Low-load deformation of InP under contact loading ...

PHILOSOPHICAL MAGAZINE A, 2002, VOL. 82, N O. 10, 1953±1961

Low-load deformation of InP under contact loading; comparison with GaAs G. Patriarche Laboratoire de Photonique et de Nanostructures, UniteÂ Propre de Recherche associeÂe au CNRS, 196 Avenue Henri Ravera, BP 29, 92222 Bagneux Cedex, France

and E. Le Bourhisy UniversiteÂ de Poitiers, Laboratoire de MeÂtallurgie Physique, UniteÂ Mixte de Recherche associeÂe au CNRS 6630, SP2MI, TeÂ leÂport 2, boulevard Marie et Pierre Curie, BP 30179, 86962 Futuroscope-Chasseneuil Cedex, France [Received 15 July 2001 and accepted in revised form 1 March 2002]

Abstract Nanoindentation has been used to explore the plasticity onset under contact loading in InP. Under low indenting loads ranging between 0.2 and 10 mN, fracture of the sample is avoided and plastic zones can be observed by transmission electron microscopy. The zone-size variation with load could be measured and analysed using the work of Johnson and of Kramer et al. The results were compared with those obtained in GaAs deformed under the same conditions. Furthermore, characterization of the dislocations was made and it was shown that the twinning formation in InP di ered strongly from that in GaAs. The results are compared with previously reported arrangements obtained in the microindentation domain.

} 1. Introduction Much work has been dedicated to the mechanical behaviour of III±V semiconductors such as GaAs or InP mainly in the high-temperatur e domain (400± 700°C) using compression tests (Rabier and George 1987). These data have been very useful to improve the handling of the materials during device processing (crystal growth and thermal treatment). Nonetheless, few data are available on the roomtemperature behaviour of III±V semiconductors (Boivin et al. 1990, Azzaz et al. 1994, Suzuki et al. 1999). In fact, at temperatures close to or below the brittle±ductile transition, con®ning pressure and/or predeformation is necessary to prevent fracture of the samples. Subsequent indentation testing is a powerful tool for investigating the low-temperature behaviour (Warren et al. 1984, Levade and Vanderschaeve 1999, KoubaõÈ ti et al. 2000, Le Bourhis and Patriarche 2000a). Furthermore, the development of heterostructures for optoelectronic applications has shown that the performance of the devices is dramatically a ected by threading dislocations that extend through the active layers. The mechanism by which such dislocations appear in y Email: [email protected]. Philosophica l Magazin e A ISSN 0141±8610 print/ISSN 1460-699 2 online # 2002 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080 /0141861021013509 8

1954

G. Patriarche and E. Le Bourhis

mismatched structures is still controversial . More data on the plastic behaviour of thin layers are necessary to obtain better understanding of the phenomenon and to ®nd new solutions to relax the mis®t strain underneath the heterostructure. In that ®eld, a nanoindentatio n probe has proved to be very helpful (Gerberich et al. 1996, Williams et al. 1999, Patriarche and Le Bourhis 2000, 2001). Here, very low indenting loads have been used to investigate the onset of plasticity in InP. Transmission electron microscopy (TEM) let us determine the zone-size variation with load and characterize the structure of dislocations. The results are compared with those obtained in GaAs deformed under the same conditions and discussed in view of arrangements reported in the microindentation domain. } 2. Experimental procedure Czochralski-grown single crystals were h001i oriented. InP samples were iron doped (10 18 atoms cm 3 ) while GaAs samples were undoped. The (001) surfaces were deformed by a Berkovich diamond pyramid using a nanohardness tester machine from CSEM (Switzerland). The tests were performed in the force-control mode of the machine. Indentation arrays were made for subsequent relocalization in the TEM samples. Large indentations were used as markers. The maximum load was varied between 0.2 and 10 mN. The load±penetration curves were analysed using the method proposed by Oliver and Pharr (1992) to determine the hardness of the samples. To prepare TEM plan-view thin foils of indented samples, the undeformed side (back side) was mechanically and chemically thinned with a bromine±methanol solution until su ciently thin to transmit the electron beam. } 3. Results and discussion 3.1. Loading±unloading curves Figure 1 shows loading and unloading curves of the indenter in bulk InP as well as the curves in bulk GaAs for maximum loads of 0.5 and 5 mN. Under 0.5 mN (®gure 1 (a)), the loading and unloading curves slightly separate as permanent deformation is generated under the indenter. This could be carefully checked by TEM as will be discussed in more detail below. In fact, observations of the indent site showed systematically a plastic zone for the loads used in this study (above 0.2 mN; see for instance ®gure 2). Sometimes the loading curve showed a discontinuity (pop-in) for a critical load Fc ranging between 0.2 and 0.3 mN in InP and between 0.2 and 0.5 mN in GaAs. Discontinuities in the loading curves have already been reported for several III±V semiconducting bulk materials and were attributed to the poor density of native defects (Hainsworth et al. 1995, Gerberich et al. 1996, Le Bourhis and Patriarche 1999, 2000b, Bradby et al. 2001). Pop-in corresponds to an abrupt plastic ¯ow generated by a strong dislocation activity (nucleation and propagation). This phenomenon is statistical as the Fc magnitude varied within 40% in the same sample and the discontinuities appeared only for a fraction of tests. It should be noted that Hainsworth et al. (1995) did not observe any pop-in in InP. One reason for the di erence is that doping dramatically changes the pathway to plastic deformation (Williams et al. 1999). Secondly, the Fc magnitude is very dependent on the tip geometry (bluntness of the extremity). Results from Bradby et al. (2001) show that Fc is considerably increased when a spherical tip (radius, 4.2 mm) is used instead of a sharp Berkovich tip like ours.

