Scripta mater. 42 (2000) 687 694 www.elsevier.

com/locate/scriptamat

THICKNESS DEPENDENCE OF ELASTIC MODULUS AND HARDNESS OF ON-WAFER LOW-k ULTRATHIN POLYTETRAFLUOROETHYLENE FILMS
J. Wang and F.G. Shi
Department of Chemical and Biochemical Engineering and Materials Science, University of California, Irvine, CA 92697

T.G. Nieh
Lawrence Livermore National Laboratory, P.O. Box 808, L-350, Livermore, CA 94551

B. Zhao and M.R. Brongo

Conexant Systems Inc., Newport Beach, CA 92660

S. Qu and T. Rosenmayer
W. L. Gore & Associates, Inc., Eau Claire, WI 54701 (Received November 22, 1999) (Accepted December 3, 1999) Keywords: Thin films; Nanoindentation; Polytetrafluoroethylene; Dielectric materials; Youngs modulus; Hardness

Introduction The relatively high dielectric constant of conventional SiO2 (with dielectric constant k 4 for a typical chemical vapor deposition SiO2 film) in integrated circuits (IC) not only limits signal speed, but also creates crosstalk noise and consumes excessive power. A substitution of SiO2 with low-k dielectrics can lead to a reduction in capacity and, in turn, will make increased speed and reduced crosstalk possible. Therefore, replacing Al/SiO2 with copper/low-k interconnect technology is, at the moment, one of the most promising approaches for enabling further chip miniaturization with concurrent speed enhancement. The need for new low dielectric constant materials has already simulated great efforts [1]. The existing data indicate that single C-C and C-F bonds are among those having the lowest electronic polarizability, making fluorinated and nonfluorinated aliphatic hydrocarbons potential candidates for low-k applications [2]. The incorporation of fluorine atoms is particularly effective in lowering polarizability due to their high electronegativity, which leads to tight binding of electrons. In additon to fluorinated inorganic materials, other new materials such as porous silica and porous organic materials, and various organic polymers are under investigation [3]. It is noted that, although these new materials would enable a dramatic reduction in dielectric constant, their integration into IC applications remains to be realized: their thermal stability, chemical and mechanical properties must be shown to be compatible with the existing and/or future IC manufacturing processes. Several recent studies have demonstrated that the thermal and chemical properties of low-k amorphous polymer films are thickness dependent [3 8]. However, only limited studies have been
1359-6462/00/$see front matter. Published by Elsevier Science Ltd. All rights reserved. PII: S1359-6462(99)00421-2

688

ELASTIC MODULUS AND HARDNESS

Vol. 42, No. 7

TABLE 1 Thickness of Polytetrafluoroethylene Thin Films Used in the Present Study

Specimen Thickness (nm) A 48.1 1.5 B 94.2 1.5 C 142.7 3.6 D 208.7 3.6 E 292.3 4.3 F 528.1 8.3 G 800.4 10.5 H 1141.0 16.0

carried out on the thickness dependence of the elastic modulus, hardness, and other mechanical properties of low-k films. It has been pointed out that such studies are important to the understanding of various integration issues related to adhesion and chemical-mechanical polishing characteristics of low-k films [9]. Polytetrafluoroethylene is known to be one of the most promising low-k materials [1]. However, there are some doubts about the mechanical and thermal properties of PTFE, for example, it was recently claimed that polytetrafluoroethylene films are too soft [10]. The purpose of this paper is to report the mechanical properties of thin/ultrathin polytetrafluoroethylene films as a function of film thickness using a nanoindentation technique, in particular, dynamic contact module (DCM) in a frequency-specific depth-sensing nanoindentation was performed. The implications of the present results for microelectronics applications are also addressed.

