Load Effects on Nanoindentation Behaviour and Microstructural Evolution of Single-Crystal Silicon

Materials Transactions, Vol. 51, No. 7 (2010) pp. 1173 to 1177 #2010 The Japan Institute of Metals

Load Eﬀects on Nanoindentation Behaviour and Microstructural Evolution of Single-Crystal Silicon

Woei-Shyan Lee1;*, Tao-Hsing Chen2, Chi-Feng Lin3 and Shuo-Ling Chang1

1Department of Mechanical Engineering, National Cheng Kung University, Tainan 701, Taiwan, R. O. China 2Center for Micro/Nano Science and Technology, National Cheng Kung University, Tainan 701, Taiwan, R. O. China 3National Center for High-Performance Computing, Hsin-Shi Tainan County 744, Taiwan, R. O. China

Nanoindentation tests are performed on single-crystal silicon wafers using a Berkovich indenter and maximum indentation loads of 30 mN, 40 mN, and 70 mN. The microstructural evolutions of the indented specimens are examined using transmission electron microscopy and selected area diffraction techniques. The results show that the unloading curve of the specimen indented to a maximum load of 30 mN has a smooth profile, whereas those of the specimens indented to 40 mN or 70 mN have a pop-out feature. The hardness and Young’s modulus of the silicon specimens reduce with an increasing indentation load, and have values of 15.8 GPa and 182 GPa, respectively, under the highest indentation load of 70 mN. In addition, a strong correlation is observed between the indentation load and the microstructural change in the indentation affected area of the silicon specimens. Specifically, a completely amorphous phase is induced within the indentation zone in the specimen indented to a maximum load of 30 mN, whereas a mixed structure comprising amorphous phase and nanocrystalline phase is found in the indentation zones in the specimens loaded to 40 mN and 70 mN. The microstructural observations imply that the load-dependent nature of the unloading curves is related to the occurrence of different phase transformation mechanisms under different indentation loads. [doi:10.2320/matertrans.M2010007]

(Received January 8, 2010; Accepted April 5, 2010; Published May 26, 2010) Keywords: nanoindentation, silicon, microstructural evolution, load

1. Introduction indenter geometry and the loading-unloading rate.10–14) Silicon crystal is known to have a cubic diamond structure Silicon has excellent semiconducting properties and is under normal atmospheric conditions.15) However, during therefore widely used as a substrate material for many nanoindentation, a pressure-induced phase transformation applications in the microelectronics and optoelectronics may occur within the indentation-affected zone.16) Since the industries. The mechanical properties of the silicon substrate mechanical response and phase transformation of a silicon affect not only the mechanical performance of the device, substrate depend strongly on the magnitude of the applied but also its electrical and/or optical performance. Typically, load, a comprehensive study regarding the effects of load on silicon substrates are coated with a thin layer of silicon nitride the indentation behaviour and microstructural change is or polyimide in order to protect the circuitry from the effects required. Accordingly, the present study utilises a nano- of the ambient environment, to relieve the stress induced indentation technique to determine the loading-unloading during the flip-chip bonding process, to render the device characteristics of silicon substrates under different indenta- more robust toward the effects of wear and tear, and so on.1–3) tion loads in the range 3070 mN. The microstructural The presence of these thin films has a profound effect on evolutions of the indented specimens are then observed using the mechanical properties of the substrate, and thus many transmission electron microscopy and selected area diffrac- experimental studies have been performed to evaluate the tion techniques. Finally, the observation results are used to properties of various thin films coated on a silicon sub- clarify the load-dependent nature of the unloading curves strate.4–8) However, the mechanical properties and nano- obtained in the nanoindentation tests. indented microstructures of silicon substrates under different loads are not yet fully clear. Therefore, further investigation 2. Experimental Procedure is required into the effects of load on the nanoindentation response and microstructural evolution of silicon in order to The nanoindentation tests were performed using device establish the loading conditions which avoid the onset of grade p-type single-crystal silicon wafers with a (100) plastic deformation and therefore improve the reliability of orientation. The wafers were 0.725 mm in thickness and the device. were acquired with chemomechanical polished finishes. The Nanoindentation provides a convenient means of analyz- nanoindentation tests were performed at room temperature ing the hardness and Young’s modulus of both bulk materials in air using an MTS Nanoindenter XP system fitted with a and thin films on a substrate.9) The loading and unloading Berkovich diamond pyramid tip with a radius of 20 nm. The regions of the load-displacement curves obtained in such specimens were indented to three different maximum loads, tests commonly contain pop-in and pop-out events, respec- namely 30 mN, 40 mN and 70 mN. The loading-unloading tively. It has been reported that the occurrence of these procedure involved the following steps: (1) impressing the features is dependent on many factors, including the phase indenter until the pre-specified value of the maximum load change induced within the indented microstructure, the was attained, (2) holding the indenter in this position for 1 s; and (3) smoothly withdrawing the indenter from the speci- *Corresponding author, E-mail: [email protected] men over a period of 10 s. Five indentation tests were 1174 W.-S. Lee, T.-H. Chen, C.-F. Lin and S.-L. Chang

