• Aucun résultat trouvé

Numerical simulation of dislocation dynamics in disordered crystals with high Peierls barriers

N/A
N/A
Protected

Academic year: 2021

Partager "Numerical simulation of dislocation dynamics in disordered crystals with high Peierls barriers"

Copied!
7
0
0

Texte intégral

(1)

HAL Id: jpa-00210567

https://hal.archives-ouvertes.fr/jpa-00210567

Submitted on 1 Jan 1987

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished 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.

Numerical simulation of dislocation dynamics in disordered crystals with high Peierls barriers

I.R. Sagdeev, V.M. Vinokur

To cite this version:

I.R. Sagdeev, V.M. Vinokur. Numerical simulation of dislocation dynamics in disordered crystals with high Peierls barriers. Journal de Physique, 1987, 48 (9), pp.1395-1400.

�10.1051/jphys:019870048090139500�. �jpa-00210567�

(2)

1395

Numerical simulation of dislocation dynamics in disordered crystals with high Peierls barriers

I.R. Sagdeev and V.M. Vinokur

Institute of Solid State Physics, U.S.S.R. Academy of Sciences, Chernogolovka, Moscow region,

U.S.S.R. 142 432

(Reçu le 29 mai 198 7, accept£ le 8 juille t 1987)

Résumé.-Une simulation numérique du mouvement de dislocations dans des cristaux désordonnés avec de hautes barrières de Peierls est présentée. Une chute brutale de la mobilité des dislocations dans la région des champs

externes faibles est révélée. Quand la charge impulsionnelle est appliquée, la mobilité des dislocations décroît avec

l’accroissement de la fréquence des impulsions. Il est montré que la chute de mobilité des dislocations sous l’action des forces externes impulsionnelles et stationnaires résulte de sous-linéarités dans la propagation des marches dans un

champ de force aléatoire.

Abstract.-Numerical simulation of dislocation motion in disordered crystals with high Peierls relief is performed. A sharp drop in dislocation mobility in the region of weak external fields is revealed. When pulse loading is applied, the

dislocation mobility decreases as the pulse frequency increases. It is shown that the dislocation mobility drop under

the action of both stationary and pulse external forces results from sublinearity in kink propagation in a random force field.

J. Physique 48 (1987) 1395-1400 SEPTEMBRE 1987,

Classification

Physics Abstracts

61.70G

LE JOURNAL DE PHYSIQUE

The basic concepts of dislocation dynamics in crystals with high Peierls barriers were formed by

Lothe and Hirth [1] and Kazantsev and Pokrovsky [2].

A dislocation is treated as an elastic string in a peri-

odic potential. It is assumed that initially the entire

dislocation lies in one of the valleys. Later, thermal

fluctuations create kink-antikink pairs on the disloca- tion, which then expand under the action of the ap-

plied external force until they annihilate with kinks from other pairs. This results in a transverse displace-

ment of the dislocation. Experimental data show that dislocation mobility is also significantly effected by point defects, always present in real crystals. Specif- ically, it has been revealed [3-5] that doping of crys- tals can give rise to an increase in dislocation velocity.

This effect could be explained within the framework of the model [1,2] as a consequence of local lowering

of the Peierls barrier due to interaction between the dislocation and impurities (Petukhov [6,7,8]). Unfor-

tunately, all the attempts to compare quantitatively

theoretical and experimental data, which would make it possible to extract unambiguously the physical pa-

rameters, describing dislocations and defects, so fair

failed. One principal difficulty in comparing theoreti-

cal and experimental data is that there are too many

parameters describing dislocation dynamics : Peierls

barrier height, kink creation energy, concentration of

impurities and the energy of their interaction with the

dislocation, kink diffusion coeflicient along the dislo-

cation line, etc. At the same time virtually the only

measurable quantity is the mean (over an ensemble of

dislocations) dislocation displacement under the ac-

tion of an applied force.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019870048090139500

(3)