Experiments The polycrystalline polytetrafluoroethylene thin films used in the present study were spin-coated on the silicon wafers, baked, and sintered on a hot plate with curing temperature of 390C. Films of various thicknesses measured with Nanospec/AFT, as listed in Table 1, are characterized at room temperature using Nano Indenter XP with the Dynamic Contact Module (DCM) option. The principle of the DCM is similar to the conventional frequency-specific depth-sensing (continuous stiffness measurement CSM) technique, but with the DCM the range of load-displacement experimentation can be extended down to the surface-contact level [11,14]. Thus, DCM is a powerful technique in measuring thin-film mechanical properties [1113]. A typical indentation experiment consists of eight segments: approach the surface; load to peak load; unload 90% of peak load; reload to peak load; hold the indenter at peak load for 120 s; unload 90% of peak load; hold the indenter after 90% unload for 120 s; and completely unload. Multiple loading and unloading were performed to examine the reversibility of the deformation and thereby making sure that the unloading data used for analysis purposes are mostly elastic. The first hold segment is included to avoid the influence of creep on the unloading characteristics. The second hold segment is included to incorporate the corrections due to thermal drift [11]. A surface approach rate of 1 nm/s is utilized to minimize the kinetic effects of contact and also to allow the precise determination of the point of contact. All of the indentation measurements were conducted at an excitation frequency of 45 Hz using a constant (1/P dP/dt) load ramp of 0.05 s 1. At least ten indentations were made for each specimen. Two indentation depths were selected; one at the depth that is less than 20% of the thickness, and the other was more than 70% of the thickness. The data from the ten indentations were then discretized into finite displacement windows and averaged to obtain the mean and standard deviation for all the indentations. For data analysis, the Poissons ratio and modulus of 0.07 and 1141 GPa for the Berkovich diamond were used. Also, the Poissons ratio for the 0.46 [15,16]. polytetrafluoroethylene thin film was taken as

Vol. 42, No. 7

ELASTIC MODULUS AND HARDNESS

689

Figure 1. Load-displacement curves for polytetrafluoroethylene films of various thicknesses under nanoindentation, (a) 48.1, 94.2, 142.7, 208.7 nm, and (b) 292.3, 528.1, 800.4, 1141 nm.

Results Load-Displacement Curves The load-displacement curves measured from samples with different thicknesses are shown in Fig. 1. Two parts, elastic and plastic, are included in the load-displacement curve. It is readily observed in Fig. 1 that the displacement under a fixed load increases with increasing film thickness. For example, the maximum displacement for the 48.1 nm-thick film under the maximum load of 11.2 nN is about 8 nm, which is less than the displacement of 17.8 nm for the 94.2 nm-thick film under the maximum load of 7.5 nN. Also, the maximum displacement for the 142.7 nm-thick film under the maximum load of 8.5 nN is about 28 nm, which is less than the displacement of 29.7 nm for the 208.7 nm-thick film at the same maximum load. However, it is noted that, when the film thickness is over 528.1 nm, the displacements at the same maximum load are virtually the same. This is illustrated by the fact that, under a fixed load of 32.8 nN, the indentation displacements for the films with thicknesses of 528.1, 800.4 and 1141 nm are 75.2, 75.6 and 76.2 nm, respectively. It suggests that, when a film is thinner than 528.1 nm, both the stiffness and hardness increase with decreasing thickness. However, when the film thickness is greater than 528.1 nm, the stiffness and hardness values are essentially constant. This is also manifested in Fig. 1a by the fact that the irrecoverable displacement (i.e. plastic strain) in the films produced during a loading-unloading cycle increase with increasing film thickness. In contrast, the irrecoverable displacement remains practically the same when the film thickness is over 528.1 nm (Fig. 1b). Youngs Modulus and Hardness Youngs modulus as a function of normalized displacement ( ), i.e. the ratio of displacement over film thickness, for polytetrafluoroethylene films with different thicknesses is shown in Fig. 2. Typical error bars are shown on the data for the 528.1 nm film and represent 1 . (The error bars at large displacement is shorter than the data points, thus no error bars can be observed at displacement greater than 528.1 nm.) The moduli for all films decrease rapidly with increasing indentation depth rapidly in the initial stage ( 0.05 ), followed by a modulus plateau ( 0.05 0.4 ), and a sharp rise ( 0.4 ). The initial decrease in modulus with displacement has been often observed during nanoindentation of very thin films [17]. The modulus value in the plateau region is considered to be the real modulus of the film; in the present study, the value at 0.2 is taken as the modulus. The sharp increase in modulus

690

ELASTIC MODULUS AND HARDNESS

Vol. 42, No. 7

Figure 2. Youngs modulus as a function of displacement for polytetrafluoroethylene films of 8 different thicknesses.

at the normalized displacement over 0.4 is caused by the substrate effect. It is evident in Fig. 2 that Youngs modulus is thickness dependent; a thinner film has a higher modulus. The hardness of the polytetrafluoroethylene films as a function of normalized displacement is shown in Fig. 3. Typical scatter bars representing 1 are again shown on the 528.1 nm film data. Similar to modulus-displacement curves, all hardness-displacement curves exhibit a 3-stage U shape, namely, an initial decrease, plateau, and the final increase. For each film, the functional relationship between hardness and normalized displacement is essentially similar to that between modulus and normalized displacement. Also, it is obvious that the hardness of the films is dependent upon thickness; a thinner film has a greater hardness. The thickness dependence of the modulus is shown in Fig. 4. Error bars representing 1 are also shown on the curve. It can be observed that the modulus decreases sharply with thickness when the film thickness is less than 500 nm. For films thicker than 500 nm, the modulus is almost independent of the thickness, with a saturated value of about 2.3 GPa. Shown in Fig. 5 is a plot showing the hardness of polytetrafluoroethylene films as a function of the film thickness. Again, the hardness decreases sharply with film thickness when the film is thinner than 500 nm. Once the film thickness is over 500 nm, the hardness becomes independent of the thickness, with a saturated value of about 58 GPa.