80 Table 1 Hardness and Young’s modulus values of silicon specimens indented to diﬀerent maximum loads. 70 RT Load 60 30 mN 40 mN 70 mN Hardness (GPa) 16.63 16.14 15.70 50 Young’s Modulus (GPa) 192.40 187.69 182 /mN

F 40

Load, 30

0 0 100 200 300 400 500 600 Depth, D/nm

Fig. 1 Typical load-displacement curves obtained during indentation to maximum loads of 30 mN, 40 mN and 70 mN, respectively. performed under each experimental condition, and the corresponding hardness and Young’s modulus values of the silicon specimen were then calculated from the load- displacement curves using the Oliver and Pharr method.9) Following the nanoindentation tests, thin foil specimens for TEM inspection were prepared using an FEI Nova 200 focused ion beam (FIB) milling system with a Gaþ ion beam and an operating voltage of 30 kV. The cross-sectional Fig. 2 Cross-sectional TEM micrograph of specimen indented to maximum load of 30 mN. microstructures of the various specimens were then observed using a Philips Tecnai F30 Field Emission Gun Transmission Microscope operated at 300 kV. of the hardness and Young’s modulus under a lower maximum indentation load can be attributed to the inden- 3. Results and Discussion tation size effect.5) Figure 2 presents a bright field cross-sectional TEM Figure 1 presents typical loading-unloading curves ob- micrograph of the silicon specimen indented to a maximum tained when indenting the silicon specimens to maximum load of 30 mN. It can be seen that the indentation-affected loads of 30 mN, 40 mN and 70 mN, respectively. It is zone (indicated by the dotted line) has a uniform micro- observed that for each curve, the loading region has a structure and is separated from the surrounding area of the smooth, continuous profile with no pop-in events. However, silicon specimen by a clear boundary. (Note that the long notably different features are observed in the unloading stripes in the region of the specimen outside of the nano- regions of the three curves. For example, for a maximum indentation-affected zone are simply interference fringes indentation load of 30 mN, the unloading curve contains a caused by a bending of the TEM sample during the FIB slight elbow (i.e. a gradual change in slope), which indicates preparation process.) Figure 3(a) presents a high-magnifica- a transformation from the original diamond cubic structure tion view of the morphology of the indentation-affected zone to an amorphous structure.10) By contrast, for maximum in Fig. 2. The inset in the lower right corner of Fig. 3(a) indentation loads of 40 mN and 70 mN, respectively, a well- shows the TEM diffraction pattern of the indentation-affected defined pop-out feature is observed in the unloading region of zone. The presence of halo rings in this diffraction pattern each load-displacement curve. In other words, the critical indicates that the microstructure within the indented area is load for the occurrence of pop-out is around 40 mN for the characterised by an amorphous phase. Figure 3(b) presents a current silicon specimens. The pop-out features observed in high-magnification micrograph of the microstructure in the nanoindentation tests have been attributed to many different boundary region between the indented zone and the silicon factors.10,11,14) In the present single-crystal silicon specimens, substrate (corresponding to the square region indicated by the pop-out feature is thought to be the result of a phase label B in Fig. 2). The micrograph clearly shows that the transformation within the indentation-affected zone. Apply- indented zone has an amorphous structure, while the ing the Oliver and Pharr method to the experimental data substrate has a crystalline structure. presented in Fig. 1, it is found that both the hardness and the Figure 4 presents a cross-sectional TEM micrograph of the Young’s modulus decrease slightly as the maximum inden- silicon specimen indented to a maximum load of 40 mN. It tation load is increased from 30 mN to 70 mN (i.e. from can be seen that the indented microstructure changes from a 16.63 GPa to 15.70 GPa (hardness) and from 192.40 GPa to fully amorphous state to a mixed amorphous/nanocrystalline 182 GPa (Young’s modulus), see Table 1). The higher values state as the maximum indentation load is increased from Load Effects on Nanoindentation Behaviour and Microstructural Evolution of Single-Crystal Silicon 1175

