PROOF COPY 001702JVA

Influence of the external solenoid coil arrangement and excitation mode on plasma characteristics and target utilization in a dc-planar magnetron sputtering system X. B. Zhang, J. Q. Xiao, Z. L. Pei, J. Gong, and C. Suna兲 Division of Surface Engineering of Materials, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, People’s Republic of China

共Received 24 May 2006; accepted 28 November 2006兲

OF

O PR

The influence of external solenoid coil arrangement and excitation mode on plasma characteristics and target utilization was investigated, with an axially external magnetic field Bext being superimposed to the original field of the magnetron source in a dc-planar magnetron sputtering system. The magnetic field configuration was simulated using the finite-element method. Langmuir probe measurement showed that the coil position had a strong effect on the near-substrate plasma parameters, resulting from the variations of magnetic field configuration in the substrate region. Furthermore, it was a relatively simple and effective method to improve the target utilization by supplying a low-frequency ac power to the external coil situated around the magnetron. The target utilization efficiency was highly sensitive to the coil position. The coil was more suitable to be placed in the vicinity of the magnetron for improving the target utilization or near the substrate for increasing the near-substrate plasma density. Finally, a novel method was proposed to simultaneously realize both objects by choosing an appropriate solenoid coil arrangement and excitation mode. © 2007 American Vacuum Society. 关DOI: 10.1116/1.2426980兴

CO

I. INTRODUCTION

PY

Magnetron sputtering techniques are of particular interests in studies of ion-assisted thin-film growth, in which the magnetic field is used to alter the ion irradiation conditions on the growing film surface. The critical issues during the film growth by magnetron sputtering techniques include the individual control of plasma growth parameters, improvement of the target utilization, uniformity of deposited films, etc.1,2 However, planar sputtering magnetrons still have some limitations in these aspects, in spite of their significant developments during the past decades.3,4 One of the most important aspects of planar magnetron sputtering systems is to find an easy way to change and control the plasma growth conditions in a broad range. For example, the ion current density Jion and ion energy Eion on the substrate surface are expected to be individually varied for optimal structure and properties of the films. A popular method is to use planar magnetrons with external electromagnet coil as a means of increasing the ionization density in the region near the substrate. This method has been investigated in single-cathode systems,5,6 as well as in dualcathode systems for multilayer depositions.7,8 In onemagnetron systems, the effect of an additional magnetic field Bext created by a solenoid dc coil positioned on the axis between the magnetron and the substrate has been investigated by Petrov et al.5 and Ivanov et al.6 An increase in the ion/neutral arrival flux, from 0.1 to 6 and from 0.1 to 5, was obtained, respectively. The effects of the strength and direction of Bext on Jion were well investigated, while the influence

of solenoid coil arrangement and excitation mode was not discussed in their work. It is necessary and useful to know how to optimize these factors to control the magnetic filed configuration and plasma characteristics. Thus, the first goal of this work is to determine the influence of solenoid coil position and excitation mode on the near-substrate plasma parameters and the I-V characteristic of magnetron. Another critical aspect of planar magnetron sputtering is to improve the low target materials utilization which usually occurred in a conventional magnetron. Typically, the utilization efficiency is less than 30%.4 The magnetic and electric fields in front of a planar magnetron are not uniform, which results in an inhomogeneous confined plasma and “racetrackshape” erosion profile on the target surface. Currently, there are some new designs aimed to achieve a higher target utilization by keeping the erosion trench moving on the target.9–11 A popular method of rotating the target or magnets is applied in these designs. However, it remarkably increases the complexity and production difficulties of the magnetron. In addition, some authors introduce a compressed coil to improve the target utilization by changing the direction and the intensity of current flowing around the target.12–15 In this work, we adopted a method to improve the target utilization by applying the external coil in the ac excitation mode, which has an advantage in making the erosion trench move on the target surface in a simple way by the alternate coil current instead of the mechanical movement of magnets or target. Therefore, the second goal of this work is to determine the influence of solenoid coil arrangement and excitation mode on the target materials utilization. Finally, based on the understanding of the external coil arrangement and excitation mode’s effects on the plasma characteristics and target utilization, a novel method of plac-

A

PROOF COPY 001702JVA

0734-2101/2007/25„2…/1/0/$23.00

JV

J. Vac. Sci. Technol. A 25„2…, Mar/Apr 2007

02

1

17

Author to whom correspondence should be addressed; electronic mail: [email protected]

00

a兲

©2007 American Vacuum Society

1

PROOF COPY 001702JVA 2

Zhang et al.: Influence of the external solenoid coil arrangement and excitation mode

