Thermomechanical properties of amorphous ... - Springer Link

2 downloads 0 Views 79KB Size Report
Abstract. Thermomechanical properties of amorphous hy- drogenated carbon-germanium alloys prepared by the rf sput- tering technique were determined for ...
Appl. Phys. A 71, 633–637 (2000) / Digital Object Identifier (DOI) 10.1007/s003390000573

Applied Physics A Materials Science & Processing

Thermomechanical properties of amorphous hydrogenated carbon–germanium alloys F.C. Marques∗ , J. Vilcarromero, R.G. Lacerda Instituto de F´ısica “Gleb Wataghin”, Universidade Estadual de Campinas, UNICAMP, CP 6165, 13083-970, Campinas, SP, Brazil Received: 9 May 2000/Accepted: 10 May 2000/Published online: 13 July 2000 –  Springer-Verlag 2000

Abstract. Thermomechanical properties of amorphous hydrogenated carbon-germanium alloys prepared by the rf sputtering technique were determined for films in the 0 at. % to 100 at. % carbon content range. The stress, thermal expansion coefficient, and elastic modulus were obtained using the thermally induced bending technique. The stress was related to the concentration of hydrogen and argon, to the difference in the Ge−Ge and Ge−C bond length, and to the carbon hybridization. The thermal expansion coefficients of pure amorphous germanium and amorphous carbon are higher than that of their corresponding crystalline counterparts, which was attributed to the compressive stress of the films. The biaxial modulus, on the other hand, are always smaller than that of their crystalline counterparts, but increases as the concentration of carbon increases due to the substitution of Ge−Ge bonds by energetically stronger Ge−C and C−C bonds. PACS: 62.20.Dc; 81.05.Gc; 65.70.+y; 78.66.Jg Recently, there has been an increasing interest in the influence of the deposition parameters on the mechanical properties of materials used in the microelectronic industries. Parameters such as stress, elastic modulus and thermal expansion coefficient have been studied more frequently since they are important for the development of stable devices. For instance, tensile stress exceeding the strength of the film produces cracks, and large compressive stress delaminates the films off the substrate. Therefore, the determination of thermomechanical properties of thin amorphous films is very important for the establishment of the preparation conditions in which the best films are obtained, as well as to understand some phenomenon taking place in amorphous semiconductors. Amorphous semiconductors are generally deposited with some residual internal stress, which is one of the great limitations for the production of long-life devices. Internal stress can appear due to the difference in the thermal expansion coefficient between the film and the substrate (thermal stress). Unfortunately, the thermal expansion coefficient of most ∗ Corresponding

author. (Fax: +55-19/7885-376, E-mail: [email protected])

amorphous thin films is not known and usually it does not match to the thermal expansion of the substrate, thus a thermal stress is usually present. However, the most important stress is the one associated with the physical nature of the films and with the chemical and structural changes occurring during the film growth, the so-called intrinsic stress. The origin of the intrinsic stress in amorphous films is fundamentally related to the variation of the density of a film during or following deposition. For instance, in amorphous semiconductors and thin metal films, tensile stress is observed in films rich in voids [1, 2], and hence provides a contraction of the film. On the other hand, the explanation of compressive stress is controversial. Compressive stress is usually associated with impurities, such as oxygen and argon, and for a-Si:H and a-Ge:H it has been frequently attributed to the incorporated hydrogen [3, 4]. For a more general presentation of stress in thin films, see [5–9]. The carbon atoms have the ability of bonding in the sp3 , sp2 and sp hybridizations, which make it possible to obtain a-C:H films with different structures such as polymeric-, graphitic-, and diamond-like. It is well known that sp 2 bonds represent a strong limitation for the improvement of a-C and a-SiC films. Hence, studies of the carbon influence on the properties of a new network, such as in a germanium matrix, can also help to get a better comprehension of the physical properties of the carbon atoms. Some thermomechanical properties of a-C and a-SiC have already been studied [10–12], but similar studies have not yet been performed in a-Ge1−x Cx :H alloys, though some optical, electrical and structural properties have already been reported for films prepared under different conditions and deposition techniques [13–21]. The present paper reports, for the first time, a study of the carbon influence on the thermomechanical properties of a-Ge1−x Cx :H films prepared by rf sputtering. 1 Experimental The a-Ge1−x Cx :H films were prepared by reactive rf cosputtering germanium (99.9995%)–graphite(99.995%) composed targets, in an Ar (99.997%) plus H 2 (99.9995%) atmosphere. Targets composed of a 3-in-diameter crystalline germanium (or graphite) wafer covered by small pieces of graph-

