the high pumping speed experiment. After reaching the maximum oxygen flow, the flow rate is stepwise decreased until the initial conditions. Sufficient time is allowed to stabilize the process before the oxygen flow is changed. During the experiment the discharge voltage and total pressure is measured. The latter allows to calculate the oxygen partial pressure by subtraction of the constant argon pressure (0.5 Pa). Fig. 1 summarizes the obtained results. Fig. 1(a) shows the variation of discharge voltage as oxygen flow rate increases and decreases at pumping speed of 120 L/s. As the supply of oxygen increases from 0 to about 1.7 sccm, the discharge voltage remains at a high value of about 400 V. Above this oxygen flow, the discharge voltage decreases abruptly to around 200 V. When the oxygen flow rate is subsequently decreased, the discharge voltage remains at a low value of about 200 V up to a flow rate of 1.3 sccm. The change in the discharge voltage corresponds to a change of the oxygen partial pressure. This behavior of both the discharge voltage and oxygen partial pressure is well known during the reactive sputtering from a metallic target. The low oxygen partial pressure regime corresponds with the so-called metallic mode. At high oxygen partial pressure, and low discharge voltage the target is poisoned (an oxide layer covers Y target surface) which can be evaluated from the low deposition rate (see further, Fig. 2). Both modes are separated from each other by the transition region (an oxide layer partially covers target surface). Using only a flow controller, it is impossible to access stable conditions within the transition region. To circumvent the hysteresis phenomenon, several methods have been suggested: nitrogen addition [32], increasing pumping speed [23,31], changing target area [23]. In this work, we obtained a stable process in the transition zone by using a high pumping speed of 500 L/s. Fig. 1(b) and (d) illustrate the behavior of the discharge voltage and oxygen partial pressure at high pumping speed. It is clear that under these conditions the deposition process can be stabilized, allowing to perform depositions at intermediate oxygen partial pressures. Based on these measurements, 11 deposition conditions were chosen. Five points (A1 to A5) were performed at low pumping speed. Six points (B1 to B6) were selected at high pumping speed (see red markers in Fig. 1).

3.2. Deposition rate Fig. 2 shows the measured deposition rate at high and low pumping speed. In metallic mode, a deposition rate of approximate 50 nm/min was obtained. The deposition rate abruptly drops to values as low as 1 nm/min in poisoned mode. The low rate in poisoned mode can be understood from the large difference in sputter yield of the Y metal (~ 0.6) and the oxide (~ 0.015) [33,34].

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

41

closed markers: oxygen flow increase open markers: oxygen flow decrease

discharge voltage (V)

450

450

A1

400

(a)

(b) 400

A2

B1

350

350

300

300

250

B2 B3

250 A4

B4

A5

200

B5

200

B6

A3 150 0.0

1.0

2.0

3.0

-2

oxygen pressure (x10 Pa)

4

150

3

2

2

1

1

2.0

4

6

8

(d)

3

1.0

2

4

(c)

0 0.0

0

3.0

0

0

2

4

6

8

oxygen flow (sccm) Fig. 1. Hysteresis experiment for two pumping speeds: 120 L/s (left panels, (a) and (c)), and 500 L/s (right panels (b) and (d)). The upper panels (a) and (b) show the change of the discharge voltage as function of the oxygen flow, while the bottom panels (c) and (d) show the oxygen partial pressure as a function of the reactive gas flow rate. The indicated points in red, labeled A1 to A5 for the low pumping speed experiments, and B1 to B6 for the high pumping speed experiments, represent the deposition conditions for the thin film deposition.

