Solid solution strengthening and phase transformation in high-temperature annealed Si80Ge20 alloy

(1)

Solid solution strengthening and phase transformation

in high-temperature annealed Si

80

Ge

20

alloy

Tun-Yuan Chiang

a,1

, Hua-Chiang Wen

b,1

, Wu-Ching Chou

b

, Chien-Huang Tsai

c,n a

Department of Mechanical Engineering, Chin-Yi University of Technology, Taichung 400, Taiwan, ROC

b_{Department of Electrophysics, National Chiao Tung University, Hsinchu 300, Taiwan, ROC} c

Department of Automation Engineering, Nan Kai University of Techonology, No. 568, Zhongzheng Road, Caotun Township, Nantou 54243, Taiwan, ROC

a r t i c l e i n f o

Article history: Received 3 August 2013 Received in revised form 20 October 2013

Accepted 10 December 2013 Communicated by A.G. Ostrogorsky Available online 22 December 2013 Keywords:

A1. Characterization A1. Crystal structure

B2. Semiconducting silicon compounds

a b s t r a c t

This investigation demonstrates the temperature-dependent mechanical properties of Si80Ge20alloy

ﬁlms via a nanoindenter in the indentation depth of 100 nm. The roughly equal root mean square roughness (Rrms) values and repeatable load–displacement (P–

δ

) curves for the samples ensure the

mechanical performances mainly contributed from the inﬂuences of annealing temperatures. The hardness (H) values of samples increase with the temperatures of an initial annealing in the range from RT to 9001C, and, conversely, decrease for annealing temperatures over 900 1C. Accordingly, both E/H and hf/hmaxvalues, exhibiting an inverse tendency in the above temperature range, hints that the solid

solution strengthening effect and the softening phenomenon occur for the initial-annealing and over-annealing stages, respectively. In addition, grazing incidence X-ray diffraction (GIXRD) analysis demon-strates the lattice expansion and the broadened peak that attribute to the solid solution strengthening of samples and the segregation of Ge, respectively. Through observing the value of the (200) lattice spacing of 5.624 Å for a 9001C-annealed sample by transmission electron microscopy (TEM) analysis, it is veriﬁed that the segregation of Ge is responsible for the decreased hardness for the 10001C-annealed sample.

1. Introduction

Given several attractive features including tunable band gaps, high carrier densities, a complete solid solution character, and high thermal conductivity, silicon–germanium (Si–Ge) alloys are con-sidered as potential materials for the applications for solar cells[1], photodetectors[2], light-emitting diodes [3], integrated optical[4]and photovoltaic[5]devices, heterojunctions[6], one-dimensional devices [7], standard resistors[8], and single-grain thin-film transistors[9]. Owing to the high quality of thinfilms, an ultrahigh-vacuum chemical deposition (UHV-CVD) is one of the popular procedures for the fabrication of Si–Ge films. Because of a thermal dissociation of precursors (e.g., silane and germane) and rapid deposition, UHV-CVD yielded metastable defects, and many researchers made efforts to overcome this problem. Kringhøj et al.

[10] relaxed the lattice strain of the Si0.915Ge0.085 alloy film by annealing samples at 11001C. Parnis et al. [11]averaged magni-tudes of fluctuations of X-ray diffraction spectra using a novel simulation procedure, and quantitatively characterized the degree of a structural disorder for the Si0.78Ge0.22film. For the Si0.5Ge0.5

film, Whiteaker et al.[12] showed an increase of the averaged order parameter with increasing thickness, but then a decrease with the thickest film of 1000 Å. These features elucidate the importance of the phase- and the defect- controls for ultra-thin Si–Ge alloy films. Except for the physical properties mentioned above, the mechanical performances for Si–Ge films are also influenced from the micro- or nano-scaled imperfections. Under the annealing treatments from RT to 10001C, this study seeks to investigate the effects of detailed structural transformations on mechanical behaviors of a Si80Ge20film. Predictably, the mechan-ism of mechanical variations could be inferred via a grazing incidence X-ray diffraction (GIXRD) and transmission electron microscopy (TEM).

