Thin Solid Films 351 (1999) 158–163
`
`Properties of SiO2 and Si3N4 layers deposited by MF twin magnetron
`sputtering using different target materials
`
`M. Ruskea,*, G. Bra¨uera, J. Pistnera, J. Szczyrbowskia, M. Weigertb
`
`aLEYBOLD SYSTEMS GmbH, Wilhelm-Rohn-Strasse 25, D-63450 Hanau, Germany
`bLEYBOLD MATERIALS GmbH, Wilhelm-Rohn-Strasse 25, D-63450 Hanau, Germany
`
`Abstract
`
`Si3N4 and SiO2 layers can be deposited by reactive sputtering in a stable manner by using MF twin magnetron systems. The increasing
`demand for these materials for industrial applications makes it necessary to find new solutions for the target material. The up to now mostly
`used boron-doped poly-crystalline silicon suffers from serious drawbacks. In this paper, properties of Si3N4 and SiO2 layers deposited by
`using casted Si/Al alloy targets as well as conventional Si targets are compared. The advantages of using casted alloy targets are presented.
`q 1999 Elsevier Science S.A. All rights reserved.
`
`Keywords: Optical coatings; Silicon oxide; Silicon nitride; Sputtering; Sputter targets
`
`1. Introduction
`
`Since the introduction of MF (sine wave) powered twin
`magnetron sputter systems, highly insulating dielectrics
`such as Si3N4 and SiO2 can be deposited in a stable process
`[1–3]. The demand for silicon targets for reactive sputtering
`of these materials is dramatically increasing [1]. Mainly the
`usage of SiO2 layers in antireflective coating systems and
`the application of Si3N4 layers in modern low emissivity
`layer stacks require large amounts of reliable sputter targets
`at low cost for the end user [4,5].
`Especially with the increased number of planar MF twin
`magnetron systems (e.g. LEYBOLD TwinMagw) installed
`in mass production lines, the necessity of a more economic
`solution for flat sputter targets becomes urgent. Up to now
`practically all planar silicon cathodes were working with
`boron doped high purity silicon targets [6]. These sputtering
`targets were produced from large blocks of polycrystalline,
`so called ‘solar silicon’. This material was 99.999% pure
`silicon with a boron doping of a few ppm to achieve suffi-
`cient conductivity for DC sputtering. Because of the compli-
`cated cutting and grinding processes for this target material
`and because of the necessity of solder bonding of these
`targets, the target costs for reactive sputtering of silicon
`compounds in mass production were comparably high.
`Another disadvantage of the high purity silicon is the high
`
`* Corresponding author. Tel.: 149-618-134-1143; fax: 149-618-134-
`1850.
`E-mail address: ruskemanf@leybold-systems.de (M. Ruske)
`
`danger of target cracking as soon as the target is exposed to
`permanent thermal cycling. This target cracking can cause
`undesired particle generation in the sputter chamber.
`This paper investigates a new class of less brittle and less
`expensive silicon alloys. These new alloy targets (SISPAe,
`LEYBOLD MATERIALS) promise to overcome the
`economical,
`logistical and technical problems with the
`conventionally used high purity silicon targets.
`
`2. Experimental
`
`The SISPAe alloys [7] are silicon based with aluminum
`as major alloying component. With respect to the major
`alloying components the target material has a purity of
`99.9%. The microstructure of SISPAe can be seen in Fig.
`1. A finely dispersed grain structure of a primary Silicon
`phase is surrounded by a ductile second phase to minimize
`the brittleness of the target material.
`The SISPAe alloys have to be melted under vacuum and
`they are cast into near net shape molds. The chemical homo-
`geneity and the fine grain dispersion is achieved by very
`rapid cooling and by the addition of grain refining additives
`(,0.1%). Because of its good ductility the alloy can be
`machined easily into any desired planar target geometry.
`The essential characterizing data of the SISPAe target
`material are listed in Table 1. The major differences in
`comparison with pure silicon are a much lower electrical
`resistance, a slightly higher coefficient of thermal expansion
`and a lower bending strength and thus less brittleness.
`
`0040-6090/99/$ - see front matter q 1999 Elsevier Science S.A. All rights reserved.
`PII: S0 0 4 0 - 6 0 9 0 ( 9 9 ) 0 0 1 5 7 - 1
`
`Page 1 of 6
`
`APPLIED MATERIALS EXHIBIT 1035
`
`

