See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323610792

Dependences of the Vacancy Concentration and the Self-Diffusion Coefficient

on the Size and Shape of a Nanocrystal

Article  in  Nanotechnologies in Russia · July 2017

DOI: 10.1134/S1995078017040140

CITATIONS READS

0 38

1 author:

Mahach Magomedov
Daghestan Scientific Centre of the Russian Academy of Sciences
143 PUBLICATIONS   926 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Mahach Magomedov on 08 March 2018.

The user has requested enhancement of the downloaded file.

ISSN 1995-0780, Nanotechnologies in Russia, 2017, Vol. 12, Nos. 7–8, pp. 416–425. © Pleiades Publishing, Ltd., 2017.
Original Russian Text © M.N. Magomedov, 2017, published in Rossiiskie Nanotekhnologii, 2017, Vol. 12, Nos. 7–8.

Dependences of the Vacancy Concentration

and the Self-Diffusion Coefficient on the Size
and Shape of a Nanocrystal
M. N. Magomedov
Institute for Geothermal Research, Dagestan Scientific Center, Russian Academy of Sciences, Makhachkala, 367030 Russia
e-mail: mahmag4@mail.ru
Received September 8, 2016; accepted for publication June 21, 2017

Abstract—Based on the generalized Rectangular Parallelepiped (RP) model of a nanocrystal, an expression

for the Helmholtz free energy is derived and the equation of state for the nanocrystal, which contains both
lattice vacancies and delocalized (diffusing) atoms, is calculated. The dependences of the probability of for-
mation vacancies (φv) and the probability of delocalization of an atom (xd) on the size and shape of a nano-
crystal at different temperatures (T) and pressures (P) are studied. The calculations are performed for the
BCC lattice of iron under the condition of isothermal compression of a nanocrystal along the isotherms at
300 and 1000 K. The size dependences are studied at the atmospheric pressure (P = 1 bar) and at P = 100 kbar.
It is shown that the isothermal–isobaric growth of a nanocrystal at the atmospheric pressure and T = 300 K
gives a lower number of vacancies per atom in a nanocrystal in comparison with a macroscopic crystal, but
dispersing the latter at T = 1000 K increases the probability of formation of vacancies. Upon reducing the size
of a nanocrystal, the probability of delocalization of an atom (the same as the self-diffusion coefficient)
increases at any pressure and temperature. The φv/xd ratio decreases with a decrease in the size of a nanocrys-
tal and the out-of-vacancy self-diffusion is observed at the size at which the number of delocalized atoms is
larger than the number of vacant sites in the nanocrystal lattice. When the shape of a nanocrystal deviates
from the most optimal shape (from the cubic shape in the case of the RP model), the size dependences of the
nanocrystal lattice properties become stronger.

DOI: 10.1134/S1995078017040140

INTRODUCTION where αshape is a parameter which depends on the

In view of the absence of a statistical theory for a
nanocrystal of a simple one-component substance, it
is still not entirely clear how changes the activation
parameters such as the vacancy formation probability
and the fraction of delocalized (diffusing) atoms upon
an isobaric decrease in the size of a nanocrystal. In
approaches to this task known to date (e.g., see [1–4]),
the correlations of the specific binding energy of a
nanocrystal (Ec), the Debye temperature (Θ), and the
melting temperature (Tm) with the activation process
energy, namely, with either the vacancy formation
energy (Ev) or the self-diffusion energy (Ed), are used.
Assuming that the interatomic distance, the crystal
structure, and the surface shape do not change upon a
decrease in the diameter (dia) of a nanocrystal, the
following heuristic dependences are used [1–4]:
2
Ec (dia) ⎡Θ(dia)⎤ T (dia)
=⎢ ⎥ = m
Ec (∞) ⎣ Θ(∞) ⎦ Tm(∞)
E (dia) Ed (dia) α
= v = = 1 − shape ,
Ev (∞) Ed (∞) dia
shape of a nanocrystal and which is usually deter-
mined from the size dependence of the melting tem-
perature, since the Tm(dia) function is most easily
(and, therefore, most accurately) measured in the
experiments.
The dependences expressed in Eq. (1) raises many
questions, one of which is the issue concerning the
constancy of interatomic distances upon a decrease in
size and the corresponding increase in surface pres-
sure. In addition, the question arose about the correct-
ness of Eq. (1) when the size decreases along the high-
pressure isobar because of the nanostructurization of a
macroscopic crystal under high compressive. How-
ever, the main question is why do metallic nanocrys-
tals produced in the experiments have faintly defective
stable crystal structures [5]? According to the criterion
given by Eq. (1), nanocrystals should be more appre-
ciably activated by vacancies than macroscopic crys-
tals [1–3], which should lead to the instability of their
(1) structures.
The issue concerning the dependence of the activa-
tion parameters of a nanocrystal on the size and shape

