Magnetic field strength influence on the reactive magnetron sputter deposition of Ta2O5
R. Hollerweger, D. Holec, J. Paulitsch, R. Rachbauer, P. Polcik, P. H. Mayrhofer
11 Magnetic field strength influence on the reactive magnetron sputter deposition of Ta O R Hollerweger , D Holec , J Paulitsch , R Rachbauer , P Polcik and P H Mayrhofer Christian Doppler Laboratory for Application Oriented Coating Development at the 5 Department of Physical Metallurgy and Materials Testing, Montanuniversität Leoben, 6 Franz-Josef-Str. 18, A-8700 Leoben, Austria 7 Department of Physical Metallurgy and Materials Testing, Montanuniversität 8 Leoben, Franz-Josef-Str. 18, A-8700 Leoben, Austria Institute of Materials Science and Technology, Vienna University of Technology, A-10 1040 Vienna, Austria OC Oerlikon Balzers AG, Iramali 18, LI-9469 Balzers, Principality of Liechtenstein Plansee Composite Materials GmbH, Siebenbürgerstr. 23, D-86983 Lechbruck am 13 See, Germany 14 15 [email protected] 16 17
Abstract.
Reactive magnetron sputtering enables the deposition of various thin films to be used for protective as well as optical and electronic applications. However, progressing target erosion during sputtering results in increased magnetic field strengths at the target surface. Consequently, the glow discharge, the target poisoning, and hence the morphology, crystal structure and stoichiometry of the prepared thin films are influenced. Therefore, these effects were investigated by varying the cathode current I m between 0.50 and 1.00 A, the magnetic field strength B between 45 and 90 mT, and the O /(Ar+O ) flow rate ratio between 0 and x oxide forms at the metallic Ta target surface which further transfers to a non-conductive tantalum pentoxide Ta O , impeding a stable DC glow discharge. These two transition zones (from Ta to TaO x and from TaO x to Ta O ) shift to higher oxygen flow rates for increasing target currents. Contrary, increasing the magnetic field strength (e.g., due to sputter erosion) mainly shifts the TaO x to Ta O transition to lower oxygen flow rates while marginally influencing the Ta to TaO x transition. To allow for a stable DC glow discharge (and to suppress the formation of non- conductive Ta O at the target) even at = 100% either a high target current (I m low magnetic field strength (B
60 mT) is necessary. These conditions are required to prepare stoichiometric and fully crystalline Ta O films. PACS classification number: 81.15.Cd Submitted to: Journal of Physics D: Applied Physics Introduction
39 Reactive magnetron sputtering is a widely spread technique to deposit thin compound films on a huge 40 variety of substrate materials. However, to reproducibly deposit high quality coatings a consolidated 41 knowledge on process influencing effects like the target poisoning behaviour and its impact on the 42 plasma conditions is necessary [1–6]. Therefore, many investigations are focusing on improving 43 process controlling parameters such as power supply, power density, reactive gas flow, total pressure, 44 pumping speed or magnetic field strength [2,7,8]. 45 Determining a cathode-voltage-hysteresis as a function of the reactive gas flow gives an easy access to 46 investigate the poisoning behaviour of the cathode material. In general two operating modes can be 47 distinguished: The metallic sputtering mode, established at low reactive gas contents, in which the 48 target surface still indicates the metallic phase, and the poisoned sputtering mode at higher reactive gas 49 contents, in which a compound layer is formed at the target surface [9,10]. Such compounds can 50 negatively affect the discharge properties due to changes in sputter yields, secondary electron emission 51 or conductivity [11,12]. 52 Magnetron sputtering was introduced in the 70ies as it enables a stable glow-discharge at lower 53 chamber pressures than known for diode-sputtering [13]. The magnetic field of the magnetron is 54 responsible for trapping secondary electrons, which are essential for ionizing the gaseous species and 55 consequently for the plasma density. Nevertheless, progressing sputter erosion of the target material 56 results in an increase of the magnetic field strength and hence to a changed discharge behaviour of the 57 target material [14,15]. 