SCT-20588; No of Pages 4 Surface & Coatings Technology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat
Adjustment of plasma properties in magnetron sputtering by pulsed powering in unipolar/bipolar hybrid pulse mode Stephan Barth ⁎, Hagen Bartzsch, Daniel Glöß, Peter Frach, Matthias Gittner, Rainer Labitzke Fraunhofer Institute for Organic Electronics, Electron Beam and Plasma Technology, Winterbergstrasse 28, 01277 Dresden, Germany
a r t i c l e
i n f o
Article history: Received 15 June 2015 Revised 18 September 2015 Accepted in revised form 19 September 2015 Available online xxxx Keywords: Pulse magnetron sputtering Sputter deposition Plasma properties Stress control AlN
a b s t r a c t A new method of pulsed powering the magnetron discharge using a pulsed switching of the anode has been developed. Practically, this hybrid pulse mode is a combination of the conventional unipolar and bipolar pulsed powering, where the time slices of both pulse modes can be freely adjusted at a time scale smaller than 1 ms, i.e. much shorter than necessary for the deposition of one monolayer. This allows varying the average plasma parameters freely between the typical values of unipolar and bipolar pulse mode. During deposition of piezoelectric AlN, the film stress could be shifted in a wide range by changing the pulse mode ratio. In combination with the other known process parameters, it was possible to shift the film stress between compressive and tensile while maintaining piezoelectric properties. Hence, in addition to classical deposition parameters such as pressure or temperature, this new parameter gives an additional degree of freedom for optimization of film properties independent from sputtering power and deposition rate. © 2015 Elsevier B.V. All rights reserved.
1. Introduction For sputtering of thin films, there is a wide range of possible processes available, e.g. DC-sputtering, MF-pulsed sputtering or RF-sputtering. In pulsed magnetron sputtering it is known, that the pulse mode, either unipolar pulsed or bipolar pulsed, may have a significant influence on plasma density and electron temperature in substrate vicinity [1]. These plasma properties determine the energetic ion bombardment of the substrate. The ion flux density is determined by plasma density and ion energy mainly associated with electron temperature considering ambipolar diffusion [2]. Especially in bipolar pulse mode the energetic ion bombardment may dominate the variety of contributions to thermal substrate load, such as energetic neutrals, photons, heat of condensation, heat of reaction [1,3]. Table 1 gives values of plasma density and electron temperature in unipolar and bipolar pulse mode for SiO2 sputtering (from [4]). Because of the differences, the bipolar mode may be used to deposit more dense films due to the stronger substrate bombardment, whereas the unipolar pulse mode is more suited to deposit films on temperature sensitive substrates with lower stress. For many thin film applications, the desired process and film properties lead to contradictory requirements on the level of energetic ion bombardment of the substrate. One example is the deposition of very dense films on plastic substrates, for example as barrier films. The higher film density needs a strong substrate bombardment, while a more moderate substrate bombardment is necessary to prevent thermal heating of the substrate to reach a critical level. Another example ⁎ Corresponding author. E-mail address:
[email protected] (S. Barth).
is the deposition of piezoelectric thin films, e.g. for SAW, BAW or Energy Harvesting applications. These films require a specific film stress and crystalline quality. The film stress increases towards more compressive stress with higher substrate bombardment. This effect is caused by “atomic peening” or “ionic peening” [5]. The crystal quality of AlN needs a minimum of energy input, either through substrate bombardment or higher energy of deposited atoms [6,7,8]. Thus, the substrate bombardment must be high enough to reach good crystal quality but low enough to prevent high film stress from causing delamination or cracks. In our previous work, we demonstrated the effect of unipolar and bipolar pulse mode, as well as pressure during deposition, on film stress for highly piezoelectric AlN thin films [9]. The aim of this work was to find a new method for fine adjustment of energetic substrate bombardment by plasma ions by an unipolar/bipolar hybrid pulse mode. 2. Material and methods 2.1. Coating equipment and processes All depositions were done at a cluster tool by reactive pulse magnetron sputtering using the Double Ring Magnetron DRM 400 sputter source developed by Fraunhofer FEP [10]. This type of magnetron combines two concentric discharge targets to deposit uniform films on substrates with a diameter of up to 200 mm. Fig. 1 shows a schematic diagram of the deposition set-up. In the case of AlN depositions, pure Al targets (5 N) were used, whereas the Si depositions were done with pure Si targets (5 N). Argon (5 N) and nitrogen (5 N) were used as inert gas and as reactive gas, respectively.
http://dx.doi.org/10.1016/j.surfcoat.2015.09.037 0257-8972/© 2015 Elsevier B.V. All rights reserved.