2

OF

O PR PY

CO 02

17

00 FIG. 1. Schematic drawings of the deposition systems: 共a兲 type I and 共b兲 type II.

II. EXPERIMENT A schematic drawing of the planar magnetron sputtering system in this work is shown in Fig. 1. The system consisted of a cylindrical stainless-steel chamber with a diameter of 600 mm, in which one 76 mm circular magnetron was placed on the side wall. The unbalanced level of the magnetron was estimated by using a coefficient of geometrical unJ. Vac. Sci. Technol. A, Vol. 25, No. 2, Mar/Apr 2007

PROOF COPY 001702JVA

balance 共KG兲, KG = Z0 / 2R, where R is an average radius of the erosion zone and Z0 is the distance of null point, and was determined by numerical simulation of the magnetic field distribution using the finite-element method 共FEM兲 or measured by a Hall probe Gauss meter.16 The value of KG of this magnetron source is approximately 1.5, corresponding to the middle balanced magnetron.17 The substrate was positioned on a rotating substrate table with a resistive heater. The target-to-substrate distance was 70 mm. There were two types of coil arrangement used in this experiment, which were denoted by type I and type II. In type I, a solenoid coil situated on the Z axis between the magnetron and the substrate was utilized to create a Bext, as shown in Fig. 1共a兲. The coil with an inner diameter of

A

ing a pair of coils with the appropriate excited mode was presented to simultaneously improve the target utilization and control the near-substrate plasma density over a large range.

JV

FIG. 2. Jion and V f as a function of Bext as well as Icoil, for the coil was situated in the vicinity of 共a兲 the magnetron 共Zcoil = 0 cm兲 and 共b兲 the substrate holder 共Zcoil = 8 cm兲.

PROOF COPY 001702JVA 3

Zhang et al.: Influence of the external solenoid coil arrangement and excitation mode

OF

O PR

100 mm was made of Kapton insulated Cu wire. With a total Cu wire of 850 turns and a coil dc of 2 A, an external magnetic flux of 100 G was obtained. Moreover, the coil could be excited not only by dc power supply 共dc mode兲, but also by a low-frequency ac power supply 共ac mode, the working frequency was set to 50 Hz兲. In type II, a pair of solenoid coils, same to that used in type I, was separately situated in the vicinity of the magnetron and the substrate along the Z axis 关Fig. 1共b兲兴. The coil around the magnetron was excited in the ac mode, while another one in the substrate vicinity was excited in the dc mode or ac mode. By changing the coil current strength and direction, a variety of plasma growth conditions could be controlled. Simulations of the magnetic field in the experimental setup were carried out by applying POISSON SUPERFISH,18 a two-dimensional finite-element software. Since the analysis model of the magnetron sputtering system was assumed to be symmetrical about the Z axis, as illustrated in Fig. 1, only the right cross section of the simulated magnetic field configuration was given in the following text. Plasma characteristics were measured with a circular flat Langmuir probe, which worked well for determining the saturation ion flux and for measuring the floating potential 共self-bias兲 V f .19 The circular probe had a radius of 4 mm and was mounted on the center of substrate holder. The sputtering system was used for the preparation of ZnO:Al films, an important transparent conducting oxide films. The target was a Zn–Al alloy 共the content of Al was 2 wt %兲. The vacuum system was pumped down to a base pressure of 3 ⫻ 10−3 Pa, using a turbo molecular pump. During the measurement, the oxygen partial pressure was always maintained at 0.04 Pa, a typical value for the deposition of ZnO:Al films by reactive sputtering.20 The sputtering gas argon was directly introduced into the chamber, to a total pressure of 0.4– 0.8 Pa, as monitored by a capacitance manometer. The discharge voltage was kept constant at 320 V, resulting in a typical discharge current of 0.15– 0.17 A.

3

FIG. 3. Simulations of the magnetic field configuration for the coil situated in the vicinity of 共a兲 the magnetron 共Zcoil = 0 cm兲 and 共b兲 the substrate holder 共Zcoil = 8 cm兲, with Bext = 100 G assisting the outer pole of the magnetron.