2. Experimental details

An ultra-high vacuum chemical vapor deposition (UHV-CVD) system was applied for depositing Si–Ge alloy ﬁlms on the P-type Si(100) substrates. The substrates were pretreated using a stan-dard Radio Corporation of American (RCA) procedure[13]and a 15-s immersion in the solution mixed with HF and H2O (1:50 in volume ratio). Initially, a 120-nm Si–Ge layer was deposited at 5001C for 43 min from the mixed atmosphere of SiH4(85 sccm) and GeH4 (15 sccm) maintained at 107mbar. Consequently, a Contents lists available atScienceDirect

journal homepage:www.elsevier.com/locate/jcrysgro

Journal of Crystal Growth

http://dx.doi.org/10.1016/j.jcrysgro.2013.12.013

n_{Corresponding author. Tel./fax:}_{þ886 49 2563489x3606.}

E-mail address:[email protected](C.-H. Tsai).

1_The_{ﬁrst and second author contributed equally to this work.}

(2)

10-nm Si buffer layer was deposited on the Si substrate at 5001C for 100 min from pure SiH4(5 sccm) at 107mbar. The SiGe and Si buffer layers were deposited following four cycles till the total thickness offilm reached 530 nm. After film fabrication, a compo-sition of films was confirmed as Si80Ge20 via an electron probe microanalyzer (EPMA, JEOL JAX-8800). The temperatures of annealing treatments for Si80Ge20 films were RT, 800, 900, and 10001C for 30 min in argon atmosphere. A nanoindentation measurement system (Hysitron) equipped with atomic force microscopy (AFM, Digital Instruments Nanoscope III) was applied for recording the load–displacement (P–

δ

) curves of the alloy ﬁlms. The test strain rate was at 0.05 s1_{, and each P}_–

_δ

_{curve was} repeated at least three times to ensure reproductivity. The surface morphologies of as-synthesized and indented samples were observed using the conjunct AFM. The grazing incidence X-ray diffraction (GIXRD) instrument used was a Shimadzu XRD-6000 with Cu K_αradiation operated at 30 kV and 20 mA. The scanning speed was 11/min. After the sample preparation by a focus ion beam (FIB), the cross-sectional transmission electron microscopy (TEM) images of a 9001C-annealed ﬁlm were recorded using a JEOL JEM-2100F operating at 200 keV.

3. Results and discussion

In order to exclude a measurement error resulting from the excessive surface deviation, AFM was applied to inspect the topological morphology for the samples before the indentation measurement. Fig. 1(a) displays the AFM image for the RT-annealed sample with the root mean square roughness (Rrms) value of 37.40 nm. A 3D-curved surface with no abnormal cracks represents theﬁlms well grown via the UHV-CVD system. An AFM analysis obtained from the samples is as follows. The Rrmsvalues of 8001C-, 900 1C-, and 1000 1C-annealed ﬁlms are 35.30, 36.45, and 33.83 nm, respectively. Notice that these Rrms values are much

smaller than the maximum indentation depth (100 nm). There-fore, one can rule out the extrinsic deviations of mechanical properties which were varied with the surface condition of the sample. Next, we take the RT-annealed ﬁlm, for example, to represent the closely experimental results for the samples in this study. Fig. 1(b) demonstrates the 2D AFM image for the RT-annealed sample after the indentation test. As observed in the central indentation zone, a regular triangle with no crack, shows that the pressure is uniformly applied on the sample. Fig. 1

(c) reveals the load_{–displacement (P–}

δ

) curves obtained from the nanoindenter. As the repeated curves are well stacked with each other, it implies a good reproductivity of our measurements. As the arrows indicated, hmax, hc, and hf are the maximum indentation depth, the contact depth, and thefinal depth, respec-tively. Herein, the parameters related to the mechanical perfor-mance in this article are referred as above. To probe mechanical characteristics in advance,Table 1lists parameters as revealed by elastic modulus (E), hardness (H), E/H, thefinal displacement after complete unloading (hf), the maximum indentation depth (hmax), and hf/hmax. One can see that the H value increases with the annealing temperature in the initial stage (from RT to 9001C), and further decreases with that in thefinal stage (4900 1C). Besides the primitive H value, complex parameters including E/H and hf/hmax values yield useful information to elucidate an interior mechanical character of materials. In comparison with H values, both E/H and hf/hmaxvalues appear to have an inverse tendency as a function of annealing temperature (See Fig. 2). The most important difference between E/H and bare H values is that E input to E/H truly specifies the substantial elastic recovery for a material. Owing to the insignificant influence of E, the ratio (E/H) does not vary much for different pure brittle solids [14]. In this study, the E/H values strongly depend on the annealing tempera-tures. This helps in providing the phase transformation insight into the hard metal matrix after a thermal treatment. Zhou et al.[15]

fabricated alloys with the composition of AlCoCrFeNiTix(x¼0, 0.5, 1.0, 1.5), and gradually increased E values with adding a Ti element. Based on the BCC solid solution lattice, the lattice distortion

Fig. 1. (a) 3-d AFM image of RT-annealed Si80Ge20alloyﬁlms before the

indenta-tion, (b) 2-d AFM image of RT-annealedﬁlm after the indentation, and (c) the P–δ curves repeated for three times for RT-annealed sample in the depth of 100 nm, where hc, hf, and hmaxare the contact depth, theﬁnal depth, and the maximum

indentation depth, respectively.