Please cite this article as: S. Barth, et al., Surf. Coat. Technol. (2015), http://dx.doi.org/10.1016/j.surfcoat.2015.09.037
2
S. Barth et al. / Surface & Coatings Technology xxx (2015) xxx–xxx
Table 1 Results of Langmuir Probe and temperature measurements in substrate vicinity for SiO2 sputtering at 7.5 kW (from [4]). Pulse mode
Unipolar
Bipolar
Plasma density [1/cm3] Electron temperature [eV] Thermal substrate load [W/cm2]
1.8 · 1010 10 0.15
11.0 · 1010 6 0.75
The Argon gas flow (30…60 sccm) was used to adjust the pressure in the range of 0.3…0.7 Pa. For reactive AlN depositions, a closed loop reactive gas control was applied to stabilize the process in the transition mode [10]. The resulting nitrogen flow rates were between 25 sccm and 40 sccm. There was no additional substrate heating or cooling. The base pressure before depositions was 10−6 mbar. Using the pulse unit UBS-C2 developed by FEP and standard dc power supplies, the pulse mode of the pulse magnetron sputtering process can be adjusted as either unipolar or bipolar. The principle of both pulse modes is shown in Fig. 1. In the unipolar pulse mode, a pulsed dc is applied between each of the two targets and the separate anode component. In the bipolar pulse mode, a voltage with alternating polarity is applied between the two targets of the DRM 400. The targets act alternately as anode and cathode of the discharge. The separate anode component is not connected to the discharge. The unipolar/bipolar hybrid pulse mode was realized by a hardware solution, the anode pulse unit. It includes an electronic switch, which periodically connects and disconnects the separate anode component to and from the discharge. If the anode component is connected to the discharge, it is operated in unipolar mode, if disconnected in bipolar mode. The time of one connecting/disconnecting cycle is 1 ms. This is schematically depicted in Fig. 1. The resulting voltages and currents over time are schematically shown in Fig. 2, showing the transition from unipolar to bipolar pulse mode for two complete cycles. As independent parameter, the share of bipolar pulse mode Sb is chosen, i.e. the ratio between the time slice of bipolar mode tb with respect to the total time of one cycle of unipolar/bipolar hybrid pulse mode, i.e. Sb = tb / (tu + tb). Sb can be adjusted between 0 and 100%. The time slices of unipolar (tu) or bipolar pulse (tb) mode are therefore each less than or equal to 1 ms. This is much faster than the time necessary for the deposition of one monolayer (typically N20 ms), but still slower than the 50 to 10 μs between the pulses corresponding to the respective discharge pulse frequency of 20…100 kHz. 2.2. Characterization Measurements of time averaged plasma properties were carried out using a double Langmuir probe. The plasma density ne and electron temperature Te were then calculated according to the method of Sonin [11]. The stress measurements were done using the wafer curvature method based on Stoney's equation [12]. The substrates were polished
Fig. 2. schematic depiction of voltage of outer target, inner target and anode pulse unit and current at anode pulse unit for hybrid pulse mode during two complete cycles with unipolar (tu) and bipolar (tb) time share.
4 in. Si-Wafers (orientation 100, thickness 525 μm). The measurements were done with a surface profiler P15-Ls (KLA Tencor, San Jose, CA). For the evaluation of piezoelectric properties, a simple ultrasound transducer test layout consisting of a three-layer structure was used. A circular aluminum bottom electrode with 10 mm diameter and a contact pad was deposited on an isolated silicon wafer. An AlN circular layer with a diameter of 13 mm was deposited, followed by the aluminum top electrode with 10 mm diameter to fabricate the transducer. The piezoelectric measurements to determine the piezoelectric charge constant d33 were done using a PM300 piezometer (Piezotest Ltd., London, UK). 3. Results 3.1. Plasma characterization The characterization of the plasma properties in substrate vicinity was done for the sputter process of metallic Si at target powers of 2.0 kW and 0.5 kW for the outer and inner target, i.e. at total target power of 2.5 kW and deposition pressure 0.4 Pa. The results for the plasma density ne depending on the share of bipolar pulse mode Sb are shown in Fig. 3. The plasma density increases linearly from 7.7 × 109 cm− 3 in the pure unipolar pulse mode to 2.5 × 1010 cm−3 in the pure bipolar pulse mode. The electron temperature Te shows a slight decrease with increasing Sb, as seen in Fig. 4. The electron temperature in the pure bipolar pulse
Fig. 1. Deposition setup for sputter deposition by DRM 400 for unipolar, bipolar and unipolar/bipolar hybrid pulse mode.