1396

In this context one can view as a major advance

the series of works by Nikitenko, Farber and Iunin

[9,10], who investigated motion of dislocations in Peierls relief under the action of a pulse force. The technique employed made it possible not only to track

the displacement of individual dislocations, but to

trace (independently) diffusion of kinks along the dis-

locations line. A sharp drop in dislocation mobility

was reported, as the frequency of the external pulse

force increases (for constant period-to-pulse duration ratio), as well as considerable slowing down of dislo-

cation motion as the spacing between pulses increases (their duration being unchanged). Furthermore, in- terpretation of the experimental data, obtained in

[9,10] corroborated the notion that dislocations in

crystals with high Peierls barriers move by successive

transitions from one valley to another. The principal

aim of our work was to simulate numerically dislo-

cation dynamics in the presence of defects. We have tested the commonly accepted view on the character

of dislocation dynamics in the presence of defects. In the work of one of the authors [111 it was shown that

moderate external forces impurities and defects can

play the leading part in forming dislocation mobility.

Specifically, if the external force is such that the quan-

tity q = 2TujcV2 1 ( T is the temperature, a is the external applied strain, V is the energy of interaction of a defect with the dislocation, c is the concentration of defects), then drift of kinks along the dislocation becomes nonlinear in time (sublinear law) : w - tq (x

is the displacement of a kink over time t), and, as a

consequence, the mobility of the dislocation sharply

decreases. We have studied, among things, disloca-

tion dynamics under the action of a pulse force, and

found that the decrease in mobility with the increase in pulse frequency might really be a consequence of the IIX _ tqll -effect.

As a model of a dislocation for numerical simula- tion we chose that of an elastic string, whose equation

of motion is of the form (we believe the friction in the system to be large, so that we can neglect the inertial

term) :

It is assumed here that the dislocation initially lies along the x axis, u is its transverse displacement, x is

the elastic constant, Uo is the height of the Peierls

barrier, r is the kinetic coefficient, -fez) is external

force per unit lenght of dislocation, U1(x, u) is a ran-

dom quantity, describing distortion of the Peierls bar- rier by defects, is the lattice parameter. The last term in the rhs of (1) is a Langevin force, which is a source

of thermal fluctuations of the dislocation line. The

angular brackets (...) denote thermodynamical aver- aging. The values of the coefficients in (1) were chosen

close to their actual experimental values.

Discretizing the dislocation into N elements (each of

them may be viewed as a lattice site), we bring equa- tion (1) to a finite difference form :

where u(i) is the displacement of the itch element at the time instant t, u’(i) is its displacement at the time

t + At, At is the discrete time step, ci = x .r, f = F.r, u(u(i)) = r.(gU/au. The external field is either

stationary or a train of square pulses with the period

T and duration tp.

The thermal force £ was approximated by a se-

quence of pulses of random amplitude with normal distribution and dispersion - T. The spacing between

the pulses had an exponential distribution with the mean Tf .

The lattice potential U was chosen in the form U = Uo sin (y/27ra). Defects were placed in randomly

chosen sites with the probability i = ac (where a is

the lattice parameter, c, the volume concentration of

defects) and where assumed to be short-range (acting withing one lattice spacing).

In the course of our numerical experiment we

simulated motion of dislocations of length N =1000

-

4000 lattice parameters, in the absence, as well

as in presence of defects, and both in stationary an pulse external fields. We computed mean dislocation

velocity, v, kink concentration, Nk, and kink mobility,

u.

In order to test the numerical model and the program, we studied dislocation dynamics in a per- fect crystal and compared our numerical results at the temperature T = 0.15 Uo with the analytical re- sults, obtained by Buttiker and Landauer [12], who

had studied the model (1) without defects. Figure

1 presents the dislocation velocity as a function of external force according to [12] (dashed curve) and

our numerical result. We also obtained the disloca- tion velocity v and kink concentration Nk as func-

tions of a stationary external force (the solid curves

in Fig. 2). The commonly accepted view is that at

large forces the kink concentration and dislocation ve-

locity should depend on force exponentially [1]. For

force F Uo kink concentration is almost force-

independent and the dislocation velocity should be simply proportional to F (linear law). Our numer-

ical results are in fairly good agreement with these predictions, the transition from exponential to linear

law taking place at FIUO c-- 0.12, confirming the the-

(4)

Fig. 1.-Comparison of our numerical data and analytical results [12] (dashed line). Dislocation velocity as a function of applied

force.