58 Tantalum pentoxide, Ta O , is a promising candidate for optical, electronic, corrosion-protective and 59 biocompatible applications [16–18]. However, the reactive sputtering process is very sensitive to the 60 reactive gas partial pressure used. Therefore Schiller et al. have studied the influence of the sputtering 61 power, total pressure as well as partial pressure on the sputter rate of Ta and Ti in an Ar/O glow 62 discharge. Based on their results, they have derived a model for the formation of a microscopic 63 (conductive) or macroscopic (insulating) oxide-compound layer on the target surface to describe the 64 radial oxide coverage and the discharge area [19,20]. 65 Within this study, we investigate the influence of the magnetic field strength, the sputtering current 66 and the oxygen gas flow on the poisoning behaviour of a tantalum target. Our results emphasize the 67 need of a well-balanced process especially when considering the change in magnetic field strength due 68 to continuous sputter erosion of the target surface and high oxygen flow rates which are necessary to 69 obtain crystalline and stoichiometric Ta O films even at substrate temperatures of 500°C. 70 71 Experimental:
72 All depositions and cathode-voltage-hysteresis experiments were performed in a Leybold Heraeus 73 A400-VL laboratory magnetron sputtering device equipped with a 70 dm chamber and a Leybold 74 Turbovac 361 turbo pump (nominal pumping speed of 345 l/sec (nitrogen)). A tantalum target with a 75 diameter of 75 mm (thickness 6mm) and a purity of 99.9% was used and powered by a Leybold/ELAN 76 SSV1.8kW/2-27 DC generator in constant current mode at I m = 0.50, 0.75 and 1.00 A. 77 In order to decrease the radial magnetic field strength B at the target surface from 90 to 60 and 45 mT, 78 austenitic steel plates were used as spacers between the cathode and the permanent magnets, see figure 79 1. The magnetic field strength was measured using a portable LakeShore 410 Gauss-meter. 80 Prior to the hysteresis experiments and without ignited plasma, we have calibrated the flows of Ar ( f Ar ) 81 and O ( f O2 ) to reach a constant total pressure of 0.4 Pa. Therefore, f O2 was stepwise increased from 0 82 to 23.7 sccm to linearly increase the O partial pressure while simultaneously f Ar was decreased from 83 17 to 0 sccm. 30 separate data points were recorded per voltage hysteresis for increasing and 84 subsequently decreasing flow rate , with 85 .
86 The holding time for every single point was 5 minutes to approach equilibrium discharge conditions. 87 Tantalum oxide films were deposited on Si stripes ((100) orientation, 20x7x0.38 mm ) using a target 88 current of 0.75 A, a radial magnetic field strength of 60 mT, and flow rates of = 50, 77 and 100%. 89 The substrate temperature was set to 500 °C and the distance to the target surface was 4.5 cm. 90 Structural investigations were conducted by X-ray diffraction (XRD) analysis in Bragg Brentano 91 geometry using a Brucker D8 diffractometer equipped with a CuK radiation source. The chemical 92 composition was determined by elastic recoil detection analysis (ERDA) using Cl ions with an 93 acceleration voltage of 35 MeV by evaluating the resulting spectrum according to Barrada et al. [21]. 94 Fracture cross-sectional scanning electron microscopy (SEM) investigations with an acceleration 95 voltage of 15 kV were conducted with a Zeiss EVO50 for coating morphology studies and thickness 96 evaluations. Prior to these SEM investigations, a thin Au-layer was deposited to increase the 97 conductivity of our samples. 98 Energies of formation and density of states (DOS) of Ta (NaCl, Fm−3m ( 5), [22]), TaO (rutile, 99 P42/mnm (136), [23]) and Ta O (orthorhombic structure proposed by Lehovec [24]) were calculated 100 using the Vienna ab initio software package (VASP) [25,26] employing projector augmented wave 101 pseudopotentials [27] and the generalized gradient approximation [28]. The unit cells of TaO, TaO