Please cite this article as: S. Barth, et al., Surf. Coat. Technol. (2015), http://dx.doi.org/10.1016/j.surfcoat.2015.09.037
S. Barth et al. / Surface & Coatings Technology xxx (2015) xxx–xxx
Fig. 3. Plasma density ne for metallic sputtering of Si at 2.5 kW and 0.4 Pa.
mode has a value of 2.1 eV. This is approx. 80% of the electron temperature measured in the pure unipolar mode, which is 2.7 eV. Compared to the plasma density, which in the unipolar mode is just 30% of the density in bipolar mode, this is a relatively small change. 3.2. Characterization of AlN Prior work regarding the influence of process parameters of the AlN sputter deposition process using the DRM 400 on film properties is described in [4,9]. The main parameters determining good piezoelectric properties of the films were the pulse mode, the pressure during deposition, the target power and the reactive working point. A very narrow window of process parameters was found for each pure pulse mode to achieve good piezoelectric thin films. Films deposited in bipolar pulse mode typically exhibit much stronger compressive stress than those deposited in unipolar pulse mode with otherwise same process conditions and thickness. This behavior is caused by the much higher substrate bombardment compared to the unipolar pulse mode. This effect is also known as “atomic peening” or “ionic peening” [5]. The effect of the hybrid pulse mode on the film stress is shown in Fig. 5 in the case of 0.5 μm thick films on Si, deposited at pressure of 0.3 Pa. Every other process parameter was kept constant, so only the direct effect of Sb on film stress can be observed. Therefore, there was no regard towards achieving good piezoelectric properties or stress optimization. In the case of unipolar pulse mode, the stress in the film was 1.2 GPa compressive. The deposition done in bipolar pulse mode resulted in a film with much higher compressive stress, which had a value of
Fig. 4. Electron temperature Te for metallic sputtering of Si at 2.5 kW and 0.4 Pa.
3
Fig. 5. Film stress σ for 0.5 μm thick AlN layers.
2.9 GPa. The stress in the films for the hybrid pulse mode showed a linear dependency on the pulse mode ratio. When combined with the other known parameters for deposition of highly piezoelectric thin films, the stress can be shifted to lower compressive or even tensile stress. The pulse mode ratio represents an additional parameter to influence film properties. Thus, the narrow process windows found at both pure pulse modes can be significantly expanded and it is possible to better optimize film properties with competing requirements. This is demonstrated in Fig. 6, where the film stresses of 1.55 μm thick AlN films deposited on Si are shown. By adjusting the deposition pressure and using the hybrid pulse mode at Sb = 20%, it was possible to deposit films with variable stress between several hundred MPa compressive and tensile while still achieving piezoelectric coefficients d33 higher than 6 pC/N and deposition rates of more than 150 nm/min. 4. Discussion The unipolar/bipolar hybrid pulse mode is realized by an electronic switch to switch on/off the anode periodically within one cycle of 1 ms. Thus, the time slices of unipolar (tu) or bipolar pulse (tb) mode are less than or equal to 1 ms. Since the frequency of anode switching is much slower than the discharge pulse frequency, it was expected, that the mean plasma parameters would be shifted according to the pulse mode ratio. Langmuir probe measurements of a metallic Si sputter process showed a gradual shift of plasma density depending on pulse mode
Fig. 6. Film stress σ for 1.55 μm thick AlN with high piezoelectric properties at share of bipolar pulse mode Sb = 20%.
