Benefits of the controlled reactive high-power impulse magnetron sputtering of stoichiometric ZrO2 films

Accepted Manuscript Benefits of the controlled reactive high-power impulse magnetron sputtering of stoichiometric ZrO2 films J. Vlček , J. Rezek , J. Houška , T. Kozák , J. Kohout PII:

S0042-207X(14)00389-3

DOI:

10.1016/j.vacuum.2014.12.004

Reference:

VAC 6486

To appear in:

Vacuum

Received Date: 20 August 2014 Revised Date:

21 November 2014

Accepted Date: 2 December 2014

Please cite this article as: Vlček J, Rezek J, Houška J, Kozák T, Kohout J, Benefits of the controlled reactive high-power impulse magnetron sputtering of stoichiometric ZrO2 films, Vaccum (2015), doi: 10.1016/j.vacuum.2014.12.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Benefits of the controlled reactive high-power impulse magnetron sputtering of stoichiometric ZrO2 films

J. Vlček*, J. Rezek, J. Houška, T. Kozák, J. Kohout

RI PT

Department of Physics and NTIS – European Centre of Excellence, University of West Bohemia, Univerzitní 8, 30614 Plzeň, Czech Republic *

Corresponding author. Tel.: +420 377632200; fax: +420 377632202.

SC

E-mail address: [email protected] (J. Vlček)

M AN U

Abstract

High-power impulse magnetron sputtering with a pulsed reactive gas (oxygen) flow control was used for high-rate reactive depositions of densified, highly optically transparent, stoichiometric ZrO2 films onto floating substrates. The depositions were performed using a strongly unbalanced magnetron with a directly water-cooled planar Zr target of 100 mm

TE D

diameter in argon-oxygen gas mixtures at the argon pressure of 2 Pa. The repetition frequency was 500 Hz at the deposition-averaged target power density from 5 Wcm-2 to 53 Wcm-2. The

EP

voltage pulse durations ranged from 50 µs to 200 µs. The target-to-substrate distance was 100 mm. An optimized location of the oxygen gas inlets in front of the target and their orientation

AC C

toward the substrate made it possible to improve the quality of the films due to minimized arcing at the sputtered target and to enhance their deposition rates up to 120 nm/min at the deposition-averaged target power density of 52 Wcm-2, a voltage pulse duration of 200 µs and a substrate temperature less than 120 °C. The films exhibited a hardness of 16 GPa, a refractive index of 2.19 and an extinction coefficient of 2×10-3 (both at the wavelength of 550 nm). Under these optimized conditions, we measured the highest (Zr+ + Zr2+) and (O2+ + O+) ion fractions in the total fluxes of positive ions, and a low population of high-energy O- ions at the substrate position.

1

ACCEPTED MANUSCRIPT

Keywords: Controlled reactive HiPIMS; Pulsed reactive gas flow; High deposition rates;

AC C

EP

TE D

M AN U

SC

RI PT

Densified ZrO2 films; High optical transparency; Positive ions; O- ions

2

ACCEPTED MANUSCRIPT

1. Introduction

In recent years, high-power impulse magnetron sputtering (HiPIMS) systems have been used for deposition of films (see, for example, [1–3]). The target power density in a pulse of these discharges with a peak value of up to several kWcm−2 is orders of magnitude

RI PT

higher than a typical target power density (usually less than 20 Wcm−2) applied in conventional dc magnetron sputtering. The high target power density leads to generation of very dense discharge plasmas with high degrees of ionization of sputtered atoms.

SC

Consequently, film deposition can be carried out at highly ionized fluxes of the target material atoms. This is of significant interest for deposition on complex-shaped substrates [4–6], for

M AN U

substrate-coating interface engineering [7–9], and ion-assisted growth of films [10–14]. HiPIMS systems have also been applied in the preparation of various optically transparent non-conductive metal oxides, for example, TiO2 [15–19], ZrO2 [20,21], Ta2O5 [21], HfO2 [22] and Nb2O5 [23], of thermochromic VO2 films [24] and optically transparent conductive oxides (TCO), such as InSnO [25], Al-doped ZnO [26], and RuO2 [27]. It has been

TE D

reported that a higher film density, higher index of refraction and lower surface roughness can be obtained for the optical coatings deposited by HiPIMS in comparison with those prepared

EP

by conventional dc magnetron sputtering [15, 16]. Other studies have also shown a lower electrical resistivity of the HiPIMS-deposited TCO films than that of their counterparts

AC C

prepared by conventional dc magnetron sputtering [25, 26]. Many of the technologically important dielectric oxide films are sputtered from metal

targets in a reactive Ar + O2 gas atmosphere to ensure industrially relevant coating deposition rates. In spite of several successful applications of the HiPIMS systems to reactive sputter depositions of dielectric oxide films, there are still substantial problems with arcing on target surfaces during the reactive deposition processes at high target power densities, particularly for voltage pulses longer than 40 µs, and with low deposition rates achieved.

3

ACCEPTED MANUSCRIPT

To avoid these problems, we have proposed a pulsed reactive gas flow control (RGFC) of the reactive HiPIMS processes [21]. It is able to promote sputter deposition of the dielectric stoichiometric films in the transition region between the metallic mode and the compound mode (a fully covered target), and to utilize the following exclusive benefits of the HiPIMS

RI PT

discharges in preparation of films: (i) intense sputtering of atoms from the target resulting in a substantially increased deposition rate, (ii) very high degrees of dissociation of RG molecules in the flux onto the substrate resulting in a higher incorporation of RG atoms into materials,

SC

(iii) a strong “sputtering wind” of the sputtered atoms resulting in a reduced flux of the RG particles onto the target and their enlarged flux onto the substrate, and (iv) highly ionized

M AN U

fluxes of particles onto the substrate and enhanced energies of the ions bombarding the growing films resulting in their structural changes and densification without a substrate bias [1, 2]. The advantage of the proposed pulsed RGFC method is that it does not require any additional measurement or monitoring devices, such as for example plasma emission monitoring system, mass spectrometer or Lambda sensor [28, 29], and that it is applicable to

TE D

all magnetron sputtering discharges.

Fundamental aspects of dc reactive magnetron sputtering, its various applications and

EP

the methods used to control the processes are discussed in Refs. [28–30]. An extensive review of reactive magnetron sputtering is given in Ref. [31]. Valuable results concerning the control

AC C

of reactive HiPIMS processes are presented in Ref. [32]. In our recent paper [33], we reported the use of the pulsed RGFC to a high-rate

reactive HiPIMS of Ta-O-N films with a tunable composition, structure, and optical, electrical and mechanical properties. The present paper is focused on high-rate reactive depositions of densified, highly optically transparent, stoichiometric ZrO2 films using HiPIMS with the pulsed RGFC. The main aim of the study is to investigate the effects of two different configurations (different locations and opposite orientations) of the RG (oxygen) inlets, placed in front of the sputtered

4

ACCEPTED MANUSCRIPT

target, on fundamental deposition characteristics, namely, the deposition rate of films and the deposition-averaged oxygen flow rate, and on energy distribution functions of dominant positive and negative ions at the substrate position. Moreover, we show and explain the effects of the increased target power densities (up to 2.3 kWcm-2) applied during shortened

RI PT

voltage pulses (from 200 µs to 50 µs) on the deposition and ion flux characteristics for both O2 inlet configurations. Motivation for this work is to demonstrate the great potential of the controlled reactive HiPIMS for high-rate depositions of densified dielectric oxide films onto

SC

floating substrates at low substrate temperatures in coating technologies and in microelectronics.

M AN U

The ZrO2 films were chosen in our investigations as an example of a technologically important dielectric material. It should be mentioned that ZrO2 exhibits a high oxidation resistance and thermal stability, high refractive index, broad region of low absorption from the near-UV (above 240 nm) to the mid-IR (below 8 µm), and good ionic conductivity in the Ystabilized cubic phase. This combination of properties is attractive for a wide range of

TE D

applications including laser mirrors, broadband interference filters and ionic conductors [34– 37]. In addition, the ZrO2 thin films with a high dielectric constant of 15-30 are considered as

EP

potential candidates for high-k dielectrics in microelectronics [38]. A high water repellency of

AC C

the ZrO2 films deposited onto ceramics surfaces is presented in Refs. [39, 40].