3.3. Composition To ensure full oxidation of the deposited thin films, even at low oxygen flow, a local reactive gas flow pointing to the substrate was used. XPS was employed to determine the element composition of the yttrium oxide films. The binding energy of the spectra was calibrated using the C1s line (at 284.6 eV) of adventitious carbon. Fig. 3(a) and (b) show the O1s and Y3d region for a sample deposited at an oxygen flow rate of 1.0 sccm and a pumping speed of 120 L/s. The O1s core

2 100

8 6

A1

4

oxygen flow 500 L/s experiment (sccm) 4 5 6

3 A2

7

8

120 L/s experiment (bottom axis) 500 L/s experiment (top axis)

B1

deposition rate (nm/min)

2

10

B2 8 6

B3

4

B4 B5

2

1

A3

3.4. Thin film texture

B6

8 6

A5

4

A4

2

0.1 1.0

1.2

level spectra can be de-convoluted into two peaks at 529.2 and 531.7 eV, corresponding to O–Y bond of yttrium oxide and HO–Y bond [35,36]. The presence of the hydroxide can be understood from its lower heat of formation (− 1435 kJ/mol per Y) as compared to the oxide (− 953 kJ/mol per Y) [35]. Sputter profiling for 5 s and 300 s (not shown here) confirms that the presence of the hydroxide is a pure surface effect, which finds its origin in the exposure to ambient before the XPS analysis was performed. Accordingly, XPS spectra of Y3d core level in Fig. 3(b) show the Y3d3/2 (158.6 eV) and Y3d5/2 (156.6 eV), corresponding to Y–O bond [36], while the other counterpart Y3d3/2 (160.1 eV) and Y3d5/2 (158.1 eV), is in agreement with the Y–OH bond [29]. Similar variation of Y3d spectra upon sputtering can be observed. The chemical composition was calculated from the peak area using the corresponding sensitivity factors. After 300 s of sputter profiling it is clear that the same chemical composition of films can be obtained irrespective of the different oxygen flow rate and target modes using local oxygen supply as shown in Fig. 3(c) and (d). XPS results confirm the fully oxidization of films even in metallic modes.

1.4 1.6 1.8 2.0 2.2 oxygen flow 120 L/s experiment (sscm)

2.4

2.6

Fig. 2. The deposition rate of yttrium oxide thin films prepared at (a) 120 L/s and (b) 500 L/s as a function of oxygen flow rate. The labels correspond to the selected deposition conditions (see Fig. 1). Please note the logarithmic scale for the deposition rate.

In order to investigate the film texture, yttrium oxide films deposited on amorphous glass substrates at different oxygen flow rates and pumping speeds were examined by XRD. Fig. 4(a) and (b) show the XRD patterns of yttrium oxide films with a thickness of approximately 550 and 1000 nm deposited at low pumping speed at 120 L/s. The samples with a thickness of 250 nm showed similar spectra as for the 550 nm. At low oxygen flow rate, i.e., in metallic mode (samples A1 (1.0 sccm) and A2 (1.3 sccm)) the films have single cubic phase (ICDD file no. 00-041-1105) with preferential (222) orientation as well as the cubic (332) and (440) peaks. All the peaks shift approximately 1° to lower diffraction angle as compared to the ICDD card, which can be

42

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

Fig. 3. XPS spectra of a typical sample (thickness 550 nm, deposition conditions A1, see Fig. 1). The (a) O1s and (b) Y3d core level regions after 0 s of sputter profiling. The calculated O/Y ratios for the different deposited samples at (c) low pumping speed and (d) high pumping speed after 0 s, 5 s and 300 s sputtering.

due to the peening effect [37], resulting in a compressive stress and a larger out-of-plane lattice parameter (see also further). Several authors also reported this phenomenon [38]. In contrast, in poisoned mode (samples A3 (1.5 sccm), A4 (1.8 sccm) and A5 (2.5 sccm)), the peaks are located almost at the same positions of the ICDD card, which could be attributed to a lower stress level. At 1.5 sccm, two phases coexists, i.e., the monoclinic phase (ICDD file no. 00-044-0399) and the cubic phase which becomes clearer for thicker films. Based on the structure factor, it is possible to show that in the mixture the monoclinic phase is the most abundant phase. Further increasing the oxygen flow rate, again results in single cubic phase thin films, but with strongly preferred

