ISSN 1063-7761, Journal of Experimental and Theoretical Physics, 2006, Vol. 103, No. 5, pp. 768–774.

© Pleiades Publishing, Inc., 2006.

Original Russian Text © E.V. Zavitaev, A.A. Yushkanov, 2006, published in Zhurnal Éksperimental’noœ i Teoreticheskoœ Fiziki, 2006, Vol. 130, No. 5, pp. 887–894.

ELECTRONIC PROPERTIES
OF SOLIDS

Dependence of the Electric Conductivity

of a Thin Cylindrical Wire in a Longitudinal Magnetic Field
on the Character of Electron Reflection
E. V. Zavitaev and A. A. Yushkanov
Moscow State Regional University, Mytishchi-5, Moscow oblast, 141005 Russia
e-mail: [email protected]
Received May 15, 2006

Abstract—The electric conductivity of a cylindrical metal wire in a longitudinal magnetic field is analyzed
with regard to the specular–diffuse boundary condition for electrons. Based on the analytic solution obtained,
a theoretical interpretation is given to the experimental data on the magnetoresistance of a thin cylindrical wire.
An efficient method for determining the specular reflection coefficient of metal surfaces is proposed.
PACS numbers: 78.67.-n
DOI: 10.1134/S1063776106110124

1. INTRODUCTION gral conductivity as a function of the external magnetic

The electric properties of conductors with linear field, the ratio of the wire diameter to the mean free
dimension R comparable to the mean free path Λ of path of electrons, frequency, and the specular reflection
electrons are essentially different from those of bulk coefficient.
conductors [1, 2].
In the fundamental study [2], the author calculated 2. MATHEMATICAL MODEL
the electric conductivity of a thin cylindrical wire with AND CALCULATIONS
radius much smaller than the length. The conductivity
of a thin cylindrical wire in a magnetic field was dis- Consider a straight cylindrical wire made of a non-
cussed in [3–7]. In these papers, the authors applied a magnetic metal in a magnetic field with induction B.
method based on the solution of the Boltzmann kinetic Let R be the radius of the wire and L be its length;
equation for electrons in a metal. assume that L
R. Suppose that an ac voltage of fre-
In [4], the effect of a longitudinal magnetic field on quency ω is applied across the ends of the wire. In this
the electric conductivity of a cylindrical metal wire was case, the directions of the electric and magnetic fields
considered in the case of a purely diffuse boundary con- coincide with the symmetry axis of the wire. We
dition. The author of [4] could not obtain an analytic neglect the skin effect (assume that R < δ, where δ is the
expression for the conductivity; using numerical meth- skin depth).
ods, he calculated the conductivity only for certain spe-
A uniform time-periodic electric field
cific values of the parameters of the problem. More-
over, the results of [4] do not agree with experimental
– iωt
data. E = E0 e (1)
Quantum mechanical phenomena associated with
the interaction of electrons with an ac electromagnetic acts on the conduction electrons (which are considered
field were considered in [8]. to be a degenerate Fermi gas) inside the wire and results
Note that the problems concerning the conductivity in a deviation f1 of their distribution function f from the
of thin metal wires become especially important in equilibrium Fermi function f0:
view of the significant progress in microelectronics,
where such wires are widely used.
f ( r, v ) = f 0 ( ε ) + f 1 ( r, v ),
In the present paper, we apply the kinetic method to
calculate the distribution function describing the linear
response of electrons in a homogeneous cylindrical here, ε = mv2/2, where r is the radius vector (the origin
wire in a longitudinal magnetic field to the ac electric of coordinates is chosen to lie on the wire axis), v is the
field applied along the symmetry axis of the wire. electron velocity, and m is the electron effective mass in
Using this distribution function, we calculate the inte- the metal.

768
 DEPENDENCE OF THE ELECTRIC CONDUCTIVITY OF A THIN CYLINDRICAL WIRE 769

This field gives rise to a high-frequency current of is a velocity vector that turns into the vector v⊥ under
density the specular reflection of the wire from the internal sur-
face; and q is the specular reflection coefficient (the
2d ( mv )
3
m 3 probability of specular reflection), 0 ≤ q ≤ 1.
∫ - = 2e ⎛ ----⎞ v f 1 d v ,
∫
3
j = e vf --------------------- (2)
h
3 ⎝ h ⎠ When q = 0, we obtain the condition of diffuse
reflection of conduction electrons from the internal sur-
where e is the electron charge and h is the Planck con- face of a metal wire; when q = 1, we obtain the condi-
stant. Here, we use the standard normalization of the tion of purely specular reflection. When q ≠ 0 and q ≠ 1,
distribution function f, in which the density of electron we have different types of mixed (specular–diffuse)
states is equal to 2/h3. For the equilibrium function reflection of electrons.
f0(ε), we will use the following step-function approxi- We will solve the kinetic equation (3) subject to the
mation [9]: boundary condition (4) by the method of characteris-
tics [12].
⎧ 1, 0 ≤ ε ≤ εF , The boundary condition (4) allows one to trace the
f 0( ε ) = θ( εF – ε ) = ⎨ variation of the function f1 along a specularly reflected
⎩ 0, ε F < ε,
trajectory. At the point of reflection t' = tn , the function
f1(t') has a jump
where εF = m v F /2 is the Fermi energy (vF is the Fermi
2