2. Experimental details 2.1. Film preparation

The films were deposited using a strongly unbalanced magnetron source with a

directly water-cooled planar zirconium target (99.9 % Zr purity, diameter of 100 mm and thickness of 6 mm) in a standard stainless-steel vacuum chamber (diameter of 507 mm and length of 520 mm), which was evacuated by a diffusion pump (2 m3s-1) backed up with a rotary pump (30 m3h-1). The base pressure before deposition was 10-3 Pa. A detailed

5

ACCEPTED MANUSCRIPT

characterization of the magnetic field and the degree of its unbalance is given in Ref. [41]. The strongly unbalanced magnetron system was used to enhance the transport of ionized sputtered atoms along the target normal to the substrate, to reduce the ion losses to chamber walls and to provide an additional electron impact ionization of the sputtered metal atoms and

RI PT

the RG atoms and molecules, and an additional electron impact dissociation of the RG molecules in the bulk of the plasma and the substrate’s vicinity. The water cooling of the target, keeping its surface temperature under 350 °C during the depositions, is described in

SC

Ref. [21].

The magnetron was driven by a high-power pulsed dc power supply (HMP 2/1,

M AN U

Hüttinger Elektronik). In this work, the repetition frequency, fr, was 500 Hz and the voltage pulse duration, t1, ranged from 50 to 200 µs with the corresponding duty cycle t1/T = 2.5 % 10 %, where the pulse period T = 1/fr.

A reactive gas (oxygen) was admitted into the vacuum chamber from a source via mass flow controllers and two corundum conduits (Fig. 1). Two O2 inlets with a diameter of 1

TE D

mm were placed symmetrically above the target racetrack at the same distance of 20 mm from the target surface and oriented toward the target (denoted as “to-target O2 inlet” configuration)

EP

or at the same distance of 25 mm from the target surface and oriented toward the substrate (denoted as “to-substrate O2 inlet” configuration). Here, it should be mentioned that to-target

AC C

oriented RG inlet systems are widely used in industrial deposition devices for reactive magnetron sputtering of dielectric films. The to-substrate O2 inlet configuration with the larger distance of the O2 inlets from the target surface is a result of optimization based on our extensive experiments. They were carried out using the pulsed RGFC system at the average target power density of 50 Wcm-2 during a deposition and the voltage pulse length of 200 µs with the aim to maximize the deposition rate of densified stoichiometric ZrO2 films at a minimized arcing on a compound part of the metal target. Prior to the admission of O2 into the system, the Ar flow rate was set to 30 sccm and

6

ACCEPTED MANUSCRIPT

the pumping speed of the diffusion pump was adjusted to attain the argon partial pressure, par, at the same value of 2 Pa for all the depositions. The settings of the Ar flow rate and the pumping speed were not changed during the experiments. In all cases the total pressure of the argon–oxygen gas mixtures, measured with the accuracy of approximately 1 %, was close to 2

RI PT

Pa (Tables 1 and 2). Waveforms of the magnetron voltage, Ud(t), and the discharge current, Id(t), were monitored, respectively, using a voltage probe (GE 3421, General Elektronik) and a current

SC

probe (TPC 303, Tektronix), which was connected to an amplifier (TCPA 300, Tektronix). The voltage probe and the current probe amplifier were connected to a digital oscilloscope

in a discharge pulse, Sda, given by t

1 1 Sda = ∫ U d (t)J t (t)dt. t1 0

M AN U

(M 621, ETC.) and our own software evaluated the time-varying average target power density

(1)

TE D

Here, the target current density Jt(t) = Id(t) / At, where At is the total area of the target (78.54 cm2 in our case). The deposition-averaged target power density, , was evaluated with the use of the formula

te

∫U

(t)J t (t)dt,

EP

1 < Sd >= te − ts

d

(2)

ts

AC C

where ts and te are the start and end times of the deposition. An analogous integral expression was used to calculate the deposition-averaged oxygen flow rate, <Фox>, from the monitored dependences of Фox(t), which is the instantaneous total oxygen flow rate into the vacuum chamber. During the deposition, Фox(t) = const or 0, see the dotted lines in Fig. 2. The timevarying average discharge current in a period of the power supply, I d , was evaluated using the formula T

1 I d = ∫ I d (t)dt. T0

(3) 7

ACCEPTED MANUSCRIPT

Prior to a given deposition, we fix the nominal target power at an essentially constant magnetron voltage during discharge pulses, and a preset argon partial pressure, total oxygen flow rates in both conduits, the O2 inlet configuration, and pre-selected critical values (Fig. 2) of the average discharge current in a period of the power supply, I d (t ) , which was chosen to

RI PT

be the control process parameter in this work [21]. During the deposition, a process controller used (a programmable logical controller) then provides a control feed-back signal to the two O2 mass flow controllers to adjust the pulsed O2 flow rate into the vacuum chamber by

SC

adjusting the duration (and shape) of the O2 flow rate pulses by means of the pre-selected critical values of I d (t ) .

M AN U

In the present experiments, we used the preset deposition-averaged target power densities = 5.2 Wcm-2, 5.4 Wcm-2 and 50-53 Wcm-2 at a fixed argon partial pressure par = 2 Pa (Tables 1 and 2). For depositions with the to-substrate O2 inlet, we applied constant values of the total oxygen flow rate Фox = 5.6-16.8 sccm in an oxygen flow pulse (Table 2). In

TE D

the case of the to-target O2 inlet, single Фox pulses with constant values of Фox = 4.0-10.0 sccm were used except for = 51 Wcm-2 and t1 = 200 µs where a two-level (19 sccm/9 sccm) Фox pulse needed to be applied (Table 1 and Fig. 2). The values of Фox were

EP

measured with the accuracy of approximately 1 %. A basic principle of the pulsed RGFC is illustrated in Fig. 2. As can be seen, the

AC C

durations of the preset Фox pulses are defined by the pre-selected critical values of I d (t ) , which is monitored by the process controller. For the to-substrate O2 inlet at = 52 Wcm-2 and t1 = 200 µs, the critical values of C1 and C2 = C1 determine the switching-on and switching-off of the Фox flow, respectively. For the to-target O2 inlet at = 51 Wcm-2 and t1 = 200 µs, the Фox flow is switched on when I d (t ) = C1 , switched over to a lower value (9 sccm) when I d (t ) = C 2 = C1 and switched off when I d (t ) = C3 . Under these conditions, the amount of oxygen injected into the discharge is sufficiently low to minimize arcing on the

8

ACCEPTED MANUSCRIPT

compound part of the metal target and to avoid a substantial reduction in the deposition rate of films, but it is sufficiently high to achieve a sufficient incorporation of the oxygen atoms into the films (stoichiometric ZrO2 composition). A detailed description of the control loop is given in Ref. [21].

RI PT

An additional grounded ring-shaped anode, which was periodically cleaned, was mounted into the system around the insulated substrate holder to suppress the “disappearing anode” effect [29].

SC

The ZrO2 films were deposited onto silicon substrates at a floating potential. The target-to-substrate distance, d, was 100 mm. The substrate temperature, Ts, achieved during

M AN U

depositions without an external heater, was in the range from 85 to 200 °C for the to-target O2 inlet, and from 60 to 130 °C for the to-substrate O2 inlet.

The time-averaged energy distributions of positive and negative ions were measured with an energy-resolved mass spectrometer (EQP 300, Hiden Analytical) placed at the targetsubstrate axis in a position directly facing the target surface at the distance of 100 mm from

TE D

the target (see Fig. 1). To perform these measurements under the conditions of very large ion fluxes at high target power densities during pulses, an additional grounded cylindrical

EP

shielding shutter with a 10 mm diameter hole in the center was located 5 mm in front of a special end cap of the mass spectrometer barrel. The diameter (110 mm) of the shielding

AC C

shutter was the same as that of the substrate holder. We used a modified configuration of the cap in which an extraction electrode with a 100 µm diameter orifice in the center was placed 2.5 mm behind a grounded front electrode of the spectrometer with a central input aperture of 1 mm diameter [42]. Only two different parameter settings of the spectrometer (one for all positive ions and one for all negative ions) were used for all discharge regimes investigated in order to have a consistent comparison of signal intensities. In the case of the positive ions, all isotopes of zirconium (90-92,94,96Zr), two isotopes of argon (36Ar and

40

Ar) and one isotope of

oxygen (16O) were considered, together with the corresponding molecular species. Owing to

9

ACCEPTED MANUSCRIPT

the same behaviour of the different isotopes, sums of the corresponding isotope contributions were presented. Each energy distribution of the positive ions was measured up to 180 eV and the integral fluxes of the individual ionic species were determined by a direct integration of the respective time-averaged energy distributions in the range from -5 eV to 180 eV.

