Nanoindentation Studies of Metal Organic Vapor ...

Copyright © 2013 by American Scientific Publishers All rights reserved. Printed in the United States of America

Energy and Environment Focus Vol. 2, pp. 85–89, 2013 (www.aspbs.com/efocus)

Nanoindentation Studies of Metal Organic Vapor Phase Epitaxy Grown Ge/Si Heterostructures P. Joice Sophia1 , D. Arivuoli1, ∗ , G. Attolini2 , M. Bosi2 , E. Buffagni2 , C. Ferrari2 , and F. Rossi2 1

Crystal Growth Centre, Anna University, Chennai 600025, India Istituto dei Materiali per l’Elettronica ed il Magnetismo, IMEM-CNR, Parco Area delle Sciencze, 37 A, 43124 Parma, Italy

2

ABSTRACT

KEYWORDS: Ge, Si, MOVPE, AFM, HRXRD, Nanoindentation.

Delivered by Publishing Technology to: Guest User IP: 166.111.120.71 On: Wed, 11 Feb 2015 02:32:37 Copyright: American Scientific Publishers defects like threading dislocations and misfit dislocations 1. INTRODUCTION are seen in Ge/Si thin films because of the large lattice In recent years, due to the increasing interest in nanomismatch (∼ 4.2%) between Ge and Si atoms.18 Oliver technology, deformation behavior of materials at nanoscale et al.8 studied the thickness dependent phase transforgains much attention in the fields of micro- and nanomations of germanium thin films of different thicknesses electromechanical systems (MEMS, NEMS). The future ranging from 50–600 nm grown on Si (100) substrate. challenge for information technology is to achieve The poor quality of the Ge thin film deposited probably innovative device structures using novel materials. Germadegrades the performance of thin films and also an obstanium (Ge) is recommended for transport channel device cle for future improvements. In the present investigation, fabrication purposes because of its higher carrier injection Ge/Si samples were grown by MOVPE technique at difvelocity and mobility than silicon (Si).1 Hence, a detailed ferent growth temperatures. The surface morphology and investigation and improvement of Ge based epitaxy will be roughness were studied using AFM. The HRXRD studof much significance for future device fabrication. Ge/Si ies have been carried out to study the structural propersystem is interesting and technologically important due to ties. The nanoindentation technique was used to investigate its applications in the field of photovoltaics, thermophothe defects induced phenomenon on the Ge over Si hettovoltaics, optoelectronics, computing, photodetectors for erostructures using Berkovich and Vickers indenters. telecommunications, high-speed transistors, wireless and broadband communication.1–3 Ge is easily adaptable with 2. MATERIALS AND METHODS most of the prevailing Si fabrication techniques compared 4 to the other III–V and II–VI semiconducting materials. A home-made horizontal MOVPE reactor without subThe response of Ge/Si to the depth sensing indentastrate rotation was used to deposit Ge films over n-type tion analysis has gained much attention in the past decade doped silicon (100) substrates. All the growths were carbut there are very few reports available on nanoindentaried out at a pressure of 60 mbar with hydrogen as cartion studies of Ge/Si.8 10–13 The strain relaxation induced rier gas and isobutylgermane (iBuGe) was used as Ge source. iBuGe was directly injected into the growth cham∗ Author to whom correspondence should be addressed. ber with a partial pressure of 1941 × 10−5 bar. Ge was Email: [email protected] heteroepitaxially grown on Si at different temperatures of Received: 5 November 2012 Accepted: 5 December 2012 550 C, 600 C and 650 C and represented by symbols Energy Environ. Focus 2013, Vol. 2, No. 1

2326-3040/2013/2/085/005

doi:10.1166/eef.2013.1031

85

ARTICLE

The Ge/Si epilayers were grown by metal organic vapor phase epitaxy (MOVPE) technique at different growth temperatures. The hardness (H) and the reduced modulus (Er were found to be 6.4–6.8 GPa and 109.9–119.9 GPa respectively using Berkovich indenter whereas Vickers indentation shows 10.1–11.0 GPa and 117.0–134.7 GPa respectively for the samples grown at different temperatures. The structural quality of the Ge/Si epilayers was investigated by high resolution X-ray diffraction (HRXRD). The full width at half maximum (FWHM) of Ge/Si epilayers were measured for all the samples with respect to growth temperature. The broadening of the diffraction profile indicates the presence of inhomogeneous strain between the substrate and epilayers. From the atomic force microscopy (AFM) studies, presence of small grains on the surface was evidenced.

ARTICLE

Nanoindentation Studies of MOVPE Grown Ge/Si Heterostructures

Sophia et al.