`

`M. Ruske et al. / Thin Solid Films 351 (1999) 158–163
`
`159
`
`Fig. 2. Surface of a sputtered SISPAe 10 alloy target.
`
`Leybold Z600 coater using three different target materials:
`conventional B-doped poly-Si, Si/Al(<5 at.%) alloy
`(SISPAe 5) and Si/Al(<10 at.%) alloy (SISPAe 10).
`The used target plates had a size of 488 £ 88 mm2. Two
`cathodes, each equipped with the target plates, were
`arranged side by side and powered with a MF sine wave
`power supply at a frequency of 40 kHz. The applied power
`density to the target surface was about 5 W/cm2. The resi-
`dual gas pressure was about 3 £ 1024 Pa. For the deposition
`of SiO2 and Si3N4, an Ar/O2 and Ar/N2 gas mixture was
`used,
`respectively. The total gas pressure was varied
`between 0.15 and 1.2 Pa. The flow of the reactive gas
`component was adjusted to obtain fully transparent layers
`at the highest possible deposition rate. Slide glass with a
`thickness of 1 and 3 inch Si wafers were used as substrates.
`The temperature during deposition was
`the ambient
`temperature. The substrates were placed on a carrier and
`transported with constant speed, passing the activated twin
`magnetron arrangement horizontally several times until the
`desired layer thickness was reached.
`
`Fig. 3. Density of Si3N4 layers deposited at different total pressures with
`different target materials.
`
`Fig. 1. Microstructure of a SISPAe 10 alloy target.
`
`During sputtering from a SISPAe alloy, a much rougher
`target surface than known from the pure silicon will develop
`(see Fig. 2). Because of this rough surface, the Si3N4 or SiO2
`redeposit at the target edges is sticking much better to the
`target, thus reducing the amount of particles falling down.
`Another interesting feature is the possibility to produce
`clamp versions of the target, which makes it very easy to
`recycle the used targets. Also larger target tiles can be casted
`compared to the 100 £ 100 mm2 standard size of solar sili-
`con.
`SiO2 and Si3N4 single layers have been deposited in a
`
`Table 1
`Material data of Si/Al( < 10 at.%)-alloy (SISPAe 10) in comparison to
`boron-doped pure silicon
`
`SISPAe 10
`
`High purity B-
`doped poly-Si
`
`Purity (%)
`Melting point (8C)
`Density (g/cm3)
`Grain size (mm)
`Thermal conductivity (W/m K)
`Electrical resistivity V cm
`Thermal expansion, /K (1008C)
`Bending strength (MPa)
`
`99.9
`580–1360
`2.31–2.36
`, 1
`90
`1.4 £ 1024
`4.1 £ 1026
`38 ^ 3
`
`99.999
`1414
`2.33
`10–100
`126
`0.5–5
`2.6 £ 1026
`145 ^ 25
`
`Page 2 of 6
`
`

`

`160
`
`M. Ruske et al. / Thin Solid Films 351 (1999) 158–163
`
`Fig. 4. SEM picture of the cross section of a Si3N4 layer deposited from a SISPAe 10 alloy target at a pressure of 0.15 Pa.
`
`The layer densities were determined by measuring the
`film thickness and the mass of the substrates before and
`after deposition with a micro balance with a resolution of
`10 mg. The layer thickness was measured using the stylus
`method. The surface morphology and cross sections of the
`
`layers were examined with scanning electron microscopy
`(SEM). The microhardness was determined with the inden-
`tation method using a Vickers diamond. The internal stress
`of the films was calculated using the film thickness and the
`radius of curvature of the Si wafers before and after deposi-
`
`Fig. 5. SEM picture of the cross section of a Si3N4 layer deposited from a SISPAe 10 alloy target at a pressure of 1.10 Pa.
`
`Page 3 of 6
`
`

`

`M. Ruske et al. / Thin Solid Films 351 (1999) 158–163
`
`161
`
`Index of refraction for a wavelength of 550 nm of Si3N4 layers
`Fig. 6.
`deposited at different total pressures from different target materials.
`
`tion. These examinations were carried out using samples
`with a film thickness of about 2 mm. The index of refraction
`was determined by spectral ellipsometry (Sentech SE 800)
`using samples of about 150 nm film thickness.
`
`3. Results
`
`3.1. Si3N4
`
`The density of the deposited layers decreases in general
`with increasing total pressure (Fig. 3). For all target materi-
`als, the measured value ranges from 2.85 g/cm3 at 0.15 Pa to
`about 2.65 g/cm3 at 1.1 Pa. This behavior is typical for
`sputter deposited layers and can be attributed to a change
`in the structure. At higher total pressures, the sputtered
`particles are scattered. The deposition rate is lower and
`particles arrive at the substrate with lower energy and a
`high oblique component, thus changing the growth mechan-
`isms and the morphology of the layer. SEM pictures of a
`cross section of a Si3N4 layer deposited at low and high total
`pressure using a SISPAe 10 target are shown in Figs. 4 and
`
`Fig. 8. Internal stress of Si3N4 layers deposited at different total pressures
`for different target materials.
`
`5, respectively. A transition from a homogeneous structure
`to a fibrous morphology can be seen. This is similar for the
`other target materials used in this work.
`refraction
`According to the density,
`the index of
`measured for a wavelength of 550 nm decreases for increas-
`ing total pressures (Fig. 6). For the high purity poly-Si
`target, the value ranges from about 2.05 to 1.93, which is
`typical for plasma-deposited Si3N4 [8]. Layers which were
`deposited using the SISPAe 5 targets show almost the same
`values, but a clear increase of the index of refraction can be
`seen for the SISPAe 10 target, perhaps due to an increasing
`AlN content in the layer.
`Also the microhardness of the deposited layers decreases
`with increasing total pressure due to the changing density
`and morphology (Fig. 7). The values are significantly higher
`than for uncoated slide glass and range from 25–18 GPa.
`The mechanical stability of layer systems can be improved
`if this material is used as the top layer. There is no clear
`dependency on the target material.
`The internal stress of the deposited layers depends very
`strongly on the total pressure (Fig. 8). High values of
`compressive stress are measured for low total pressures up
`to about 0.4 Pa. The stress can be reduced by increasing the
`total pressure. Then it turns to the tensile region. Care has to
`be taken if conclusions are drawn for optical layer systems
`
`Fig. 7. Microhardness of Si3N4 layers deposited at different total pressures
`from different target materials.
`
`Fig. 9. Density of SiO2 layers deposited at different total pressures with
`different target materials.
`
`Page 4 of 6
`
`

