A modified DC sputter process is presented, which enables the independent control of the primary ion energy for sputtering and, by this, the control of the energy of the sputtered atoms forming the layer. Moreover, the energy flux of neutral reflected sputter gas atoms is controlled by this way. For this, a sputter magnetron was developed, where the target area of the erosion zone is isolated from the residual target and powered by an additional negative voltage. It could be shown with ion energy measurement of the primary sputtering ions at the erosion zone, that this voltage increases the primary ion energy. An estimation of the sputter yield from the deposition speed and the target currents shows a good agreement with the theoretical sputter yield for several target materials. First depositions of copper layers with varying primary ion energy show the influence of the sputter conditions controlled in this way by changing of electrical resistance and of mass density.
In direct ion beam sputtering it is a well-known effect, that the ion energy of the sputtering ions influences the particle energy of the sputtered target atoms. Earlier [1] and later publications [2] of this show that the energy distribution of sputtered target atoms is shifted to higher energies by varying the primary ion energy between 250 and 1.000 eV. By this way the mean energy of the sputtered particles grows typically from 3 to 6 eV up to 6 to 10 eV, having influence on the corresponding layer growth. Moreover, the energy of reflected neutral atoms in dependence on the primary ion energy is increased and reach a remarkable high energy flux also influencing the layer growth [3,4].
This effect is well known in direct ion beam sputtering. The primary ion energy in magnetron sputtering is determined by the strength of the permanent magnets und some plasma parameters and is normally in the range between 250 and 500 eV [5]. The aim of this work is to show that with a modified magnetron (Double Target Magnetron, DTM) an extended and controlled primary ion energy range up to 1.000 eV like in direct ion beam sputtering can be realized.
Like shown at Figure1, the Double Target Magnetron consists of a conventional magnetron magnet system generating a conventional coaxial or linear (oval) sputter plasma. It should not be confused with the Double Ring Magnetron [6] or with the Twin Magnetrons [7] which consist both of two independent sputter magnetrons in one arrangement.
Fig. 1: Principal arrangement of sputtering with the Dual Target Magnetron in cross section
Fig. 2: Target currents of the Dual Target Magnetron in dependence from the additional negative voltage U2 applied on target 2 (argon, 10–2 mbar, 80 W)Experimental
Figure 1 shows the used setup in cross section. A modified sputter magnetron (Dual Target Magnetron DTM) with two target parts is used. The sputter plasma was ignited and generated by the DC-generator U1, applied between anode and target 1. The other target part (erosion zone, target 2), where most sputtering occurs, was isolated against target 1 and powered by an additional negative voltage U2.
To determine the ion energy distribution of the primary ions at target 2 a Retarding Field Analyzer (RFA) [8] was integrated into target 2. The substrate was located direct over target 2 (distance 40 mm) together with a quartz microbalance sensor.
The image on page 362 shows the used Dual Target Magnetron in operation (power 80 W, pressure 1×10–2 mbar). A rectangular target with dimensions of 135 × 110 mm was used so that a linear sputter magnetron was created. It shows the oval shaped sputter plasma together with the erosion zone consisting of target 2. Both targets had been made of copper. While increasing U2, the plasma did not change its typical oval shape.
Figure 2 shows the target currents I1 and I2 (corresponding to Fig. 1), depending on the additional negative target voltage U2 at target 2. Both currents stay constant, which shows that there is no change in the sputter plasma, while increasing the ion acceleration to target 2.
Results Primary ion energy
To demonstrate the effect of primary ion acceleration in dependence from target voltage U2, a Retarding Field Analyzer [9] was integrated into target 2 in such a way that the first analyzer grid was electrically connected with target 2 and made of 1 mm thick copper sheet to ensure a pure copper sputter process. Ion current densities up to 3 mAcm–2 could be analyzed with this special RFA.
Figure 3 shows the measured ion energy distributions while varying only the voltage U2 at target 2. In the case of common magnetron sputtering (U2 = 0 V), the primary sputtering ions have a mean ion energy around 280 eV. That is around 40 V lower than the corresponding DC sputter generator voltage of 320 V. The difference of 40 V can be explained from the potential distribution in the plasma as a typical anode potential [10].
While the voltage U2 is increased up to 300 V, the ion energy distribution peak is shifted to higher energies by nearly the amount of U2. Some more measurements of this shift lead to an estimation formula <1> for the mean primary ion energy in dependence from the voltages applied at the Dual Target Magnetron:
Wion = e*U1 - e*Uanode + C*e*U2<1>
with Uanode = 40 V, C = 0.9 and e = 1.610–19 As
In the following, equation <1> was used to determine directly the actual mean primary ion energy from the voltages applied at the Dual Target Magnetron (see Fig. 1).
Sputter yield
The thin film deposition at the substrate is a superposition of sputtered atoms, coming from target 1 (sputtered with sputter yield Y1(Wion1) and coming from target 2 (sputtered with Y2(Wion2). By arranging the substrate direct over target 2 at a close distance of 40 mm, mostly the sputter process from target 2 contributes to the layer growth. If R = F/ρtar2*Δf/Δt is the measured growth rate at the substrate (measured with quartz microbalance, F – constant factor, ρtar2 – layer mass density and Δf/Δt – quartz frequency shift) and I2 is the current at target 2, then the sputter yield Y2 at target 2 is nearly proportional to R/I2 <2>:
Y2(Wion2)= ΔNtar2/ΔNion2 ~ R(Wion2)/I2(Wion2)<2>
By applying equation <2>, a value for characterization of the sputter yield Yexp(Wion) in nm/(min*mA) can be determined and compared with the corresponding sputter yield from literature [11] in one diagram (Fig. 4).