(420) orientation. Furthermore, no change of the film texture with increasing thickness is noticed for the samples deposited at low pumping speed. The results for the deposition series at high pumping speed of 500 L/s are shown in Fig. 5. Similar to the low pumping speed experiment, at low oxygen flow (sample B1 (2.5 sccm) and B2 (3.2 sccm)) only the cubic phase can be noticed, irrespective of the film thickness. Again the peaks are shifted to lower diffraction angles. Some asymmetry is noticed for the films prepared at 2.5 sccm, which can be attributed to a disorder of the oxygen network [38]. For thickest films the (440) peak becomes clear. From 3.5 to 5 sccm, the mixture of cubic and

Fig. 4. XRD patterns of yttrium oxide thin films with a thickness of approximately (a) 550 and (b) 1000 nm deposited at different oxygen flows and at constant current 0.5 A, constant argon pressure (0.5 Pa) and a fixed pumping speed of 120 L/s.

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

43

Fig. 5. XRD patterns of yttrium oxide thin films with a thickness of approximately (a) 550 and (b) 1000 nm deposited at different oxygen flows and at constant current 0.5 A, constant argon pressure (0.5 Pa) and a fixed pumping speed of 500 L/s.

monoclinic phases can be clearly seen in thick films (1000 nm). As the oxygen flow rate reaches 7.1 sccm, only the cubic phase with strong (420) orientation can be observed. From Fig. 1 one could get the impression that the deposition conditions at low and high pumping speeds are drastically different. However, the two series of depositions have one common deposition parameter, i.e., the oxygen partial pressure. Fig. 6 shows a compilation of the texture analysis discussed above but now as a function of the oxygen partial pressure rather than the oxygen flow. For reasons of convenience, the oxygen flow axes are added to this phase diagram. This way of presenting the data clearly shows that the presence of the cubic or monoclinic phase and the out-of-plane preferential orientation directly correspond to the oxygen pressure. The cubic phase with strong (222) out-of-plane orientation is observed at low oxygen atmosphere, while the mixture of the monoclinic phase is present in the range from 0.84 to 1.25 × 10− 2 Pa. Then, further increasing oxygen pressure results in cubic phase with a preferential (420) orientation. The presence of the cubic phase with (222) out-of-plane orientation

has also been reported by S. Zhang et al. during the deposition of yttria at low oxygen pressure with pulsed laser deposition [26]. This texuture diagram summarizes the crystallographic structure distribution of yttrium oxide films deposited using DC magnetron sputtering via local oxygen supply. It is interesting to notice that the monoclinic phase appears at mild oxygen pressure. Although the real mechanism for the observed results is still under investigation, the role of negative ion bombardment, relative to the deposition rate, can be expected to play a role [39]. Indeed, some authors reported on the presence of the monoclinic phase due to negative ion bombardment during the deposition of erbium sesquioxide [40]. Moreover, several articles reported that ion bombardment can induce unexpected phase transformation and/or a change of the preferential orientation [41,42], in comparison to growth under thermodynamic conditions [20,43]. However, one must also realize that changing from metallic to poisoned mode induces many other changes such as a higher energy per arriving atom which play an important role in the film growth. 3.5. Grain size and microstructure

Fig. 6. Texture diagram of yttrium oxide thin films prepared at high and low pumping speed. The top axis shows the oxygen flows used during the deposition at low pumping speed. The bottom axis corresponds to the oxygen flows used for the high pumping speed experiments. The corresponding oxygen partial pressure during both series of experiments is shown on the middle axis.