416
 DEPENDENCES OF THE VACANCY CONCENTRATION 417

of its surface under different isobaric and isothermal 1/3

6v (φ = 0)⎤
1/3
⎛ 6k V ⎞
conditions is addressed in the present studies on the co = ⎜ p ⎟ = ⎡ a v , (6)
basis of a three-phase statistical model of simple sub- ⎝ πN ⎠ ⎢
⎣ π ⎥⎦
stances, which has been proposed in our earlier studies where c = co(1 – φv)1/3 is the distance between the cen-
[6–8]. ters of adjacent cells, kp is the packing coefficient of
the structure comprised of (N + Nv) spherical cells,
THREE-PHASE MODEL and co is the distance between the centers of adjacent
OF SIMPLE SUBSTANCES cells in the original (not relaxed into the state activated
by vacancies) vacancy-free (at Nv = 0) system (labeled
Let us assume that a macrosystem forms a lattice with the index “o”).
structure comprised of N + Nv spherical cells of the Reckoning that the center of a delocalized atom
same volume, among which Nv cells are vacant and N can migrate all over the system volume and the center
cells are occupied by identical spherically symmetric of a localized atom only within the cell in which it is
atoms, each with a mass equal to m. Let the system localized, the following expression for the specific (per
have a volume of V and a temperature of T; moreover, atom) Helmholtz free energy of a macroscopic system
no restrictions are imposed on the values of the spe- (in which N → ∞ and V → ∞ at N/V = const) was
cific volume (v = V/N) and T. Among N atoms, only obtained [6–8]:
some of the atoms (N – Nd) are localized in cells, while
the others (Nd) are delocalized; i.e., they can migrate fH = fi + fs + fw + fd. (7)
all over the volume V. Furthermore, a localized atom Here, fi is the specific free energy of the translational
can become delocalized when excited and vice versa. motion of delocalized atoms, which is equal to
Let atoms interact through the Mie–Lennard–
Jones potential: ⎡⎛ ⎞ 3/2
V ⎤,
fi = − xd kBT ⎢⎜ T ⎟ ⎥ (8)
⎡ r b r ⎤
a
⎣⎢⎝ Ad ⎠ V0 ⎦⎥
ϕ(r ) = D ⎢a ⎛⎜ o ⎞⎟ − b ⎛⎜ o ⎞⎟ ⎥ , (2)
(b − a) ⎣ ⎝ r ⎠ ⎝r⎠ ⎦ where
where D and ro are the depth and coordinate of the 2/3
⎛ ⎞⎛ ⎞ ⎛ ⎞
Ad = ⎜ 2π ⎟ ⎜ N ⎟ and V0 = N ⎜ π ⎟ ro3 .
2
potential minimum and b > a > 1 are the parameters.
If the lattice contains Nv vacancies uniformly dis- ⎝ mkB ⎠ ⎝ eV0 ⎠ ⎝ 6k p ⎠
tributed over the volume, then the first coordination Here,  is the Planck constant, e = 2.718, and xd is the
number (i.e., the number of the nearest neighbor fraction of delocalized atoms, i.e., the probability that
atoms) is equal to an atom has a kinetic energy higher than atom-delo-
calization energy Ed [6, 8, 11, 12], expressed as follows:
knoN
kn = = kn (1 − φv ),
o
(3) ∞
N + Nv N 2
xd = d = 1/2 ∫
t exp(−t )dt .
1/2
(9)
where kno is the number of cells (both occupied and N π Ed /(kBT )
vacant) closest to the given atom; i.e., this is the first The second summand in Eq. (7) is the specific free
coordination number at Nv = 0, and φv is the probabil- energy of the static interaction of all atoms with each
ity of a vacancy creature in the lattice of a simple sub- other, which takes the following form when the “inter-
stance [6, 9, 10]: action of only the nearest neighbors” approximation is
∞ applied:
Nv 2
φv = = ∫
N + N v π1/2 (E /k T )1/2
exp( −t 2 )dt, (4) ⎛ko ⎞
f s = ⎜ n ⎟ (1 − φv )DU o(R) , (10)
v B
⎝2⎠
where kB is the Boltzmann constant and Ev is the where R = ro/c is the relative linear density of a crystal,
energy of creation of a vacant site in the lattice.
The system volume is equal to the sum of volumes U o(R) = 1 (aR b − bR a )
(va) per one cell (either occupied or vacant), the shape (b − a)
of which is considered spherical with taking into and
account the lattice structure:
1/3
v ro ⎡ V0 ⎤
V = π (N + N v )c3 = a N , R= = .
6k p k p (1 − φv ) c ⎢⎣V (1 − φv )⎥⎦
va = π c = V k p (1 − φv ) ,
3
(5) The third summand in Eq. (7) is the specific free
6 N energy of vibrational motion of localized atoms

NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8 2017

418 MAGOMEDOV

according to the Einstein crystal model (ΘE is the Ein- Here, Θ is the Debye temperature, which is connected
stein temperature): with the Einstein temperature by the following rela-

{ }
Θ Θ tionship: Θ = (4/3)ΘE [15]. The Θ function is
fw = 3(1 − xd )kBT 0.5 E + ln ⎡1 − exp ⎛⎜ − E ⎞⎤
⎟ . (11) expressed as follows [6, 13, 14]:
T ⎢
⎣ ⎝ T ⎠⎦⎥
The fd member is the free energy of dynamic inter- ⎡ ⎛ 8D ⎞ ⎤
1/2

action of delocalized atoms due to displacement from Θ = Awξ ⎢−1 + ⎜1 + 2⎟ ⎥, (18)