Fig.2.-Kink concentration Nk (per thousand lattice parame-

ters), dislocation velocity v and kink mobility p = v / Nk / F as

functions of exernal force F in perfect crystals (solid curves).

a = 0.5, Ud = 0.5 Uo. The force is in units of Peierls barrier.

Units of mobility and velocity are their values at F = 0.1 Uo.

T = 0.5 Uo.

oretical estimates of [1]. Furthermore, we studied the

dependence of v and Nk on the frequency of a pulse

external force (in the nonlinear region), for a constant period-to-duration ration 2:1 (Fig.2). It is clear that

the dislocation velocity under the action of a high- frequency force should be approximately equal to that

under the action of a half-amplitude low-frequency

force. Our numerical results (Fig.2) agree well with

this idea. Note that in the case under consideration

(nonlinear region, no defects) the frequency depen-

dence of dislocation velocity is completely determined by that of kink concentration.

Kink concentration, dislocation velocity and kink mobility as functions of external force in a crystal

with defects are depicted in figure 2 (dashed lines).

In the range of external forces 0.15 Uo 0.4 Uo the

dislocation velocity in disordered crystals is larger,

while outside of this range, smaller, than in a crys- tal without defects (Fig.2b). Such a behaviour of the

velocity could be easily explained within the frame- work of the model [111. At forces 0.4 Uo F Uo

the kink creation energy is sufficiently small, and the

contribution of defects to generation of kinks is in-

significant. In this case the defects only slightly slow

down the motion of kink along the dislocation line. At F 0.4Uo a majority of kinks are created by clusters

of defects, locally lowering the Peierls barrier. Thus,

at F 0.4 Uo defects promote generation of kinks

(Fig.2a), while their braking of kink motion is still

insignificant and the dislocation velocity is greated

than in a perfect crystal. As the external force fur- ther decrease, an abrupt change in the character of kink motion along the dislocation takes place in disor-

dered crystals, according to [11] (at a threshold force F = Fo, when q = 2 FoT/cUd =1) : the law of kink motion becomes sublinear. The sharp drop in dis-

location mobility (and, therefore, velocity) observed

at F 0.15 Uo in disordered crystals (Fig.2c), is a

consequence of the above-mentioned slowing down of

kink spreading at F 0t15 Uo. The experimental

value of the parameter q =1.2 agrees well with that

predicted in [111 (q = 1). It is easy to see that the mobility drop takes place within the force range, in which the kink concentration Nk depends no more on

the applied force i.e., within the linear range.

In order to study the influence of defects on dislo- cation dynamics under the action of a nonstationary

force in the nonlinear region, we applied to the dis-

location a train of square pulses with the amplitude

F = 0.5 Uo. The period-to-pulse duration ratio was

chosen to be 2:1. This was exactly the same kink of

pulse force as that used in [9,10]. The temperature

T was chosen to be T = 3 Uo, the concentration of

defects and their potential c =0.3 and Ud± O.lx Uo,

respectively (the sign is random). It turned out that

the frequency dependence of dislocation velocity in

(5)

1398

the presence, as well as in the absence of defects, ac- tually copies that of kink concentration Nk. What’s

more, in the presence of defects the kink concentra-

tion, and, therefore, dislocation velocity are 1.5 times greater than in a perfect crystal. Our results show that the role of defects in the nonlinear region is re-

stricted to increasing the nucleation rate : it is eas- ier for kink-antikink pairs to be born at clusters of

barrier-lowering defects. However, the defects do not

change the general character of the frequency depen-

dence of dislocation velocity.