634

ite (or germanium) were used for the preparation of low (or high) carbon content alloys. By this method it was possible to vary the carbon content from 0 at. % to 100 at. %. The system was pumped down to 1 × 10 −6 mbar prior to the film deposition. The substrate temperature and bias were kept constant at 230 ◦ C and 640 V, respectively. All the films were prepared under the same hydrogen 1.5 × 10 −3 mbar and argon 1.3 × 10−2 mbar partial pressure conditions. These are the deposition conditions used in the preparation of a-Ge:H films with good optoelectronic qualities in our laboratory [22, 23]. The morphology of the films was examined by X-ray diffraction. No evidence of crystalline phase has been observed in the X-ray diffraction spectra of the a-Ge 1−x Cx :H samples in the entire carbon concentration range. Rutherford backscattering (RBS) (2.0-MeV He+ ) was used for measuring the carbon, germanium and argon relative concentrations. The backscattered particles were detected at an angle of 165◦ with respect to the incident beam direction. The errors on the carbon, germanium and argon concentrations are estimated to be of the order of 8%. The total hydrogen content was measured using elastic recoil detection analysis (ERDA), with a probe beam of 2.2-MeV He+ incident on the sample at an angle of 75◦ with respect to the surface normal. The detection of recoiled protons was performed at an angle of 30 ◦ . A 10-µm Mylar foil covered the detector to prevent the detection of the scattered alpha particles. The experimental error is of the order of 11%. Stress measurements were taken from films deposited onto 3 × 25 × 0.4 mm (111)Si bars using a Dektak profilometer to determine the radius of curvature of the film–substrate composite. The stress was then calculated through the Stoney equation [24]: σ = [E/(1 − ν)](t 2/6dR) ,

(1)

where E, ν and t are the Young’s modulus, Poisson’s ratio, and thickness of the substrate, R and d are the radius of curvature and the thickness of the film, respectively. The thermal expansion coefficient, α, and the biaxial modulus, E/(1 − ν), were determined by the thermally induced bending (TIB) technique, based on the determination of the curvature of the film + substrate composite as a function of temperature. The temperature dependence of the stress is given by the relation [24] dσ/dT = [E f /(1 − νf )](αs − αf ) ,

The band gap of the films, Fig. 2, increases only in the 0 < x < 0.2 range, being basically constant for x > 0.2. The extremely low deposition rate of films with high carbon contents, Fig. 1, limited the measurements of the optical properties of films with x > 0.4. Those films are thinner than 0.4 µm, which does not allow a proper calculation of the optical constants by the Swanepoel method [26]. So, although samples in the 0 at. % to 100 at. % carbon content range were prepared, the optical properties of the a-Ge 1−x Cx :H films were determined only in the x ≤ 0.4 range. The hydrogen concentration of the a-Ge 1−x Cx :H films increases from 8 at. % to about 25 at. % in the 0 < x < 0.2 range, Fig. 3a. In the samples with carbon content x > 0.2, it

Fig. 1. Deposition rate of a-Ge1−x Cx :H films as a function of the carbon content (measured by RBS). The carbon concentration of few samples (triangles) of this and of the other figures, were estimated by using the Ge/C sputtering yield ratio obtained from the RBS measurements taken of the other samples and the Ge/C target area ratio. The dashed lines on this and in all the other figures are only used as a guide for the eyes

(2)

where α is the thermal expansion coefficient, and the “s” and “f” subscripts refer to substrate and film, respectively. Detailed description of the used apparatus and procedure is found elsewhere [25]. 2 Results The deposition rate of the a-Ge1−x Cx :H films as a function of the carbon content is greatly reduced as the carbon concentration is increased, Fig. 1, since the sputtering yield of carbon is much smaller than that of germanium. The deposition rate of the non-alloyed a-C:H film is about 2 orders of magnitude smaller than that of the a-Ge:H films, limiting the deposition of thick films with high concentration of carbon.