the centers of cells during migration throughout the ⎢⎣ ⎝ kB Awξ ⎠ ⎥⎦
volume. Within the “interaction of only the nearest where
neighbors” approximation, the following expression
b+2
was obtained for this function [6–8]: 5k ab(b + 1) ⎛ ro ⎞
ξ = 9o , Aw = K R n ⎜ ⎟ , and
xd kn (1 − φv )abD
o
kn 144(b − a) ⎝ c ⎠
fd =
2(b − a) KR =  2 .
2
(19)
(12)
⎧ ⎡l (b, ξ p )⎤ ⎡l (a, ξ p )⎤⎫ kBro m
× ⎨R b ⎢ 3 ⎥ − Ra ⎢ 3 ⎬,
⎩ ⎣ b ⎦ ⎣ a ⎥⎦⎭ It has been shown that such definitions of the Ev,
where Ed, and ΘE functions allow one to achieve a good
k −2 k −2 agreement of the calculated results with the experi-
(1 + t ) − (1 − t )
l3(k, t ) = −1 mental data. Moreover, the use of the Ev, Ed, and
2(k − 2)t(1 − t 2 )k −2 ΘE functions included in Eq. (7) in the form of
and (13) Eqs. (14)–(19) leads both to the fulfillment of the
0.5 Third Principle of thermodynamics in the “strong”
ξp = < 0.5. formulation of Max Planck and to the consistency of
31/2 k1/3
p the thermal and caloric equations of state for a macro-
The Ev, E d, and ΘE functions included in scopic system. It has also been shown that the function
Eqs. (3)–(12) were defined for a macroscopic system given by Eq. (7) transforms into the expressions for a
in [6, 9–14]. The energy of formation of a vacancy in gas (at xd = 1) or crystal (at xd = 0) fully considering
the structure of a simple substance is defined as fol- “communal entropy.” Let us note that reckoning the
lows [6, 9, 10]: “communal entropy” in the approaches developed by
EL other authors towards the creation of a three-phase
Ev = , (14) model remains problematic [16, 17].
1 + xd [(CD E L /kBT ) − 1]
where
RP MODEL OF A NANOCRYSTAL
ck Θ Θ
2
4kno
E L = mo ⎛⎜ o B Eo ⎞⎟ ⋅ f y ⎛⎜ Eo ⎞⎟ , CD = 2/3 , WITH VACANCIES
kn ⎝ 2 ⎠ ⎝ T ⎠ 3k p Previously we developed the Rectangular Paral-
[1 − exp(− yw )] lelepiped (RP) model of a nanocrystal without vacan-
f y ( yw ) = 2 . (15) cies or the delocalized atoms [6, 18, 19]. A number of
yw [1 + exp(− yw )]
new results that still cannot be obtained in the com-
The index “o” in the co and ΘEo functions means puter models (like the Monte Carlo or molecular
that the functions are calculated for the original (not dynamics models) were obtained within this RP
been relaxed into the state activated by vacancies) model. For example, we have obtained the tempera-
vacancy-free virtual lattice (at φv = 0). ture dependences of the surface energy and surface
The following expression was obtained for the pressure, the construction of which in computer mod-
delocalization energy of an atom [6, 11, 12]: els is problematic.
2 Let us extend the formalism from Eqs. (3)–(19) to
⎛c k Θ ⎞
Ed = ⎛⎜ 3 2 ⎞⎟ ⋅ m ⋅ ⎜⎜ o B1/3Eo ⎟⎟ ⋅ f y ( ywo ) the case of a nanocrystal comprised of N + Nv =
⎝ 8π ⎠ ⎝ k p ⎠
(16) N/(1 – φv) identical cells, among which Nv vacant cells
= Ed1 ⋅ f y ( ywo ) = Cld ⋅ E L, are uniformly distributed over the nanocrystal volume.
Like in [6, 18, 19], we assume that a nanocrystal with
where the free surface takes the shape of a rectangular paral-
o lelepiped with a square base and is limited by faces of
= ⎛⎜ 9 2 ⎞⎟ CD
3kn
Cld = the (100) type with the Gibbs geometric surface. The
2π k p
2 2/3
⎝ 8π ⎠ cells form a crystalline structure with the packing
and (17) coefficient equal to kp. The value f = Nps/Npo =
o o
⎛ 3cokBΘo ⎞ .
2
N ps / N po is the shape parameter, which is determined
Ed1 = Ed ( f y ( ywo ) = 1) = 3m2/3 ⎜ ⎟
8k p ⎝ 4π ⎠ o
by the ratio of the number N ps of atoms (or the number

NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8 2017

 DEPENDENCES OF THE VACANCY CONCENTRATION 419

o
Nps = N ps /(1 – φv)1/3 of cells) on the side edge to the The shape function Zs( f ) = (1 + 2f )/(3f 2/3) reaches
o o
the minimum value equal to one at f = 1, i.e., for a
number N po of atoms (or the number Npo = N po /(1 – cubic shape. For platelike ( f < 1) or rodlike ( f > 1)
φv) of cells) on the edge of the square base. For a
1/3
shapes, the shape function value is greater than one;
rodlike shape, f > 1; for a cube, f = 1; and, for a plate- i.e., Zs( f ≠ 1) > 1. Therefore, the kn( f )* function at any
like shape, f < 1. N + Nv value has a maximum at f = 1, i.e., for the most
The number of cells and atoms in a nanocrystal of energetically stable cubic shape of a rectangular paral-
a specific shape (at given f ) is equal to lelepiped.
The volume and surface area are expressed as fol-
N 3po (N 3po )3 o 3
(N po ) lows within the generalized RP(vac) model:
N + Nv = f = f , N = f , (20)
α α(1 − φv ) α V = (N po ) fc /(1 − φv )
o 2 3