RI PT

Assuming the same characteristics for the transmission of all the positive ionic species through the instrument and its energy-independent acceptance cone, the composition of the total flux of positive ions onto the substrate was calculated using the particular integral ion

SC

fluxes determined. The same way was used to normalize the time-averaged energy distribution functions of positive ions measured. In the mass spectrum of the negative ions

M AN U

only O- and O -2 species were observed. However, the corresponding populations of the O -2 ions were 20-40 times lower. Therefore, only the time-averaged energy distributions of the Oions, measured up to 900 eV, are presented in this work.

TE D

2.2. Film characterization

The film thickness (typically between 500 and 1100 nm) was measured at the film edge by profilometry (Dektak 8 Stylus Profiler, Veeco) using a 380 µm thick removable Si

EP

step, and at various places of the film by ellipsometry (described in more detail below). The maximum measurement error of both of these techniques was well below 1%. The

AC C

corresponding deposition rate, aD, was determined as the ratio of the film thickness at the center of the film edge, where the two aforementioned techniques provide almost identical results, to the deposition time. The film thickness, measured by ellipsometry, decreased with an increasing radial distance from the target-substrate holder axis. It was reduced by 9 % at the radial distance of 20 mm. The position of the film edge at the substrate holder was the same in all depositions with the center of the film edge 10 mm from the axis. The thickness uniformity was not optimized in this work. The elemental composition of the films was determined by the Rutherford

10

ACCEPTED MANUSCRIPT

backscattering spectrometry (RBS) and the elastic recoil detection (ERD) methods [43]. The contents of Zr and O were measured by RBS while the content of H was measured by ERD. The RBS and ERD spectra were evaluated by computer codes GISA [44] and SIMNRA [45], respectively. The accuracy of the measurements is approximately 0.1-0.2 at.% for hydrogen

RI PT

and 1–2 at.% for the other elements detected. The surface morphology of the films was determined by atomic force microscopy (AFM) using a SmartSPM Microscope (AIST-NT) with a silicon tip (nominal radius of 10 nm) in non-contact mode. The average roughness of the surface, Ra, was computed from a

SC

randomly selected square area of 5 × 5 µm2.

M AN U

The refractive index, n, and extinction coefficient, k, were determined by variable angle spectroscopic ellipsometry (VASE) using the J.A. Woollam Co. Inc. instrument. The measurements were performed using angles of incidence of 65°, 70° and 75° in reflection. The optical data were fitted in the 300-2000 nm range using the WVASE software and an optical model consisting of a c-Si substrate, a ZrO2 layer described by the Cauchy dispersion

TE D

formula with an Urbach absorption tail and a surface roughness layer. Below we discuss n and k values at specific wavelengths given in the subscript, e.g. n550 and k550 for 550 nm. The

EP

relative measurement error of n550 is similar to that of thickness, i.e. below 1 % (see the discussion above). The relative measurement error of k550 decreases with an increasing value

AC C

of k550 itself, and it is below 10 % for the highest k550 values shown in this work.

(

)

The film hardness, H, and the effective Young's modulus, E * = E 1 − υ 2 , where E

and υ are the Young's modulus and the Poisson's ratio, respectively, were determined using an ultramicroindenter (Fischerscope H-100B) according to the ISO 14577-1:2002 E standard. The measurements were performed with a preset maximum load of 20 mN. The relative measurement errors, determined from 25 measurements at different places of 20 × 20 mm2 samples, are 6 % and 4 % for the H and E* values, respectively.

11

3. Results and discussion

ACCEPTED MANUSCRIPT

In the following, we present the results obtained for high-rate reactive depositions of densified stoichiometric ZrO2 films onto floating substrates using HiPIMS with a pulsed RGFC. The elemental compositions (in at.%) of the films can be characterized as Zr32-34O65-67

RI PT

with low contamination level (H content less than 1 at.%). For two different O2 inlet configurations, i.e. for a to-target O2 inlet configuration and an optimized to-substrate O2 inlet configuration with the larger distance of the O2 inlets from the target surface, the effects of the

SC

increased target power densities (up to 2.3 kWcm-2) applied during shortened voltage pulses (from 200 µs to 50 µs) on deposition and ion flux characteristics, and on film properties are

M AN U

systematically investigated. First, we show and explain the discharge characteristics (Figs. 3 and 4, and Tables 1 and 2) and the fundamental deposition characteristics, namely, the deposition rate of films, aD (Fig. 5), the deposition-averaged oxygen flow rate, <Фox>, and the aD/<Фox> ratio (Fig. 6). Then, we show and explain the normalized time-averaged energy distribution functions of dominant positive ions (Figs. 7 and 8, and Table 3), compositions of

TE D

the total fluxes of the positive ions (Fig. 9) and the time-averaged energy distribution functions of O- ions at a substrate position (Fig. 10). Lastly, we present the optical and

EP

mechanical properties of the ZrO2 films (Tables 1 and 2). In addition, the surface morphology of the ZrO2 films prepared using the optimized to-substrate O2 inlet configuration is shown

AC C

(Fig. 11).

3.1. Discharge characteristics Figs. 3 and 4 show the time evolution of the magnetron voltage, Ud(t), and the target

current density, Jt(t), for the to-target O2 inlet at the preset deposition-averaged target power density = 51 Wcm-2 and for the to-substrate O2 inlet at = 52 Wcm-2, respectively, at a voltage pulse duration t1 = 200 µs relating to the minimum (dashed lines) and maximum (solid lines) values of the average discharge current in a period of the power supply, I d (t ) ,

12

ACCEPTED MANUSCRIPT

which controls the deposition processes (Fig. 2). Thus, Figs. 3 and 4 show the ranges of the waveforms of Ud(t) and Jt(t) during these controlled depositions. As expected, smaller oscillations of the oxygen density in front of the target (Fig. 2) for the to-substrate oriented O2 inlets at a larger distance from the target surface lead to a much narrower range of the average

RI PT

target power density in a discharge pulse, Sda, being between 370 and 540 Wcm-2. Taking into account that the secondary-electron emission coefficient of a Zr target partly covered by an oxide increases with the target coverage [46], the significantly lower values of Sda for the

SC

optimized to-substrate O2 inlet can be explained mainly by a significantly lower target coverage by a Zr oxide during the deposition. The reason for the decreased target coverage by

M AN U

a Zr oxide is that the O2 rarefaction in front of the sputtered target [3] is not compensated by the O2 injection directed (and shifted) toward the substrate in this case. For a discharge pulse with the highest target coverage by a Zr oxide achieved at the maximum pox (maximum I d (t ) ) during the deposition using the to-target O2 inlet, Fig. 3 shows enlarged values of Jt(t) and

TE D

decreasing values of Ud(t), which are caused predominantly by a higher secondary electron emission and a consequent decrease of the plasma impedance. 3.2. Deposition characteristics

EP

As can be seen in Fig. 5, the optimized to-substrate O2 inlet resulted in substantially enhanced (1.6-5.3 times) deposition rates (80-118 nm/min) of stoichiometric ZrO2 films at

AC C

= 51-53 Wcm-2 compared with those achieved using the to-target O2 inlet at = 5053 Wcm-2. This is a consequence of the decreased target coverage by a Zr oxide. An increased metallic fraction of the target results in a more intense sputtering of Zr atoms. This is clearly seen from the big difference between the sputtering yield of Zr atoms from a non-reacted Zr surface and from a ZrO2 surface, being 0.67 and 0.03, respectively, for the energy of dominant Ar+ ions incident on the target Ei = 550 eV (see the values of Uda in Tables 1 and 2). The sputtering yields were calculated using the SDTrimSP program [47]. Note that the decreased target coverage by a Zr oxide is of key importance also for stabilization of the deposition

13

ACCEPTED MANUSCRIPT

processes, as it results in a reduced arcing on the target surface. Assuming that the sputtering of Zr atoms from the metallic fraction of the target is the dominant process determining the deposition rate of the ZrO2 films for the to-substrate O2 inlet at = 51-53 Wcm-2 and that the total flux of secondary electrons and negative

RI PT

oxygen ions (see the low population in Fig. 10) from the sputtered target is negligible in comparison with the flux of positive ions onto the target under these conditions, the corresponding dependence of aD on the duty cycle t1/T in Fig. 5 can be qualitatively explained

SC

using the relation applicable to a non-reactive HiPIMS of metals [48, 49]

aD ∝ α (U d )U d−0.5 , Sd

(4)

M AN U

where Sd is the target power density during a sputter deposition at the magnetron voltage Ud, and α(Ud) is the normalized rate coefficient referring the sputtering and transfer of sputtered target material atoms and ions to the substrate for a high-power dc magnetron sputter deposition to the sputtering and transfer of the sputtered atoms for a hypothetical conventional

atoms.