Figure 3. Hardness as a function of displacement for polytetrafluoroethylene films of 8 different thicknesses.

Vol. 42, No. 7

ELASTIC MODULUS AND HARDNESS

691

Figure 4. Youngs modulus as a function of film thickness.

Discussion Experimental results from the present study demonstrate that both the Youngs modulus and hardness of polytetrafluoroethylene thin films ( 500nm) depend strongly upon the film thickness. However, when the films are thicker than 500 nm, the modulus and hardness are independent of thickness and have values of 2.3 GPa and 58 MPa, respectively. Lucas et al. [16] also measured the modulus and hardness of polytetrafluoroethylene thin films (500 to 1500 nm thick). The reported values for films with thicknesses of 500 and 1500 nm are 1 GPa and 30 55 MPa, respectively. These values are noted to be lower than that observed in the present experiments. The relatively high values observed in the present experiments may be caused by different structure and indentation heads used. Specifically, Lucas et al. [16] were using conventional XP indentation head, instead of DCM. Nevertheless, they also reported a constant modulus and hardness for films thicker than 500 nm, which is consistent with the present observation. It is of interest to compare the mechanical properties of polytetrafluoroethylene in thin-film and bulk forms. Experimental tensile properties of bulk polytetrafluoroethylene (at a strain rate of 0.2 min 1) and free-standing polytetrafluoroethylene films at room temperature are available and now presented in Table 2. A tensile modulus of about 0.4 GPa was observed in both bulk sample and the free-standing film. Comparing with the data obtained by nanoindentation, the moduli of bulk and free-standing samples are only about 1/6. This is probably because the strength of polymers normally has a high pressure dependence. For example, it has been reported that the modulus of polytetrafluoroethylene film has a pressure-dependence parameter (dE/dP) of 3.3 at room temperature [18]. Thus, the modulus is

Figure 5. Hardness as a function of film thickness.

692

ELASTIC MODULUS AND HARDNESS

Vol. 42, No. 7

TABLE 2 Tensile Properties of Bulk Sample and Free-Standing Films at Room Temperature
Materials Bulk sample 15 mm free-standing film 15 mm free-standing film Modulus (GPa) 0.41 0.4 0.4 Yield Strength (MPa) 9.0 10.3 17 Reference [19] [20] [16]

expected to be higher in compression (e.g. indentation) than in tension. In fact, Lucas et al. [16] observed a higher modulus from indentation as compared to that measured from tension of the same films. Youngs modulus is an intrinsic material property, which depends upon the nature of interatomic or molecular bond. The molecular organization in ultrathin (thicknesses less than 100 nm) and thin (thicknesses between 100 and 1000 nm) polymer films may differ substantially from that of bulk polymers, which can result in significant differences in thermophysical and mechanical properties. For example, ultrathin poly (3-methyl-4-hydroxy styrene) film has been found to contain a substantial amount of residual solvent, which produced a strong hydrogen-bonded network [21]. Resulting from this network, the ultrathin film has a higher glass transition temperature as compared to the thick film. A reduced glass transition temperature Tg of poly(metheylmathacrylate) films on SiO2 substrate with increasing film thickness was also observed [7]. Although no information is available on the structure of the present polytetrafluoroethylene thin/ultrathin films, it is not unreasonable to suggest that the structure of the ultrathin and thin polytetrafluoroethylene films is substantially different from their bulk counterpart. A stronger atomic or molecular bonding can result in a higher Youngs modulus. A decrease in Youngs modulus with increasing film thickness has been also observed in metal films such as MnBi, MnAl [22], MnTe, MnSe [23] and Cu [24]. In the case of MnTe and MnSe films, it was argued to be caused by a decrease in interatomic spacing [23]. In the case of Cu, on the other hand, the enhanced Youngs modulus at low thickness was attributed to surface roughness [24]. There is yet another explanation for the higher modulus in ultrathin films. Strain energy density is generally higher in a thin sample than that in a thick sample. Assuming that the strain is the same for all thicknesses, then, the stress is larger in the thinner sample; this can also result in a higher modulus in the thinner sample. At an indentation depth greater than about 0.4 , both the modulus and hardness were observed to rise sharply as penetration proceeds. This is caused by the fact that the substrate begins to interfere the stress fields around the indenter head. Based on a finite element calculation [25,26], the plastic zone radius ahead of the indenter is about 1.6 2.5 times the radius of the indentation for polytetrafluoroethylene which has a modulus to yield stress ratio, E/Y, of about 39. It is particularly noted that, resulting from a smaller indenter head, the elastic/plastic boundary is larger in the DCM mode than that in the conventional Nanoindenter XP mode [16], which is about 4.5 6 times the depth of the indent. This is demonstrated in Fig. 2. There is another factor that may, in part, contribute to increase in hardness. It is material pile-up around indenter, especially for soft materials. The pile-up of material around indenter in soft films on hard substrates can sometimes cause an overestimation of modulus and hardness up to 50% and 100%, respectively. The overestimation results from an inaccurate calculation of the contact area from analytical models [2729]. The present findings of the thickness-dependence of the elastic modulus and hardness of polytetrafluoroethylene film have a direct implication for the integration of this low-k film into IC fabrication. For example, the hardness variation with thickness in polytetrafluoroethylene films is expected to affect their chemical-mechanical polishing (CMP) characteristics [30 33].