S1, S2 and S3. Growth time was 15 minutes for all the obtained by considering the geometric constant as 0.75. samples. The dependence of the growth rate of the samThe hardness is calculated using the relation given by ples on the growth temperature shows that the growth rate P increases with growth temperature, indicating a kinetic(3) H = max Ahc limited process. HRXRD have been performed using a Philips X’Pert The reduced modulus Er is calculated by the equation Pro diffractometer with an incident beam collimated by √ four Ge (220) reflections, a beam divergence of 12 arcsec −5 E = ×S (4) √ r and a wavelength dispersion / < 10 . The surface 2 Ahc morphology of the thin films was analyzed by AFM with a Digital Nanoscope IIIa in contact mode. The mechani3. RESULTS AND DISCUSSION cal properties such as hardness and elastic modulus were studied by scanning probe microscopy (SPM) coupled 3.1. HRXRD Studies Hysitron TI950 Triboindenter (Hysitron Inc.). Depth controlled nanoindentation tests was performed to The structural qualities of the Ge/Si samples were investiunderstand the mechanical properties of the samples. Two gated using HRXRD. Figure 1 shows the XRD spectrum different tip geometries such as Berkovich (R ∼ 50 nm) of the Ge/Si epitaxial layer grown at a temperature of and Vickers (R ∼ 100 nm) were used and the load was 600 C carried out in (400) symmetric mode. The epitaxial continuously applied and released in a single cycle without Ge (400) peaks indicate the significant shift 1% towards any holding period. Berkovich and Vickers indenters have higher 2 angles, indicating the existence of tensile strain the centerline-to-face angles as 65.35 and 68 respecin these epilayers due to the thermal mismatch between Si tively. Hardness data obtained with Berkovich indenter can and Ge.2 be transformed to Vickers hardness because the shape of The FWHM values of the samples S1, S2 and S3 a three-sided pyramid, gives the same projected area-toare presented in Table I. The Ge epilayers quality was depth ratio as that of Vickers indenter.4 The substrate influalso investigated by omega scans performed on the Ge ence is an important criterion Delivered that makesbyone to study Technology Publishing to: Guest User the detector angular position peak position by keeping the mechanical properties of film alone because ofOn: theWed,fixed. 16 IP: 166.111.120.71 11 Feb 02:32:37 In2015 this manner, the amount of micro-domain misCopyright: American Publishers substrate effect in indentation studies. In order to obtain Scientific orientation (mosaic spread) of the samples are measured. “film-only” properties and to avoid the substrate effect, Mosaic spread is indeed correlated with the dislocation a commonly used rule of thumb is to limit the indentation depth to 10–25% of the film thickness.5 Hence, in our case the maximum penetration depth is limited to 50 nm. As the indenter penetrates to a maximum load, the deformation process of the Ge/Si films can be derived from the load– displacement curve. The Oliver and Pharr method5 was followed to determine the hardness and elastic modulus from the load–displacement curves. In order to evaluate the hardness and reduced modulus from the load–displacement (P –h) curve the following relations were used. The curve fitting is made by the power law relation: P = Ah − hf m

(1)

where P is the peak load, h is the displacement relative to the initial undeformed surface, hf is the permanent depth of penetration after the indenter is fully unloaded and A is the projected contact area. In order to obtain the contact stiffness (S), the derivative of the power law relation with respect to h is evaluated at the maximum load. The contact depth (hc ) is determined as P hc = hmax − max S

Table I.

(2)

where Pmax is the maximum load used and hmax is the maximum penetration depth. To account for edge effects, the deflection of the surface at the contact perimeter is 86

Fig. 1. HRXRD spectrum of epitaxial Ge layer on Si (100) grown at 600 C.

Sample S1 S2 S3

FWHM of the Omega scan profiles of Ge/Si samples. Growth temperature ( C)

FWHM ( )

550 600 650

0.352 0.437 0.306

Energy Environ. Focus, 2, 85–89, 2013

Sophia et al.

density of the epilayer.17 The increase in intrinsic stress causes an increase in XRD peak broadening attributed to the inhomogeneous strains. In order to optimize film properties, it is necessary to understand the relationship between the deposition parameters, residual stresses and hardness.


(a)

3.2. AFM Studies The average root mean square roughness have been found to be 22, 47 and 24 nm respectively for the samples S1, S2 and S3 using AFM (Fig. 2). This shows that the surface is generally rough with 3D islands, irrespective of variation in growth conditions such as temperature, pressure and partial pressure of isobutyl germane. The sample S2 shows increase in surface roughness compared to the samples S1 and S3 which can be attributed to the influence of growth conditions. The AFM results also show the presence of small grains on the surface that probably might have resulted from the nucleation of islands due to large lattice mismatch between Si and Ge.