TE D

dc sputter deposition at the same magnetron voltage Ud and no ionization of the sputtered

A shortening of the voltage pulses from 200 to 50 µs at approximately constant values

EP

of = 51–53 Wcm-2 for the to-substrate O2 inlet (Table 2) resulted in rising values of the

AC C

target power density in a discharge pulse from Sda = 370-540 Wcm-2 at the average magnetron voltage in a discharge pulse Uda = 484-513 V to Sda = 1700-2100 Wcm-2 at Uda = 673-679 V. The rise in Sda leads to rapidly increasing probabilities of ionization of sputtered atoms in front of the target. As a consequence, the rate coefficient α(Ud) in relation (4) decreases mainly due to a higher backward flux of ionized sputtered atoms to the target which may be combined with higher losses of target material ions, compared with neutrals, to chamber walls [48, 49]. Moreover, the values of U da−0.5 in relation (4) also decrease. For comparison, the deposition rates of stoichiometric ZrO2 films achieved for the to-

14

ACCEPTED MANUSCRIPT

target O2 inlet at = 5.2 Wcm-2 and t1 = 200 µs with Sda = 44-51 Wcm-2, and for the tosubstrate O2 inlet at = 5.4 Wcm-2 and t1 = 200 µs with Sda = 35-58 Wcm-2 are also given in Fig. 5. In the latter case, we achieved a very low deposition rate of 5 nm/min, which is even lower than usual values of 10-15 nm/min achieved for stoichiometric ZrO2 films using a

RI PT

conventional dc reactive magnetron sputtering [50]. The larger distance of the to-substrate oriented O2 inlets from the target surface (25 mm), proposed for the high-rate reactive HiPIMS of the ZrO2 films at the much higher = 50 Wcm-2, proved to be unsuitable at a

SC

low value of = 5.4 Wcm-2. It resulted in a weak dissociation of O2 molecules in the flux onto the substrate. As a consequence, the relative consumption of the oxygen gas, needed to

M AN U

ensure the stoichiometric composition of the ZrO2 films, significantly increased (see the lowest value of aD/<Фox> = 3.4 nm min-1/sccm in Fig. 6).

Fig. 6 shows the deposition-averaged oxygen flow rate, <Фox>, characterizing the consumption of oxygen during a reactive deposition, and the deposition rate per depositionaveraged oxygen flow rate, aD/<Фox>, characterizing the efficiency of the oxygen utilization

TE D

in a reactive deposition, for both O2 inlet configurations. As seen in Fig. 6, except for a deposition with the to-target O2 inlet at = 50 Wcm-2 and t1 = 100 µs, we achieved a

EP

higher efficiency of the oxygen utilization at systematically higher values of <Фox> for the depositions with the to-substrate O2 inlet at = 51-53 Wcm-2 than with the to-target O2

AC C

inlet at = 50-53 Wcm-2. A maximum value of aD/<Фox> = 14.2 nm min-1/sccm measured during the mentioned deposition of the ZrO2 film using the to-target O2 inlet at = 50 Wcm-2 and t1 = 100 µs may be explained by a somewhat lower mass density of the film (see

n550 = 2.16 and H = 13 GPa in Table 1, and discussion in Section 3.4.) resulting in a relatively high deposition rate aD = 64 nm/min (Fig. 5) at a low value of <Фox> = 4.5 sccm (Fig. 6).

15

3.3. Ion flux characteristics

ACCEPTED MANUSCRIPT

3.3.1. Positive ions The normalized time-averaged energy distribution functions of dominant positive ions at the substrate position are shown in Figs. 7 (to-target O2 inlet) and 8 (to-substrate O2 inlet).

RI PT

Fig. 9 presents the compositions of the total fluxes of the positive ions for both O2 inlet configurations. In spite of the simplifying assumptions applied in the determination of the contributions of individual ionic species to the total fluxes of positive ions onto the substrate

SC

(see Section 2.1.), the obtained results provide valuable qualitative information on changes in the fractions of individual positive ions in the total fluxes of the positive ions at the substrate

M AN U

position under the experimental conditions investigated.

As can be seen in Figs. 7-9 and 5, we observed a correlation between the (Zr+ + Zr2+) ion fractions in the total fluxes of positive ions and the deposition rates of the ZrO2 films. As expected, the ( O +2 + O+) ion fractions in the total fluxes were systematically higher for the to-

TE D

substrate O2 inlet except for a deposition at = 53 Wcm-2 and t1 = 50 µs, when we achieved the same ( O +2 + O+) ion fractions (Fig. 9) for both O2 inlet configurations. For these shorter voltage pulses, we observed a predominance of the O+ ions over the O +2 ions in the

EP

flux onto the substrate (Figs. 7-9) due to a very strong dissociation of O2 molecules in a discharge plasma at high values of Sda being close to 2 kWcm-2 (Tables 1 and 2).

AC C

As is typical for HiPIMS discharges [3], the energy distributions of the positive ions

measured at the substrate position for the average target power density close to 50 Wcm-2 during a deposition are extended to higher energies (Figs. 7 and 8). This results, together with the enlarged fractions of ionized sputtered metal atoms in the total fluxes of positive ions onto the substrate (Fig. 9), in a densification of the films prepared (Section 3.4.). The high-energy parts of the time-averaged energy distributions of Ar+ and O+ ions are more pronounced for the to-substrate O2 inlet, particularly at = 53 Wcm-2 and t1 = 50 µs (Fig. 8). We believe that localized, 10-50 V potential humps formed in ionization zones, 16

ACCEPTED MANUSCRIPT

which are driven by energetic drifting electrons in front of the sputtered target, can contribute to this effect [51-53]. However, the main contribution is probably given by a backscattering (reflection) of light O (mO = 16.00) and Ar (mAr = 39.95) atoms from the sputtered Zr target (mZr = 91.22) with a small coverage by a Zr oxide. According to our model predictions

RI PT

(unpublished), the coverages of the Zr target by ZrO2 at = 50 Wcm-2 and t1 = 200 µs are approximately 15 % and 30 % for the to-substrate and the to-target O2 inlet, respectively. Both these values increased by approximately 5 % at = 50 Wcm-2 and t1 = 50 µs.

SC

In Table 3, we give the backscattering probability of atoms, B, and the mean energy of backscattered atoms at the target surface, Eb, calculated for Ar, Zr and O atoms backscattered

M AN U

from Zr and ZrO2 targets after their bombardment by the respective Ar+, Zr+ and O+ ions. The calculations were carried out using the SDTrimSP program [47]. As can be seen in Table 3, for the energy of ions incident on the target Ei = 600 eV (see Uda = 576-619 V for the to-target O2 inlet in Table 1 and Uda = 673-679 V for the to-substrate O2 inlet in Table 2 at = 53 Wcm-2 and t1 = 50 µs), 18 % of the incident Ar+ ions and 30 % of the incident O+ ions leave a

TE D

pure Zr target surface as neutral atoms with a mean energy of 88 eV for Ar atoms and even 203 eV for O atoms. In the case of a fully oxidized ZrO2 target, the backscattering

EP

probabilities would decrease almost 3 times and the mean energies of the backscattered atoms would decrease by almost 25 %. Let us recall that the occurrence of O atoms and O+ ions in

AC C

the discharge plasma increases significantly at = 53 Wcm-2 and t1 = 50 µs due to the strong dissociation of O2 molecules at high Sda values. Note that we observed a pronounced high-energy part of the normalized time-averaged

energy distribution of Ar+ ions for the to-substrate O2 inlet also at the low value of = 5.4 Wcm-2 (Fig. 8), when the total number of not only the Ar+ ions but particularly of the O+ ions in the discharge plasma is significantly reduced.

17

ACCEPTED MANUSCRIPT

3.3.2. O- ions

The time-averaged energy distribution functions of the O- ions at the substrate position are shown in Fig. 10. For the to-target O2 inlet, when the coverages of the sputtered Zr targets by a Zr oxide are supposed to be higher at ≈ 50 Wcm-2 than those achieved with the to-

RI PT

substrate O2 inlet, we observed a typical structure of the O- energy distributions measured in reactive magnetron sputtering [56-59]. Three ion populations were identified: low, medium and high energy ions. The high-energy peaks coincide with absolute values of the corresponding target potentials during voltage pulses, see Uda = 403 – 408 V at = 5.2 W

M AN U

619 V at = 53 Wcm-2 and t1 = 50 µs (Table 1).

SC

cm-2 and t1 = 200 µs, Uda = 453 – 516 V at = 51 Wcm-2 and t1 = 200 µs, and Uda = 576 –

The high-energy O- ions are formed as a consequence of electron tunneling [58, 60] or attachment to O atoms at the target surface. The surface O- ions are subsequently sputtered (possibly backscattered) and accelerated by the full target potential in the cathode sheath. They have gained additional kinetic energy provided by the sputtering or backscattering

TE D

events and lose energy due to their collisions with the background gas (par = 2 Pa in our case) during their transport to the substrate [58, 59]. It should be mentioned that the energy of these

EP

O- ions is high enough to cause defects and resputtering of growing films [56, 59]. The O- ions contributing to the low-energy peak with an energy less than 100 eV (Fig.