Fig. 2. E04 and Tauc’s optical band gap of a-Ge1−x Cx :H films as a function of carbon content. In this range E04 ≈ E Tauc + 0.1 eV

635

is observed that the hydrogen content remains basically constant, except for the unalloyed a-C:H film (x = 1), which has about 13 at. % hydrogen. Some few percent of argon is also present in the films with small concentration of carbon, and tends to zero in films with high carbon concentration, Fig. 3b. The reason for the decreasing of the argon content as the carbon concentration increases is not clear. It may be related to

the structure of the films. The films with low carbon content are compact and thus are more likely to trap argon. As the carbon content increases the films may become less compact which allows the argon to diffuse more easily from the bulk of the film. The stress of the a-Ge1−x Cx :H films depends on the carbon concentration, as can be observed in Fig. 4. Positive values represent compressive stress. As the carbon concentration increases the stress decreases and becomes slightly tensile for alloys with carbon content in the 20 at. % to 60 at. % range. For higher carbon concentration the stress becomes compressive again. Figure 5 displays the stress as a function of temperature for a film with 26 at. % carbon. The negative slope of both curves indicates that the thermal expansion of the films is higher than that of the substrates. In Fig. 6 the slope dσ/dT of the curves of Fig. 5 is plotted against the thermal expansion of

Fig. 3. a Hydrogen concentration as a function of carbon content obtained by a 2.2-MeV He ERDA. b Argon concentration as a function of carbon content in a-Ge1−x Cx :H alloys Fig. 5. Temperature dependence of the stress of a film with 26 at. % carbon concentration deposited into (111)-crystalline silicon and 7059 Corning glass

Fig. 4. Intrinsic mechanical stress of a-Ge1−x Cx :H films as a function of carbon content. Positive values represent compressive stress

Fig. 6. dσ/ dT of the curves of Fig. 5 as a function of the thermal expansion coefficients of the substrates

636

Fig. 7. Thermal expansion coefficient as a function of carbon content for the a-Ge1−x Cx :H alloys

Fig. 8. Biaxial modulus E/(1 − ν) as a function of carbon content of a-Ge1−x Cx :H alloys

the substrates. The slope of the solid curve in Fig. 6 gives the biaxial modulus, whereas its intersection with the α axis gives the thermal expansion coefficient of the film, according to (2). Adopting this procedure we determined the thermal expansion, Fig. 7, and the biaxial modulus, Fig. 8, for some films. A decreasing of the thermal expansion coefficient is observed as the carbon content increases up to 20 at. %, followed by a small increase for the pure carbon sample. The biaxial modulus increases monotonically as the carbon content increases. 3 Discussion The stress of thin films depends on a number of factors such as the concentration of hydrogen (and argon, in sputtered films), density of voids, and microstructure. It has been observed that amorphous silicon (a-Si:H) and germanium

