ON SOME ASPECTS OF THE MOBILITY OF DISLOCATIONS IN SLIGHTLY n-DOPED InSb SINGLE CRYSTALS

(1)

HAL Id: jpa-00223076

https://hal.archives-ouvertes.fr/jpa-00223076

Submitted on 1 Jan 1983

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

ON SOME ASPECTS OF THE MOBILITY OF DISLOCATIONS IN SLIGHTLY n-DOPED InSb

SINGLE CRYSTALS

J. Di Persio, R. Kesteloot

To cite this version:

J. Di Persio, R. Kesteloot. ON SOME ASPECTS OF THE MOBILITY OF DISLOCATIONS IN SLIGHTLY n-DOPED InSb SINGLE CRYSTALS. Journal de Physique Colloques, 1983, 44 (C4), pp.C4-469-C4-474. �10.1051/jphyscol:1983456�. �jpa-00223076�

(2)

ON SOME ASPECTS OF THE MOBILITY OF DISLOCATIONS IN SLIGHTLY n-DOPED InSb SINGLE CRYSTALS

J. Ui Persio and R. Kesteloot

Laboratoire de S t r u c t u r e e t Propridtds de Z f B t a t S o l i d e ( L . A . 2 3 4 ) ,

UniversiL6 des S c i e n c e s e t Techniques de L-IZZe, 59655 ViZleneuve d f A s c q Cedex, France

R6sum6 - La technique de topographie aux rayons X est utilis6e pour 6tudier les configurations dedislocations dSvelopp6es sous contrainte dans des 6prouvettes d'InSb faiblement dopges de type n. Les r6sultats confirment la vitesse beau- coup plus grande des dislocations a par rapport aux dislocations vis. Par con- tre, les dislocations f3 sont excessivement peu mobiles, apparemment P toute contrainte et ?I toute tempsrature. Ceci a pour cons6quence :

I) d'affecter la mobilit6 des dislocations vis dont le mouvement semble contr6- 16 par la propagation de crans simples a.

2) de bloquer le fonctionnement de sorlrces internes efficaces, sauf pr2s des surfaces.

Ceci conduit 2 une plasticit6 qui semble trZs spgcifique aux compos6s 111-V.

Abstract - X-Ray Lang topography is used to investigate the dislocation substructure developped under stress in slightly n-doped InSb samples. Results confirm the much higher velocity of a-dislocations compared to screws. On the other hand, f3-dislocations are excessively little mobile, apparently at all stress and temperature. This has for consequence :

I) to affect the mobility of screw dislocations whose motion looks to be con- trolled by the propagation of single a-kinks.

2) to prevent the development of efficient internal sources, except near surfaces.

This behaviour induces a plasticity which appears to be very specific to 111-V compounds.

I - INTRODUCTION

In 111-V compounds, because of the non-centrosymmetry of the sphalerite structure, one has to distinguish, beside the screw dislocation, two types of 60" dislocations, labelled A and B. Indeed, depending on the sense of the Burger's vector, the extra-half plane may end either by a row of trivalent A atoms, or by a row of pentavalent B atoms. It is now well established that the difference in the corestruc- tures which results, contrary to the elemental tetravalent semiconductors, induces strong differences in the d~namical behaviour of these dislocations, specially in their mobility, and in the related electrical properties.

In these compounds, mainly GaAs and InSb, most of experimental work about mobility has been devoted to velocity measurements performed on samples deformed almost exclusively by bending 11 to 6 1 , in some cases by creep 1 7 , 8, 9 1 , the double etching technique being extensively used to follow dislocation motion. A remarkable feature;

which comes out from these results and obviously common to all 111-V compounds, is the much higher velocity of A dislocations compared to the ones of B cr screw dislocations, with a ratio which depends sensitively on temperature and doping. On the other hand, the results also show a remarkable dispersion of experimental values, velocities and corresponding activation energies, probably connected to the techniques used.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1983456

(3)

C4-470 JOURNAL DE PHYSIQUE

We have choosen a somewhat different experimental approach, with the aim to study dislocation mobility in nearly perfect, slightly n-doped (% 10'~/cm 3 ) InSb samples.

This approach is based on a combination of dynamical compression experiments (Instron machine) performed in the preyield region up to various stress levels, and X-Ray Lang topography /lo/ of thin slices (300 Ua thick) cut from the deformed samples 111, 121.

Of course, our results confirm the strong asymmetry of dislocation velocities in InSb. However, owing to the advantage of the technique used, they also show addition- nal features which bring out original aspects of the mobility.of A, B and screw dis-

locations. The aim of this paper is to show, and discuss, some of these aspects. De- tails of the experimental work can be found in 1121.

In the following, we use the "SHUFFLE" convention. A and B dislocations are ter- med a and 6 respectively.