AC C

10) are either generated at the target surface and undergo energy losses due to collisions or are formed in the discharge plasma via electron attachment to sputtered or backscattered O atoms. Low-energy O- ions may be also produced inside the target sheath or pre-sheath via electron attachment or dissociative electron attachment processes rather than at the target surface and are hence accelerated by only a fraction of the cathode potential [57, 59]. The medium-energy O- ions may be generated by a collision induced dissociation of O2- ions accelerated by the full target potential. The energy of the O2- ions is then equally shared (based on the mass ratio) between the O atom and the O- ion [57]. Therefore, the O-

18

ACCEPTED MANUSCRIPT

ions are detected at the energies which correspond to half of the respective target potential (see Fig. 10). Fig. 10 also shows a changed shape of the time-averaged energy distributions of the Oions measured at the substrate position for the optimized to-substrate O2 inlet with ≈

RI PT

50 Wcm-2. A strongly reduced (approximately 10 times) population of high-energy O- ions observed at = 52 Wcm-2 and t1 = 200 µs, in comparison with that obtained for the totarget O2 inlet at = 51 Wcm-2 and t1 = 200 µs, is a natural consequence of the decreased

SC

target coverage by a Zr oxide [56, 58]. In spite of a relatively high noise level of the spectrometer for negative ion detection (see Fig. 10 and Ref. [56]), it seems that energy of the

M AN U

high-energy O- ions is shifted to higher values in the case of the to-substrate O2 inlet at = 52 Wcm-2 and t1 = 200 µs, when Uda = 484 – 513 V, and at = 53 Wcm-2 and t1 = 50 µs, when Uda = 673 – 679 V (Table 2). This might be caused by the additional energy of the O-

3.4. Film properties

TE D

ions at the target surface provided by the backscattering of O atoms from the target (Table 3).

Fig. 11 presents typical AFM images of surfaces of the stoichiometric ZrO2 films

EP

prepared using the to-substrate O2 inlet for which the amount of the high-energy O- ions at the substrate is low (Fig. 10). We observed a correlation between the energies of positive ions

AC C

bombarding the growing films (Fig. 8) and the roughness of the films. The average surface roughness increased from Ra = 1.3 nm and 1.4 nm for = 5.4 Wcm-2 and 52 Wcm-2, respectively, at t1 = 200 µs to Ra = 9.5 nm for = 53 Wcm-2 at t1 = 50 µs for which we measured the pronounced high-energy parts of the Ar+ and O+ energy distributions at the substrate. Optical and mechanical properties of the ZrO2 films are presented in Tables 1 (totarget O2 inlet) and 2 (to-substrate O2 inlet). The high values of n550 up to 2.21 (or n633 up to 2.19) prove that the films prepared at ≈ 50 Wcm-2 are fully densified for the to-substrate

19

ACCEPTED MANUSCRIPT

O2 inlet at t1 ≥ 100 µs and for the to-target O2 inlet at t1 = 200 µs. The densification is confirmed by a comparison of the presented refractive indices with those reported for both bulk ZrO2 (n550 = 2.17 [61]) and ZrO2 thin films (n = 2.11 [62], n550 ≤ 2.15 [63], n550 ≤ 2.17 [64], n633 ≤ 2.19, being the limit of a dependence of the refractive index on packing density

RI PT

[65], and n550 ≤ 2.22 [37]), in the literature. In parallel, the low values of k550 (1×10-3 - 5×10-3 at n550 = 2.18-2.21) prove that the films are stoichiometric (i.e. the high n values are indeed due to the densification) and highly optically transparent at least in the aforementioned

SC

thickness range (Section 2.2.). For example, note that for a 1 µm thick slab with n550 = 2.20 the transmittance of a normal beam with a wavelength of 550 nm would be 75 % at k550 = 0

M AN U

and 67 % at k550 = 5×10-3 without considering the interference. For a 500 nm thick slab, the transmittance would increase to 71 %. It means that even for this highest k achieved, the absorption would still be clearly lower than reflection taking place at any k under consideration. Despite the very high deposition rates, the k550 values achieved are on the same order as e.g. k550 ≤ 2.5×10-3 reported for ZrO2 films with similar n550 ≤ 2.22 prepared using rf

TE D

magnetron sputtering [37]. Even lower values of k550 on the order of 1×10-4 were achieved (particularly when using the to-substrate O2 inlet; see Table 2) for = 5.4 Wcm-2 and t1 =

EP

200 µs at the expense of a much lower deposition rate (Fig. 5) or for = 53 Wcm-2 and t1 = 80 µs at the expense of a moderate densification (n550 = 1.97).

AC C

Furthermore, Tables 1 and 2 show an almost monotonic relationship between the refractive index (measure of the densification) and material hardness (these relationships are similar for both O2 inlet configurations) and between the refractive index and the effective Young's modulus (at higher E* values for the to-substrate O2 inlet). The H values range from 9 GPa (exhibited by the least densified sample with n550 = 1.97) to 16 GPa (exhibited by the three most densified samples with n550 = 2.19-2.21). The maximum H of 16 GPa is comparable even with values reported for Y-stabilized high-density phases of ZrO2 (H ≤ 17 GPa for the cubic phase in Ref. [66], H ≤ 14 GPa for a not fully densified tetragonal phase

20

ACCEPTED MANUSCRIPT

in Ref. [67] and H ≤ 17 GPa for the tetragonal phase densified by a slight Al doping in Ref. [67]). This constitutes another confirmation that the films with n550 = 2.19 - 2.21 are fully densified. In spite of the pox oscillations during film depositions (Tables 1 and 2), no multilayer

RI PT

structures of the films or periodic changes in their compositions and properties were observed.

4. Conclusions

SC

We have proposed a pulsed reactive gas flow control (RGFC) of the reactive highpower impulse magnetron sputtering (HiPIMS) to avoid substantial problems with arcing on

M AN U

target surfaces during reactive sputter depositions of dielectric films at high target power densities and with low deposition rates achieved [21]. Using this process control, we are able to maintain a sputter deposition of the dielectric stoichiometric films in the region between a more and less metallic mode, and to utilize exclusive benefits of the HiPIMS discharges in preparation of the films.

TE D

In this work, HiPIMS with the pulsed RGFC was used for high-rate reactive depositions of densified, highly optically transparent ZrO2 films onto floating substrates at the

EP

distance d = 100 mm from the target. An optimized location of the oxygen gas inlets in front of the target and their orientation toward the substrate made it possible to improve quality of

AC C

the films due to minimized arcing at the sputtered target and to enhance their deposition rates up to 120 nm/min at the deposition-averaged target power density = 52 Wcm-2, the voltage pulse duration t1 = 200 µs and the substrate temperature Ts < 120 °C. The films exhibited a hardness of 16 GPa, a refractive index of 2.19 and an extinction coefficient of 2×10-3 (both at the wavelength of 550 nm). We have thus demonstrated a great potential of the controlled reactive HiPIMS for high-rate depositions of densified dielectric oxide films onto floating substrates at low substrate temperatures in coating technologies and in microelectronics.

21

ACCEPTED MANUSCRIPT Acknowledgment This work was supported by the Grant Agency of the Czech Republic under Project

RI PT

No. GA14 – 03875S.

References

[1] K. Sarakinos, J. Alami, S. Konstantinidis, High power pulsed magnetron sputtering: A

SC

review on scientific and engineering state of the art, Surf. Coat. Technol. 204 (2010) 16611684.

M AN U

[2] D. Lundin, K. Sarakinos, An introduction to thin film processing using high-power impulse magnetron sputtering, J. Mater. Res. 27 (2012) 780-792. [3] J. T. Gudmundsson, N. Brenning, D. Lundin, U. Helmersson, High power impulse magnetron sputtering discharge, J. Vac. Sci. Technol. A30 (2012) 030801. [4] V. Kouznetsov, K. Macák, J.M. Schneider, U. Helmersson, I. Petrov, A novel pulsed

122 (1999) 290-293.

TE D

magnetron sputter technique utilizing very high target power densities, Surf. Coat. Technol.