(a-Ge:H) with tensile stress are usually defective, with high density of voids, columnar structure, and other void-like defects [27]. These films pick up contamination when exposed to the atmosphere. On the other hand, compact films are usually compressive and stable. The origin of the stress in a-Si:H, a-Ge:H and a-C:H is still in debate. The explanation is more complicated for compressive films. In a-Si:H and a-Ge:H films the compressive stress has been mainly associated with the concentration of hydrogen. Apparently the compressive stress is more likely to be associated with the unbonded (or molecular) hydrogen either interstitial or trapped in void-like defects [27]. In carbon films it is generally agreed that the compressive stress is created by the sub-implantation of carbon in the sp3 C−C hybridization [28, 29] The analyses of the stress a-Ge1−x Cx :H allows, Fig. 4, are even more complicated, since in addition to the presence of hydrogen and argon it also involves a change in stoichiometry, and differences in the carbon hybridization. The compressive stress of the films with low carbon content (x < 0.1) suggests that they have good structural properties. It is likely that, in this range, the stress is associated with the hydrogen and argon concentration. The influence of sp 3 C−C sites must not be significant in this range of concentration since most carbon atoms are bonded to germanium as observed in a study of Raman, XPS and infrared spectroscopy [30]. As the carbon concentration increases, the compressive stress decreases and eventually turns to tensile. As we have mentioned above, tensile a-Si:H and a-Ge:H films are usually defective and pick up contamination when exposed to the atmosphere. However, we did not find any indication of contamination in the films studied here, either using XPS or infrared analysis, suggesting that the tensile stress may not be due to void-like defects. In this region the stress is probably due to the substitution of some Ge−Ge (2.45 Å [31]) bonds by bonds with a shorter length, sp 3 Ge−C (1.98 Å [32]) and sp2 C−C (1.48 Å [33]), which shrink the film and thus give a tensile stress contribution. The relatively low compressive stress of the films with high carbon content might be related to the carbon hybridization. Diamond-like films (high density of sp 3 hybridization) have high compressive stress (typically in the 20–150 kbar range), due to the sub-implantation process which change the carbon hybridization inside the film as it grows [28, 29]. On the other hand amorphous carbon films with low stress are graphitic- and/or polymeric-like. The hydrogen concentration of the a-Ge1−x Cx :H films, Fig. 3, can give some information about their structure. As is well known, polymeric films are highly hydrogenated (> 30 at. %). However all films reported in this work have hydrogen content lower than 30 at. %. Thus, it seems that the amount of sp2 hybridized carbon is relatively high in films with high carbon content, which is a characteristic of graphitic-like films. It has been observed that the few thermal expansion coefficient data reported in literature for amorphous semiconductors are different from their crystalline counterpart. It has been theoretically predicted that they depend on the bond strain [34], and recently, we observed experimentally that the thermal expansion of a-Si:H and a-Ge:H depends on the internal stress [25]. In this work, it was found that the thermal expansion increases as the stress changes from tensile to compressive. The non-alloyed a-Ge:H, for instance, has high compressive stress and its thermal expansion is much higher

637 −6

−1

than that of crystalline germanium (≈ 6 × 10 K ). Thus, we expect that the thermal expansion of the alloys, Fig. 7, may be influenced by the internal stress. The thermal expansion of the films with high carbon concentration may also depend on their structure and concentration of sp 2 and sp3 sites. The thermal expansion of the pure a-C:H is close to the value already reported for a-C:H [35], but higher than that of diamond (≈ 1 × 10 −6 K−1 ). The biaxial modulus increases as the carbon content increases, which is related to the stronger C−C (82.6 kcal /mole [36]) and C−Ge (56.6 kcal/mole [37]) energy bond as compared to the Ge−Ge (38.2 kcal/mole [38]) energy bond. The increase of the biaxial modulus is mainly due to a relative increase of the Young’s modulus since the Poisson’s ratio usually does not change enough to account for such a substantial increasing in the biaxial modulus.

4 Conclusion Stress, thermal expansion coefficient and elastic modulus of amorphous germanium–carbon alloys were determined. The compressive stress was associated with hydrogen and argon content (for films with low carbon content), and to a high density of sp2 hybridization (graphitic-like) for films with high carbon content. For intermediate carbon concentration the tensile stress may be related to the incorporation of Ge−C and C−C bonds. The thermal expansion of pure a-Ge:H and a-C:H films is higher than that of crystalline germanium and diamond, and was attributed to the high compressive stress for a-Ge:H and to the graphitic nature of the carbon-rich films, respectively. The increasing of the elastic modulus as the concentration of carbon increases is associated with the incorporation of strong C−Ge and C−C bonds as compared with Ge−Ge bonds. Acknowledgements. The authors are grateful to Dr. F.L. Freire, Jr. for RBS measurements. This work has been supported by the Brazilian agencies CNPq, and FAPESP.

References 1. 2. 3. 4.