102 and Ta O contained 8, 6, and 14 atoms, respectively and the used cut-off energy of 800 eV and more 103 than 2000 k-points·atom ensure a minimum accuracy of 1 meV/atom. The energies of formation were 104 calculated as, 105 where E TaOx , E Ta and E O2 are the total energies of the crystalline TaO x , bcc Ta and molecular O , 106 respectively. 107 108 Results and Discussion:
Voltage hysteresis
110 Figure 2 shows a voltage hysteresis, at a target current of I m = 0.75 A and a magnetic field strength of 111 B = 90 mT as a function of the oxygen gas flow . Between = 0 and ~65% (first regime, (a)) a 112 voltage peak of ~475 V can be observed at about = 50%. The increase to this peak value can be 113 attributed to the linear reduction of the Ar flow, oxygen gettering by evaporated metal atoms [29], 114 chemisorption and implantation of oxygen atoms at the target [29,30], and also to the constriction of 115 the discharge region due to partial poisoning [19]. At the peak, the discharge voltage and the ion 116 density have reached the optimum conditions for restricting further oxidation of the target surface and 117 the constriction of the discharge area has reached its maximum. If even more oxygen is introduced to 118 the chamber, the target coverage increases, the discharge area broadens again and the voltage 119 decreases until the poisoning processes are completed for the whole target surface. Subsequently stable 120 discharge conditions, characterized by an almost constant voltage regime (b), can be observed up to
121 = 90%, figure 2, and indicate that the formed compound is stable. For higher flow rates, between = 122 90 and 100% (c), the third regime is characterized again by an increase of the discharge voltage, due to 123 the formation of a different compound layer at the target surface further influencing the target 124 resistance. Upon decreasing the O flow rate the hysteresis can be clearly identified for regime (a) and 125 (c) in figure 2. Moreover, a non-stable DC glow discharge at = 100% suggests a close to 126 stoichiometric Ta O compound formed on the target surface with a low conductivity, whereas the 127 stable conditions within regime (b) suggest that the target surface is fully covered by a conductive 128 substoichiometric TaO x phase. 129 Possible candidates for substoichiometric TaO x phases are metastable TaO and TaO [31]. Therefore, 130 ab initio calculations were performed to obtain information on the conductivity, stability and affinity to 131 oxygen. These calculations yield energies of formation of −1.192 eV/atom (−230 kJ/mol) for TaO, 132 −2.990 eV/atom (−865 kJ/mol) for TaO and −3.158 eV/atom (−1079 kJ/mol) for TaO (for 133 comparison with the other phases we have normalized Ta O to TaO ). These results are in excellent 134 agreement with experimentally obtained values of −1023 kJ/mol [32] and − 7 kJ/mol [33] for 135 TaO . According to our calculations, the energy gain from TaO and TaO towards the stable TaO is 136 849 kJ/mol and 214 kJ/mol for TaO and TaO , respectively. The DOS for these three phases, figures. 137 3a, b, and c, indicate a shift in the (pseudo) bandgap (marked by the black arrow) towards the Fermi 138 level with increasing oxidation state of Ta from TaO to TaO to Ta O , respectively. For TaO and 139 TaO there are occupied states at the Fermi level suggesting conductivity. On the other hand, Ta O
140 exhibits a bandgap (~2.5 eV) at the Fermi-level, characteristic for an insulating material (please note 141 that density functional theory (DFT) calculations tend to underestimate the bandgap width). 142 Due to these calculations and the much higher energy of formation for TaO as compared with TaO, 143 combined with the result that regime (b) of figure2 is characterized by a stable discharge over a wide 144 range of oxygen flow rate, we propose that the properties of the tantalum oxide layer responsible for 145 this regime are comparable to metastable TaO
146 Moreover, the voltage increase (figure 2) from ~310V at = 0% to ~410V at = 75% to ~475V at
147 = 100%, indicates decreasing secondary electron emission (SEE) yields for Ta, TaO x and Ta O
148 covered target surfaces, respectively, which is known for the Ta – Ta O system [34]. 149 Current variation.