(b)

The results of nanoindentation measurements are displayed in Figures 3 and 4. The mechanical properties such as Delivered by Publishing hardness and elastic modulus have been estimated using Technology to: Guest User IP: 166.111.120.71 On: Wed, 11 Feb 2015 02:32:37 Berkovich and Vickers indenters with continuous stiff- Scientific Publishers Copyright: American ness measurement procedure. Figures 3(a) and (b) shows the typical load versus displacement (P –h) curves from nanoindentation tests performed on Ge/Si for Berkovich and Vickers indenters respectively at a maximum pene(c) tration depth of 50 nm. The change in the slope of the unloading curve gradually is called as the elbow phenomenon and is seen in both the Berkovich and Vickers P –h curves of the samples S2 and S3 in Figures 3(a) and (b). The similar behaviour as observed on Si wafers explained by Domnich et al.9 is also found in the present indentation analysis and attributed to the amorphization of the material during pressure release. There is a strong correlation between the shape of the unloading curve and the structural changes occurring in Ge during nanoindentation. The elbow in the unloading part of the P –h curve may be contributed by the volume expansion during indentation process or due to the formation of an amorphous phase of Ge in which a reverse phase transformation occurs in Ge from metallic phase.10 The elbow results Fig. 2. AFM images of the Ge/Si samples (a) S1 (b) S2 and (c) S3. from the material’s slow transformation from the metallic to the amorphous semiconducting phase and it can also The inelastic response in Ge is widely based on the cause minimal indenter uplift indicating an ideal elastic– applied load and the thickness of the film13 where the plastic deformation without pile-up of material. In case of phase transformation occurs in the films which are thin Ge, some deformation hardening may be needed in the and shear plasticity in the films which are thick. In all the metallization in which during unloading a reverse phase P –h curves obtained by both the indenters, no pop-in was transformation from metallic Ge occurs. The plastic deforobserved during loading and no pop-out during unloadmation in the indentation process is dependent on the loading. This phenomenon has been attributed to the absence ing conditions.9 10 Energy Environ. Focus, 2, 85–89, 2013

87

ARTICLE

3.3. Nanoindentation Analysis


Sophia et al.

(a)

ARTICLE

(b)

Fig. 4. P –h curve of sample S3 for both Berkovich and Vickers indenters at a maximum penetration depth of 50 nm.

tip. It is clear that the deformation will be a shear when the film’s shear stress is high and conversely, the deformation will be pressure-induced phase transformation when the shear stress of the film is low.13 The tip geometry and film thickness plays an important role for this kind of behavior. Delivered by Publishing Technology to: Guest User modulus values are tabulated The hardness and reduced IP: 166.111.120.71 On: Wed,in11Table Feb II 2015 and 02:32:37 compared with the previous experimenCopyright: American Scientific Publishers tal values. The variation in the hardness values obtained from Berkovich and Vickers indenters are different due to the dependence of hardness values on the geometry of the indenter and also depend on the friction between the indenter and the specimen. Figure 4 shows the typical load versus displacement (P –h) curves for Berkovich and Vickers indenters of sample S3. The graphical representation of the growth temperaFig. 3. P –h curves of the samples S1, S2 and S3 at a maximum penetration depth of 50 nm using (a) Berkovich indenter and (b) Vickers ture versus hardness and FWHM is given in Figures 5(a) indenter. and (b) shows the correlation between the two parameters (hardness and FWHM) with respect to the growth of onset of plasticity which needs a sufficient applied temperature. Ge thin films deform plastically by deformaload to nucleate the dislocations. The sudden change in tion mechanism such as twinning, dislocation slip, etc., the load–indentation depth curve like pop-in, pop-out and when indented with Berkovich or Vickers indenter. The elbow events may occur during indentation loading when increased FWHM and hardness values of S2 grown at the indenter tip pulls out the top layer of the threading dislocations and it causes a sudden increase in plasticity. The homogenous nucleation of threading dislocations beneath the indenter produced due to the high local stress which leads to the displacement discontinuities.14 The response to sharp contact deformation in Ge is to deform by conventional dislocation slip and twinning with the unloading curves generally featureless at normal loads.8 12 Hence, it can be stated that the mechanical behavior of Ge may be different from Si in device manufacturing process for machining and handling. Oliver et al.13 from their simulation studies reported that the preferred deformation mode relies on the spatial distributions of the shear and hydrostatic stresses under the 88

Table II. Comparison of hardness and reduced modulus values with reported results. Reference

Material

H (GPa)

Chen et al.15 Bulk Ge 8–11 (Hv As-grown Ge 8–12 Bradby et al.12 He et al.4 SiGe/Si 13.9–15.2 Present study S1 6.6 as-grown Ge/Si S2 6.8 S3 6.4 S1 10.1 S2 11.0 S3 10. 9

Er (GPa) Indenter type – – 190–207 109.9 119.9 114.3 117.0 134.7 130.3

– Berkovich Berkovich Berkovich

Vickers

Energy Environ. Focus, 2, 85–89, 2013

Sophia et al.