PY

CO

traveled toward the substrate. The higher density and energy of electrons in the substrate vicinity ionized more gases. Figure 2共b兲 also displays that Jion increased from ⬃0.2 to 2 mA cm2 as Bext increased in the same range, when the coil was positioned in the substrate vicinity. However, compared with Fig. 2共a兲, the maximum Jion exhibited a twofold increase. This difference originated from the change of magnetic field configuration in the substrate and target area with the different coil positions, as indicated by FEM simulation. Figure 3 gives the simulation patterns of magnetic field configuration for the different coil position used in Fig. 2, with Bext = 100 G assisting the outer pole of the magnetron. Comparing Fig. 3共a兲 with Fig. 3共b兲, the magnetic field lines of the former moved away from the target, rapidly attenuating, and returned to the magnetron. In contrast, the latter had much more magnetic field lines extending toward the substrate, and the intensity of the vertical magnetic component was much higher near the substrate region, which meant that the magnetic filed was strong enough to constrain the ultimate electrons to closely follow the magnetic lines on helical orbits toward the substrate. In addition, the coil position also had an influence on V f . The changes of V f became flatter when the coil was situated near the substrate, mainly due to the weaker influence of external electromagnetic field on the original magnetic field of the target region. Moreover, Fig. 2 indicates that Jion and V f were slightly affected by the sputtering pressure in the range of 0.4– 0.8 Pa. Figure 4 clearly illustrates the Jion dependence on the position of the external coil Zcoil with Bext of 100 G assisting the outer pole of the magnetron. The ion current density Jion showed the different tendencies of the changes in three distinct ranges of the coil position: 共i兲 Jion had a slight increase in the magnetron region 共Zcoil ⬍ 2 cm兲, 共ii兲 then rapidly increased until the coil moved along the Z axis up to the substrate region 共2 cm⬍ Zcoil ⬍ 6 cm兲, 共iii兲 and gradually decreased further moving the coil away. Therefore, it can be concluded that the coil was more suitable to be placed in the vicinity of the substrate than the magnetron for changing Jion over a larger range.

A

PROOF COPY 001702JVA

JV

JVST A - Vacuum, Surfaces, and Films

02

In type-I systems, variations of the coil position were found to have a strong effect on the plasma parameters near the substrate. Figures 2共a兲 and 2共b兲 show the dependence of Jion and V f on the coil current Icoil and Bext when the coil was situated in the vicinity of the magnetron and substrate, respectively. It is observed from Fig. 2共a兲 that Jion increased from ⬃0.1 to 1 mA cm2 and V f also increased from −8 to − 23 V, as Bext increased from −25 to 100 G 共correspondingly, Icoil increased from −0.5 to 2 A兲. The ion current density near the substrate exhibited a tenfold increase. The increase in Jion and V f has been previously explained by Ivanov et al.6 By applying an increased Bext enhancing the outer pole of the magnetron, the magnetic field lines tend to disperse and extend toward the substrate. Consequently, more electrons gyrated around them in a spiral pattern and

17

A. Effects of solenoid coil arrangement and excitation mode on plasma characteristics

00

III. RESULTS AND DISCUSSION

PROOF COPY 001702JVA 4

Zhang et al.: Influence of the external solenoid coil arrangement and excitation mode

4

O PR

FIG. 4. Jion as a function of Zcoil with Bext of 100 G assisting the outer pole of the magnetron. The lines in the figure are drawn as a guide for the eye.

OF

tween the sheath and the presheath. The ionization collisions occur just above the sheath area.14 The deepest point of the erosion profile corresponds to the place where Br is at its maximum and Bz is zero.15 Figure 6共a兲 shows the 兩Bz / B兩 distribution of the magnetic field at Z = 1 mm under the different Bext. It is observed that the 兩Bz / B兩 profile moved from the outer part to the inner part over a large region in the lateral direction of the target, when Bext changed from −150 to 150 G. This was attributed to the fact that the magnetic field in front of target was compressed toward the inner part 共simulation patterns of the magnetic field were not shown here兲. Consequently, it implied that the confined plasma would shift back and forth on the target surface with the change of Bext which led to a notable broadening of the erosion profile. Furthermore, the deepest points where the magnetic field lines were parallel to the target surface could also shift in the radial direction by changing Bext, as shown in Fig. 6共a兲. It is well known that the deepest point of the target erosion profile determines the failure time of a target during the sputtering. The deepest point of 兩Bz / B兩 共corresponding to the deepest point of target erosion兲 had a large shift from R = 15– 25 mm, which meant a significant improvement of the target life and the utilization efficiency. In this experiment, a method to improve the target materials utilization is presented by placing an additional solenoid coil excited in the ac mode. Therefore, the confined plasma in front of the target would quickly move, as the coil was excited in the ac mode. Compared with other methods, this is relatively simple and effective to improve the target utilization. Variations of the coil position were also found to have a significant influence on the 兩Bz / B兩 erosion profile of the target surface. Comparing Fig. 6共a兲 with Fig. 6共b兲, it shows that the shifting range of the deepest point of 兩Bz / B兩 sharply decreased from 10 to 3 mm, when the coil was moved from the