The grain size of the deposited films can be evaluated according to the well-known Scherrer's formula. For samples A1, A2, B1, and B2, the cubic (222) peak was selected to calculate the grain size, while this was done using the cubic (420) for samples A4–A5, B6, and using the cubic (222) and monoclinic (202) for samples A3, and B3–B5. Fig. 7 shows the grain size for films with different thickness deposited at the two pumping speeds. For 120 L/s (see Fig. 7(a)), in metallic mode, the films have large grain size of about 60 nm. In poisoned mode the grain size reaches a minimum value (~10 nm), then slightly increases with at higher oxygen flow rates. Films prepared at 500 L/s illustrate similar behavior. As shown in Fig. 7(b), at 2.5 sccm, a grain size of ~30 nm can be obtained, reaching a minimum of about 10 nm, and then slightly increases when oxygen flow rate increases to 7.1 sccm. The similarity between the trends for both pumping speeds is remarkable. Based on the idea discussed in the previous paragraph (see paragraph 3.4, film texture), the grain size has been plotted in

44

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

grain size (nm)

30

cubic (222)

250 nm 550 nm 1000 nm

monoclinic cubic

70

cubic (420)

60

20 50 10 (b) 500 L/s 0 2.5 3.5 4.5 5.5 6.5 7.5 70 250 nm 60 550 nm 50 1000 nm 40 30 20 10 (a) 120 L/s 0 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 0.0 oxygen flow (sccm)

40 30

grain size (nm)

grain size (nm)

40

20 10 (c) 0 0.5 1.0 1.5 2.0 2.5 -2 oxygen pressure (x10 Pa)

Fig. 7. Grain size of the yttrium oxide thin films with a thickness of 250, 550 and 1000 nm prepared at (a) 120 L/s and (b) 500 L/s as a function of oxygen flow rate. Figure (c) shows the same data as function of the oxygen pressure. The striped line is not a fit but merely a guide for the eye.

Fig. 7(c) not as a function of the oxygen flow, but as a function of the oxygen partial pressure. Although the large error on the measured grain size, and the difficulty to control the exact partial oxygen pressure, it is clear that the oxygen pressure seems to control the grain size. Probably the competition between the two different crystal structures is the driving force for the low grain size. The cross-section images of films at 120 L/s are shown in Fig. 8. The films show clear columnar structure at metallic mode, however, inside the poisoned-mode regime, the column structure disappears. This microstructure transformation can have different causes typically for DC magnetron sputtering [44], which needs further investigation. 3.6. Optical and mechanical properties After discussing the chemical and microstructure of the films, linking properties to the structure is an interesting and critical issue for industrial applications. The optical properties of the deposited films were measured by spectroscopic ellipsometry. Building an appropriate model is necessary to extract thickness and optical constants. Firstly, a RCA-cleaned silicon substrate was measured, then, a thermal oxide/Si substrate model was used to fit the measured data. As all the films were fully oxidized and hence transparent films, a Cauchy dispersion function is appropriate for yttrium oxide film. Thus,

a model with four layers (Cauchy/SiO2/INTR/Si substrate) was built to fit all the experimental data from SE and reach a good agreement. The values of refractive index at a wavelength of 550 nm for the samples deposited at 120 L/s were plotted in Fig. 9(a) and (b). Again the same trend is noticed for both experiments. When plotted as a function of the oxygen partial pressure (not shown) a similar conclusion as for the grain size can be drawn, i.e., a correlation between the oxygen partial pressure and the refractive is noticed. Moreover, it is clear that in poisoned mode, i.e., at a high oxygen flow and for a poisoned target, the refractive index is much lower as compared to the bulk value (1.97). This indicates a lower density for these films. A difference between the high energy species bombarding the target is of importance. Indeed as shown by S. Mahieu et al. [45,46] there is a clear relationship between the momentum flux and the film density. In poisoned mode the contribution of reflected neutrals can be ruled out, as the average mass of the atoms in the top layer will drastically decrease. Also the momentum of the sputtered atoms will be lower as the discharge voltage drastically decreases on poisoning the target. However, in poisoned mode, the presence of high energy negative ions has been shown by several authors (see above). Nevertheless, their influence on the density and stress build-up seems to be less important. Indeed, it could be expected that these high energy atoms could also affect the film density. Besides a possible lower flux towards the substrate, the lower efficiency

Fig. 8. Cross-section images of yttrium oxide thin films prepared at 120 L/s deposited at different oxygen flows.

