residual stress reduction in sputter deposited thin · pdf fileresidual stress reduction in...
TRANSCRIPT
Residual Stress Reduction in Sputter Deposited Thin Films
by Density Modulation
Journal: 2009 MRS Fall Meeting
Manuscript ID: 1224-FF05-22.R1
Symposium: Symposium FF
Date Submitted by the
Author:
Complete List of Authors: Alagoz, Arif; University of Arkansas at Little Rock, Department of Applied Science Kamminga, Jan-Dirk; Materials Innovation Institute; Delft University of Technology, Department of Materials Sciences and Engineering Grachev, Sergey; Saint-Gobain Recherche Lu, Toh-Ming; Rensselaer Polytechnic Institute, Physics Karabacak, Tansel; University of Arkansas at Little Rock, Department of Applied Science
Keywords: sputtering, Ru, microstructure
RESIDUAL STRESS REDUCTION IN SPUTTER DEPOSITED THIN FILMS BY
DENSITY MODULATION
Arif S. Alagoz1, Jan-Dirk Kamminga
2,3, Sergey Yu. Grachev
4, Toh-Ming Lu
5 and Tansel
Karabacak1
1 Department of Applied Science, University of Arkansas at Little Rock, Little Rock, AR 72204,
U.S.A. 2 Materials Innovation Institute M2i, Mekelweg 2, 2628 CD Delft, Netherlands
3 Department of Materials Sciences and Engineering, Delft University of Technology, Mekelweg
2, 2628 CD Delft, Netherlands 4
Saint-Gobain Recherche, 39 quai Lucien Lefranc, B.P. 135, 93303 Aubervilliers, France 5 Department of Physics, Applied Physics and Astronomy, Rensselaer Polytechnic Institute,
Troy, NY 12180, U.S.A.
ABSTRACT
Control of residual stress in thin films is critical in obtaining high mechanical quality
coatings without cracking, buckling, or delamination. In this work, we present a simple and
effective method of residual stress reduction in sputter deposited thin films by stacking low and
high material density layers of the same material. This multilayer density modulated film is
formed by successively changing working gas pressure between high and low values, which
results in columnar nanostructured and dense continuous layers, respectively. In order to
investigate the evolution of residual stress in density modulated thin films, we deposited
ruthenium (Ru) films using a DC magnetron sputtering system at alternating argon (Ar)
pressures of 20 and 2 mTorr. Wafer’s radius of curvature was measured to calculate the intrinsic
thin film stress of multilayer Ru coatings as a function of total film thickness by changing the
number of high density and low density layers. By engineering the film density, we were able to
reduce film stress more than one order of magnitude compared to the conventional dense films
produced at low working gas pressures. Due to their low stress and enhanced mechanical
stability, we were able to grow these density modulated films to much higher thicknesses without
suffering from buckling. Morphology and crystal structure of the thin films were investigated by
scanning electron microscopy (SEM), transmission electron microscopy (TEM), and X-ray
diffraction (XRD). A previously proposed model for stress reduction by means of relatively
rough and compliant sublayers was used to explain the unusually low stress in the specimens
investigated.
INTRODUCTION
It is well known that residual (also called intrinsic) stress naturally evolves during growth of
sputter deposited thin films. The control of this stress is essential for high mechanical quality
coatings without cracking, buckling, or delamination [1-2]. Dependence of residual stress on
process parameters and microstructure is systematically investigated by many research groups.
Haghiri-Gosnet et al. [3] and Windischmann [4] showed strong correlation between residual
stress and Thornton’s structure zone model (SZM) [5-9] for sputtered thin films. Low density
thin films deposited at high argon pressure and low temperature (regime Zone I) have almost
zero residual stress. However, these films exhibit poor electrical properties due to their columnar
Page 1 of 6
microstructure. Decreasing argon pressure develops tensile stress and further decrease in
pressure results in a dramatic change from tensile to compressive stress finally forming a dense
thin film (regime Zone T). In order to deposit low stress thin films, setting working pressure
between tensile to compressive stress transition region is not feasible since the working pressure
window is too narrow and the deposition is not stable. Karabacak et al. [10] demonstrated that
compressive stress build-up in a film can be reduced by modulating between relatively rough and
compliant sublayers, and dense sublayers. Overall thin film stress was reduced in sputter
deposited tungsten thin films by successively changing sputtering conditions leading to a density
modulated multilayer thin film composed of low and high density layers. The low density layer
consisted of nano-columns deposited under high pressure is believed to be acting as compliant
layer. It was proposed that a rough compliant underlayer delays the development of compressive
stress in the following dense layer over a thickness in the order of the underlayer’s roughness.
In this paper we present stress reduction in multilayer ruthenium (Ru) sputtered thin films of
more than one order of magnitude compared to the conventional dense films by engineering the
film density. It will be shown that the low density films exhibit a significant roughness and are
likely to be acting as a compliant layer, thus explaining the low stresses observed. It is also
shown that the deposition pressure has an important effect on crystal orientation, where high and
low pressure depositions results in a different crystal texturing in density modulated thin films.