Table 1

Typical mechanical parameters acquired from the P–δ plots of the samples annealed at RT, 8001C, 900 1C, and 1000 1C, respectively.

Sample condition E (GPa) H (GPa) E/H hf(nm) hmax(nm) hf/hmax

RT annealing 182.570.8 12.370.2 14.83 13.8 50.8 0.271 8001C annealing 188.370.4 12.970.1 14.59 13.3 50.8 0.261 9001C annealing 190.170.7 13.170.1 14.51 13.0 50.6 0.256 10001C annealing 186.170.5 12.570.2 14.88 13.5 50.9 0.265

Fig. 2. The temperature-dependent plots of H, E/H, and hf/hmaxvalues for the

samples, where H, E, hf, and hmaxare the hardness, the elastic modulus, theﬁnal

displacement after complete unloading, and the maximum indentation depth, respectively.

(3)

energy will increase signiﬁcantly and the solid solution strength-ening will be enhanced because a Ti atom, with a larger atomic radius, occupies the lattice sites. Hence, E undoubtedly increases with x in AlCoCrFeNiTix. Inferred from the Si–Ge phase diagram[16], the incensement of E/H ratio with elevating the annealing temperature from RT to 9001C can be explained via the same mechanism since Si80Ge20 is basically a diamond-like solid solution. However, the mechanism of the descending hardness while temperature surpassing 9001C needs detailed veriﬁcation. The hf/hmax acquired from the unloading P_–

δ

curve can be used to clarify the sample in the nanoindentation experiment; the natural limits for the hf/hmax para-meter are 0rhf/hmaxr1, where the lower limit and the upper one correspond to the full elastic deformation and a rigid-plastic behavior

[17]. On the basis of hf/hmaxo0.7, dominated by the sink-in effect[18], the decreasing of hf/hmaxvalues from RT to 9001C could be suspected for the solid solution strengthening which is considered as a weak type of hardening.Fig. 3shows the XRD patterns for the Si80Ge20alloy ﬁlms after RT-, 800 1C-, 900 1C-, and 1000 1C-annealing treatments, respectively. Three main groups of peaks are locating around 66.6– 67.0, 67.8, and 69.21, corresponding to pure Ge, Si–Ge alloy, and Si substrate, respectively. As the red-dashed line indicated, the peaks are

attributed to the Si–Ge(400) plane-shift leftward with the tempera-tures ranging from RT to 9001C, and correspondingly show a right shift for temperatures higher than 9001C. The lattice parameters of RT-, 8001C-, 900 1C-, and 1000 1C-annealed samples are 5.519, 5.521, 5.524, and 5.515 Å, respectively. It implies that the lattice parameters of the investigated Si–Ge films increase with the temperatures from RT to 9001C but then reduce with temperatures greater than 900 1C. The expansion of the lattice parameters strongly supports the solid solution strengthening mechanism to cause the hardness increasing that varies with the annealing conditions. The processing of Si_–Ge films involves thermal dissociations of SiH4 and GeH4 gases. Una-voidably, a small amount of pure elemental Ge or Si exists in the as-synthesizedfilm. The annealing treatment promotes entire mixing of alloying elements, and accompanies the solid solution strengthening effect. Notice that the lattice parameters of Si and Ge are 5.43[19]and 5.65 Å[20], respectively. The peak shifting leftward signifies increas-ing of the lattice parameter which warrants dissolvincreas-ing the elemental Ge atoms into the Si-based matrix in the annealing process. Thus rather instead of a pure Si lattice, an enlarged lattice parameter occurs. On the other hand, the over-annealing treatment (4900 1C) demotes the hardness effect. As the experimental segregation enthalpy of Si–Ge alloy is only 5.370.5 kJ mol1_[21]_{, the} precipita-tion of elemental Ge causes the decayed hardness, after Ge atoms get enough thermal energies to overcome the activation barrier. As displayed inFig. 3, the broaden Ge peak at 9001C points to the presence of amorphous Ge, while the raised Ge peak at 1000_1C displays the nanoscaled Ge. The precipitation of Ge not only weakens the solid solution strengthening but also creates the right shift of the peak.Fig. 4(a) shows the bright-field (BF) TEM micrograph including the Si substrate, the Si–Ge alloy films, and the ultra-thin silicon buffer layers for the samples after the 9001C annealing.Fig. 4(b) exhibits the high resolution (HR) lattice image obtained from the blue rectangle marked inFig. 4(a). As shown in the upper-inset figure, the lattice spacing with the value of 2.812 Å roughly satisfies the spacing of the (200) plane for the Si–Ge alloy.Fig. 4(d) demonstrates the selected area diffraction patterns (SADP) obtained from the region shown in