Fig. 4 Cross-sectional TEM micrograph of specimen indented to maximum load of 40 mN.

from a diamond cubic structure to amorphous phase.18,19) Zarudi et al.20) indicated that the microstructure of the transformation zone of indented silicon is characterised by amorphous phase at a maximum indentation load of 30 mN. Yan et al.21) also reported that for single crystal silicon, the critical load for phase transformation and the occurrence of pop-out events is around 30 mN. It is observed in Fig. 4 that a crack extends from the bottom of the phase transformation region into the silicon substrate, indicating that the critical load for micro-fracture initiation has been exceeded. In addition, it can be seen that the bend contour in the surrounding area partly expands to the indentation affected zone. This feature again confirms that the microstructure of the indentation affected zone is characterised by a mixed Fig. 3 High-resolution TEM micrographs of regions indicated by: structure comprising amorphous phase and nanocrystalline (a) square A in Fig. 2; (b) square B in Fig. 2. phase. The mixed structure of the indentation-affected zone is further confirmed by the high-resolution TEM micrograph presented in Fig. 5(a). Figure 5(b) presents a high-magnifi- 30 mN to 40 mN. This microstructural change is thought to be cation TEM micrograph of region B in Fig. 4. As shown, the responsible for the pop-out event in the unloading region of silicon atoms within the region of the substrate outside the the corresponding load-displacement curve shown in Fig. 1. indentation-affected zone are arranged in a regular lattice. Under a Berkovich indenter, the hydrostatic stress generated Furthermore, the corresponding diffraction pattern shown in at the maximum indentation load initiates the formation of the inset of Fig. 5(b) contains only diffraction spots. There- crystalline phase, while the octahedral shear stress breaks the fore, it is confirmed that the silicon substrate has a perfect bonds and leads to the formation of amorphous phase. Thus, diamond cubic single-crystalline structure without any phase in the present indentation experiments, the different max- transformation. imum indentation loads result in different stress distributions Figure 6 presents a cross-sectional TEM micrograph of the under the Berkovich indenter, which in turn lead to different silicon specimen indented to a maximum load of 70 mN. As microstructural distributions of the amorphous and crystal- shown in Fig. 7(a), corresponding to the square region line phase within the transformation zone. The phase indicated by label A in Fig. 6, the indentation-affected zone transformation of the mixed microstructure beneath the contains a mixed structure of amorphous phase and nanoindenter is accompanied by a sudden volume release, which crystalline phase. Meanwhile, Fig. 7(b), corresponding to the leads to the pop-out phenomena observed in Fig. 1. square region indicated by label B in Fig. 6 shows that the As discussed above, the microstructure of a transformed upper region of the indent has a pure amorphous structure. material is dependent on the maximum indentation load.17) Comparing Figs. 4 and 6, it is observed that while both During the indentation of monocrystalline silicon, the microstructures contain a mixture of amorphous phase and presence of an elbow feature in the unloading region of the nanocrystalline phase in the nanoindentation-affected area, load-displacement curve is attributed to a transformation the density of the nanocrystalline phase in the specimen 1176 W.-S. Lee, T.-H. Chen, C.-F. Lin and S.-L. Chang

Fig. 5 High-resolution TEM micrographs of regions indicated by: Fig. 7 High-resolution TEM micrographs of regions indicated by: (a) square A in Fig. 4; (b) square B in Fig. 4. (a) square A in Fig. 6; (b) square B in Fig. 6.

indented to a maximum load of 70 mN (i.e. Fig. 6) is signiﬁcantly higher than that in the specimen indented to a lower maximum load of 40 mN (i.e. Fig. 4).