Fig.3.-Kink concentration (triangles) and dislocation velocity (circles) as functions of the period of pulse external force in perfect crystal at T = 0.5 Uo, F = 0.5 Uo (all units same as

in Fig.2, except for the pulse period, normalized by time of displacement of dislocation by 1 lattice parameter under sta- tionary external force).

In the linear region, as noted above, a more sig-

nificant role is played by braking of kink motion by

defects. We expect this effect to be particularly im- portant in the case of a dislocation, moving under

the action of pulse force in the linear region, as in

the experiments [9,10]. Depicted in figure 4 are the

kink concentration and dislocation velocity as func-

tions of the period of constant-amplitude and con-

stant period-to duration ratio pulses in a crystal with

defects. The amplitude of the pulse force was chosen slightly higher than its threshold value for a station-

ary force (F = O.lC/o)’ One can see that at low pulse frequencies the kink concentration and dislocation ve-

locity are the same as in the case of stationary force (when normalized to active time). However, as the

Fig.4.-Kink concentration (triangles) and dislocation velocity (circles) as functions of external force pulse period in crystal

with defects (E = 0.5, Ud = 0.5 Uo).

pulse force frequency increases, a sharp drop of dislo-

cation velocity is observed. A peculiarity of this drop

is that it is not accompanied (as was the case in the

linear region) by a decrease in the kink concentra- tion. On the contrary, the latter even grows as the

frequency increases. Thus, the above data show that the decrease in dislocation velocity is due to braking

of kink motion by defects, rather than to suppression

of kink generation at high frequencies.

In addition to the frequency dependences of dis-

location velocity and mean kink concentration, we

also obtained the dependences of these quantities on

the period-to pulse duration of the force, for constant

duration of the pulses. These dependences (not nor-

malized to active time) are shown in figure 5 and are

in qualitative agreement with the experimental re-

sults of [9,10].

Fig.5.-Kink concentration (triangles) and dislocation velocity (circles) as functions of spacing between pulses, normalized to constant pulse duration in crystal with defects (E = 0.5, Ud =

0.5 Uo).

We have studied, by means of numerical simula-

tion, the influence of defects on dislocation dynam-

ics in Peierls relief under both stationary and pulse loading. It was demonstrated that in the case of sta-

tionary and pulse loading. It was demonstrated that

in the case of stationary external force defects lower

dislocation mobility in the linear region (F 0.2 Uo),

(6)

by braking kink propagation, while at greater forces

(F > 0.2 !7o) clusters of barrier-lowering defects rise nucleation rate, there by increasing dislocation mobil-

ity and plasticity of the crystal. The threshold force

value, at which a sharp drop in dislocation mobility

is observed was found to be close to the critical force

Fo, at which the transition from linear x N t to sub- linear x N t7, q 1 law, predicted in [111, takes place.

A study of the dependence of the velocity of an in-

dividual kink on a stationary external force showed that in crossing the threshold f = Fo the charac- ter of the kink motion undergoes a radical change,

and in the case of low forces a kink spends most of

the time in traps (formed by defect clusters). It was

demonstrated that defects do not alter the character of dislocation dynamics in the case of pulse loading

with large amplitudes, corresponding to the nonlin-

ear region. However, at small amplitudes of external

force, close to Fo (the sublinear kink motion thresh-

old), a sharp drop in dislocation velocity as frequency

increases is observed. The results obtained enable us

to come to the conclusion that the sharp drop in dis-

location mobility, observed experimentally [9,10] at

high frequencies of the pulse loading, might be a con-

sequence of braking of kink propagation by defects,

rather than of suppression of kink generation at high frequencies. The latter effects is altogether absent in

the linear region. Indeed, dislocation velocity in a perfect crystal is [1,2]