With increasing sputtering current I m from 0.50 to 0.75 to 1.00 A the three 150 discharge regimes are clearly shifted to higher oxygen flow rates, figure 4. Furthermore, the regime 151 with a stable discharge voltage (between the first and the second hysteresis) covers a broader oxygen 152 flow rate with increasing sputtering current. For I m = 1 A, the second hysteresis effect due to the 153 formation of an insulating Ta O cannot be observed allowing for stable discharge conditions even for 154 100% of oxygen flow rate. Additionally, the second hysteresis, due to the formation of Ta O , clearly 155 widens with decreasing target currents. This can be explained by the reduced particle bombardment 156 (density) and energy caused by decreasing target current and voltage, and hence Ta O stabilizes 157 earlier during increasing flow rates, and (when formed) is stable to lower values during decreasing 158 flow rates. The shift of the first hysteresis (formation of TaO x ) to lower oxygen flow rates with 159 decreasing sputtering current is also based on the decreasing particle bombardment and consequently 160 decreasing sputtering rate. 161 Magnetic field strength variation.
The effect of progressing target erosion is investigated by 162 modifying the magnetic field strength B at the surface (see experimental). Figures 5a and b show the 163 discharge voltage behaviour for radial magnetic field strengths B of 45, 60, and 90 mT at a constant 164 target current I m of 0.50 and 0.75 A, respectively. Increasing magnetic field strengths at constant target 165 currents shifts the target voltage characteristics to lower values as more electrons are captured and 166 more particles are ionized. Whereas the magnetic field strength variation (within the investigated 167 ranges) has no significant influence on the position of the first hysteresis voltage peak, the second 168 hysteresis is strongly influenced in shape and position. For both target current conditions of I m = 0.50 169 and 0.75 A no second hysteresis can be observed for the lower magnetic field strengths B of 45 and 60 170 mT, see figures 5a and b. This indicates that under these conditions no Ta O compound layer is 171 formed at the target surface. In this case we assume the cohesive energy of the present phase on the 172 target surface, which is in general much lower for a metal than for the particular oxide, to play a major 173 role. Consequently, mainly the density of the impinging ions is influencing the flow rate position of 174 the Ta to TaO x hysteresis. Furthermore, the data also suggests that the second transition zone from 175 TaO x to Ta O is additionally to the density also influenced by the energy of the target impinging 176 particles.
177 In agreement to the voltage dependence of the TaO x formation, which suggests decreasing SEE yields 178 from Ta to TaO x to Ta O (last paragraph in section 3.1) the above discussed magnetic field strength 179 variation suggests decreasing secondary electron energies from Ta to TaO x . Thus, the dependence of 180 the discharge voltage on the magnetic field strength is more pronounced for the metallic Ta target 181 surface at = 0% as compared with the oxidized state. 182 These investigations clearly demonstrate that the discharge and consequently the poisoning behaviour 183 of the target are substantially influenced by the target erosion induced increase of the magnetic field 184 strength, especially when operating at higher oxygen flow rates and lower sputtering currents. 185 186 Film deposition
187 Based on the previous studies we have chosen the settings of I m = 0.75 A and B = 60 mT for the 188 preparation of tantalum oxide films using oxygen flow rates of 50, 77 and 100%, as these allow for a 189 stable deposition process even at higher oxygen flow rates. The comparison of the cathode voltage and 190 deposition rate as a function of the oxygen flow rate, figure 6, clearly exhibits a maximum deposition 191 rate of 110 nm/min, when using the oxygen flow rate slightly above the peak voltage. Increasing the 192 oxygen flow rate to 77 and 100% causes the deposition rate to decrease to 15 and 10 nm/min, 193 respectively. The decrease from 110 to 15 nm/min can mainly be related to the different target surface 194 condition, which is still metal dominated at the voltage peak for = 50%, but completely covered by a 195 TaO x compound layer for = 77%. Further decreased deposition rates of 10 nm/min when using = 196 100% are probably caused by different sputtering yields of Ar and O , as proposed by Yamamura et. 197 al. [35], as well as due to changes in crystallinity and stoichiometry of the coatings, see next 198 paragraph. 