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

45

Fig. 9. (a) the variation of refractive index of films with three different thicknesses deposited at 120 L/s as a function of oxygen flow rate; (b) the change of refractive index of films with different thicknesses deposited at 500 L/s as a function of oxygen flow rate.

in energy transfer due to low mass of the oxygen ions is a possible mechanism which needs further investigation. Fig. 10 gives the load–displacement curves of yttrium oxide films (550 nm) under different modes, which shows the difference in response to diamond tip impression into films. The hard film needs the larger force to get a certain depth, whereas the softer film has the lower force. This feature could reflect the film hardness, and is consistent to the reports [47]. The area between loading and unloading curves corresponds to the energy of plastic deformation. Obviously, the films in metallic mode have larger energy of plastic deformation than that of films in poisoned mode. When the indentation tip moves into the Si substrate, the pop-in phenomenon occurs due to the defects gathering in the front of the indentation [48]. Fig. 11 shows the measured mechanical properties of yttrium oxide films of 550 nm at different target modes. The hardness of films with columnar structure has higher hardness (~ 12 GPa) than those (~6 GPa) of films without columnar structure in Fig. 11(a). The modulus

Fig. 11. (a) the hardness, (b) the modulus and (c) the plasticity of yttrium oxide (~550 nm) films under different modes at 120 L/s.

demonstrates the same trend as well as that of the hardness. Although many reports declared that the crystal phase or chemical bonds played the important role in the film hardness [4], in this work we found that the microstructure of the films give a more critical criterion for the mechanical properties of films. Due to the anisotropy of mechanical properties for columnar-like films, the films with (222) out-of-plane texture give the single crystal-like mechanical properties along the tip moving direction. The superior mechanical properties are mainly attributed to the high density and the high elastic moduli along the (222) crystallographic direction. However, in poisoned mode, the films processing low density and more grain boundaries which is easy to slip along boundaries, resulting in low hardness and modulus. Several studies also reported that the columnar structure of films or composite decided the high mechanical properties [49,50]. The plasticity of materials, defined by the ratio of hardness to the modulus, was evaluated to the erosion quality. We see again the same trends as discussed before, and therefore it is clear that the oxygen partial pressure is an important parameter. 4. Conclusion

Fig. 10. The load–displacement curves of yttrium oxide films (~550 nm) at 120 L/s as a function of oxygen flow rate.

A series of yttrium oxide thin films were prepared on silicon and glass substrates by unbalanced DC magnetron sputtering deposition at different target conditions (metallic, transition and poisoned). XPS shows that the films have the same element composition irrespective of oxygen flow rate. Based on XRD analysis, a phase diagram is formed that intrinsically correlates the structure of the films with the oxygen partial pressure rather than oxygen flow or pumping speed. In metallic mode