Fig. 4(b). As can be seen, the [111], [200], [111] spots from the [110] zone axis conﬁrms existence of the diamond cubic structure. The lattice spacing calculated from the (200), the experimental lattice parameter inFig. 4(b) is 5.624 Å, evidencing segregation of Ge (Notice

Fig. 3. The GIXRD patterns for the samples annealed at different temperatures.

Fig. 4. The cross-sectional TEM morphology of 9001C annealed film. (a) The bright filed image of the region including the Si substrate, Si–Ge alloy films, and ultra-thin silicon buffer layers. (b) The HR lattice image obtained from the blue rectangle marked in (a). (c) The HR lattice image enlarged from the red rectangle marked in (a). (d) The selected area diffraction patterns (SADP) of (b). (For interpretation of the references to color in thisfigure legend, the reader is referred to the web version of this article.)

T.-Y. Chiang et al. / Journal of Crystal Growth 390 (2014) 92–95 94

(4)

that the lattice parameter of pure Ge is 5.650 Å[20]). This result validates our conjecture that the decreasing hardness is contributed by the Ge cluster once again. In addition,Fig. 4(c) shows the HR lattice image (The upper-insetﬁgure displays in enlarged scale) for the area marked with the red rectangle in Fig. 4(a). Corresponding to the quartz (151) plane, the referred lattice spacing (2.913 Å, JCPDS: 2-471) and the experimental one (2.905 Å) are close to each other. It expresses the presence of SiO2 on the top layer of the alloy ﬁlm, even though the sample was annealed in the Ar atmosphere.

4. Conclusion

This study investigates the mechanical performances of Si80Ge20 alloyﬁlms annealed at RT, 800 1C, 900 1C, and 1000 1C, respectively. The samples were indented in the depth of 100 nm to probe the intrinsic mechanical properties in avoidance of the extrinsic distur-bances including the abnormal crack and the excess deviation of roughness. Recorded from the load–displacement curves, the hard-ness value increases with the annealing temperature ascended from RT to 9001C, and further deceases with that elevated higher than 9001C. Accordingly, both E/H and hf/hmaxvalues appearing an inverse tendency in the above temperature range hints the occurrence of the solid solution strengthening, and implies a soft phase segregating at a temperature higher than 9001C. Meanwhile, XRD analysis demon-strates that the lattice expansion and the broadened peak attribute to the solid solution strengthening of Si80Ge20and the segregation of Ge, respectively. Expected as a symptom of the abruptly descended hardness for 10001C-annealed sample, TEM analysis evidences the segregation of Ge for the 9001C-annealed one in the presence of the (200) lattice spacing of 5.624 Å.

Acknowledgments

The authors thank Prof. C.P. Chou for technical support (Depart-ment of Mechanical Engineering, National Chiao Tung University). We also thank Dr. H.C. Wen and Prof. W.C. Chou for technical suggestions, and Prof. Y.R. Jeng for the SPM equipmentused in this study.

References

[1] Y. Yukimoto, M. Aiga, a-SiGe:H alloy and its application to tandem-type solar cell, in: MRS Online Proceedings Library, 70, 1986.

[2]F.Y. Huang, X. Zhu, M.O. Tanner, K.L. Wang, Normal-incidence strained-layer superlattice Ge0.5Si0.5/Si photodiodes near 1.3mm, Appl. Phys. Lett. 67 (1995)

566–568.