Vol. 42, No. 7

ELASTIC MODULUS AND HARDNESS

693

Conclusions Elastic modulus and hardness of on-wafer low-k dielectric polytetrafluoroethylene films in the thickness range of 48.1 to 1141 nm have been investigated using the dynamic contact module (DCM) in a frequency-specific depth-sensing indentation. It was observed that, when the film thickness is less than 500 nm, both the modulus and hardness of the films decrease sharply as the thickness of the film increases. However, when the films are thicker than 500 nm, the modulus and hardness are practically independent of the film thickness, and equal to approximately 2.3 Pa and 58 MPa, respectively. These values are about 5 times of the values obtained from the corresponding bulk sample. The significant increase in mechanical strength is probably resulted from a molecular reorganization in thin/ultrathin polymer films, often observed in polymeric materials. The present findings suggest that the chemicalmechanical polishing (CMP) characteristics of polytetrafluoroethylene films are thickness dependent.

Acknowledgment Supports of this study by California Microelectronics Innovation & Computer Research Opportunities (MICRO) and by Lawrence Livermore National Laboratory under the auspices of the U.S. Department of Energy by LLNL (contract No. W-7405-Eng-48) are gratefully acknowledged. We are also thankful to H. K. Kim and H. Zhou for their assistance, and Professor S. Lee for technical discussion.

References
1. 2. 3. 4. 5. 6. M. Morgan, E. T. Ryan, J. Zhao, C. Hu, T. Cho, and P. S. Ho, JOM. 51, 37 (1999). G. Hougham, G. Tesoro, and J. Shaw, Macromolecules. 27, 3642 (1994). T. E. Wade, Semicond. Int. 22, 125 6, 128, 130, 132 (1999). H. Y. Tong, F. G. Shi, B. Zhao, S. Q. Wang, M. Brongo, and P. K. Vasudev, Appl. Phys. A. 65, 287 (1997). H. Y. Tong, F. J. Szalkowski, F. G. Shi, B. Zhao, M. Brongo, S. Q. Wang, and P. K. Vasudev, in Low-Dielectric Constant Materials III, ed. C. Case, P. Kohl, T. Kikkawa, and W. W. Lee, pp. 75 80, Materials Research Society, Pittsburgh (1997). D. T. Hsu, M. Iskandar, H. Y. Tong, F. G. Shi, S. Lopatin, Y. Shacham-Diamand, B. Zhao, M. Brongo, and P. K. Vasudev, in Dielectric Material Integration for Microelectronics, ed. W. D. Brown, S. S. Ang, M. Loboda, B. Sammakia, R. Singh, and H. S. Rathore, pp. 103112, Electrochemical Society, Pennington (1998). D. T. Hsu, H. K. Kim, F. G. Shi, B. Zhao, M. Brongo, P. Schilling, and S. Q. Wang, in SPIE Conference on Multilevel Interconnect Technology III, pp. 60 67 (1999). D. T. Hsu, F. G. Shi, S. Lopatin, Y. Shacham-Diamand, B. Zhao, M. Brongo, and P. K. Vasudev, Mater. Sci. Semicond. Process. 2, 19 (1999). E. T. Ryan, A. J. McKerrow, J. Leu, and P. S. Ho, MRS Bull. 22, 49 (1997). R. D. Miller, Science. 286, 421 (1999). W. C. Oliver and G. M. Pharr, J. Mater. Res. 7, 1564 (1992). G. M. Pharr and W. C. Oliver, MRS Bull. 17, 28 (1992). J. Li, E. T. Thostenson, T. W. Chou, and L. Riester, Trans. ASME, J. Eng. Mater. Technol. 120, 154 (1998). B. N. Lucas, W. C. Oliver, and J. E. Swindeman, in Fundamentals of Nanondentation and Nanotribology, Materials Research Society Symposium Proceedings, vol. 522, ed. N. R. Moddy, W. W. Gerberich, N. Burnham, and S. P. Baker, pp. 314, Materials Research Society, Warrendale, PA (1998). C. T. Rosenmayer, J. W. Bartz, and J. Hammes, in Low-Dielectric Constant Materials III, ed. C. Case, P. Kohl, T. Kikkawa, and W. W. Lee, pp. 231236, Materials Research Society, Pittsburgh, PA (1997). B. N. Lucas, C. T. Rosenmayer, and W. C. Oliver, in Thin-FilmsStresses and Mechanical Properties VII, ed. R. C. Cammarata, M. Nastasi, E. P. Busso, and W. C. Oliver, pp. 97102, Materials Research Society, Warrendale, PA (1998). N. R. Moody, W. W. Gerberich, N. Burnham, and S. P. Parker, ed., Fundamentals of Nanoindentation and Nanotribology, Materials Research Society Symposium Proceedings, vol. 522, Materials Research Society, Warrendale, PA (1998). B. Harmann, in Handbook of Polymer Science and Technology, ed. N. P. Cheremisinoff, vol. 2, pp. 101125, Marcel Dekker, New York (1989).