46

P. Lei et al. / Surface & Coatings Technology 276 (2015) 39–46

(low oxygen pressure) films have a cubic phase with strong (222) orientation, together with a clear columnar structure. In transition mode (increasing oxygen pressure), there coexist cubic and monoclinic phases in the films with small grain size and no-columnar structure. Further oxygen pressure increase (full poisoned) induces a cubic phase with (420) orientation. The structural features also determine the film properties. Films have a higher refractive index and also high packing density in metallic mode, while the opposite occurs for poisoned mode. Compared with films at poisoned mode, under metallic mode, films have superior mechanical properties, such as high hardness, modulus and strong erosion resistance due to the columnar structure. Acknowledgments The authors thank the short visiting term from Harbin institute of Technology. In addition, this work was supported by the National Natural Science Foundation of China (Grant no. 51372053), the National Natural Science Excellent Young Foundation of China (Grant no. 51222205), the Doctoral Program Foundation of the Ministry of Education of China (20112302110036), and the Heilongjinag outstanding youth science fund (JC201305). References [1] E. Courcot, F. Rebillat, F. Teyssandier, C. Louchet-Pouillerie, Stability of rare earth oxides in a moist environment at elevated temperatures—experimental and thermodynamic studies part II: comparison of the rare earth oxides, J. Eur. Ceram. Soc. 30 (2010) 1911–1917. [2] P. de Rouffignac, J.S. Park, R.G. Gordon, Atomic layer deposition of Y2O3 thin films from yttrium tris(N, N′-diisopropylacetamidinate) and water, Chem. Mater. 17 (2005) 4808–4814. [3] M. Aghazadeh, A.A.M. Barmi, H.M. Shiri, Cathodic electrodeposition and characterization of nanostructured Y2O3 from chloride solution part I: effect of current density, Russ. J. Electrochem. 49 (2013) 344–353. [4] S.A. Barve, Jagannath, N. Mithal, M.N. Deo, A. Biswas, R. Mishra, R. Kishore, B.M. Bhanage, L.M. Gantayet, D.S. Patil, Effects of precursor evaporation temperature on the properties of the yttrium oxide thin films deposited by microwave electron cyclotron resonance plasma assisted metal organic chemical vapor deposition, Thin Solid Films 519 (2011) 3011–3020. [5] K. Verhiest, A. Almazouzi, N. De Wispelaere, R. Petrov, S. Claessens, Development of oxides dispersion strengthened steels for high temperature nuclear reactor application, J. Nucl. Mater. 385 (2009) 308–311. [6] S. Ukai, M. Fujiwara, Perspective of ODS alloys application in nuclear enviroments, J. Nucl. Mater. 307–311 (2002) 749–757. [7] P. Lei, J.Q. Zhu, Y.K. Zhu, C.Z. Jiang, X.B. Yin, Yttrium oxide thin films prepared under different oxygen-content atmosphere: microstructure and optical properties, Appl. Phys. A 108 (2012) 621–628. [8] S.J. Pearce, G.J. Parker, M.D.B. Charlton, J.S. Wilkinson, Structural and optical properties of yttrium oxide thin films for planar waveguiding application, J. Vac. Sci. Technol. A 28 (2010) 1388–1392. [9] O. Pons-Y-Moll, J. Perriere, E. Millon, R.M. Defourneau, D. Defourneau, B. Vincent, A. Essahlaoui, A. Boudrioua, W. Seiler, Structural and optical properties of rare-earthdoped Y2O3 waveguides grown by pulsed-laser deposition, J. Appl. Phys. 92 (2002) 4885–4890. [10] F. Yan, Z.T. Liu, W.T. Liu, The preparation and properties of Y2O3/AlN anti-reflection films on chemical vapor deposition diamond, Thin Solid Films 520 (2011) 734–738. [11] D. Dosev, B. Guo, I.M. Kennedy, Photoluminescence of Eu3+: Y2O3 as an indication of crystal structure and particle size in nanoparticles synthesized by flame spray pyrolysis, J. Aerosol Sci. 37 (2006) 402–412. [12] J.H. Hao, S.A. Studenikin, M. Cocivera, Blue, green and red cathodoluminescence of Y2O3 phosphor films prepared by spray pyrolysis, J. Lumin. 93 (2001) 313–319. [13] H.J. Quah, K.Y. Cheong, Effects of post-deposition annealing ambient on Y2 O 3 gate deposited on silicon by RF magnetron sputtering, J. Alloys Compd. 529 (2012) 73–83. [14] R. Addou, A. Dahal, M. Batzill, Growth of a two-dimensional dielectric monolayer on quasi-freestanding graphene, Nat. Nanotechnol. 8 (2013) 41–45. [15] H.L. Xu, Z.Y. Zhang, H.T. Xu, Z.X. Wang, S. Wang, L.M. Peng, Top-gated graphene field-effect transistors with high normalized transconductance and designable Dirac point voltage, ACS Nano 5 (2011) 5031–5037. [16] H.L. Xu, Z.Y. Zhang, Z.X. Wang, S. Wang, X.L. Liang, L.M. Peng, Quantum capacitance limited vertical scaling of graphene field-effect transistor, ACS Nano 5 (2011) 2340–2347. [17] W.H. Chang, P. Chang, W.C. Lee, T.Y. Lai, J. Kwo, C.H. Hsu, J.M. Hong, M. Hong, Epitaxial stabilization of a monoclinic phase in Y2O3 films on c-plane GaN, J. Cryst. Growth 323 (2011) 107–110.