PY

CO

In another aspect, the solenoid coil position and excitation mode also influenced the I-V characteristic of magnetron. The dependence of target current It on Icoil in different positions and excitation modes are shown in Fig. 5. During the probe measurement, the discharge operated in a constant voltage of 320 V. While the coil excited with a dc of 2 A moved from the substrate to the magnetron region along the Z axis, It gradually decreased from 0.17 to 0.14 A. This decrease was due to the variation of the plasma impedance which resulted from the strengthened influence of the additional magnetic field on the original magnetic field in the target region. Moreover, it was worth noting that the variation of It with the increase of Icoil was obviously different in a different coil excitation mode. In the dc mode, It slowly decreased with increasing Icoil. The decrease can be attributed to the further dispersion of the magnetic field lines in the target region with the increase of Bext, which weakened the ability of confining the electrons on the target surface and increased the plasma impedance. Nevertheless, in the ac mode, It remained approximately constant, which was possibly owing to the quickly alternate external magnetic field. Therefore, the coil operated in the ac mode was beneficial to maintain the stability of the sputtering process.

FIG. 5. Target current It as a function of Icoil and Zcoil for different coil excitation modes. The working pressure is 0.8 Pa.

A

JV

PROOF COPY 001702JVA

02

J. Vac. Sci. Technol. A, Vol. 25, No. 2, Mar/Apr 2007

17

The inhomogeneous magnetic and electric fields in front of a planar magnetron lead to a nonuniform confined plasma and racetrack-shape erosion profile on the target surface. The erosion profile correlates with the patterns of the magnetic field. E ⫻ B electron drift motion occurs in the region where the magnetic filed lines are parallel to the target surface. Therefore, it is desirable to obtain a uniform transverse component of magnetic field over the target surface for higher utilization of the target. The erosion profile can be qualitatively represented by the lateral distribution of 兩Bz / B兩 at Z = 1 mm, where it corresponds to the boundary position be-

00

B. Effects of solenoid coil arrangement and excitation mode on target utilization

PROOF COPY 001702JVA 5

Zhang et al.: Influence of the external solenoid coil arrangement and excitation mode

5

O PR

FIG. 7. Shifting range of the deepest points of 兩Bz / B兩 profiles as a function of Zcoil, where an ac of 2 A passed through the coil.

OF

PY

CO

target utilization but unfavorable to enhance the nearsubstrate ion density. On the other hand, the position of the coil near the substrate would favorably control the ion density but was disadvantageous in increasing the target utilization. In order to take advantage of both coil arrangements, a pair of solenoid coils were placed in the vicinity of the magnetron and the substrate along the Z axis, respectively. The coil around the magnetron was passed through a lowfrequency ac current, while the coil in the substrate vicinity was excited in the dc or ac mode. This novel method was designed to not only improve target materials utilization but also control the near-substrate plasma parameters over a large region. Figure 8 shows the simulations of magnetic field configuration in the case when a pair of coils were positioned near the substrate and the target, with a dc of 2 A and an AC of 冑2 A, respectively. The simulations were conducted in two typical moments of the ac corresponding to the external field of the near-target coil 共a兲 Bext = −100 G and 共b兲 Bext = 100 G.

A

JV

magnetron region to the substrate region. Figure 7 clearly indicates the high sensitivity of the shifting range of the deepest point of 兩Bz / B兩 to the coil position. The value initially increased sharply, reached a maximum at Zcoil = 0 cm, then quickly decreased when the coil was further moved toward the substrate region. Hence, it can be concluded that the coil position should be carefully adjusted and it was more appropriate to place the coil in the vicinity of the magnetron than the substrate, as viewed from improving the target utilization.

02

17

00 FIG. 6. 兩Bz / B兩 profiles of the magnetic field at Z = 1 mm under different Bext. The coil was situated in the vicinity of the 共a兲 magnetron 共Zcoil = 0 cm兲 and 共b兲 substrate holder 共Zcoil = 6 cm兲.

C. Type II

As discussed above, the position of the coil supplied with the ac power near the target was beneficial to increase the JVST A - Vacuum, Surfaces, and Films

PROOF COPY 001702JVA

FIG. 8. Simulations of magnetic field configuration in two typical moments of the AC corresponding to the external field of the near-target coil: 共a兲 Bext = −100 G and 共b兲 Bext = 100 G.