7. 8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18.

694

ELASTIC MODULUS AND HARDNESS

Vol. 42, No. 7

19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33.

J. A. Sauer, D. R. Mears, and K. D. Pae, Eur. Polym. J. 6, 1015 (1970). S. C. Sun, Y. C. Chiang, C. T. Rosenmayer, J. Teguh, and H. Wu, in Low-Dielectric Constant Materials II, ed. A. Lagendijk, H. Treichel, K. J. Uram, and A. C. Jones, pp. 8590, Materials Research Society, Pittsburgh, PA (1997). C. W. Frank, V. Rao, M. M. Despotopoulou, R. F. W. Pease, W. D. Hinsberg, R. D. Miller, and J. F. Rabolt, Science. 273, 912 (1996). M. Angadi and V. Thanigaimani, Mater. Sci. Eng. B. B25, L7 (1994). M. A. Angadi and V. Thanigaimani, J. Mater. Sci. Lett. 13, 703 (1994). R. Whiting, M. A. Angadi, and S. Tripathi, Mater. Sci. Eng. B. B30, 35 (1995). A. Bolshakov, W. C. Oliver, and G. M. Pharr, J. Mater. Res. 11, 760 (1996). T. Y. Tsui, C. A. Ross, and G. M. Pharr, in Materials Reliability in Microelectronics VII, ed. J. J. Clement, R. R. Keller, K. S. Krisch, J. E. Sanchez, Jr., and Z. Suo, pp. 57 62, Materials Research Society, Pittsburgh, PA (1997). A. Bolshakov, W. C. Oliver, and G. M. Pharr, in Thin Films: Stresses and Mechanical Properties VI, ed. W. W. Gerberich, H. Gao, J. E. Sundgren, and S. P. Baker, pp. 141146, Materials Research Society, Pittsburgh, PA (1997). A. Bolshakov and G. M. Pharr, J. Mater. Res. 13, 1049 (1998). T. Y. Tsui, W. C. Oliver, and G. M. Pharr, in Thin Films: Stresses and Mechanical Properties VI, ed. W. W. Gerberich, H. Gao, J. E. Sundgren, and S. P. Baker, pp. 207212, Materials Research Society, Pittsburgh, PA (1997). C.-W. Liu, B.-T. Dai, and C.-F. Yeh, J. Electrochem. Soc. 142, 3098 (1995). C.-W. Liu, B.-T. Dai, W.-T. Tseng, and C.-F. Yeh, J. Electrochem. Soc. 143, 716 (1996). C. G. Kallingal, M. Tomozawa, and S. P. Murarka, J. Electrochem. Soc. 145, 1790 (1998). F. G. Shi, B. Zhao, and S. Q. Wang, in Proceedings of the IEEE 1998 International Interconnect Technology Conference, pp. 7375, IEEE, New York (1998).