[18] P.S. Das, G.K. Dalapati, D.Z. Chi, A. Biswas, C.K. Maiti, Characterization of Y2O3 gate dielectric on n-GaAs substrates, Appl. Surf. Sci. 256 (2010) 2245–2251. [19] R.J. Gaboriaud, F. Paumier, M. Jublot, B. Lacroix, Ion irradiation-induced phase transformation mechanisms in Y2O3 thin films, Nucl. Instrum. Methods in Phys. Res. Sect. B 311 (2013) 86–92. [20] M. Zinkevich, Thermodynamics of rare earth sesquioxides, Prog. Mater. Sci. 52 (2007) 597–647. [21] D. Djurovic, M. Zinkevich, F. Aldinger, Thermodynamic modeling of the yttrium– oxygen system, Calphad 31 (2007) 560–566. [22] K. Ellmer, T. Welzel, Reactive magnetron sputtering of transparent conductive oxide thin films: role of energetic particle (ion) bombardment, J. Mater. Res. 27 (2012) 765–779. [23] S. Berg, T. Nyberg, Fundamental understanding and modeling of reactive sputtering processes, Thin Solid Films 476 (2005) 215–230. [24] S.E. Webster, R. Kumaran, S. Penson, T. Tiedje, Structural analysis of thin epitaxial Y2O3 films on sapphire, J. Vac. Sci. Technol. B 28 (2010) (C3A20–C3A23). [25] R.N. Sharma, S.T. Lakshmikumar, A.C. Rastogi, Electrical behavior of electron-beamevaporated yttrium oxide thin films on silicon, Thin Solid Films 199 (1991) 1–8. [26] S.Q. Zhang, R.F. Xiao, Yttrium oxide films prepared by pulsed laser deposition, J. Appl. Phys. 83 (1998) 3842–3848. [27] M.B. Korzenski, Ph. Lecoeur, B. Mercey, D. Chippaux, B. Raveau, R. Desfeux, PLDgrown Y2O3 thin films from Y metal: an advantageous alternative to films deposited from Yttria, Chem. Mater. 12 (2000) 3139–3150. [28] C.V. Ramana, V.H. Mudavakkat, K.K. Bharathi, V.V. Atuchin, L.D. Pokrovsky, V.N. Kruchinin, Enhanced optical constants of nanocrystalline yttrium oxide thin films, Appl. Phys. Lett. 98 (2011) 031905. [29] S. Barve, M. Deo, R. Kar, N. Sreenivasan, R. Kishore, A. Biswas, B. Bhanage, M. Rao, L.M. Gantayet, D. Patil, Microwave ECR plasma assisted MOCVD of Y2O3 thin films using Y(tod)3 precursor and their characterization, Plasma Process. Polym. 8 (2011) 740–749. [30] A.F. Jankowski, L.R. Schrawyer, J.P. Hayes, Sputter deposition of yttrium–oxides, J. Vac. Sci. Technol. A 11 (1993) 1548–1552. [31] I. Safi, Recent aspects concerning DC reactive magnetron sputtering of thin films: a review, Surf. Coat. Technol. 127 (2000) 203–219. [32] D. Severin, O. Kappertz, T. Kubart, T. Nyberg, S. Berg, A. Pflug, M. Siemers, M. Wuttig, Process stabilization and increase of the deposition rate in reactive sputtering of metal oxides and oxynitrides, Appl. Phys. Lett. 88 (2006) 161504. [33] M. Saraiva, V. Georgieva, S. Mahieu, K. Van Aeken, A. Bogaerts, D. Depla, Compositional effects on the growth of Mg(M)O films, J. Appl. Phys. 107 (2010) 034902. [34] K. Strijckmans, W.P. Leroy, R. De Gryse, D. Depla, Modeling reactive magnetron sputtering: fixing the parameter set, Surf. Coat. Technol. 