where α = π/(6kp) is the structure parameter. = (N + N v )αc3 = (N po

It is obvious that shape parameter f may vary
within the following limits:
2 (N + N v )α
≤ f ≤ , (21)
[(N + N v )α/2]1/2
8 packing density of atoms on the face (i.e., on the Gibbs
where the left value refers to a plate and the right value
to a rod of two cells in thickness.
The limitation of a system by the surface will give
rise to the breakage of bonds in the boundary. There-
fore, if the “interaction of only the nearest neighbors”
approximation is used, then the mean 〈kn〉 value (aver-
aged over the entire nanosystem) of the coordination
number, which will depend on both the size and shape
of a nanosystem, should be taken rather than the first
coordination number (kn). Furthermore, the system
structure is believed to be unchanged; i.e., kp = const.
This model of a nanocrystal in the form of a rectangu-
lar parallelepiped, whose shape can be varied using the
shape parameter f, is called here the RP(vac) model.
Within the RP(vac) model, the variation of the
normalized mean value of the first coordination num-
ber upon the variation of N + Nv and f is described by
the following expression [6, 18, 19]:
〈kn(N + N v , f )〉 ⎛ ⎞
= 1 − Z s( f ) ⎜ α
2
kn* = ⎟
kn(∞) ⎝ N + Nv ⎠
1/3
(22)
⎛ α2 ⎞
= 1 − Z s ( f ) ⎜ (1 − φv ) ⎟ ,
⎝N ⎠ the kn(N) value, the dependence of these properties
where kn(∞) is the first coordination number (i.e., the
number of occupied cells nearest to the given atom)
expressed as follows for a macroscopic crystal:
kno(∞)N
kn(∞) = = kn (∞) ⋅ (1 − φv ).
o
N + Nv
Therefore, the dependence of the first coordina-
tion number on N + Nv and f takes the following form:
kn(N , φv )
⎡ ⎛ 2
1/3
⎞ ⎤ (24)
= kno(∞) ⋅ (1 − φv ) ⎢1 − Z s ( f ) ⋅ ⎜ α (1 − φv ) ⎟
⎢⎣ ⎝N ⎠ ⎥⎦
o 3
) fco3,
(25)
Σ = 6c α s[(N + N v )α] Z s ( f )
2 2/3
= 6co α s (N α) Z s ( f ),
2 2/3
where αs is the coefficient that takes into account the
geometric surface) of a nanocrystal (αs ≅ α2/3). It is
easy to see that the nanocrystal volume depends on the
system shape only through the dependence of the co
value or c = co(1 – φv)1/3 on the size and shape of a
nanocrystal.
The cubic shape can be achieved only when there is
a certain number of cells from which a cube can be
o 3
built, namely, (N + Nv)cub = (N po ) /[α(1 – φv)], where
o
N po = 2, 3, 4, … . When the number of cells is “non-
cubic,” i.e., N + Nv ≠ (N + Nv)cub, the parallelepiped
may have either a platelike or rodlike shape upon ful-
filling the following inequality:
kn((N + Nv)cub ± 1)* < kn(N + Nv)cub*.
Therefore, the isomorphic (i.e., calculated at f =
const) kn(N + Nv) dependence monotonically
decreases at N + Nv → Nmin = 8/[α(1 – φv)], but the
overall kn(N + Nv) dependence has an oscillating shape
1/3 with the maxima at the kn(N + Nv)cub points corre-
sponding to nanocrystals with a cubic shape and with
the minima at such N + Nv ≠ (N + Nv)cub values from
which only a rod can be built. Considering that many
properties of a nanocrystal are determined namely by
from the N value should also have an oscillating shape.
Thus, the use of Eq. (24), along with the formalism
given by Eqs. (3)–(19) at kp = const allows one to
obtain the dependences of the Helmholtz free energy
both on the size (on the number N of atoms) and shape
(23) of a nanocrystal at given values of the temperature and
specific volume, v = V/N, for the nanocrystal with the
free Gibbs geometric surface.
CALCULATION RESULTS
FOR THE BCC LATTICE OF IRON
⎥. Let us take an iron crystal (m(Fe) = 55.847 a.m.u.)
having a body-centered cubic (BCC) structure with

NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8 2017

420 MAGOMEDOV
the parameters kno (∞) = 8 and kp = 0.6802 for the cal-
culations. The parameters of the Mie–Lennard–
Jones paired interatomic potential given by Eq. (2) for
the BCC iron lattice were determined and tested in the
calculation of the properties of a macroscopic crystal
in [20]. They are as follows:
ro = 2.4775 × 10–10 m, D/kB = 12576.7 K,
a = 2.95, and b = 8.26. (26)
We have compared the thermodynamic parameters
calculated by the method used in the present studies
with the experimental assessments and with the results
of computer simulations in [6, 8, 9, 11] both for the
formation process of electrically neutral vacancies and
for the process of self-diffusion of atoms in a BCC iron
macrocrystal.
It should be noted that the size characteristics of
properties of a BCC iron nanocrystal were studied in
[19] within the RP model of the ideal nanocrystal,
which did not contain lattice vacancies or atoms capa-
ble of migrating all over the volume. The size proper-
ties of a BCC iron nanocrystal in the present studies
are calculated for the generalized RP(vac) model,
which takes into account the presence of both vacan-
cies and delocalized atoms in the structure.
Based on Eq. (26), the model parameters at v/vo =
(co/ro)3 = 1 are as follows:
vo = [π/(6kp)] ro3 = 7.0494 cm3/mol, KR = 0.1415 K,
Aw(1) = 1.6703 K, Xw(1) = 4.605 × 10–3,
ΘE(1) = 306.055 K, Θ(1) = 408.073 K,
γ(1) = 1.702, and q(1) = 7.874 × 10–3,
where, in accordance with Eq. (18), the first (γ) and
second (q) Grünaisen parameters, as well as the Xw
function characterizing the role of quantum effects,
are equal to
( )
γ = − ∂ ln Θ = b + 2 ,
∂ ln V T 6(1 + X w )
( )
∂ ln γ X (1 + 2 X w ) Aξ
q= =γ w , Xw = w . (27)
∂ ln V T (1 + X w ) Θ upon the isomeric–isothermal deviation of the nano-
The calculations were performed for the two iso-
therms: at a low (T/Θ < 1) temperature of T = 300 K
and at a high (T/Θ > 1) temperature of T = 1000 K.
The isomorphic–isomeric (at constant f and N val-
ues) dependences of the pressure on the normalized
volume (v/vo) in a cubic ( f = 1) BCC-Fe nanocrystal
are shown in Fig. 1. In [20], the P(v/vo, T) dependence
calculated in such a way has been compared with the
experimental data for a BCC-Fe macroscopic crystal;
therefore, we do not repeat this again. The solid lines
in Fig. 1 are obtained for a macrocrystal at Npo = 106,
i.e., at N = 1.3 × 1018, and the dashed lines for a nano-
crystal at N = 665, i.e., at Npo = 8. The lower two lines
are isotherms obtained at T = 300 K, and the upper
two lines are isotherms obtained at T = 1000 K.
NANOTECHNOLOGIES IN RUSSIA
P, kbar bcc-Fe
300
250 macro
200
150 nano
T = 1000 K
100
50
T = 300 K
0
−50
0.90 0.95 1.00 1.05
v/vo
Fig. 1. (Color online) Isomorphic–isomeric dependences
of the pressure on the normalized volume in a cubic
BCC-Fe nanocrystal; the solid lines are obtained for a
macroscopic crystal and the dotted lines for a nanocrystal
at N = 665; the two lower lines are isotherms obtained at
T = 300 K and the two upper lines are isotherms obtained
at T = 1000 K.
A lower increase in the pressure upon the transition
from a macroscopic to a nanoscopic crystal points to a
decrease in the elasticity modulus, B = –V(∂P/∂V)T,
for the nanocrystal, as has already been shown by
other methods in [19, 21].
As is seen from Fig. 1, the P(v/vo) dependences for
a nanocrystal and macrocrystal intersect at a certain
value of the relative volume (v/vo)0. Thus, the surface
pressure (Psf(v) = P(v)macro – P(v)nano) equals zero
(Psf(v/vo)0 = 0) at (v/vo)0. The surface pressure com-
presses (Psf > 0) a nanocrystal at v/vo < (v/vo)0 and
stretches (Psf < 0) it at v/vo > (v/vo)0. The (v/vo)0 value
decreases both upon an isomorphic–isomeric
increase in the temperature T and upon an isomor-
phic–isothermal decrease in the N value, as well as
crystal shape from the most optimal shape (the cubic
shape for the RP models). As also follows from Fig. 1,
the pressure in a nanocrystal passes through the zero
value at a v/vo value that is larger than in a macrocrystal.
The isomorphic–isomeric baric dependences are
shown in Fig. 2a for the logarithm of the vacancy for-
mation probability and in Fig. 2b for the vacancy for-
mation thermodynamic potential (or Gibbs free
energy), gv = – kBT ln(φv) (gv in eV). The solid lines
show the data for a macroscopic crystal (at N = 1.3 ×
1018) and the dotted lines for a cube-shaped nanocrys-
tal (at N = 665 and f = 1). As is seen from Fig. 2, the
baric dependences of the log(φv) and gv values for a
macrocrystal and a nanocrystal intersect at low pres-
sures. The isotherms obtained at T = 300 K (lower
Vol. 12 Nos. 7–8 2017
 DEPENDENCES OF THE VACANCY CONCENTRATION
bcc-Fe
log φv
(a)
−20 T = 1000 K
macro
−40 nano
T = 300 K
−60
macro
−80
nano
0 1000 2000 3000 4000
P, kbar
Fig. 2. (Color online) Isomorphic–isomeric baric dependences (a) of the decimal logarithm of the formation probability of a
vacancy and (b) of the thermodynamic formation potential of a vacancy.
lines) intersect at the point with Pv = –20.07 kbar,
log(φv) = –26.993, and gv = 1.605 eV, i.e., when a crys-
tal is stretched in every direction. The isotherms
obtained at T = 1000 K (upper lines) intersect at the
point with Pv = 53.43 kbar, log(φv) = –9.72, and gv =
1.928 eV, i.e., in the state of a hydrostatically com-
pressed crystal. If P < Pv, then φv(N) > φv(N = ∞), and
φv(N) < φv(N = ∞) at P > Pv. This means that the
vacancy formation probability upon an isothermal
increase in the pressure decreases in a nanocrystal
more appreciably than in a macrocrystal and the
vacancy formation probability in a nanocrystal at
P > Pv becomes lower than in a macrocrystal. The Pv
value increases with an increase in the temperature
and transits from the area of stretching (where Pv < 0)
to the area of compression. It also follows from the
gv(P) dependence that the vacancy formation volume,
vv = (∂gv/∂P)T, increases upon a decrease in the size of
a nanocrystal.
The isothermal–isomorphic dependences of the
logarithm of the vacancy formation probability on the
number of atoms (N) in a cubic BCC-Fe nanocrystal
are shown in Fig. 3. The dependences obtained at the
atmospheric pressure (P = 1 bar) are shown (Fig. 3a)
and obtained at P = 100 kbar (Fig. 3b). The line with
cubes is an isotherm obtained at T = 300 K and the
line with circles is an isotherm obtained at T = 1000 K.
The symbols on the isomorphs (here and hereinafter)
indicate the positions of the permissible values for the
o 3
number of atoms, specifically, N = f ⋅ (N po ) /α in a
o
cubic-shaped nanocrystal (i.e., at f = 1 and N po = 2, 3,
4, …) at the uniform (i.e., equilibrium) distribution of
vacancies and migrating atoms over the volume and
NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8
421
gv, eV
2.0 (b)
1.9 T = 1000 K
nano
1.8 macro
macro
nano
1.7
T = 300 K
1.6
−20 0 20 40 60
P, kbar
surface of a nanocrystal with the Gibbs geometrical
surface.
As follows from Fig. 3, an isothermal–isobaric
increase in the N value at the atmospheric pressure
gives rise to an increase in the φv(N) function at T =
300 K and to its decrease at the T = 1000 K function.
That is, a small nanocrystal contains fewer vacancies
per atom at T = 300 K in comparison with a macro-
crystal. However, dispersing a macrocrystal at T =
1000 K leads to an increase in the vacancy-formation
probability. Under the condition of the isothermal–
isobaric growth of a nanocrystal at a pressure of
100 kbar, the φv(N) increases upon an increase in the
N value at both 300 and 1000 K. That is, a small nano-
crystal grown at 100 kbar contains fewer vacancies per
atom in comparison with a macrocrystal at both T =
300 and 1000 K. Therefore, more perfect nanocrystals
are formed either at low pressures and low tempera-
tures or at high static pressures and any temperatures.
As is seen from the left diagram in Fig. 3, a certain
temperature should exist at P = 1 bar within the range
300 K < Tnv < 1000 K, at which the φv(N) function
does not depend on the N value. As follows from
Fig. 3, the Tnv value increases with an increase in the
pressure value, and the following inequality is satisfied
at 100 kbar: Tnv > 1000 K.
The calculations showed that the isothermal-iso-
morphic dependences of the delocalization probabil-
ity of an atom (xd) and of the self-diffusion coefficient
(Df ∼ xd) monotonically decrease upon an increase in
the pressure. Furthermore, as is seen from Fig. 4a, the
log(xd) dependences (the same as the log Df [cm2/s]
dependences) for a macrocrystal (at N = 1.3 × 1018, the
solid lines) and a cubic-shaped nanocrystal (at N = 665
2017
422 MAGOMEDOV