Please cite this article as: S. Barth, et al., Surf. Coat. Technol. (2015), http://dx.doi.org/10.1016/j.surfcoat.2015.09.037
4
S. Barth et al. / Surface & Coatings Technology xxx (2015) xxx–xxx
ratio. The relative value went from 30% in pure unipolar mode to 100% in pure bipolar mode. The electron temperature also exhibits a gradual shift from 100% at pure unipolar pulse mode to 80% at pure bipolar pulse mode. Thus, it is possible to finely adjust the mean plasma properties between those of pure unipolar and bipolar pulse mode. Furthermore, since the frequency of anode switching is much faster than the time necessary for the deposition of one monolayer, it was expected, that the film properties could be influenced by changing Sb. This was proven by characterizing AlN films deposited at different shares of bipolar pulse mode. There was a linear shift of film stress towards higher compressive by increasing Sb. This is congruent with previous work, where it was shown, that the bipolar pulse mode results in more compressive stress in the films compared to unipolar pulse mode [4,9]. Thus, the hybrid pulse mode offers an additional parameter for the deposition process of AlN, significantly increasing the possible parameter field to achieve good piezoelectric properties and film stress at the same time. 5. Summary and conclusion This paper describes the results of using unipolar/bipolar hybrid pulse mode in the pulsed magnetron sputtering process. By varying the main parameter of this new pulse mode, the share of bipolar pulse mode Sb, it was possible to adjust values of plasma parameter and sensitive film properties such as stress rather linearly between the value of the pure pulse modes. In the reactive magnetron sputter deposition of piezoelectric AlN films, the hybrid pulse mode, in combination with the other deposition parameters that are known to influence the properties, allowed to deposit highly piezoelectric thin films at high deposition rates above 150 nm/min with the feasibility to adjust the film stress between several hundred MPa compressive and tensile. In summary, the unipolar/bipolar hybrid pulse mode is another possibility for influencing plasma and therefore film properties. In contrast to influencing energetic substrate bombardment by substrate bias which determines ion energy the new technique mainly influences plasma density and hence ion flux density.
Acknowledgments This work has been partially supported by a Grant-in-Aid for Technology Funding by the European Regional Development Fund (ERDF) (Grant-number: 13555/2317) 2007–2013 and the State of Saxony.
References [1] H. Bartzsch, P. Frach, K. Goedicke, Anode effects on energetic particle bombardment of the substrate in pulsed magnetron sputtering, Surf. Coat. Technol. 132 (2000) 244–250. [2] B. Chapman, Glow Discharge Processes, John Wiley & Sons, New York, 1980. [3] H. Kersten, G.M.W. Kroesen, R. Hippler, On the energy influx to the substrate during sputter deposition of thin aluminium films, Thin Solid Films 332 (1998) 282–289. [4] H. Bartzsch, M. Gittner, D. Gloess, P. Frach, T. Herzog, S. Walter, H. Heuer, Properties of piezoelectric AlN layers deposited by reactive pulse magnetron sputtering, Society of Vacuum Coaters, 54th Annual Technical Conference Proceedings, 2011. [5] J. Tranchant, P.Y. Tessier, J.P. Landesman, M.A. Djouadi, B. Angleraud, P.O. Renault, B. Girault, P. Goudeau, Relation between residual stresses and microstructure in Mo(Cr) thin films elaborated by ionized magnetron sputtering, Surf. Coat. Technol. 202 (2008) 2247–2251. [6] X.-H. Xu, H.S. Wu, C.-J. Zhang, Z.-H. Jin, Morphological properties of AlN piezoelectric thin films deposited by DC reactive magnetron sputtering, Thin Solid Films 388 (2001) 62–67. [7] J.X. Zhang, H. Cheng, Y.Z. Chen, A. Uddin, S.J. Shu Yuan, S. Geng, Zhang, Growth of AlN films on Si (100) and Si (111) substrates by reactive magnetron sputtering, Surf. Coat. Technol. 198 (2005) 68–73. [8] H. Jin, J. Zhou, S.R. Dong, B. Feng, J.K. Luo, D.M. Wang, W.I. Milne, C.Y. Yang, Deposition of c-axis orientation aluminum nitride films on flexible polymer substrates by reactive direct-current magnetron sputtering, Thin Solid Films 520 (2012) 4863–4870. [9] S. Barth, H. Bartzsch, D. Gloess, P. Frach, T. Herzog, S. Walter, H. Heuer, Sputter deposition of stress controlled piezoelectric AlN and AlScN films for ultrasonic and energy harvesting applications, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 61 (8) (2014). [10] P. Frach, C. Gottfried, H. Bartzsch, K. Goedicke, The double ring process module - a tool for stationary deposition of metals, insulators and reactive sputtered compounds, Surf. Coat. Technol. 90 (1–2) (1997) 75–81. [11] A. Sonin, Free-molecule Langmuir probe and its use in flow-field studies, J. Am. Inst. Aeronaut. Astronaut. 4 (1966) 1588–1596. [12] G. Stoney, The tension of metallic films deposited by electrolysis, Proc. R. Soc. Lond. A 82 (553) (1909) 92–172.
Please cite this article as: S. Barth, et al., Surf. Coat. Technol. (2015), http://dx.doi.org/10.1016/j.surfcoat.2015.09.037