I1 - MOBILITY OF a AND 6 DISLOCATIONS

Figure 1 below is a good illustration of the dislocation substructure which is present in samples deformed at a stress slightly above the lower yield point stress.

This substructure is composed of half loops, strongly elongated in the direction

F i g m e 1 : n d d L o o p w i t h a-@~ovLtr, induced by n f i e s b . The ubhuded dace -in an t h e Le6.t. T = 4 9 5 K ( x 3 0 ) .

of the <I102 Burgers vector. The elongated, remarkably straight, parts are of pure screw character. The leading parts are therefore 60" fronts, which have not any character. From crystallographic considerations based on the polar nature of t h a 1 1 1 ) faces of deformed samples, we have developped an analysis of the possible configurations of superficial sources which, under uniaxial compression, lead to half-loops expanding through the bulk 1121. This analysis shows that, in the "shuffle" set, dislocation half-loops such as those observed in figure 1 do have 60" fronts of a-type exclusively. (They would be of 6- type in the "GLIDE" set. Experiment does not allow to decide between the two sets. If the accepted hypothesis does hold, that a-dislocations, with dangling bonds of In-type, are the more mobile, then our observations are

consistent with a "shuffle" motion).

The very elongated and narrow shape of the half-loops interprets quite well the much higher velocity of 60' segments, here of a-type, an observation which corrobo-

(4)

All these half-loops with a-fronts are almost exclusively emitted by superficial sources, which can be selectively introduced (case of figure 1) by specific grinding of the surfaces 113, 141. As we show below, internal sources appear to be scarcely distributed and of little efficiency, their development being prevented by the quite low mobility of 6 dislocations.

As a matter of fact, the essential feature which has to be emphasized here is the almost complete absence of 6 dislocations, issued either from internal sources or from selectively introduced superficial sources 1131, at least at the topographic scale. In particular, dislocation half-loops with a-fronts have never been observed to move from the surfaces, in a wide range of stress and temperature. Of course, 6

dislocations do exist. Figure 2 shows a typical example of an internal source whose aspect recalls the one of a Frank-Read source, that is a row of concentric loops which are nucleated here by a source which is located at the very left of the figure.

This picture shows 6 segments which are located very near the source, the expansion of the loops being exclusively observed on the a side. It confirms the very low mobility of B dislocations compared to the a one. It explains also the absence of B

fronts nucleated at the surfaces. Because of their low mobility, very high stresses

Figune 2 : IvLtetrnde bounce i n InSb, obnehved by X-Ray topoghaphq.

The bouhce LA Located aL A. T = 4 9 5 K ( x 1 0 5 ) .

are required in order to move them subsequently. At "normal" stresses, a-dislocations are already active. The observation of such a motion would need therefore to block the nucleation of "a" sources, which seems to be improbable in a F.C.C. structure where the number of slip systems able to develop a-dislocations is potentially high.

This renders doubtful the possibility to introduce selectively 6-dislocations 1131, with intent to study their electrical properties for exemple, even at higher temperature where their mobility would be normally increased. Very recent experiments we made in this sense at a temperature of 623 K ( 0 , 8 Tm) have not shown any mobile 6-

dislocations.

111- MOBILITY OF SCREW DISLOCATIONS

In previous studies by the double etching method, the motion of screw dislocations has never been clearly analyzed and results on velocity measurements are very dispersed. The reason is that the half-loop which develops from a scratch has its screw part parallel to the etched surface, so the motion of the screw segment cannot be followed easily, except under particular geometrical conditions 131. Our own observations of the actual motion of screw dislocations reveal in fact two features : on the one hand, a very slow motion associated with a strongly rectilinear character of the line (figure l), on the other hand, a wide displacement of segmented screw lengths when these can get single kinks of a-type. From a fondamental point of view, these observations raise questions about the nature of the exact mechanism which con- trols the propagation of a screw dislocation in a 111-V compound, and correlatively, about the physical origin of the difference between screw and a-mobilities.

In crystals with a high lattice friction like semiconductors, it is well accepted that the displacement of a dislocation line from a Peierls valley to the next operates by the nucleation of a double kink and lateral propagation of the two kinks

(5)

thus created. In tetravalent semiconductors, the two kinks are identical, and so mi- grate at the same velocity. The situation is different in 111-V compounds, especially when a double kink is nucleated on a screw segment. Figure 4 illustrates this point for a half-loop with leading a-fronts.

screw

Figwie 4 : Double k i n k nudeated d o n g a ~ l v c a t i o n hded loop uli;th a-dho&.

[nchematicl In

Screw

In this case, the nucleation of a double kink on the screw part involves two kinks with dangling bonds of different character, a and B. We may therefore expect a different lateral propagation for the two kinks, because they do not experiment the same Peierls friction. If the @-kink is effectively much less mobile than the a-one, then we would expect screw dislocations loosing their rigid character by a rapid ac- cumulation of the faster a-kinks in the direction of propagation of the a-60' front.