Low-load deformation of InP under contact loading

Figure 1.

1955

Loading and unloading curves of a Berkovich indenter in InP and GaAs at room temperature: (a) under 0.5 mN; (b) under 5 mN.

Comparison of the loading curves under the same load shows that InP deforms much more than GaAs (®gure 1 (b)). After the test is completed, the residual depths could be determined to be about 100 and 130 nm for GaAs and InP respectively. In fact, InP is determined to be much softer than GaAs (hardnesses of 5.2 GPa for InP and 7.5 GPa for GaAs) in good agreement with the work of Bradby et al. (2001).

1956


Figure 2. TEM plan view of the indent site made at room temperature (a), (b) in bulk InP under 0.26 mN maximum load and (c), (d) in bulk GaAs under 0.28 mN maximum load: (a), (c) g ˆ ‰220Š; (b), (d) g ˆ ‰2220Š.

3.2. Plastic-zone size evolution with load Figure 3 plots the plastic-zone radius c as a function of the square root of the maximum indenting load F max . The diameter of the plastic-zone projection on the (001) plane was measured by TEM considering only the central part of the plastic zone and was then averaged along two orthogonal h110i directions parallel to the 1=2 indented surface. Such plots show proportionality between c and Fmax when high indenting loads are used (Le Bourhis and Patriarche 2000b, Patriarche and Le Bourhis 2001). This behaviour could be well described using the works of Johnson (1985) and Kramer et al. (1998) and allowed to determine the ¯ow stress Y using

cˆ

3 2pY

1=2

1=2 Fmax :

…1†


1957

Figure 3. Plastic-zone radius c as a function of the square root of the maximum load Fmax . The maximum load Fmax was measured on the loading curves (see for example ®gure 1 (b)). The plastic-zone radius was determined by TEM in plan view. Fth is determined 1=2 by extrapolating linearly the data to the Fmax axis.

These models cannot be employed straightforwardly in the light-load range used here as a plastic-¯ow threshold is observed. Linear extrapolation of the data yields threshold loads Fth of about 0.016 mN for bulk InP and 0.030 mN for bulk GaAs (®gure 3). Both values are found to be lower than the Fc value determined at pop-in. In fact, TEM observations showed plastic deformation under all the loads used here. Therefore, the threshold load is to be placed below the lowest load used here (Fth < 0:2 mN). As a ®rst approximation, we used equation (1) with c values determined under the highest maximum load used in this study (Fmax ˆ 10 mN). In fact, we can assume in this case that Fth Fmax is satis®ed. This let us determine the ¯ow-stress values Y to be about 1.3 GPa for bulk InP and about 2.3 GPa for bulk GaAs. The ratio of the hardness to the ¯ow stress was determined to range between 3 and 4, in good agreement with the work of Johnson (1985). 3.3. Plastic zone structure As stated before, plastic zones were observed systematically under all the loads used for this study. This could be carefully checked as markers were used to map the indent sites observed in the microscope. Figure 4 shows indent sites generated under 3 mN in InP and in GaAs. As observed under lighter loads (®gure 2), the dislocation density is very high in the plastic zone centre where individual dislocations cannot be observed. The use of two di raction conditions (®gure 4) allowed us to observe stacking faults in InP that were aligned along both h110i directions parallel to the indented surface. Stacking faults could be observed already under a maximum load of about 0.5 mN (®gure 5). Only a few perfect dislocations were observed at the edge of the plastic zone and vanished using one of the di raction conditions (®gure 4).

1958


Figure 4. TEM plan view of the indent site made under 3 mN maximum load at room temperature (a), (b) in bulk InP and (c) (d) in bulk GaAs: (a), (c) g ˆ ‰220Š; (b), (d) g ˆ ‰2220Š:

Using the vanishing condition g· b ˆ 0, we determined that the Burgers vector of these dislocations is parallel to the sample surface. In contrast, under the same load (®gure 4), only a few stacking faults were observed in GaAs. Moreover, they were aligned along only one of the h110i directions parallel to the sample surface. Perfect-dislocation segments were observed at the edge of the plastic zone. Most of those extending along the [1110] direction vanished under g ˆ ‰220Š while most of the loops extending along the [110] direction vanished under g ˆ ‰2 220Š. Therefore, these dislocations have long segments with screw character. In fact, we have observed here the ®rst developments of rosette arms that are observed in the microindentation domain (Warren et al. 1984, Levade and Vanderschaeve 1999, KoubaõÈ ti et al. 2000).