EP

[5] J. Alami, P.O.A. Persson, D. Music, J.T. Gudmundsson, J. Bohlmark, U. Helmersson, Ionassisted physical vapor deposition for enhanced film properties on nonflat surfaces, J. Vac.

AC C

Sci. Technol. A 23 (2005) 278-280. [6] K. Bobzin, N. Bagcivan, P. Immich, S. Bolz, J. Alami, R. Cremer, Advantages of nanocomposite coatings deposited by high power pulse magnetron sputtering technology, J. Mater. Process. Technol. 209 (2009) 65-70. [7] A.P. Ehiasarian, W.-D. Münz, L. Hultman, U. Helmersson, I. Petrov, High power pulsed magnetron sputtered CrNx films, Surf. Coat. Technol. 163-164 (2003) 267-272. [8] P.E. Hovsepian, C. Reinhard, A.P.Ehiasarian, CrAlYN/CrN superlattice coatings deposited by the combined high power impulse magnetron sputtering/unbalanced magnetron sputtering

22

ACCEPTED MANUSCRIPT

technique, Surf. Coat. Technol. 201 (2006) 4105-4110.

[9] A.P.Ehiasarian, J.G. Wen, I. Petrov, Interface microstructure engineering by high power impulse magnetron sputtering for the enhancement of adhesion, J. Appl. Phys. 101 (2007) 054301.

RI PT

[10] A.P. Ehiasarian, P.E. Hovsepian, L. Hultman, U. Helmersson, Comparison of microstructure and mechanical properties of chromium nitride-based coatings deposited by high power impulse magnetron sputtering and by the combined steered cathodic

SC

arc/unbalanced magnetron technique, Thin Solid Films 457 (2004) 270-277.

[11] J. Alami, P. Eklund, J. Emmerlich, O. Wilhelmsson, U. Jansson, H. Högberg, L. Hultman,

M AN U

U. Helmersson, High-power impulse magnetron sputtering of Ti-Si-C thin films from a Ti3SiC2 compound target, Thin Solid Films 515 (2006) 1731-1736. [12] J. Alami, P. Eklund, J.M. Andersson, M. Lattemann, E. Wallin, J. Bohlmark, P. Persson, U. Helmersson, Phase tailoring of Ta thin films by highly ionized pulsed magnetron sputtering, Thin Solid Films 515 (2007) 3434-3438.

TE D

[13] J. Alami, K. Sarakinos, F. Uslu, M. Wuttig, On the relationship between the peak target current and the morphology of chromium nitride thin films deposited by reactive high power

EP

pulsed magnetron sputtering, J. Phys. D: Appl. Phys. 42 (2009) 015304. [14] G. Greczynski, J. Jensen, J. Böhlmark, L. Hultman, Microstructure control of CrNx films

AC C

during high power impulse magnetron sputtering, Surf. Coat. Technol. 205 (2010) 118-130. [15] S. Konstantinidis, J.P. Dauchot, M. Hecq, Titanium oxide thin films deposited by highpower impulse magnetron sputtering, Thin Solid Films 515 (2006) 1182-6. [16] K. Sarakinos, J. Alami, M. Wuttig, Process characteristics and film properties upon growth of TiOx films by high power pulsed magnetron sputtering, J. Phys. D: Appl. Phys. 40 (2007) 2108-14. [17] V. Straňák, M.Quaas, H. Wulff, Z. Hubička, S. Wrehde, M. Tichý, R. Hippler, Formation of TiOx films produced by high-power pulsed magnetron sputtering, J. Phys. D: Appl. Phys.

23

ACCEPTED MANUSCRIPT

41 (2008) 055202.

[18] J. Alami, K. Sarakinos, F. Uslu, C. Klever, J. Dukwen, M. Wuttig, On the phase formation of titanium oxide films grown by reactive high power pulsed magnetron sputtering, J. Phys. D: Appl. Phys. 42 (2009) 115204.

RI PT

[19] A. Amin, D. Köhl, M. Wuttig, The role of energetic ion bombardment during growth of TiO2 thin films by reactive sputtering, J. Phys. D: Appl. Phys. 43 (2010) 405303.

[20] K. Sarakinos, J. Alami, C. Klever, M. Wuttig, Process stabilization and enhancement of

SC

deposition rate during reactive high power pulsed magnetron sputtering of zirconium oxide, Surf. Coat. Technol. 202 (2008) 5033-5.

M AN U

[21] J. Vlček, J. Rezek, J. Houška, R. Čerstvý, R. Bugyi, Process stabilization and a significant enhancement of the deposition rate in reactive high-power impulse magnetron sputtering of ZrO2 and Ta2O5 films, Surf. Coat. Technol. 236 (2013) 550-556. [22] K. Sarakinos et al., On the phase formation of sputtered hafnium oxide and oxynitride films, J. Appl. Phys. 108 (2010) 014904.

TE D

[23] M. Hála, J. Čapek, O. Zabeida, J.E. Klemberg-Sapieha, L. Martinu, Hysteresis-free deposition of niobium oxide films by HiPIMS using different pulse management strategies, J.

EP

Phys. D: Appl. Phys. 45 (2012) 055204.

[24] J.-P. Fortier, B. Baloukas. O Zabeida, J.E. Klemberg-Sapieha, L. Martinu,

AC C

Thermochromic VO2 thin films deposited by HiPIMS, Sol. Energy Mater. Sol. Cells 125 (2014) 291-299.

[25] V. Sittinger, F. Ruske, W. Werner, C. Jacobs, B. Szyszka, D.J. Christie, High power pulsed magnetron sputtering of transparent conducting oxides, Thin Solid Films 516 (2008) 5847-59. [26] F. Ruske, A. Pflug, V. Sittinger, W. Werner, B. Szyszka, D.J. Christie, Reactive deposition of aluminium-doped zinc oxide thin films using high power pulsed magnetron sputtering, Thin Solid Films 516 (2008) 4472-7.

24

ACCEPTED MANUSCRIPT

[27] D. Benzeggouta, M.C. Hugon, J. Bretagne, Study of a HPPMS discharge in Ar/O2 mixture: II. Plasma optical emission and deposited RuOx film properties, Plasma Sources Sci. Technol. 18 (2009) 045026. [28] I. Safi, Recent aspects concerning DC reactive magnetron sputtering of thin films: a

RI PT

review, Surf. Coat. Technol. 127 (2000) 203-218. [29] W. D. Sproul, D. J. Christie, D.C. Carter, Control of reactive sputtering processes, Thin Solid Films 491 (2005) 1-17.

SC

[30] J. Musil, P. Baroch, J. Vlček, K. H. Nam, J. G. Han, Reactive magnetron sputtering of thin films: present status and trends, Thin Solid Films 475 (2005) 208-218.

science, vol. 109, Berlin 2008.

M AN U

[31] D. Depla and S. Mahieu (Eds.), Reactive Sputter Deposition, Springer Series in Materials

[32] M. Audronis, V. Bellido-Gonzales, B. Daniel, Control of reactive high power impulse magnetron sputtering processes, Surf. Coat. Technol. 204 (2010) 2159-2164. [33] J. Rezek, J. Vlček, J. Houška, R. Čerstvý, High-rate reactive high-power impulse

TE D

magnetron sputtering of Ta-O-N films with tunable composition and properties, Thin Solid Films 566 (2014) 70-77.

EP

[34] V.M.M. Mercier, P. Van der Sluis, Toward solid-state switchable mirrors using a zirconium oxide proton conductor, Solid State Ionics 145 (2001) 17-24.

AC C

[35] S. Venkataraj, O. Kappertz, H. Weis, R. Drese, R. Jayavel, M. Wuttig, Structural and optical properties of thin zirconium oxide films prepared by reactive direct current magnetron sputtering, J. Appl. Phys. 92 (2002) 3599-3607. [36] P.T. Gao, L.J. Meng, M.P. Dos Santos, V. Teixeira, M. Andritschky, Characterisation of ZrO2 films prepared by rf reactive sputtering at different O2 concentrations in the sputtering gases, Vacuum 56 (2000) 143-148. [37] H.H. Zhang, C.Y. Ma, Q.Y. Zhang, Scaling behavior and structure transition of ZrO2 films deposited by RF magnetron sputtering, Vacuum 83 (2009) 1311-1316.

25

ACCEPTED MANUSCRIPT

[38] G.D. Wilk, R.M. Wallace, J.M. Anthony, High-k gate dielectrics: Current status and materials properties considerations, J. Appl. Phys. 89 (2001) 5243-5275. [39] Y. Ohtsu, M. Egami, H. Fujita, K. Yukimura, Preparation of zirconium oxide thin film

RI PT

using inductively coupled oxygen plasma sputtering, Surf. Coat. Technol. 196 (2005) 81 - 84.