206 (2012) 3666–3675. [35] J.G. Tao, M. Batzill, Ultrathin Y2O3(111) films on Pt(111) substrates, Surf. Sci. 605 (2011) 1826–1833. [36] X.C. Zhang, H.M. Yang, A.D. Tang, Optical, electrochemical and hydrophilic properties of Y 2 O 3 doped TiO 2 nanocomposite films, J. Phys. Chem. B 112 (2008) 16271–16279. [37] F.M. D'Heurle, J.M.E. Harper, Note on the origin of intrinsic stresses in films deposited via evaporation and sputtering, Thin Solid Films 171 (1989) 81–92. [38] R.J. Gaboriaud, F. Paumier, F. Pailloux, P. Guerin, Y2O3 thin films: internal stress and microstructure, Mater. Sci. Eng. B 109 (2004) 34–38. [39] S. Mahieu, W.P. Leroy, K. Van Aeken, D. Depla, Modeling the flux of high energy negative ions during reactive magnetron sputtering, J. Appl. Phys. 106 (2009) 093302. [40] C. Adelhelm, T. Pickert, M. Balden, M. Rasinski, T. Plocinski, C. Ziebert, F. Koch, H. Maier, Monoclinic B-phase erbium sesquioxide (Er2O3) thin films by filtered cathodic arc deposition, Scripta Mater. 61 (2009) 789–792. [41] S. Takayanagi, T. Yanagitani, M. Matsukawa, Unusual growth of polycrystalline oxides film induced by negative ion bombardment in the capacitively coupled plasma deposition, Appl. Phys. Lett. 101 (2012) 232902. [42] T. Yanagitani, M. Kiuchi, Control of in-plane and out-of-plane texture in shear mode piezoelectric ZnO films by ion-beam irradiation, J. Appl. Phys. 102 (2007) 044115. [43] Z.Y. Zhang, M.G. Lagally, Atomistic processes in the early stages of thin-film growth, Science 276 (1997) 377–383. [44] Z.G. Xie, A.M. Allen, M. Chang, P. Wang, T. Gung, Control of bombardment energy and energetic species toward a super dense titanium nitride film, J. Vac. Sci. Technol. A 28 (2010) 1326–1329. [45] S. Mahieu, W.P. Leroy, K. Van Aeken, M. Wolter, J. Colaux, S. Lucas, G. Abadias, P. Matthys, D. Depla, Sputter deposited transition metal nitrides as back electrode for CIGS solar cells, Sol. Energy 85 (2011) 538–544. [46] S. Mahieu, K. Van Aeken, D. Depla, Quantification of the ion and momentum fluxes toward the substrate during reactive magnetron sputtering, J. Appl. Phys. 104 (2008) 113301. [47] J. Musil, J. Vlček, Magnetron sputtering of hard nanocomposite coatings and their properties, Surf. Coat. Technol. 142–144 (2001) 557–566. [48] S.-R. Jian, C.-Y. Huang, W.-C. Ke, Nanoindentation responses of InN thin films, J. Alloys Compd. 609 (2014) 125–128. [49] P. Zeman, R. Čerstvý, P.H. Mayrhofer, C. Mitterer, J. Musil, Structure and properties of hard and superhard Zr–Cu–N nanocomposite coatings, Mater. Sci. Eng. A 289 (2000) 189–197. [50] J. Lin, I. Dahan, B. Valderrama, M.V. Manuel, Structure and properties of uranium oxide thin films deposited by pulsed dc magnetron sputtering, Appl. Surf. Sci. 301 (2014) 475–480.