different temperatures have been carried out. The atomic force microscopy studies show the change in surface roughness with the growth temperatures. High resolution X-ray diffraction analysis revealed a good crystalline quality of Ge/Si thin films with a weak dependence of the Ge layer mosaic spread with growth temperature. The mechanical deformation behaviour of the germanium grown on (100) silicon has been investigated by nanoindentation technique using Berkovich indenter and Vickers indenter. Hardness and reduced modulus values estimated from the load–displacement curve as a continuous function of contact depth were compared. The hardness and the reduced modulus have been found to be 6.4 to 6.8 GPa and 109.9 to 119.9 GPa respectively using Berkovich whereas 10.1 to 11.0 GPa and 117.0 to 134.7 GPa respectively using Vickers nanoindentation. The variation in the hardness value may be due to the material property variation relative to the depth, surface chemical effects and surface roughness of the indented Ge/Si films.

(a)

(b)

Delivered by Publishing Technology to: Guest User Notes IP: 166.111.120.71 On: Wed,References 11 Feb 2015and 02:32:37 Copyright: American Scientific Publishers 1. M. Bosi and G. Attolini, Prog. 2. 3. 4. Fig. 5. Plot of growth temperature versus (a) hardness obtained from nanoindentation studies and (b) FWHM obtained from HRXRD.

600 C is due to the relaxed behavior of the epilayers at that particular temperature and also attributed to the degraded quality of the interface. This is also evidenced from the increased surface roughness of the sample S2 obtained from AFM studies. It is evident that a correlation exists between the FWHM and hardness of the Ge epilayers. The FWHM is in turn connected with the dislocation density, the higher the FWHM the higher the dislocation density and hence the hardness is higher. Normally dislocated materials are harder than the non dislocated one and this seems quite interesting. Cross sectional transmission electron microscopy and micro-Raman studies of the indentations are desirable to better understand the deformation mechanism and hence helpful to study the transformation process.

4. CONCLUSION The growth of germanium over silicon epitaxial films using metal organic vapor phase epitaxy (MOVPE) at Energy Environ. Focus, 2, 85–89, 2013

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

18.

Cryst. Growth Charact. Mater. 56, 146 (2010). J. M. Hartmann, A. M. Papon, V. Destefanis, and T. Billon, J. Cryst. Growth 310, 5287 (2008). D. J. Paul, Adv. Mater. 11, 191 (1999). B. C. He, C. H Cheng, H. C. Wen, Y. S. Lai, P. F. Yang, M. H. Lin, W. F. Wue, and C. P. Chou, Microelectronics Reliability 50, 63 (2010). W. C. Oliver and G. M. Pharr, J. Mater. Res. 7, 1564 (1992). D. Arivuoli, N. S. Lawson, A. Krier, G. Attolini, and C. Pelosi, Mater. Chem. Phys. 66, 207 (2000). A. C. Fischer-Cripps, Nanoindentation, 3rd edn., Springer, New York, Dordrecht, Heidelberg, London, LLC (2011). D. J. Oliver, J. E. Bradby, and J. S. Williams, J. Appl. Phys. 101, 043524 (2007). V. Domnich, Y. Gogotsi, and S. Dub, Appl. Phys. Lett. 76, 2214 (2000). Y. G. Gogotsi, V. Domnich, S. N. Dub, A. Kailer, and K. G. Nickel, J. Mater. Res. 15, 871 (2000). J. C. Jamieson, Science 139, 762 (1963). J. E. Bradby, J. S. Williams, J. Wong-Leung, M. V. Swain, and P. Munroe, Appl. Phys. Lett. 80, 2651 (2002). D. J. Oliver, J. E. Bradby, J. S. Williams, M. V. Swain, and P. Munroe, Nanotechnology 19, 475709 (2008). R. Navamathavan, S. J. Park, J. H. Hahn, and C. K. Choi, Mater. Charact. 59, 359 (2008). X. Q. Chen, H. Niu, D. Li, and Y. Li, Intermetallics 19, 1275 (2011). V. M. Kaganer, O. Brandt, H. Riechert, and K. K. Sabelfeld, Physical Review B 80, 033306 (2009). C. Ferrari and C. Bocchi, Characterization of Semiconductor Heterostructures and Nanostructures, edited by CarloLamberti, Elsevier B.V (2008), Chap. 4, pp. 93–132. P. I. Gaiduk, A. N. Larsen, and J. L. Hansen, Thin Solid Films 367, 120 (2000).

89

ARTICLE

Acknowledgments: One of the authors Joice Sophia Ponraj gratefully acknowledges the “International Center for Theoretical Physics, Trieste, Italy” for the fellowship through the Training and Research in Italian Laboratories (TRIL) Programme.