[40] Y. Ohtsu, Y. Hino, T. Misawa, H. Fujita, K. Yukimura, M. Akiyama, Influence of ionbombardment-energy on thin zirconium oxide films prepared by dual frequency oxygen

SC

plasma sputtering, Surf. Coat. Technol. 201 (2007) 6627 – 6630.

M AN U

[41] P. Kudláček, J. Vlček, K. Burcalová, J. Lukáš, Highly ionized fluxes of sputtered titanium atoms in high-power pulsed magnetron discharges, Plasma Sources Sci. Technol. 17 (2008) 025010.

[42] J. Lazar, J. Vlček, J. Rezek, Ion flux characteristics and efficiency of the deposition

(2010) 063307.

TE D

processes in high power impulse magnetron sputtering of zirconium, J. Appl. Phys. 108

[43] In: J. R. Tesmer, M. Nastasi (Eds.), Handbook of Modern Ion Beam Material Analysis,

EP

Material Research Society, Pittsburgh, PA, 1995. [44] J. Saarilahti, E. Rauhala, Interactive personal-computer data analysis of ion

AC C

backscattering spectra, Nucl. Inst. Methods Phys. Res. B 64 (1992) 734-738. [45] M. Mayer, SIMNRA, User’s Guide, Forschungscentrum Julich, Institute for Plasma Physics, Germany, 1998. [46] D. Depla, S. Heirwegh, S. Mahieu, J. Haemers, R. De Gryse, Understanding the discharge voltage behavior during reactive sputtering of oxides, J. Appl. Phys. 101 (2007) 013301.

26

ACCEPTED MANUSCRIPT

[47] W. Eckstein, R. Dohmen, A. Mutzke, R. Schneider, Sdtrimsp: A monte-carlo code for calculating collision phenomena in randomized targets, Technical report, Max-PlanckInstitute fur Plasmaphysik, 2007. [48] J. Vlček, K. Burcalová, A phenomenological equilibrium model applicable to high-power

RI PT

pulsed magnetron sputtering, Plasma Sources Sci. Technol. 19 (2010) 065010. [49] T. Kozák, J. Vlček, Š. Kos, Transport and ionization of sputtered atoms in high-power impulse magnetron sputtering discharges, J. Phys. D, Appl. Phys. 46 (2013) 105203.

SC

[50] S. Venkataraj, O. Kappertz, H. Weiss, R. Drese, R. Jayavel, M. Wuttig, Structural and optical properties of thin zirconium oxide films prepared by reactive direct current magnetron

M AN U

sputtering, J. Appl. Phys. 92 (2002) 3599-3607.

[51] A. Anders, M. Panjan, R. Franz, J. Andersson, P. Ni, Drifting potential humps in ionization zones: The “propeller blades” of high power impulse magnetron sputtering, Appl. Phys. Lett 103 (2013) 144103.

[52] M. Panjan, R. Franz, A. Anders, Asymmetric particle fluxes from drifting ionization

TE D

zones in sputtering magnetrons, Plasma Sources Sci. Technol. 23 (2014) 025007. [53] C. Maszl, W. Breilmann, J. Benedikt, A. von Keudell, Origin of the energetic ions at the

EP

substrate generated during high power pulsed magnetron sputtering of titanium, J. Phys. D: Appl. Phys. 47 (2014) 224002.

AC C

[54] W. D. Wilson, L. G. Haggmark, J. P. Biersack, Calculations of nuclear stopping, ranges, and straggling in the low-energy region, Phys. Rev. B 15 (1977) 2458-2468. [55] W. Moller, W. Eckstein, J.P. Biersack, Comput. Phys. Commun. 51 (1988) 355-368; W. Moller, M. Posselt, TRIDYN_FZR User Manual. [56] M. Zeuner, H. Neumann, J. Zalman, H. Biederman, Sputter process diagnostics by negative ions, J. Appl. Phys. 83 (1998) 5083-5086. [57] S. Mráz, J.M. Schneider, Influence of the negative oxygen ions on the structure evolution of transition metal oxide thin films, J. Appl. Phys. 100 (2006) 023503.

27

ACCEPTED MANUSCRIPT

[58] S. Mahieu, W.P. Leroy, K. Van Aeken, D. Depla, Modelling the flux of high energy negative ions during reactive magnetron sputtering, J.Appl. Phys. 106 (2009) 093302. [59] M. Bowes, P. Poolcharuansin, J.W. Bradley, Negative ion energy distributions in reactive HiPIMS, J.Phys. D: Appl. Phys. 46 (2013) 045204.

RI PT

[60] D. Depla, S. Mahieu, R. De Gryse, Magnetron sputter deposition: Linking discharge voltage with target properties, Thin Solid Films 517 (2009) 2825-2839.

[61] G. Buxbaum, G. Pfaff, Industrial Inorganic Pigments, Wiley-WCh, Weinheim, 2005.

SC

[62] I. John Berlin, S. Sujatha lekshmy, V. Ganesan, P.V. Thomas, K. Joy, Effect of Mn doping on the structural and optical properties of ZrO2 thin films prepared by sol–gel method, Thin

M AN U

Solid Films 550 (2014) 199-205.

[63] D. Franta, I. Ohlidal, P. Klapetek, P. Pokorny, M. Ohlidal, Analysis of inhomogeneous thin films of ZrO2 by the combined optical method and atomic force microscopy, Surf. Interface Anal. 32 (2001) 91-94.

[64] S. Zhao, F. Ma, K.W. Xu, H.F. Liang, Optical properties and structural characterization of

TE D

bias sputtered ZrO2 films, J. Alloys Compd. 453 (2008) 453-457. [65] S. Ben Amor, B. Rogier, G. Baud, M. Jacquet, M. Nardin, Characterization of zirconia

EP

films deposited by r.f. magnetron sputtering, Mat. Sci. Eng. B57 (1998) 28-39. [66] P. Yashar, J. Rechner, M.S. Wong, W.D. Sproul, S.A. Barnett, High-rate reactive

AC C

sputtering of yttria-stabilized zirconia using pulsed d.c. power, Surf. Coat. Technol. 9495(1997) 333-338.

[67] O. Vasylkiv, Y. Sakka, V.V. Skorokhod, Hardness and fracture toughness of aluminadoped tetragonal zirconia with different yttria contents, Mater. Trans. JIM 44 (2003) 22352238.

28

ACCEPTED MANUSCRIPT

RI PT

Table 1. Process parameters and material characteristics of the stoichiometric ZrO2 films for the to-target O2 inlet at the preset deposition-averaged target power densities = 5.2 Wcm-2 and 50 - 53 Wcm-2, voltage pulse durations, t1, from 50 to 200 µs, and a fixed argon partial pressure par = 2 Pa. The oxygen partial pressures, pox, oscillated between 0 and 0.08 Pa. Here, Sda is the average target power density in a discharge pulse, Uda is the corresponding average magnetron voltage in a discharge pulse, Φox is the oxygen flow rate in an oxygen flow pulse, k550 and n550 are the extinction coefficient and refractive index of the films at a wavelength of 550 nm, respectively, H is the hardness of the films and E* is their effective Young’s modulus.

Material characteristics

t1 (µs) 200

(Wcm-2) 5.2

Sda (Wcm-2) 44-51

Uda (V) 403-408

Φox (sccm) 4.0

k550 (10-3)

n550

0.6

2.02

200

51.0

350-770

516-453

19.0/9.0

100

50.0

780-1040

531-521

10.0

80

52.0

910-1420

561-547

10.0

SC

Process parameters

50

53.0

1650-2270

619-576

10.0

E* (GPa) 125

4.0

2.21

16

146

1.0

2.16

13

140

1.0

2.18

11

128

1.0

2.11

13

139

M AN U

TE D EP AC C 29

H (GPa) 10

ACCEPTED MANUSCRIPT

Process parameters

RI PT

Table 2. Process parameters and material characteristics of the stoichiometric ZrO2 films for the to-substrate O2 inlet at the preset deposition-averaged target power densities = 5.4 Wcm-2 and 51 - 53 Wcm-2, voltage pulse durations, t1, from 50 to 200 µs, and a fixed argon partial pressure par = 2 Pa. The oxygen partial pressures, pox, oscillated between 0 and 0.08 Pa. Here, Sda is the average target power density in a discharge pulse, Uda is the corresponding average magnetron voltage in a discharge pulse, Φox is the oxygen flow rate in an oxygen flow pulse, k550 and n550 are the extinction coefficient and refractive index of the films at a wavelength of 550 nm, respectively, H is the hardness of the films and E* is their effective Young’s modulus. Material characteristics

t1 (µs) 200

(Wcm-2) 5.4

Sda (Wcm-2) 35-58

Uda (V) 351-347

Φox (sccm) 5.6

k550 (10-3)

n550

0.1

200

52.0

370-540

484-513

16.8

2.0

100

51.0

810-1220

528-531

14.0

80

53.0

1030-1460

551-542

14.0

50

53.0

1700-2100

673-679

14.0

2.19

16

166

SC

E* (GPa) 147

5.0

2.20

16

164

0.1

1.97

9

135

1.0

2.07

10

145

M AN U

TE D EP AC C 30

2.10

H (GPa) 14

ACCEPTED MANUSCRIPT

Backscattering probability, B (%) Ar Zr O

Energy of backscattered atoms, Eb (eV) Ar Zr O

SC

Energy of incident ions, Ei (eV)