velocity). It is assumed that the Fermi surface is sphe- f 1( tn + 0 ) = q f 1( tn – 0 ) (5)

rical.
The problem reduces to finding the deviation f1 , due (the signs + or – denote the right or left limit of the
to the high-frequency field (1), of the distribution func- function f1 at the reflection point tn with respect to the
tion of electrons from the equilibrium function f0 . In a transit time).
linear approximation in the electric field, the function f1 The difference tn – tn – 1 is independent of the num-
satisfies the kinetic equation [9–11] ber n of the reflection point:
∂f ∂f n ∈
,
– iω f 1 + v ⋅ --------1 + e ( v ⋅ E ) --------0 t n = nT + const,
∂r ∂ε
(3)
e ∂f f where T is the transit time of electrons with velocity v⊥
+ ---- [ v × B ] ⋅ --------1 = – -----1 , from the point rn – 1, ⊥ to the point rn, ⊥ .
m ∂v τ
A solution to Eq. (3) is given by the function
where the harmonic time-dependence f1 ∝ exp(–iωt ) is
assumed and the collision integral is taken in the – νt' A
approximation of electron relaxation time τ: f 1 = Ce – ---,
ν

⎛ d--------
f 1⎞ f
= – -----1 . where
⎝ dt ⎠ s τ
1 ∂f0
To determine the function f1 uniquely, one should ν = --- – iω, A = ev z E z --------.
τ ∂ε
define boundary conditions for this function on the
cylindrical surface of the wire. As such conditions, we
take the condition of specular–diffuse reflection of Let us solve this equation on the interval (tn – 1, tn).
electrons from this surface [2]: At the initial moment (t' = 0), we have

f 1 ( r ⊥, v ⊥, v z ) = q f 1 ( r ⊥, v '⊥, v z ) (4) A
f 1 ( t n – 1 + 0 ) = C – ---, (6)
ν
with
whence we determine the value of the constant C:
r ⊥ = R, r ⊥ ⋅ v ⊥ < 0,
A
C = --- + f 1 ( t n – 1 + 0 ).
where r⊥ and v⊥ are the components of the radius vector ν
r and the velocity v of an electron in the plane perpen-
dicular to the symmetry axis of the wire; vz is the com- Now, we obtain a relation between the initial values
ponent of the electron velocity along the wire axis; v '⊥ of the function f1 on two adjacent segments of the tra-

JOURNAL OF EXPERIMENTAL AND THEORETICAL PHYSICS Vol. 103 No. 5 2006

770 ZAVITAEV, YUSHKANOV

jectory. Since tn – 0 = tn – 1 + T, we obtain Consider the projection of the electron trajectory

onto the plane perpendicular to the wire axis. This pro-
A – νT A jection has the form of an arc of a circle. It is clear that,
f 1 ( t n – 0 ) = --- + f 1 ( t n – 1 + 0 ) e – --- when the trajectory intersects the boundary of the wire,
ν ν we have
A – νT – νT
= – --- ( 1 – e ) + f 1 ( t n – 1 + 0 )e .
ν t' = r B ϕ/v ⊥ , (8)