`

`162
`
`M. Ruske et al. / Thin Solid Films 351 (1999) 158–163
`
`Index of refraction for a wavelength of 550 nm of SiO2 layers
`Fig. 10.
`deposited at different total pressures from different target materials.
`
`Fig. 12. Internal stress of SiO2 layers deposited at different total pressures
`for different target materials.
`
`because film thicknesses in such systems are much lower.
`The measured values do not only include the intrinsic stress
`due to the layer structure, but also the thermal stress due to
`different thermal expansion coefficients of the film and the
`substrate. As can be seen, there is no large difference
`between the different target materials.
`
`3.2. SiO2
`
`As shown in Fig. 9, the density of the deposited layers
`decreases in general with increasing total pressure. For the
`films deposited using the conventional poly-Si target, the
`value ranges from about 2.10–1.95 g/cm3. The reasons for
`this are similar to the ones for Si3N4. Interestingly, the
`density is higher for the alloy targets. The layers reach
`values up to 2.25 g/cm3 at a total pressure of 0.15 Pa.
`The obtained indices of refraction for a wavelength of
`550 nm as a function of total pressure are shown in Fig.
`10. Only the layers deposited with a poly-Si target show a
`slight decrease from about 1.47 to 1.46 with increasing
`pressure as expected because of the decreasing density of
`the layer. The layers using alloy targets have a slightly
`higher index of refraction of about 1.48 to 1.49. This can
`
`Fig. 11. Microhardness of SiO2 layers deposited at different total pressures
`from different target materials.
`
`be due to the Al2O3 content and the higher density. A depen-
`dency on the pressure was not detected.
`The microhardness of the deposited layers decreases with
`increasing total pressure (Fig. 11). For the lowest total pres-
`sures, layers from poly-Si targets reach values of uncoated
`slide glass, which is about 8 GPa. This value decreases to
`about 4 GPa at 0.9 Pa. This change correlates with the
`change in density and structure. The hardness values for
`layers from alloy targets are slightly higher. This can be
`due to the higher density of these layers.
`The determined stress values were very low. For all target
`materials, a compressive stress of about 20.2 GPa was
`measured. This value was independent of the total pressure
`(Fig. 12).
`
`4. Conclusions
`
`Casted Si/Al alloy targets can be used to replace the more
`expensive high purity boron-doped poly-Si
`targets. A
`reduced brittleness of the material and the possibility to
`produce larger tile sizes and different shapes are important
`advantages. The Si3N4 and SiO2 layer properties are not very
`different from the ones when pure Si targets are used. The
`density is similar for Si3N4 and slightly higher for SiO2. The
`index of refraction of refraction changes only slightly due to
`the different density and the Al content. The morphology
`shows the same dependency on the total pressure and also
`the values for the internal stress are almost identical.
`Si/Al alloy targets of the SISPAe type are already
`successfully used in industrial production lines for low
`emissivity coatings and may replace poly-Si targets in the
`near future completely.
`
`References
`
`[1] G. Bra¨uer, J. Szczyrbowski, G. Teschner, J. Non-Cryst. Solids 218
`(1997) 19.
`[2] J. Szczyrbowski, G. Teschner, Soc. Vac. Coaters, Proc. 38th Annu.
`Tech. Conf. (1995) 389.
`
`Page 5 of 6
`
`

`

`M. Ruske et al. / Thin Solid Films 351 (1999) 158–163
`
`163
`
`[3] M. Ruske, G. Bra¨uer, J. Pistner, J. Szczyrbowski, A. Zmelty, Mater.
`Sci. Forum 287/288 (1998) 247.
`[4] D. Beisenherz, G. Bra¨uer, J. Szczyrbowski, G. Teschner, Proc. Glass
`Processing Days, Tampere (1997) 309.
`[5] H. Schilling, G. Bra¨uer, M. Ruske, J. Szczyrbowski, Proc. Glass
`Processing Days, Tampere (1997) 203.
`
`[6] P. Wirz, German Patent DE, 3,566,274.3 (1984).
`[7] M. Weigert, U. Konietzka, LEYBOLD MATERIALS GmbH, Hanau,
`Germany, patent pending.
`[8] A.C. Adams, Plasma Deposited Thin Films, CRC Press, Boca Raton,
`FL, 1986, 129.
`
`Page 6 of 6
`
`