log φv P = 1 bar log φv P = 100 kbar

−31.5 −10.4
−27.95 (a) (b)
−8.3 T = 1000 K
−32.0
−28.00 −8.4 −10.5
T = 300 K −32.5
−28.05 −8.5 −33.0 −10.6
−28.10 bcc-Fe
−8.6 −33.5
T = 300 K −10.7
−28.15 T = 1000 K −8.7 −34.0
−34.5 −10.8
−28.20 −8.8
−35.0
1.0 1.5 2.0 2.5 3.0 3.5 4.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0
log N log N

Fig. 3. (Color online) Isomorphic–isothermal dependences of the logarithm of the formation probability of a vacancy on the
number of atoms (N) in a cubic BCC-Fe nanocrystal; the dependences obtained at the atmospheric pressure (P = 1 bar) are
shown on the left and at P = 100 kbar on the right; the line with cubes designates an isotherm obtained at T = 300 K and the line
with circles an isotherm obtained at T = 1000 K.

log xd bcc-Fe gd, eV

0 3.8 T = 1000 K
(a) (b)
−20 3.6
−40 3.4

−60 T = 1000 K nano 3.2 T = 300 K

macro 3.0 macro
−80
2.8
−100 nano
2.6
−120
T = 300 K 2.4
−140 nano
macro 2.2
0 2000 4000 6000 0 100 200 300 400
P, kbar P, kbar

Fig. 4. (Color online) Isomorphic–isomeric baric dependences of the (a) logxd and (b) gd functions; the solid lines show the
results for a macroscopic crystal and the dotted lines for a cube-shaped nanocrystal.

and f = 1, the dotted lines) do intersect within the
studied range of pressures. The two lower lines in the
Fig. 4a correspond to isotherms obtained at 300 K and
the two upper lines to isotherms obtained at 1000 K.
Figure 4b shows the baric dependences of the thermo-
dynamic potential (or the Gibbs free energy) of delo-
calization of an atom, which is expressed as follows:
gd = – kBT ln(xd), in eV. In the right diagram, the solid
and dotted lines are isotherms obtained at 300 K and
the dashed and dashed–dotted lines are isotherms
obtained at 1000 K for macroscopic and nanoscopic
crystals, respectively. As is seen from Fig. 4, the frac-
tion of delocalized atoms increases with a decrease in
the size of a nanocrystal at any pressure or tempera-
ture. This is consistent with the results obtained by
other authors [4, 22, 23].
However, as is seen form Fig. 4b, the baric depen-
dences of the thermodynamic potential of delocaliza-
tion of an atom intersect in the case of isotherms
obtained at 300 and 1000 K. For a macrocrystal, the
gd(P) isotherms intersect in the following point: Pt =
124.02 kbar and gd = 2.888 eV; for a nanocrystal, they
intersect in the point with Pt = 173.77 kbar and gd =
2.818 eV. That is, the gd(T) function decreases upon
isobaric heating at low pressures (P < Pt), and it
increases with an increase in the temperature at high
pressures (P > Pt). This means that the atom delocal-

NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8 2017

 DEPENDENCES OF THE VACANCY CONCENTRATION 423

log xd P = 1 bar log xd P = 100 kbar

(a) (b)
T = 1000 K −10 T = 1000 K
−10
−15
−15
−20
−20 −25
bcc-Fe
−25 −30

−30 −35

−35 −40
T = 300 K T = 300 K
−45
−40
1.0 1.5 2.0 2.5 3.0 3.5 4.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0
log N log N

Fig. 5. (Color online) Isomorphic–isothermal dependences of the logarithm of the delocalization probability of an atom on the
number of atoms (N) in a cubic BCC-Fe nanocrystal; the dependences obtained at the atmospheric pressure P = (a) 1 bar,
(b) 100 kbar. The line with cubes corresponds to an isotherm obtained at T = 300 K and the line with circles to an isotherm
obtained at T = 1000 K.

log(φv /xd) bcc-Fe gm, eV

(a) 2.0 (b)
50 T = 300 K macro T = 1000 K

40 1.5 macro
nano
T = 300 K
30 nano

T = 1000 K 1.0
20 macro
nano
10
0.5

0 2000 4000 6000 0 500 1000 1500 2000

P, kbar P, kbar

Fig. 6. (Color online) Baric dependences of the (a) log(φv/xd) and (b) gm = kBTln(φv/xd) functions; the solid lines show the cal-
culation results for a BCC-Fe macroscopic crystal and the dotted lines for a cubic-shaped nanocrystal (at N = 665 and f = 1).