Figure 5.

1959

TEM plan view of the indent site made at room temperature in bulk InP under 0.5 mN maximum load: (a) g ˆ ‰220Š; (b) g ˆ ‰2220Š.

Such perfect-dislocation arrangements could be observed in InP along [110] under a higher load (5 mN (®gure 6)). Nonetheless, stacking faults are still predominant along [1110]. In InP, rosette arms are well known to develop under high indenting loads and both arms show extended microtwins (Le Bourhis and Patriarche 2000a). Here the ®rst steps of microtwin formation have been observed. This deformation mode is expected to be easier in InP than in GaAs if considering only the stacking-fault energy ® (equal to 18 mJ m 2 for InP compared with 55 mJ m 2 for GaAs (Gottschalk et al. 1978)). However, GaSb (® ˆ 53 mJ m 2 ) as well as InAs (® ˆ 30 mJ m 2 ) showed microtwins in both rosette arms (Ning et al. 1995, Le Bourhis and Patriarche 2000b). Therefore, the energy consideration does not yield

1960

Figure 6.


TEM plan view of the indent site made in InP under 5 mN maximum load at room temperature: (a) g ˆ ‰220Š; (b) g ˆ ‰2220Š.

any satisfactory interpretation for the twinning anisotropy of GaAs. Further understanding of the phenomenon requires considering the di erence between the mobilities of the dislocations that form the two orthogonal rosette arms (Warren et al. 1984, Le Bourhis and Patriarche 2000a). In fact, GaAs showed strong anisotropy in rosette-arm length particularly when high indenting loads were used (®gure 4). } 4. Conclusion Plasticity onset under contact loading in InP has been investigated by nanoindentation under low indenting loads ranging between 0.2 and 10 mN. Plastic zones were observed systematically by TEM and this let us conclude that the plastic threshold is to be set below the load range investigated here. Interestingly, pop-ins have been observed, however. Therefore plasticity may be induced before pop-in in InP as well as in GaAs. The zone-size variation with load could be measured and the results


1961

were analysed using the work of Johnson (1985) and of Kramer et al. (1998). Flowstress values as well as hardness values were determined to be in good agreement with the literature values. Furthermore, characterization of the dislocations showed that twinning formation in InP di ered strongly from that in GaAs. Anisotropy was observed in GaAs while InP showed similar arrangements in both rosette arms. References Azzaz, M., Michel, J. P., and George, A., 1994, Phil. Mag. A, 69, 903. Boivin, P., Rabier, J., and Garem, H., 1990, Phil. Mag. A, 61, 619. Bradby, J. E., Williams, J. S., Wong-Leung, J., Swain, M. V., and Munroe, P., 2001, Appl. Phys. Lett., 78, 3235. Gerberich, W. W., Nelson, J. C., Lilleoden, E. T., Anderson, P., and Wyrobek, J. T., 1996, Acta mater., 44, 3585. Gottschalk, H., Patzer, G., and Alexander, H., 1978, Phys. Stat. sol. (a), 45, 207. Hainsworth, S. V., Whitebread, A. J., and Page, T. F., 1995, Plastic Deformation of Ceramics, edited by R. C. Bradt, C. A. Brookes and J. L. Toutbort (New York: Plenum), p. 173. Johnson, K. L., 1985, Contact Mechanics (Cambridge University Press). KoubaiÈ ti, S., Levade, C., Vanderschaeve, G., and Couderc, J. J., 2000, Phil. Mag. A, 80, 83. Kramer, D., Huang, H., Kriese, M., Robach, J., Nelson, J., Wright, A., Bahr, D., and Gerberich, W. W., 1998, Acta mater., 47, 333. Le Bourhis, E., and Patriarche, G., 1999, Phil. Mag. Lett., 79, 805; 2000a, Eur. Phys. J. appl. Phys., 12, 31; 2000b, Phys. Stat. sol. (a), 179, 153. Levade, C., and Vanderschaeve, G., 1999, Phys. Stat. sol. (a), 171, 83. Ning, X. J., Perez, T., and Pirouz, P., 1995, Phil. Mag. A, 72, 837. Oliver, W. C., and Pharr, G. M., 1992, J. Mater. Res., 7, 1564. Patriarche, G., and Le Bourhis, E., 2000, Phil. Mag. A, 80, 2899; 2001, J. Mater. Sci. Lett., 20, 43. Rabier, J., and George, A., 1987, Rev. Phys. appl., 22, 1327. Suzuki, T., Yasutomi, T., Tokuaka, T., and Yonenaga, I., 1999, Phil. Mag. A, 79, 2637. Warren, P. D., Pirouz, P., and Roberts, S. G., 1984, Phil. Mag. A, 50, L23. Williams, J. S., Chen, Y., Wong-Leung, J., Kerr, A., and Swain, M. V., 1999, J. Mater. Res., 14, 2338.