EXPERIMENT
All Ruthenium depositions are performed at room temperature by dc magnetron sputtering
technique. Films were deposited on n-type, 12-25 Ω·cm resistivity, (100) oriented single crystal
silicon wafers using 99.95% pure 3 inch ruthenium target. All samples were mounted parallel to
target at 18 cm distance. Before each deposition, 5 x 10-7
Torr base pressure was achieved by a
turbo-molecular pump. All depositions were performed at 200 W DC power using ultra pure
Argon plasma. Argon pressure was set to 2 mTorr in order to deposit high density single layer
films and 20 mTorr for low density columnar single layer thin films. Density modulated
multilayer thin film depositions started with 15 nm thick low density film grown at 20 mTorr
followed by 17 nm dense film at 2 mTorr. Pressure level was successively changed between
these two values until the desired thin film thickness is reached.
Intrinsic thin film stress was calculated from the Stoney equation [11]. Radius of curvature
of wafer before and after the deposition was measured by using a Flexus 2320 dual wavelength
(λ1 = 670 nm and λ2 = 750 nm) system. Thin films’ morphology and thickness were investigated
by using Carl Zeiss Ultra 1540 dual beam focused ion beam (FIB)/scanning electron microscopy
(SEM) system. Crystal structure of the films was studied by using a Scintag X-ray
diffractiometer (XRD) using a Cu target operated at 50 kV and 30 mA. Transmission electron
microscopy (TEM) was performed on cross sections of Ru multilayers, prepared by ion milling
in a Gatan PIPS 691 ion mill, using Ar. Specimens were analyzed in a CM30T Philips
transmission electron microscope operated at 300 kV.
DISCUSSION
Figure 1 shows the measured stress values of single layer dense ruthenium thin films
deposited at low argon pressure, low density films deposited high argon pressure, and multilayer
thin films deposited at successive high/low argon pressure as a function of total film thickness.
Page 2 of 6
Figure 1. Stress evolution as a function of film thickness.
During deposition, a highly compressive residual stress develops at the Ru thin films deposited at
a low argon pressure. These high material density films could not be grown thicker than 85 nm,
after which they peeled off from the substrate. In contrast, we could deposited almost 2 µm thick
low density single layer thin films at high argon pressure with close to zero tensile residual
stress. SEM images show that these films exhibit a columnar microstructure as predicted by the
structure zone model (Fig. 2 a-b). Deposition at high argon pressure reduces the directionality of
the incoming ruthenium flux to the substrate and enhances the shadowing effect during film
deposition which is responsible for the formation of a columnar structure [12]. Multilayer Ru
thin films deposited by changing argon pressure successively between high and low pressure are
relatively dense (Fig. 2 c-d) and show very low residual stress values (Fig. 1). This way, we
reduced total residual stress of ruthenium films one order of magnitude compared to the
conventional dense films. Using density modulation technique, very thick films with reasonably
low stresses can be prepared. Interestingly, unlike the single layer high density Ru films, the
stress of multilayer film increases slowly as the total thickness increases. This tendency can be
explained by thermal stress evolution during long deposition times due to the thermal expansion
coefficient mismatch between Si substrate (2.6 x 10-6
K-1
) and Ru (6.4 x 10-6
K-1
) film. Thermal
stress seems to stabilize at thicker samples (Fig. 1).
The microstucture of multilayer thin films was further examined by TEM. In Fig. 3, the
TEM image of a density modulated layer clearly shows high and low density sublayers, where
the low density layers exhibit a marked porosity. According to Ref. [10] stress build-up in dense
films is delayed by a compliant rough layer as long as the dense layer thickness is of the order of
the roughness of the interface. Figure 3 shows that the interface is sufficiently rough. The
porosity of the low density layer has a nanocrack-like appearance which suggests that the layer is
also sufficiently compliant to reduce the stress in the dense layer. (On the basis of the present
results we cannot conclude whether the nanocrack-like features were produced by deformation,
or are a direct result of the growth process).
Page 3 of 6
Figure 2. Top and cross-sectional view SEM images of (a-b) low density and (c-d) low/high
density multilayer samples. Scale bars at each image measure 500 nm.
Consequently, the findings are in good correspondence with the model proposed in [10]. In
Ref. [13] an alternative model for stress generation was proposed, in which the total stress is the
sum of a compressive growth stress, and a tensile stress created at grain boundaries. This model
could also explain the current results, because the current deposition strategy leads to small
grains. However, the model in Ref. [13] relies on the columnar microstructure of the layers. In
the present layers, a clear columnar microstructure is not observed. Therefore we attribute the
low stress to the compliance of the low density sublayers.