199 The coating prepared at = 50% exhibits a dense columnar crystalline structure for the ~10 µm thick 200 top region, preceded by a featureless, amorphous-like morphology up to a layer thickness of the first 201 ~5 µm, figure 7a. This was confirmed when decreasing the deposition time from 120 to 5 min, as 202 thereby the ~1 µm thin film exhibits only a featureless morphology, figure 7b. This featureless near-203 substrate region of the coating decreases with increasing oxygen flow rate and decreasing deposition 204 rate, and is ~0.5 µm for = 77% and essentially zero for = 100%, see figures. 8a and b. XRD 205 investigations, figure 9, confirm the crystalline nature of a highly texturized (110)/(200) orthorhombic 206 Ta O structure, in the outer most part of the coatings prepared at = 50%, pattern (a). When 207 decreasing the layer thickness to ~1 µm, only an X-ray amorphous response is obtained, pattern (b), in 208 agreement to the SEM fracture cross section, figure 7b. The coatings prepared at higher oxygen flow 209 rates, = 77 and 100%, exhibit an orthorhombic Ta O structure as well. However, with increasing 210 oxygen flow rate the preferred orientation changes towards (001), see figure 9 pattern (c) and (d). The 211 chemical compositions of our coatings, table 1, suggest that only for = 100% a stoichiometric Ta O
212 with O/Ta = 2.5, is obtained, indicating that high oxygen partial pressures are needed to form 213 crystalline stoichiometric Ta O as proposed by Ritter [36]. The coatings synthesized at = 50 and 214 77% are slightly substoichiometric, with an O/Ta ratio of 2.33 and 2.35, respectively. 215 216 Summary and conclusions
217 Voltage hysteresis of a Ta target in an Ar/O atmosphere at three different sputtering currents and 218 magnetic field strengths were investigated with respect to its target poisoning behaviour as well as 219 resulting film structure and morphology. We indicated three main sputtering regions which are 220 dominated by the target surface transition from Ta to TaO x , a stable conductive TaO x region and 221 finally the transition of TaO x to insolating Ta O . All three regions are strongly influenced by the 222 cathode current used. Variations of the magnetic field strength mainly affect the second and third 223 regime, whereas the first one does not change significantly. This can be related to the difference in 224 cohesive energies of metallic Ta and Ta-oxide. XRD investigations of deposited Ta-oxides at 50, 77, 225 and 100% oxygen flow rate exhibit a pentoxide structure. However, stoichiometric films could only be 226 obtained at = 100%, whereas the O-to-Ta rate of the coatings prepared with = 50 and 77% is 227 slightly sub-stoichiometric with 2.33 and 2.35 respectively. These coatings also exhibit an amorphous 228 phase content next to the substrate interface. 229 Our investigations demonstrate the pronounced influence of the magnetic field strength, which may 230 change due to e.g. target erosion, on the target poisoning and discharge behaviour, especially at high 231 reactive gas flow ratios, which are necessary to obtain stoichiometric crystalline Ta O coatings. 232 233 Acknowledgement
234 This work has been supported by the European Community as an Integrating Activity 'Support of 235 Public and Industrial Research Using Ion Beam Technology (SPIRIT)' under EC contract no. 227012 236 as well as by the Start Program (Y 371) of the Austrian Science Fund (FWF) and the “Christian 237 Doppler” research association. 238 0
References:
239 [1] l , l and ll 7 a al a a n a n n d n 240 reactive magnetron sputtering
Appl. Phys. Lett. Thin Solid
Films
Thin Solid Films
J. Vac. Sci. Technol.
381 248 [5] Kadlec S 1987 Influence of the pumping speed on the hysteresis effect in the reactive 249 sputtering of thin films
Vacuum J. Appl.Phys.
252 7089 253 [7] Baroch P, Musil J, Vlcek J, Nam K H and Han J G 2005 Reactive magnetron sputtering of 254 TiOx films
Surf. Coat. Technol.
Thin Solid
Films Thin
Solid Films
Surf. Coat. Technol.
Thin Solid Films
Thin Solid Films
Sputter deposition (New York: AVS) 266 [14] Furuya A and Hirono S 2000 Suppression of deposition rate variations and of structural 267 changes in the characteristics of magnetic films deposited by magnetically controlled 268 magnetron sputtering
J. Appl. Phys. J. Appl. Phys. Surf.
Coat. Technol.