ization entropy, sd = – (∂gd/∂T)P, is positive at P < Pt
and negative (sd < 0) at P > Pt. Moreover, the Pt(N)
value increases with an isomorphic decrease in the size
of a nanocrystal. It also follows from the gd(P) depen-
dence that the self-diffusion volume, vd = (∂gv/∂P)T,
increases with an increase in the temperature for both
macroscopic and nanoscopic crystals.
The isomorphic–isothermal dependences of the
logarithm of the delocalization probability of an atom
on the number of atoms (N) in a BCC-Fe nanocrystal
are shown in Fig. 5. Figure 5a shows the dependences
obtained at the atmospheric pressure (P = 1 bar), and
Fig. 5b shows the dependences at P = 100 kbar. The
line with cubes corresponds to an isotherm obtained at
T = 300 K and the line with circles to an isotherm
obtained at T = 1000 K. As is seen from Fig. 5, the
delocalization probability of an atom (the same as the
self-diffusion coefficient) increases upon a decrease in
the size of a crystal under any P–T conditions.
In Fig. 6, the baric dependences of the log(φv/xd)
function (Fig. 6a) and of the thermodynamic potential
(or the Gibbs free energy), gm = gd – gv = kBTln(φv/xd),
of the migration of an atom (Fig. 6b) are shown. The
solid lines are obtained for a macrocrystal (at N =
1.3 × 1018) and the dotted lines for a cubic-shaped
nanocrystal (at N = 665 and f = 1). As is seen from

NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8 2017

424 MAGOMEDOV

log(φv /xd) P = 1 bar log(φv /xd) P = 100 kbar

12 (a) 14 (b)
10 T = 300 K 12 T = 300 K
8 10
8 bcc-Fe
6
6
4
4
2
T = 1000 K 2 T = 1000 K
0 0
−2 −2
−4 −4
1.0 1.5 2.0 2.5 3.0 3.5 4.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0
log N log N

Fig. 7. (Color online) Isomorphic–isothermal dependences of the logarithm of the ratio of the formation probability of a vacancy
to the delocalization probability of an atom on the number of atoms (N) in a cubic BCC-Fe nanocrystal; the dependencies
obtained at the atmospheric pressure P = (a) 1 bar and (b) 100 kbar; the line with cubes corresponds to an isotherm obtained at
300 K and the line with circles to an isotherm obtained at 1000 K.

Fig. 6, the lower the temperature is, the stronger the
increase in the log(φv/xd) function is upon an increase
in the pressure and the more appreciable its decrease
is with a decrease in the size of a nanocrystal. The
baric gm(P) dependences corresponding to isotherms
obtained T = 300 and 1000 K intersect at the point
with Pm = 874.2 kbar and gm = 1.554 eV for a macro-
crystal and at the point with Pm = 1228.2 kbar and gm =
1.329 eV for a nanocrystal. That is, the gm function
decreases upon the isobaric heating at P < Pm and
increases with an increase in the temperature at
P > Pm. This means that the migration entropy, sm =
–(∂gm/∂T)P, is positive at P < Pm but negative at P > Pm.
The Pm value increases upon an isomorphic decrease
in the size of a nanocrystal. As is seen from the gm(P)
dependence, the migration volume, vm = (∂gm/∂P)T,
increases upon an increase in the temperature or upon
a decrease in the size of a nanocrytsal.
The isomorphic–isothermal dependences of the
log(φv/xd) function on the number of atoms (N) in a
cubic BCC-Fe nanocrystal are shown in Fig. 7. Fig-
ure 7a shows the dependencies obtained at P = 1 bar
and Fig. 7b at P = 100 kbar. The line with cubes corre-
sponds to an isotherm obtained at 300 K and the line
with circles to an isotherm obtained at 1000 K. As is
seen from Fig. 7, the out-of-vacancy self-diffusion [6],
in which the number of diffusing atoms is greater than
the number of vacant sites in the nanocrystal lattice,
i.e., xd > φv and gm < 0, is observed at a certain size (Nsd)
of a nanocrystal, which increases upon an increase in
the temperature (Nsd = 19–21 at T = 300 K and Nsd =
37–49 at T = 1000 K). It is easy to realize that the out-
of-vacancy self-diffusion is a consequence of enhanc-
ing the role of surface effects over bulk effects.
CONCLUSIONS
It is shown that the formation probability of vacan-
cies (φv) decreases more appreciably in a nanocrystal
than in a macrocrystal upon an isothermal increase in
the pressure, and the φv function for a nanocrystal
comprised of N atoms at a certain pressure P > Pv(N)
becomes smaller than for a macroscopic crystal. The
Pv(N) value increases upon an increase in the tem-
perature and transits from the area of stretching (in
which Pv < 0) into the area of compression. The
vacancy formation volume increases upon a decrease
in the size of a nanocrystal.
Under the condition of the isothermal–isobaric
growth of a nanocrystal at P = 1 bar, the φv(N) func-
tion increases upon an increase in the number of
atoms (N) at 300 K and the φv(N) function decreases
with an increase in the N value along the isotherm at
1000 K. That is, a nanocrystal contains fewer vacan-
cies per atom at P = 1 bar and T = 300 K than a mac-
roscopic crystal. However, dispersing a macrocrystal
at P = 1 bar and T = 1000 K increases the formation
probability of a vacancy. Under the condition of the
isothermal–isobaric growth of a nanocrystal at a pres-
sure of 100 kbar, the φv(N) function increases with an
increase in the N value at both 300 and 1000 K. That
is, a nanocrystal obtained at 100 kbar contains fewer
vacancies per atom at both 300 and 1000 K in compar-
ison with a macroscopic crystal. Therefore, a certain
temperature should exist at P = 1 bar within the range
300 K < Tnv < 1000 K, at which the φv(N) function does
not depend on the N number. With an increase in the
pressure value, the Tnv value increases and at 100 kbar
reaches the value Tnv > 1000 K.

NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8 2017

 DEPENDENCES OF THE VACANCY CONCENTRATION
The probability xd of delocalization of an atom (the
same as the self-diffusion coefficient) increases upon
a decrease in the size of a nanocrystal at any pressure
and temperature. The entropy of self-diffusion is pos-
itive at low pressures (P < Pt) and negative at high pres-
sures (P > Pt). Furthermore, the Pt(N) increases upon
an isomorphic decrease in the size of a nanocrystal.
The self-diffusion volume for both macroscopic and
nanoscopic crystals increases upon an increase in the
temperature.
The lower the temperature is, the more appreciable
the decrease in the φv/xd ratio is upon an isothermal–
isobaric decrease in the size of a nanocrystal. Out-of-
vacancy self-diffusion is observed at a certain size
(Nsd) of a nanocrystal [6], at which the number of
delocalized atoms is larger than the number of vacant
sites in the nanocrystal lattice. The Nsd value increases
with an increase in the temperature, so Nsd = 19–21 at
300 K and Nsd = 37–49 at 1000 K. The migration
entropy is positive at low pressures (P < Pm) but nega-
tive at high pressures (P > Pm). The Pm value increases
upon an isomorphic decrease in the size of a nano-
crystal. The migration volume increases with an
increase in the temperature or with a decrease in the
N value.
As has been shown in [6, 18, 19], the size depen-
dences of the nanocrystal lattice properties become
stronger when the nanocrystal shape deviates from the
most optimal shape (from the cubic shape for the RP
model). Therefore, the size dependences obtained in
the present work for a cube-shaped nanocrystal ( f = 1)
are weakest; i.e., any deviation (toward any direction)
of the shape parameter from one ( f ≠ 1) will give rise
to stronger size dependences of the properties studied
in the present work.
ACKNOWLEDGMENTS
I am grateful to E.N. Akhmedov, N.Sh. Gazanova,
Z.M. Surkhaeva, and M.M. Gadzhieva for fruitful
discussions and help. This work was supported by the
Russian Foundation For Basic Research (project
no. 16-03-00041_a) and by the Program of Presidium of
Russian Academy of Sciences (program no. I.11P(1)).
REFERENCES
1. C. C. Yang and S. Li, “Investigation of cohesive energy
effects on size-dependent physical and chemical prop-
erties of nanocrystals,” Phys. Rev. B 75, 165413 (2007).
2. M. A. Shandiz, “Effective coordination number model
for the size dependency of physical properties of nano-
crystals,” J. Phys.: Condens. Matter 20, 325237 (2008).
3. G. Guisbiers, “Schottky defects in nanoparticles,”
J. Phys. Chem. C 115, 2616–2621 (2011).
4. X. Yu and Z. Zhan, “The effects of the size of nanocrys-
talline materials on their thermodynamic and mechan-
ical properties,” Nanoscale Res. Lett. 9, 1–6 (2014).
NANOTECHNOLOGIES IN RUSSIA Vol. 12 Nos. 7–8
View publication stats
425
5. T. N. Zvonareva, A. A. Sitnikova, G. S. Frolova, and
V. I. Ivanov-Omskii, “Platinum nanoclusters encapsu-
lated in amorphous carbon,” Semiconductors 42, 325
(2008).
6. M. N. Magomedov, Study of Interatomic Interaction,
Vacancy Formation, and Self-Diffusion in Crystals
(Fizmatlit, Moscow, 2010) [in Russian].
7. M. N. Magomedov, “On the equation of simple sub-
stance state, describing three-phase equilibrium,”
Vestn. Mosk. Tekh. Univ. im. N.E. Baumana, Ser.:
Estestv. Nauki, No. 2, 28–42 (2013).
8. M. N. Magomedov, “On self-diffusion in iron at very
strong compression of crystal,” Phys. Met. Metallogr.
114, 207 (2013).
9. M. N. Magomedov, “On the probability of vacancy for-
mation,” Teplofiz. Vys. Temp. 27, 279–281 (1989).
10. M. N. Magomedov, “Parameters of vacancies’ forma-
tion in the carbon subgroup crystals,” Semiconductors
42, 1133 (2008).
11. M. N. Magomedov, “Calculation of the activation
energy for bulk self-diffusion in a simple substance,”
Phys. Met. Metallogr. 74, 319–321 (1992).
12. M. N. Magomedov, “Self-diffusion parameters in car-
bon-subgroup crystals,” Semiconductors 44, 271
(2010).
13. M. N. Magomedov, “On the determination of the
Debye temperature from experimental data,” Phys.
Solid State 45, 32 (2003).
14. M. N. Magomedov, “On the Debye temperature and
Gruneisen parameters for hexagonal close-packed
crystals consisting of p-H2 and o-D2 molecules,” Tech.
Phys. 58, 1297 (2013).
15. L. A. Girifalco, Statistical Physics of Materials (Wiley,
New York, 1973).
16. J. O. Hirschfelder, Ch. F. Curtiss and R. B. Bird,
Molecular Theory of Gases and Liquids (Wiley, New
York, 1954; Inostr. Liter., Moscow, 1961).
17. S. M. Stishov, “The thermodynamics of melting of
simple substances,” Sov. Phys. Usp. 17, 625 (1975).
18. M. N. Magomedov, “Dependence of the surface
energy on the size and shape of a nanocrystal,” Phys.
Solid State 46, 954 (2004).
19. M. N. Magomedov, “On dependences of the thermo-
elastic properties on size and shape of an iron nanocrys-
tal,” Nanotechnol. Russ. 10, 89 (2015).
20. M. N. Magomedov, “Change in the thermophysical
properties of BCC iron during isothermal compres-
sion,” Tech. Phys. 60, 1619 (2015).
21. G. Sharma and M. Kumar, “Effect of size with freedom
of pressure and temperature for nanomaterials,” Indian
J. Pure Appl. Phys. 54, 251–258 (2016).
22. Q. Jiang, S. H. Zhang, and J. C. Li, “Grain size-depen-
dent diffusion activation energy in nanomaterials,”
Solid State Commun. 130, 581–584 (2004).
23. G. Xiong, J. N. Clark, C. Nicklin, J. Rawle, and
I. K. Robinson, “Atomic diffusion within individual
gold nanocrystal,” Sci. Rep. 4, 6765 (2014).
Translated by O. Kadkin
2017