Fig. 3: Ion energy distributions of the primary ions at sputtering of copper (argon, 10–2 mbar, 80 W) measured with a Retarding Field Analyzer arranged at target 2
Fig. 4: Comparison of theoretical (right y-axis) and experimental (left y-axis) sputter yield at target 2 for different target materials in dependence on the primary ion energy (power 80 W, 10–2 mbar argon)
Several target materials, covering a wider range in sputter yield, had been installed at the Dual Target Magnetron. The typical magnetron parameters (power: 80 W, 10–2 mbar argon) had been held constant, and only the target 2 voltage U2 was varied, leading to an increase in primary ion energy from 250 up to 1.000 eV (determined by equation <1>). Although the method according to equation <2> is only a rough estimation, a good agreement between theoretical and “experimental” sputter yield can be demonstrated. After the effect of controlled primary ion energy was shown directly in Figure 3 with the measurement of the ion energy distribution, Figure 4 now provides indirect proof of this through the sputter yield behaviour.
Copper layers
To demonstrate the influence of the increased primary ion energy on layer growth as an example, some copper layers had been deposited. The Dual Target Magnetron was equipped with copper at target 1 and target 2 and approx. 200 nm thick layers had been deposited with constant plasma conditions (power 80 W, 10–2 mbar argon, substrate temperature around 50 °C). Only the primary ion energy was varied by U2 and the corresponding primary ion energy calculated by equation <1>.
The specific resistance of the layers was measured by a four point probe (Rsq) and calculated by
Rfilm = Rsq*d <3>)
with d – layer thickness, measured with a profilometer. Figure 5 illustrates that the specific resistance during sputtering with the Dual Target Magnetron decreases slightly with increasing primary ion energy.
Moreover, the mass density of the deposited copper layers could be determined in the same experiment from quartz microbalance data <4>:
ρfilm = D*Δf/d <4>
with D – constant factor and Δf – frequency change during deposition.
Figure 6 shows the so-determined mass density of approx. 200 nm thick copper layers as a function of the primary ion energy. The mass density of solid copper is 8.9 g/cm3. An increase of the mass density from around 7 g/cm3 to more than 8 g/cm3 can be demonstrated while increasing the primary ion energy.
Fig. 5: Specific resistance of copper layers as a function of the primary ion energy sputtered with the DTM (power 80 W, 10–2 mbar argon)
Fig. 6: Mass density of copper layers as a function of the primary ion energy sputtered with the DTM (power 80 W, 10–2 mbar argon)Summary/Outlook
In this work the principal function of the newly developed Dual Target Magnetron is demonstrated with:
- demonstration of the primary ion acceleration to target 2 as a function of the voltage U2 by measuring the corresponding ion energy distributions,
- demonstration of the increase of the sputter yield as a function of the primary ion energy for several target materials.
First layer deposition tests show that the layer properties of copper films could be improved by increasing primary ion energy.
Further development in this field should be based on: - More investigation of layer properties, deposited through the DTM while varying the primary ion energy,
- further magnetron development (tube target versions, optimized DTMs),
- RF-sputtering with the DTM for deposition of semiconductor layers or dielectric layers like oxides, nitrides, etc.
Literatur
[1] R. V. Stuart, G. K. Wehner, „Energy Distribution of Sputtered Cu Atoms“, J. Appl. Phys., 35, (1964), 1819 – 1824
[2] M. Stepanova, S. K. Dew, “Estimates of differential sputtering yields for deposition applications”, J. Vac. Sci. Technol. A 19, (2001), 2805-2816
[3] W. Eckstein, Computer Simulations of Ion-Solid Interactions, Springer Series in Material Science 10, 1991, p 162
[4] C. Bundesmann, R. Feder, T. Lautenschläger, H. Neumann, ”Energy Distribution of Secondary Particles in Ion Beam Deposition Process of Ag: Experiment, Calculation and Simulation”, Contrib. Plasma Phys. 55 (2015) 737-745
[5] H. Schlemm, „Ion energy analysis of sputtering ions at planar targets with DC- and RF-sputtering by Retarding Field Analysis“, Poster Contribution Conference Plasma Surface Engineering 2022, http://www.jenion.de/News/, (accessed 20 December 2024)
[6] P. Frach, K. Goedicke, T. Winkler, Chr. Gottfried, H. Walde, W. Hentsch, “Advantageous, possibilities, design aspects and technical use of double-ring magnetron sputter sources”, Surface and Coatings Technology 74-75, (1995) 85-91
[7] G. Bräuer, M. Ruske, J. Szczyrbowski, G. Teschner, A. Zmelty, “Mid frequency sputtering with TwinMag®-a survey of recent results”, Vacuum, Volume 51, Issue 4, 1 December 1998, pp 655-659
[8] C. Böhm, J. Perrin, “Retarding-field analyzer for measurements of ion energy distributions and secondary electron emission coefficients in low-pressure radio frequency discharges”, Rev. Sci. Instrum. 64 (1993) 31–44, https://doi.org/10.1063/1.1144398
[9] Product information PlasmaMon 3, http://www.jenion.de/Plasma-Analysis/ (accessed 20 December 2024)
[10] A. Pflug, M. Siemers, C. Schwanke, B. Febty Kurnia, V. Sittinger, B. Szyszka, “Simulation of plasma potential and ion energies in magnetron sputtering”, Materials Technology 2011, VOL 26, NO 1, pp 10 – 14
[11] TU Wien, Institute of Applied Physics, “A Simple Sputter Yield Calculator”, https://www2.iap.tuwien.ac.at/www/surface/sputteryield (accessed 20 December 2024)