Mater. Sci. Eng. R: Probl. At. Sci. Technol.
131 278 [19] Schiller S, Heisig U, Steinfelder K and Strümpfel J 1979 Reactive D.C. sputtering with the 279 magnetron-plasmatron for tantalum pentoxide and titanium dioxide films
Thin Solid Films
280 369–75 281 [20] Schiller S, Heisig U, Steinfelder K, Strümpfel J and Sieber W 1981 Reactive D.C. High-rate 282 Sputtering with the Magnetron/Palsmatron for Industrial Applications
Vak. Tech. App. Phys. Lett.
291 285 [22] Schönberg N, Overend W G, Munthe-Kaas A and Sörensen N A 1954 An X-Ray Investigation 286 of the Tantalum-Oxygen System.
Acta Chem. Scand. Jpn. J. Appl. Phys. J. Less Common Met. Phys. Rev. B
290 558–61 291 [26] Kresse G and Furthmüller J 1996 Efficient iterative schemes for ab initio total-energy 292 calculations using a plane-wave basis set
Phys. Rev. B Phy. Rev. B Phys. Rev. B Surf. Coat. Technol.
Surf. Coat. Technol.
Landolt-
Börnstein - Group III Condensed Matter Numerical Data and Functional Relationships in
Science and Technology vol 7b1, ed K-H Hellwege and A M Hellwege (Berlin/Heidelberg: 304 Springer-Verlag) pp 451 – 461 305 [32] Blachnik R 1998
Taschenbuch für Chemiker und Physiker (Berlin: Springer) 306 [33] Perry R and Green D 1997
Perry’s chemical engineers’ handbook (New York: McGraw-Hill) 307 [34] Ngaruiya J M, Venkataraj S, Drese R, Kappertz O, Leervad Pedersen T P and Wuttig M 2003 308 Preparation and characterization of tantalum oxide films produced by reactive DC magnetron 309 sputtering physica status solidi (a)
Appl. Surf. Sci.
J. Vac. Sci. Technol.
225 313 [37]
Powder Diffraction File 01-089-2843
PDF-2/Release 2007, International Center for Diffraction 314 Data 315 316 3
Figures:
317 318 Figure 1: Sectional schematic of the used unbalanced magnetron system. The cross-striped area 319 indicates the austenitic steel spacer sheets. 320 321 Figure 2: Cathode voltage hysteresis indicating the three main target poisoning regimes: (a) transition 322 from Ta to TaO x , (b) region of stable TaO x , and (c) transition from TaO x to Ta O . 323 4 324 Figure 3: Total density of states (DOS) of (a) TaO, (b) TaO , and (c) Ta O . The vertical arrows 325 indicate the (pseudo-) bandgap. 326 5 327 Figure 4: Cathode voltage hysteresis at a constant radial magnetic field strength of 90 mT and target 328 currents of 0.50, 0.75, and 1.00 A. 329 330 Figure 5: Cathode voltage hysteresis for magnetic field strengths of 45, 60 and 90 mT at target currents 331 I m of (a) 0.50 and (b) 0.75 A. 332 6 333 Figure 6: Cathode voltage hysteresis used for film deposition and resulting deposition rates for I m = 334 0.75 A and B = 60 mT. 335 336 Figure 7: Cross-sectional SEM micrographs of tantalum oxide films deposited at = 50% for 337 deposition times of (a) 120 min and (b) 5 min. 338 339 Figure 8: Cross-sectional SEM micrographs of tantalum oxide films deposited at (a) = 77% and (b) 340 100% for a deposition time of 120 min. 341 7 342 Figure 9: XRD patterns for tantalum oxide films deposited at (a) = 50% for 120 min, (b) = 50% 343 for 5 min, (c) = 77% for 120 min, and (d) = 100% for 120 min (indexed after Lehovec et. al. [37]). 344 The corresponding SEM cross-sections are shown in Figures. 7a, b and Figures. 8a, b, respectively. 345 8 Tables
Table 1.
ERDA evaluated chemical composition of Ta O thin films 347