W.E. Spear, M. Heintze: Philos. Mag. B 54(5), 343 (1986) J.C. Knights: J. Non-Cryst. Solids 35–36, 159 (1980) P. Paduschek, C. Hopfl: Thin Solid Films 110, 291 (1983) X. Jiang, B. Goranchev, K. Schimidt, P. Grunberg, K. Reichelt: J. Appl. Phys. 67(11), 6772 (1990)

5. R.W. Hoffman: In Physics of Thin Films, Vol. 3, ed. by D.G. Hass, R.E. Thun (Academic Press, New York 1966) 6. D.S. Campbell: In Handbook of Thin Film Technology, ed. by L.I. Maissel, R. Glang (McGraw-Hill, New York 1970) Chapt. 12 7. R.W. Hoffman: In Physics of Thin Films, Vol. 3, ed. by D.G. Hass, R.E. Thun (Academic Press, New York 1966) p. 211 8. M.F. Doerner, W.D. Nix: CRC Critical Rev. Solid State Mater. Sci. 14, 225 (1988) 9. K.L. Chopra: In Thin film Phenomena (McGraw-Hill, New York 1969) p. 266 10. J. Robertson: Prog. Solid St. Chem. 21, 199 (1991) 11. R.G. Lacerda, F.C. Marques: Appl. Phys. Lett. 73, 617 (1998) 12. F.C. Marques, R.G. Lacerda, G.Y. Odo, C.M. Lepienski: Thin Solid Films 332, 113 (1998) 13. S. Kumar, S. Kashyap, K. Chopra: J. Non-Cryst Solids 101, 287 (1988) 14. J. Shinar, H. Wu, R. Shinar, H. Shanks: J. Appl. Phys. 62, 808 (1987) 15. S. White, D. McKenzie: J. Appl. Phys. 68, 3194 (1990) 16. T. Drusedau, A. Annen, B. Schroder, H. Freistedt: Philos. Mag. B 69, 1 (1994) 17. N. Saito, T. Yamaguchi, I. Nakaaki: J. Appl. Phys. 78, 3949 (1995) 18. F.C. Marques, J. Vilcarromero, F.L. Freire, Jr.: Proc. of the MRS Spring Meeting on Amorphous Silicon Technology, ed. by E.A. Schiff, H. Hack, S. Wagner, R. Shropp, I. Shimizu, Vol. 467, pp. 537–542 (1997) 19. J. Vilcarromero, F.C. Marques, J. Andreu: J. Non-Cryst Solids 227– 230, 427 (1998) 20. J. Vilcarromero, F.C. Marques, F.L. Freire, Jr.: J. Appl. Phys. 84, 174 (1998) 21. J. Vilcarromero, F.C. Marques: Thin Solid Films 343–344, 445 (1999) 22. F.C. Marques, I. Chambouleyron: Proceedings of the 9th European Conference on Photovoltaic Solar Energy, ed. by W. Palz, G.T. Wrixon, P. Helm (Kluwer, Dordrecht 1989) pp. 1042–1045 23. A.R. Zanatta, I. Chambouleyron: Phys. Rev. B 46, 2119 (1992) 24. R.W. Hoffman: In Physics of Non-Metalic Thin Films, Vol. B14 (Plenum Press, New York 1970) 25. M.M. Lima Jr., R.G. Lacerda, J. Vilcarromero, F.C. Marques: J. Appl. Phys. 86, 4936 (1999) 26. R. Swanepoel: J. Phys. E 16, 1214 (1983) 27. F.C. Marques, P. Wickboldt, D. Pang, J.H. Chen, W. Paul: J. Appl. Phys. 84(6), 3118 (1998) 28. C.A. Davis: Thin Solid Films 226, 30 (1993) 29. J. Robertson: Phys. Rev. Lett. 68, 220 (1992) 30. J. Vilcarromero, F.C. Marques: Appl. Phys. A 70, 581 (2000) 31. G.R. Wilkinson, M.K. Wilson: J. Chem. Phys. 44, 3867 (1966) 32. L.O. Brockway, H.O. Jenkins: J. Am. Chem. Soc. 58, 2036 (1936) 33. D.R. Lide, D.E. Mann: J. Chem. Phys. 27, 868 (1957) 34. J. Fabian, P.B. Allen: Phys. Rev. Lett. 79, 1885 (1997) 35. J.S. Wang, Y. Sugimura, A.G. Evans, W.K. Tredway: Thin Solids Films 325, 163 (1998) 36. T.L. Cottrell: Strengths of Chemical Bonds, 2nd edn. (Butterworth, London 1958) pp. 157–158, 200 37. A.E. Pope, H.A. Skinner: Trans. Faraday Soc. 60, 1404 (1964) 38. S.R. Gunn, L.G. Grenn: J. Phys. Chem. 68, 946 (1964)