PROOF COPY 001702JVA 6

Zhang et al.: Influence of the external solenoid coil arrangement and excitation mode

It indicates that both the distribution of the magnetic field lines and their vertical components varied a little near the substrate, while the magnetic field above the target shifted over a large region in the radial direction. This meant that it was possible to independently control the near-substrate plasma parameters while simultaneously improving the target utilization. Moreover, the coil operated in the ac mode had an advantage in maintaining the stability of sputtering as indicated in Fig. 5. Other improvements, such as the increased uniformity of the deposited film and mitigating arcing events21 due to the target erosion broadening, could occur in the planar magnetron sputtering, which will be investigated in future work.

O PR

IV. CONCLUSIONS

OF

A dc magnetron sputtering system has been designed to operate with an external magnetic field created by a coil with a dc or an ac. It was found that the coil arrangement and the excitation mode had strong effects on the plasma characteristics and the target utilization. In order to control the ion density over a broad range, it was more suitable to place the coil near the substrate than near the target whether with an ac or a dc. This was due to the increased density of the vertical magnetic lines in the near-substrate area, originating from the changed magnetic field configuration with the changed coil position, using a finite-element method analysis. For improving the target utilization, the coil in the ac mode was more favorable to be positioned near the target than near the substrate. This was attributed to the larger shift of the confined plasma on the target surface, resulting from the different sensitivities of coil positions on the near-target magnetic field distribution. Furthermore, in order to not only improve target materials utilization but also control the near substrate

6

plasma parameters over a larger region, a novel method of placing a pair of coils with the appropriate excited mode was presented. ACKNOWLEDGMENT The authors are grateful to the National Natural Science Foundation of China 共Grant No. 50172051兲 for its financial support of this research. 2

PY

CO

J. S. Colligon, Philos. Trans. R. Soc. London, Ser. A 362, 103 共2004兲. I. Petrov, P. B. Barna, L. Hultman, and J. E. Greene, J. Vac. Sci. Technol. A 21, S117 共2003兲. 3 P. J. Kelly and R. D. Arnell, Vacuum 56, 159 共2000兲. 4 D. M. Mattox, Handbook of Physical Vapor Deposition (PVD) Processing 共Noyes, Westwood, NJ, 1998兲. 5 I. Petrov, F. Adibi, J. E. Greene, W. D. Sproul, and W. D. Munz, J. Vac. Sci. Technol. A 10, 3283 共1992兲. 6 I. Ivanov, P. Kazansky, L. Hultman, I. Petrov, and J. E. Sundgren, J. Vac. Sci. Technol. A 12, 314 共1994兲. 7 S. L. Rohde, I. Petrov, W. D. Sproul, S. A. Barnett, P. J. Rudnick, and M. E. Graham, Thin Solid Films 193/104, 117 共1990兲. 8 C. Engstrom, T. Berlind, J. Birch, L. Hultman, I. Ivanov, S. R. Kirkpatrick, and S. Rohde, Vacuum 56, 107 共2000兲. 9 D. A. Glocker and S. Ismat Shah, Handbook of Thin Film Process Technology 共Institute of Physics, Bristol, 1995兲. 10 R. Kukla, Surf. Coat. Technol. 93, 1 共1997兲. 11 R. K. Waits, J. Vac. Sci. Technol. 15, 179 共1978兲. 12 T. Hata and Y. Kamide, J. Vac. Sci. Technol. A 5, 2154 共1987兲. 13 S. Ido and K. Nakamura, Jpn. J. Appl. Phys., Part 1 32, 5698 共1993兲. 14 S. Ido, M. Kashiwagi, and M. Takahashi, Jpn. J. Appl. Phys., Part 1 38, 4450 共1999兲. 15 S. Ido, Y. Ishida, and K. Hijikata, Jpn. J. Appl. Phys., Part 1 32, 2112 共1993兲. 16 I. V. Svadkovski, D. A. Golosov, and S. M. Zavatskiy, Vacuum 68, 283 共2003兲. 17 www.gencoa.com 18 POISSON SUPERFISH, freely available from www.LANL.com 19 I. Petrov, I. Ivanov, V. Orlinov, and J. Kourtev, Contrib. Plasma Phys. 30, 223 共1990兲. 20 X. B. Zhang, Z. L. Pei, J. Gong, and C. Sun, J. Appl. Phys. 100 共2006兲, available online. 21 A. Anders, Thin Solid Films 502, 22 共2006兲. 1

A

JV

02

17

00

J. Vac. Sci. Technol. A, Vol. 25, No. 2, Mar/Apr 2007

PROOF COPY 001702JVA