As a consequence, screw dislocations would escape obliquely from their Peierls valleys. This is never observed on topographs, which show very elongated (over several millimeters length) straight lines. These lines remain remarkably straight during motion, contrary to a-60" fronts which are seen to espace freely from the <]lo> valleys (figure 5 ) , at least at sufficiently high temperature. The sketch of figure 4 shows that the propagation of a a-60' segment by the double kink mechanism does not involve any @ kink.

F i g m e 5 : RecLZLneah a p e d 06 the ~ c r r e w pant by compahinon to the tounded - nhaped domi 06 the a-&hod. T = 495 K ( x 130).

If we may ask for the efficiency of a double-kink mechanism to explain the slow propagation of an elongated rigid screw dislocation, there is no doubt that this propagation is greatly accelerated when the screw gets single a-kinks. Figure 6 shows such an example. It is easy to verify that the superkinks which are observed to mi- grate along the screw lengths are all of a-type, which is another confirmation that a segments are the more rapid. In this case, the two opposite screw arms of a half loop move quickly apart fromeachother, giving rise to wide loops sweeping the slip plane (figure 1).

A screw dislocation can readily get a single kink either when it intercepts another secant dislocation, forming a junction, or at surfaces. As a matter of fact, junctions are frequently observed on the topographs 1121. They are of pure edge character, with a <112> orientation.Like screwdislocations, they are also remarkably straight and stable against applied stress, a point which is not yet understood. The fact that junctions are very efficient to provide a-kinks is illustrated in figure 6 where two families of a half loops intersect in the same slip plane. However, the way by which a junction is actually formed is difficult to intercept. There are exam- ples on topographs which suggest that junctions could be initiated also at internal

(6)

sources (In figure 2, the source is actually at the start of a junction). A more de- tailed study of these unusual configurations will be published elsewhere. They show striking evidence of an asymmetrical propagation of screw dislocations, depending wether they can get a-kinks, or @-kinks, in order to move. This would hold also at

surfaces, where kinks of opposite nature may be formed by cross-slip of the two emer- gent arms of a dislocation half-loop, which are of opposite sense. We have no confirmation of this point for now.

IV - CONCLUSION

The main results which emerge from our observations are the nearly sessile character of f3 motion and the unusual behaviour of screws. In slightly doped InSb or GaAs, velocity measurements by double etching, although very dispersed, confirm a low mobility of f3 dislocations, as a rule slightly lower or equal to the mobility of screws. It is of course difficult to correlate these results with our own observations because of the different techniques used and the wide range of temperature, stress and doping studied, which do not always overlap. Also, the dispersion of published results may partly stem from the use of a technique (surface etching) which may give by nature erroneous estimates both of the true character of the underlying moving segment and of its true velocity, because of surface effects. This remark is illustrated by the behaviour of screw dislocations we show. Most of our experiments have been performed at a temperature of 495 K, only on slightly n-doped InSb samples.

Experiments are now planned to study by the same technique the influence of temperature and doping on the relative mobilities of the dislocations involved. There is indeed a challenging interest to investigate wether the relative mobilities of a and B dislocations are influenced by doping or not, and to what extent. Reports on GaAs /2/

indicate that the mobilities of a and f3 dislocations would be reversed by p-doping, B dislocations being the faster.

REFERENCES

CHOI S.K. and MIHARA M., J. Phys. Soc. Japan 32 (1972).

CHOI S.K., MIHARA M. NINOMIYA T., Japan J. of Appl. Physics 16, (1977) 5.

EROFEEVA S.A. and OSIP'YAN Yu A. Soviet Phys. Solid State 15 (1973) 3.

EROFEEVA S.A. and OSIP'YAN Yu A. Soviet Phys. Solid State 16 (1975) 10.

MIHARA M. and NINOMIYA T. Phys. Stat. Sol. (a) 32 (1975) 4 3 . SHIMIZU H. and SUMINO..K. Phil. Mag. 2 ^(1975).

STEINHARDT H. and SCHAFER S., Acta Met. 19 (1971).

STEINHARDT H., Thesis, University of Gijttingen (1972).

STEINHARDT H. and HAASEN P., Phys. Stat. Sol. (a) 2 (1978).

LANG A.R., J. Appl. Phys. 30 (1959).

(7)

/II/ KESTELOOT R. and DI PERSIO J., J. de Physique - ^Lettres41 ^(1980).

/12/ KESTELOOT R., These 3e cycle - Universitg de Lille (1981).

1131 FERRE D. and FARVACQUE J.L., Phys. Stat. Sol. (a) 2 (1978).

1141 FARVACQUE J.L. and FERRE D., J. de Physique 40, C6, (1979).