where u is the kink propagation velocity, W(F) is the

energy of creation of a kink-antikink pair, depending

on the stationary external force F. If the frequency

of the external force is much lower than the mean

nucleation frequency, corresponding to a stationary

force of the same amplitude, then

where the angular brackets denote averaging over a period. In the opposite limiting case of high (i.e.,

much higher that the nucleation frequency) frequen-

cies

The transition from vlif to vl f should be sufficiently gradual, as illustrated by figure 3. In the linear regime

the nucleation energy W is virtually independent of

the applied force, so that vi.f m Vh. f. In other words,

in a perfect crystal in linear region one should not ex- pect a significant change in dislocation velocity during

the transition from high (in the above sense) to low

frequencies of the external pulse force.

Note that in the linear region, as one can see

from (4) and (5), the following relation must hold (after normalization to the active time) :

where v(F) is the dislocation velocity in a stationary

external field F, equal to the amplitude of the applied pulse force. Comparing the results, obtained for the

case of stationary loading in the absence of defects

with those shown in figure 3, one can see that relation

(6) holds to a good accuracy.

All the values of parameters, used in our numer- ical simulation were close to actual experimental val-

ues for semiconductor crystals with high Peierls bar- riers, with the exception of temperature, which was by far higher than in laboratory experiment. Such

a choice of T was dictated by the limitations of the computers available.

Summing up, we would like to point out that our

results show that point defects in crystal lattice have

an extremely profound effect on dislocation dynamics.

One might expect that at lower temperatures, closer

to actual experimental values, the effects we have ob- served (dislocation mobility threshold, frequency de- pendence of dislocation velocity in nonlinear region, nonlinearity in kink motion) would be much sharper

and more pronounced.

Acknowledgements

We are grateful to V.I. Nikitenko, B.Ya. Farber and V.Ya. Kravchenko for useful discussions of the model and results.

References

[1] LOTHE, J., and HIRTH, J.P., Phys. Rev. 115 (1959) 543.

[2] KAZANTSEV, P.A., and POKROVSKY, V.L.,

ZhETF 58 (1968) 677.

(7)

1400

[3] PATEL, J.R. and CHANDHURI, A.R., Phys. Rev.

143 (1966) 601.

[4] EROFEEV, V.N., NIKITENKO, V.I. and OSVENS- KII, V.B., Phys. Status Solid 35 (1969) 79.

[5] EROFEEV, V.N.; and NIKITENKO, V.I., Fiz. Tv.

Tela 13 (1971) 146.

[6] PETUKHOV, B.V., Fiz. Tv. Tela 13 (1971) 1445.

[7] PETUKHOV, B.V., and SUKHAREV, V.Ya., Fiz.

Tv. Tela 22 (1980) ; Fiz. Tv. Tela 23 (1981)

1093.

[8] PETUKHOV, B.V., Fiz. Tv. Tela 25 (1983) 1822.

[9] NIKITENKO, V.I., FARBER, B.Ya., and LUNIN,

Yu.L., Pis’ma ZhETF 41 (1985) 103.

[10] FARBER, B.Ya., LUNIN, Yu.L., and NIKITENKO, V.I., Phys. Status Solid a 97 (1986) 469.

[11] VINOKUR, V.M., J. Physique 47 (1986) 1425.

[12] BUTTIKER, M., and LANDAUER, R., Phys. Rev.

A23 (1980) 1397.

Références

Documents relatifs

The twinning dislocation produced in the twin boundary by passage of a perfect dislocation is eviden- tly stable ; very small partial dislocations might for- mally have

2: Models used in the simulations: Oblique cell including a dislocations dipole with periodic boundary conditions (left), and cell with a single dislocation and fixed

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

STUDY OF DISLOCATION - POINT DEFECTS INTERACTION IN MgO SINGLE CRYSTALS BY INTERNAL

As the external force decreases the length of the critical nucleus increases and the contribution to nucleation energy from the random field increases too. When

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

Interesting, however, is that the radii of curvature for the trailing partials in the high stress deformed crystals in general are found too large, so that the force