4. Conclusions

This study has evaluated the nano-mechanical properties of single-crystal silicon substrates indented to maximum loads of 30 mN, 40 mN and 70 mN, respectively. The effects of the load on the microstructural changes induced within the indented specimens have been characterised using transmission electron microscopy and selected area diffraction techniques. The results have shown that the loading curve has a smooth, continuous profile for all considered values of the maximum indentation load. However, for a maximum load of 40 mN or more, a well-defined pop-out feature is observed in the unloading region of the load-displacement curve. In addition, it has been shown that both the hardness and the Fig. 6 Cross-sectional TEM micrograph of specimen indented to max- Young’s modulus reduce slightly as the indentation load is imum load of 70 mN. increased. The observation results have shown that the phase Load Effects on Nanoindentation Behaviour and Microstructural Evolution of Single-Crystal Silicon 1177 transformation induced within the indentation-affected zone 5) W. S. Lee and T. Y. Liu: Nanotechnology 18 (2007) 335701. depends on the magnitude of the indentation load. Specifi- 6) H. Pelletier, J. Krier and P. Mille: Mech. Mater. 38 (2006) 1182–1198. 7) W. S. Lee and F. J. Fong: Mater. Sci. Eng. A 475 (2008) 319–326. cally, a small indentation load of 30 mN prompts the 8) S. Ruffell, J. E. Bradby and J. S. Williams: Appl. Phys. Lett. 90 (2007) formation of a completely amorphous structure, whereas a 131901. higher load (i.e. > 40 mN) gives rise to a mixed amorphous/ 9) W. C. Oliver and G. M. Pharr: J. Mater. Res. 7 (1992) 1564–1580. nanocrystalline structure. Finally, it has been shown that the 10) V. Domnich, Y. Gogotsi and S. Dub: Appl. Phys. Lett. 76 (2000) 2214– relative proportion of the nanocrystalline phase increases 2216. 11) J. Jang, M. J. Lance, S. Wen, T. Y. Tsui and G. M. Pharr: Acta Mater. with an increasing indentation load. 53 (2005) 1759–1770. 12) X. J. Zheng, Y. C. Zhou and J. Y. Li: Acta Mater. 51 (2003) 3985–3997. Acknowledgements 13) R. Rao, J. E. Bradby, S. Ruffell and J. S. Williams: Microelectron. J. 38 (2007) 722–726. The authors gratefully acknowledge the financial support 14) W. S. Lee and F. J. Fong: Mater. Trans. 48 (2007) 2650–2658. 15) P. S. Pizani, R. G. Jasinevicius and A. R. Zanatta: Appl. Phys. Lett. 89 provided to this study by the National Science Council (NSC) (2006) 031917. of Taiwan under contract no. NSC 97-2221-E-006-047. 16) J. E. Bradby, J. S. Williams and M. V. Swain: Phys. Rev. B 67 (2003) 085205. REFERENCES 17) I. Zarudi and L. C. Zhang: Tribol. Int. 32 (1999) 701–712. 18) V. Domnich, Y. Gogotsi and S. Dub: Appl. Phys. Lett. 76 (2000) 2214– 1) J. L. Wang: Microelectron. Reliab. 42 (2002) 293–299. 2216. 2) L. Liu, S. Yi, L. S. Ong and K. S. Chia: Thin Solid Films 462 (2004) 19) H. Saka, A. Shimatani, M. Suganuma and M. Suprijadi: Philos. Mag. A 436–445. 82 (2002) 1971–1981. 3) T. H. Fang, W. J. Chang and C. M. Lin: Microelectron. Eng. 77 (2005) 20) I. Zarudi, J. Zou and L. C. Zhang: Appl. Phys. Lett. 82 (2003) 874–876. 389–398. 21) J. Yang, H. Takahashi, X. Gai, H. Harada, J. Tamaki and T. 4) R. Saha and W. D. Nix: Acta Mater. 50 (2002) 23–38. Kuriyagawa: Mater. Sci. Eng. A 423 (2006) 19–23.