Figure 3 Cross-sectional TEM image of high/low pressure multilayer thin film.
Page 4 of 6
In addition, we investigated the crystal orientation (texture) of thin films by XRD and
observed that crystal orientation of the grains is strongly affected by argon pressure level. As
shown in Fig. 4, thin films deposited at low pressure are textured along (002) and (101)
orientations. On the other hand, thin films deposited at high pressure are textured along the (100)
orientation. Different texture evolution for different argon pressure levels can be explained by
the increased shadowing effect at a high pressure and different surface mobility of adatoms on
different crystal orientations. Karabacak [14, 15] presented that at the early stages of deposition,
low surface mobility islands can grow vertically as opposed to higher surface mobility islands
which prefer lateral island growth. Due to their taller height, the shadowing effect enhances
further vertical growth of low adatom mobility islands making them the grains of dominant
crystal orientation. In a similar way, increasing argon pressure enhances the shadowing effect
and the crystal growth on low mobility crystal orientation. XRD measurements of multilayer thin
films exhibit the crystal orientation contribution of both low and high density layers. However,
the profile of the density modulated film is shown to be quite different than a simple composition
of the profiles of each layer. In the density modulated film (002) orientation becomes quite
insignificant unlike in the single layer high density films. This can be attributed to a “resetting”
effect, in which just before the (002) islands start to grow at the expense of (101) islands during
the low pressure step, Ar pressure is set to high and the (100) island growth is promoted. When
the pressure is set to low again, the process starts from the beginning again resetting the growth.
This mechanism is also believed to influence the dynamic evolution of the island size. By
resetting the growth at each layer, we keep the average grain size small before it has a chance to
grow. We are currently investigating this property in more detail.
Figure 4. Crystal orientation (texture) of Ru films.
We also note that, in Fig. 4, central peak positions in the XRD profiles of a single layer high
and low density films are shifted from their equilibrium positions, which is due to the residual
stress [16]. On the other hand, there is no such a shift in low stress density modulated thin films
of Ru.
Page 5 of 6
CONCLUSIONS
In conclusion, it has been shown that material density of a sputter deposited film can be
engineered by changing the argon pressure between high and low levels. This approach results in
a density modulated multilayer thin film having high and low density layers produced by low
and high working gas pressures, respectively. By this way, the residual compressive stress values
can be reduced more than one order of magnitude and the requirement of a critical thickness for
films without buckling may be relaxed. The low density layers act as a compliant layer to reduce
the stress of the subsequent deposited high density layer. In addition, the crystal orientation
profile of density modulated films is shown to be quite different than a simple composition of the
profiles of each layer.
ACKNOWLEDGEMENT S
F.D Tichelaar at the National Centre for HREM at Delft University of Technology is
acknowledged for the investigations by transmission electron microscopy. This work was
supported by the NSF.
REFERENCES
1. Thin Film Materials: Stress, Defect Formation and Surface Evolution, L. B. Freund and S.
Suresh, (Cambridge University Press, 2004)
2. D. C. Meyer, A. Klingner, T. Holz, and P. Paufler, Appl. Phys. A: Mater. Sci. Process. 69, 657
(1999).
3. A. M. Haghiri-Gosnet, F. R. Ladan, C. Mayeux, H. Launois, and M. C. Joncour, J. Vac. Sci.
Technol. A 7, 2663 (1989).
4. H. Windischmann, J. Vac. Sci. Technol. A 9, 2431 (1991).
5. J. A. Thornton, J. Vac. Sci. Technol. 11, 666 (1974).
6. J. A. Thornton, J. Vac. Sci. Technol. 12, 830 (1975).
7. J. A. Thornton, Annu. Rev. Mater. Sci. 7, 239 (1977).
8. J. A. Thornton, J. Vac. Sci. Technol. A 4, 3059 (1986).
9. J. A. Thornton, Proc. SPIE 821, 95 (1987).
10. T. Karabacak, C. R. Picu, J. J. Senkevich, G.-C. Wang, and T.-M. Lu, J. Appl. Phys. 96, 5740
(2004).
11. G. C. Stoney, Proc. R. Soc. London, Ser. A 82, 172 (1909).
12. R. Messier, A. P. Giri, and R. A. Roy, J. Vac. Sci. Technol. A 2, 500 (1984).
13. G.C.A.M. Janssen and J.-D. Kamminga, Appl. Phys. Lett. 85, 3086 (2004).
14. T. Karabacak, A. Mallikarjunan, J. P. Singh, D. Ye, G.-C. Wang, and T.-M. Lu, Appl. Phys.
Lett. 83, 3096 (2003).
15. D. Deniz, T. Karabacak, and J. M. E. Harper, J. Appl. Phys. 103, 083553 (2008).
16. I. C. Noyan, T. M. Shaw, and C. C. Goldsmith, J. Appl. Phys. 82, 4300 (1997).
Page 6 of 6