[3]T. Yoshida, Y. Yamada, N. Suzuki, T. Makino, T. Orii, K. Murakami, D. Geohegan, D.H. Lowndes, M.J. Aziz, Crystallinities and light- emitting properties of nanostructured SiGe alloy prepared by pulsed laser ablation in inert back-ground gases, Proceedings of SPIE 3618 (1999) 512–519.

[4]B. Schuppert, J. Schmidtchen, A. Splett, U. Fischer, T. Zinke, R. Moosburger, K. Petermann, Integrated optics in silicon and SiGe-heterostructures, J. Light-wave Technol. 14 (1996) 2311–2323.

[5] SiGe: a key to unlocking the potential of solar cells, Photovoltaics Bulletin, 2003, pp. 7–9.

[6]S.S. Iyer, G.L. Patton, J.M.C. Stork, B.S. Meyerson, D.L. Harame, Heterojunction bipolar transistors using Si–Ge alloys, IEEE Trans. Electron Devices 36 (1989) 2043–2064.

[7]Y. Wu, R. Fan, P. Yang, Block-by-block growth of single-crystalline Si/SiGe superlattice nanowires, Nano Lett. 2 (2002) 83–86.

[8]J.A. Babcock, P. Francis, R. Bashir, A.E. Kabir, D.K. Schroder, M.S.L. Lee, T. Dhayagude, W. Yindeepol, S.J. Prasad, A. Kalnitsky, M.E. Thomas, H. Haggag, K. Egan, A. Bergemont, P. Jansen, Precision electrical trimming of very low TCR poly-SiGe resistors, Electron Device Lett., IEEE 21 (2000) 283–285.

[9]V. Subramanian, M. Toita, N.R. Ibrahim, S.J. Souri, K.C. Saraswat, Low-leakage germanium-seeded laterally-crystallized single-grain 100-nm TFTs for vertical integration applications, Electron Device Lett., IEEE 20 (1999) 341–343. [10] P. Kringhøj, R.G. Elliman, J.L. Hansen, The effect of strain and strain-gradients

on the crystallisation kinetics of S1xGex Alloy Layers, in: MRS Online

Proceedings Library, 321, 1993.

[11] D. Parnis, E. Zolotoyabko, W.D. Kaplan, M. Eizenberg, N. Mosleh, F. Meyer, C. Schwebel, Structural disorder in SiGeﬁlms grown epitaxially on Si by ion beam sputter deposition, Thin Solid Films 294 (1997) 64–68.

[12]K.L. Whiteaker, I.K. Robinson, J.E. Van Nostrand, D.G. Cahill, Compositional ordering in SiGe alloy thinﬁlms, Phys. Rev. B 57 (1998) 12410–12420. [13]H.-W. Liu, W.-K. Lai, S.-Y. Yu, S.C. Huang, H.-C. Cheng, Materials science

communication effects of rca clean-up procedures on the formation of roughened poly-si electrodes for high-density drams0_{capacitors, Mater. Chem.}

Phys. 51 (1997) 195–198.

[14]B. Basu, M. Kalin, Tribology of Ceramics and Composites: Materials Science PerspectiveThe American Ceramic Society, Wiley, 2011 522 pp.

[15]Y.J. Zhou, Y. Zhang, Y.L. Wang, G.L. Chen, Solid solution alloys of AlCoCrFeNiTix

with excellent room-temperature mechanical properties, Appl. Phys. Lett. 90 (2007) 181904-1–181904-3.

[16]R.W. Olesinski, G.J. Abbaschian, The Ge–Si (Germanium–Silicon) system, Bull.. Alloy Phase Diagr. 5 (1984) 180–183.

[17] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and reﬁnements to methodology, J. Mater. Res. 19 (2004) 3–20.

[18] P. Waters, U.o.S. Florida, Stress Analysis and Mechanical Characterization of Thin Films for Microelectronics and MEMS ApplicationsUniversity of South Florida, 2008http://gradworks.umi.com/33/47/3347380.html.

[19]W.L. Bond, W. Kaiser, Interstitial versus substitutional oxygen in silicon, J. Phys. Chem. Solids 16 (1960) 44–45.

[20] S.B. Qadri, E.F. Skelton, A.W. Webb, High pressure studies of Ge using

synchrotron radiation, J. Appl. Phys. 54 (1983) 3609–3611.

[21]J. Nyéki, C. Girardeaux, G. Erdélyi, A. Rolland, J. Bernardini, Equilibrium surface segregation enthalpy of Ge in concentrated amorphous SiGe alloys, Appl. Surf. Sci. 212–213 (2003) 244–248.