RI PT

Table 3. The backscattering probability of atoms, B, and the mean energy of backscattered atoms at the target surface, Eb, calculated for Ar, Zr and O atoms backscattered from Zr and ZrO2 targets after their bombardment by the corresponding ions. Here, Ei is the energy of ions incident on the target. The calculations were carried out using the SDTrimSP program [47], which simulates the impact of energetic atoms or ions onto a target material using a binary collision approximation. We used the Kr-C potential [54] and the standard displacement energies supplied in the program's material data table (21.0 eV and 0.5 eV for Zr and O, respectively). The surface binding energy for each atom on the surface is calculated by a linear combination of the surface concentrations and surface binding energies for all possible atom pairs. These pairwise surface binding energies of 6.3 eV and 15.4 eV for Zr-Zr and Zr-O, respectively, were estimated from the known standard heats of formation of Zr and ZrO2 [55].

Zr target 23.4 19.4 17.8 16.8

0.8 1.5 1.8 2.0

34.4 31.5 30.1 29.2

M AN U

200 400 600 800

30.9 59.7 88.4 117.5

10.4 18.7 26.2 33.5

70.5 137.6 203.1 267.4

23.7 45.4 67.2 87.9

11.6 15.7 20.0 26.2

55.8 107.4 160.1 212.8

ZrO2 target

8.9 7.2 6.6 6.1

0.3 0.4 0.5 0.5

13.6 12.1 11.1 10.8

AC C

EP

TE D

200 400 600 800

31

ACCEPTED MANUSCRIPT

Figure captions

Fig.1. Schematic diagram of the experimental setup with two different locations and orientations of the O2 inlets in front of the target and an energy-resolved mass spectrometer.

RI PT

Dashed lines represent the to-target O2 inlet configuration (20 mm from the target surface), and solid lines represent the to-substrate O2 inlet configuration (25 mm from the target surface). Positions of the pressure sensor and the Ar inlet in the back side of the vacuum

SC

chamber are also shown.

M AN U

Fig.2. Time evolutions of the average discharge current in a period of the power supply, I d (t ) , and the corresponding oxygen flow rate pulses during controlled reactive sputter depositions of stoichiometric ZrO2 films with the to-target O2 inlet at = 51 Wcm-2 and t1 = 200 µs (Table 1) and with the to-substrate O2 inlet at = 52 Wcm-2 and t1 = 200 µs (Table 2). For

TE D

the to-target O2 inlet, the oxygen flow rate Фox = 19 sccm is switched on when I d (t ) = C1 , switched over to a lower value of 9 sccm when I d (t ) = C 2 = C1 and this reduced flow is switched off when I d (t ) = C3 . For the to-substrate O2 inlet, different pre-selected critical

EP

values of C1 and C2 = C1 determine the switching-on and switching-off of the oxygen flow

AC C

rate Фox = 16.8 sccm, respectively.

Fig.3. Waveforms of the magnetron voltage, Ud, and the target current density, Jt, for a preset deposition-averaged target power density = 51 Wcm−2 at a voltage pulse duration t1 = 200 µs with the average target power densities in a discharge pulse Sda = 350 Wcm−2 and 770 Wcm−2 relating to the minimum and maximum values of the average discharge current in a period of the power supply, I d (t ) , respectively, controlling the duration (and shape) of the oxygen flow rate pulses during a reactive sputter deposition of stoichiometric ZrO2 films with

32

ACCEPTED MANUSCRIPT

the to-target O2 inlet.

Fig.4. Waveforms of the magnetron voltage, Ud, and the target current density, Jt, for a preset deposition-averaged target power density = 52 Wcm−2 at a voltage pulse duration t1 =

RI PT

200 µs with the average target power densities in a discharge pulse Sda = 370 Wcm−2 and 540 Wcm−2 relating to the minimum and maximum values of the average discharge current in a period of the power supply, I d (t ) , respectively, controlling the duration of the oxygen flow

SC

rate pulses during a reactive sputter deposition of stoichiometric ZrO2 films with the to-

M AN U

substrate O2 inlet.

Fig.5. Deposition rates, aD, of stoichiometric ZrO2 films for the to-target O2 inlet (empty downward triangles) and the to-substrate O2 inlet (full upward triangles) at depositionaveraged target power densities = 50-53 Wcm-2 and = 51-53 Wcm-2, respectively,

TE D

and various voltage pulse durations, t1, ranging from 50 to 200 µs during depositions with the corresponding duty cycles from 2.5 % to 10 %. For comparison, the deposition rates achieved for the to-target O2 inlet at = 5.2 Wcm-2 and t1 = 200 µs, and for the to-substrate O2 inlet

EP

at = 5.4 Wcm-2 and t1 = 200 µs are presented (Tables 1 and 2).

AC C

Fig.6. As Fig. 5, but for deposition-averaged oxygen flow rates, <Фox>, and the aD/<Фox> ratios.

Fig.7. Normalized time-averaged energy distribution functions of dominant positive ions at the substrate position (d = 100 mm) measured for the to-target O2 inlet at = 5.2 Wcm-2 and t1 = 200 µs, = 51 Wcm-2 and t1 = 200 µs, and = 53 Wcm-2 and t1 = 50 µs (Table 1).

33

ACCEPTED MANUSCRIPT

Fig.8. Normalized time-averaged energy distribution functions of dominant positive ions at the substrate position (d = 100 mm) measured for the to-substrate O2 inlet at = 5.4 Wcm-2 and t1 = 200 µs, = 52 Wcm-2 and t1 = 200 µs, and = 53 Wcm-2 and t1 =

RI PT

50 µs (Table 2).

Fig.9. Compositions of the total fluxes of positive ions at the substrate position (d = 100 mm) determined for the to-target O2 inlet (empty columns) at = 5.2 Wcm-2 and t1 = 200 µs,

= 51 Wcm-2 and t1 = 200 µs, and = 53 Wcm-2 and t1 = 50 µs (Table 1), and for the

SC

to-substrate O2 inlet (hatched and full columns) at = 5.4 Wcm-2 and t1 = 200 µs, =

M AN U

52 Wcm-2 and t1 = 200 µs, and = 53 Wcm-2 and t1 = 50 µs (Table 2).

Fig.10. Time-averaged energy distribution functions of O- ions at the substrate position (d = 100 mm) measured for the to-target O2 inlet at = 5.2 Wcm-2 and t1 = 200 µs (dotted line), = 51 Wcm-2 and t1 = 200 µs (solid line), and = 53 Wcm-2 and t1 = 50 µs (dashed

TE D

line), and for the to-substrate O2 inlet at = 5.4 Wcm-2 and t1 = 200 µs (dotted line),

EP

= 52 Wcm-2 and t1 = 200 µs (solid line), and = 53 Wcm-2 and t1 = 50 µs (dashed line).

Fig.11. Surface morphology of the stoichiometric ZrO2 films deposited using the to-substrate

AC C

O2 inlet at = 5.4 Wcm-2 and t1 = 200 µs, = 52 Wcm-2 and t1 = 200 µs, and = 53 Wcm-2 and t1 = 50 µs.

34

ACCEPTED MANUSCRIPT

Fig.2

AC C

EP

TE D

M AN U

SC

RI PT

Fig.1

Fig.3

35

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig.4

Fig.5

36

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig.6

Fig.7 37

Fig.8

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Fig.9

38

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig. 10

Fig.11

39

ACCEPTED MANUSCRIPT

Highlights:

EP

TE D

M AN U

SC

RI PT

Controlled reactive high-power impulse magnetron sputtering was used. Very high deposition rates of densified ZrO2 films were achieved. Highly optically transparent films were produced. Enlarged substrate fluxes of ionized metal atoms were measured. Energy distributions of positive ions were extended to high energies.

AC C

-