Applying condition (5), we find where rB = mv⊥/eB is the radius of the arc and ϕ is the
central angle corresponding to this arc.
A – νT – νT
f 1 ( t n + 0 ) = q – --- ( 1 – e ) + f 1 ( t n – 1 + 0 )e . When calculating integral (2), it is convenient to
ν
pass to cylindrical coordinates both in the space of
coordinates (r⊥ , γ, z; the polar axis is the axis z and the
Using this recurrence relation, we express f1(tn – 1 + 0)
vector E0 is parallel to axis z) and in the space of veloc-
in terms of f1(tn – 2 + 0), etc., and arrive at an expression
ities (v⊥, α, vz; the polar axis is the axis vz). The wire
for f1(tn + 0) in terms of the sum of an infinite geometric
axis coincides with axis z. In the cylindrical coordi-
series with the ratio qexp(–νT). Summing this series, nates, the field (1) has only the z component:
we obtain
– νT – iωt
qA ( 1 – e ) E = Ez ez , Ez = E0 e .
f 1 ( t n + 0 ) = – -----------------------------
– νT
-.
ν ( 1 – qe )
Accordingly, the current (2) also has only the z compo-
To determine a concrete expression for the solution nent (the current lines are straight lines parallel to the z
to Eq. (3), we use the initial condition (6). For t' = 0, we axis):
obtain
– νt'
( q – 1 )e
2
3ne
∫z v z δ ( ε – ε F ) -------------------------
2 3
– νT j z = ----------------E - + 1 d v.
qA ( 1 – e ) A 4πv F ν
3
1 – qe
– νT
- = C – ---,
– -----------------------------
ν ( 1 – qe )
– νT ν
Here, we took into account that the concentration of
whence conduction electrons in metals is given by
– νT
A q(1 – e ) A 1–q 2
C = --- 1 – -------------------------
- = --- --------------------- . m 34 3
n = 2 ⎛ ----⎞ f 0 d v = 2 ⎛ ----⎞ --- πv F .
m
ν ν 1 – qe –νT ∫
– νT 3
1 – qe ⎝ h⎠ ⎝ h⎠ 3
Thus,
Using the properties of the δ function, we obtain
f 1 ( t' ) = --- ⎛ ---------------------
A 1 – q ⎞ –νt' A
e – --- 2
ν ⎝ 1 – qe –νT⎠ ν δ ( ε – ε F ) = ---- δ ( v z + v ⊥ – v F )
2 2 2

(7) m
– νt'
A ( q – 1 )e
= – --- -------------------------
-+1 . 2
ν 1 – qe –νT = ---- δ [ v z – ( v F – v ⊥ ) ]
2 2 2
m
The parameter t' in (7) has the meaning of the trav-
2
= ---- δ [ ( v z – v F – v ⊥ ) ( v z + v F – v ⊥ ) ]
2 2 2 2
eling time of an electron with velocity v along the tra-
jectory from the boundary on which reflection occurs to m
a point r. The parameter T is the period of the electron,
i.e., the time interval between two successive collisions 1
= ----------------------------
of the electron with the surface of the wire. 2 2
m vF – v⊥
Determination of the parameters t' and T is the key
moment in this problem since, due to the longitudinal
× [δ(v z – v F – v ⊥) + δ(v z + v F – v ⊥)].
2 2 2 2
magnetic field, part of electrons move along helical tra-
jectories between two successive reflections from the
wire surface, while the trajectories of other electrons do In view of the symmetry of the problem, the integra-
not intersect the boundary of the wire (t' ∞)! tion over the entire range of velocities vz is replaced by

JOURNAL OF EXPERIMENTAL AND THEORETICAL PHYSICS Vol. 103 No. 5 2006

 DEPENDENCE OF THE ELECTRIC CONDUCTIVITY OF A THIN CYLINDRICAL WIRE 771

the integration over the positive range and the result y

obtained is doubled; therefore,

v δ ⎛ v – v F – v ⊥⎞
2 2 2
3ne E z 2 z ⎝ z ⎠
2
rB
j z = ----------------
4πv F ν m
3
- ---
- ∫
----------------------------------------------------
2 2
-
ψ0
vF – v⊥ R ϕ
( q – 1 )e
– νt' ϕ0
× -------------------------
3
– νT
- + 1 d v.
1 – qe r⊥ 2π – α x

Specifying the limits of integration, we arrive at

v⊥
v F 2π ∞ v 2 δ ⎛ v –
2
vF – v ⊥⎞
2
2 z ⎝ z ⎠
3ne E z
j z = --------------------
2πmv F ν 0
3
- ∫ ∫∫ -----------------------------------------------------
2
vF – v⊥
2
0 0
– νt'
( q – 1 )e
× -------------------------
– νT
- + 1 v ⊥ dv ⊥ dαdv z . Fig. 1. Complex plane for determining the angle ϕ.
1 – qe
Integrating with respect to the variable vz , we obtain To determine the angle ϕ, consider the complex
plane z = x + iy. It is convenient to use the complex
v F 2π
2 plane because the equation of a circle with center at the
3ne E z
∫ ∫v
2 2
j z = --------------------
- ⊥ vF – v⊥ point (x0, y0) and radius rB , which corresponds to the
2πmv F ν 0
3
motion of an electron, has a rather simple expression
0 (9)
– νt'
( q – 1 )e
× -------------------------
– νT
- + 1 dv ⊥ dα. z = z 0 + r B exp [ i ( ϕ 0 + ϕ ) ],
1 – qe
For further calculations and the analysis of the where ϕ0 is the angle between the radius rB and the x
results, we introduce new variables axis at the moment when the electron is reflected from
the boundary of the wire.
r v⊥ eR
ξ = ----⊥- , ρ = ------, µ = ----------- B, At a point r, we define the angle ϕ for the electron
R vF mv F
moving with velocity v. Draw the axis x from the center
νR of the wire to the point r⊥ (Fig. 1), i.e., to the point with
∆ = ------- = ⎛ --- – iω⎞ ------
1 R
vF ⎝ τ ⎠ vF the coordinates (x, 0). At the moment when the electron
is reflected from the boundary of the wire, the relation
R Rω R Rω
= --------- – i -------- = ---- – i -------- = λ – iΩ.
τv F vF Λ vF iψ 0
Re = z 0 + r B exp ( iϕ 0 )
Then, formula (9) can be rewritten as
2 1 2π holds, where ψ0 is the angle between the projection r⊥
3ne RE
∫∫ρ 1–ρ of the radius vector of the electron and the axis x at the
2
j z = ----------------------z
2πmv F ∆ moment of reflection (in this case, |r⊥ | = R). Therefore,
0 0 (10) we obtain the following relation between the quantities
( q – 1 ) exp ( – ϕ∆/µ ) that define the position of the electron in the complex
× ------------------------------------------------ + 1 dρdα.
1 – q exp ( – ϕ*∆/µ ) plane:

In order to integrate expression (10), we must relate r ⊥ + δ exp ( iϕ 0 ) = R exp ( iψ 0 ) + r B exp [ i(ϕ 0 + ϕ) ]. (11)
the angle ϕ (the central angle of the arc of the circle
along which an electron moves in the case when its tra-
jectory intersects the boundary of the wire) and the When the coordinate axes are directed as indicated
angle ϕ* (in terms of which the period T of an electron above, we have ϕ = α – ϕ0 – π/2 (all the angles are mea-
is expressed) to the angle α (the angle between the vec- sured counterclockwise). This relation between the
tors r⊥ and v⊥) in the space of velocities. angles ϕ, ϕ0 , and α can easily be obtained by differenti-

JOURNAL OF EXPERIMENTAL AND THEORETICAL PHYSICS Vol. 103 No. 5 2006

772 ZAVITAEV, YUSHKANOV

ating the equation of circle z = z0 + rBexp[i(ϕ0 + ϕ)] with the surface of the wire, is equal to
respect to the parameter t' with regard to relation (8):
r B ϕ* ∆
∂z - = ---ϕ*,
T = ----------- ϕ* = 2π ( ϕ 2 – ϕ 1 ).
------ = iv ⊥ exp [ i ( ϕ + ϕ 0 ) ] v⊥ µ
∂t'
π Here, ϕ1 ≡ ϕ and ϕ2 are the roots of Eq. (12) that deter-
= v ⊥ exp i ⎛ ϕ + ϕ 0 + ---⎞ = v ⊥ exp ( iα ).
⎝ 2⎠ mine the points of intersection of the electron trajectory
with the boundary of the wire; moreover,
Using expression (11), we obtain
2ρξµ cos α – D
r ⊥ + r B cos ϕ 0 = R cos ψ 0 + r B cos ( ϕ 0 + ϕ ), ϕ 1 = 2 arccot ----------------------------------------
- ,
µ (ξ – 1)
2 2

r B sin ϕ 0 = R sin ψ 0 + r B sin ( ϕ 0 + ϕ ) 2ρξµ cos α + D

ϕ 2 = 2 arccot ----------------------------------------- .
µ (ξ – 1)
2 2
or
r ⊥ + r B sin ( α – ϕ ) = R cos ψ 0 + r B sin α, Integrating expression (10), we determine the total
current I through the cross section of the cylindrical
– r B cos ( α – ϕ ) = R sin ψ 0 – r B cos α. wire:
Eliminating ψ0 from the latter system, we arrive at the 2 3 1 1 2π
following equation for the angle ϕ: 3ne R E
∫
I = -----------------------z ξ dξ ∫∫ρ 1–ρ
2
mv F ∆
( r B – r ⊥ sin α ) cos ϕ + r ⊥ cos α sin ϕ 0 0 0

( q – 1 ) exp ( – ∆ϕ/µ )
r ⊥ + 2r B – 2r B r ⊥ sin α – R × ------------------------------------------------ + 1 dρdα.
2 2 2
= --------------------------------------------------------------
-, 1 – q exp ( – ∆ϕ*/µ )
2r B
or, in dimensionless variables, Then, we formally apply Ohm’s law in the form I = GU,
where U is the voltage across the ends of the wire, and
obtain a formula for calculating the conductivity G of
( ρ – ξµ sin α ) cos ϕ + ξµ cos α sin ϕ the wire (the electric field inside the wire is uniform;
ξ µ + 2ρ – 2ρξµ sin α – µ (12)
2 2 2 2 therefore, U = EzL):
= --------------------------------------------------------------------- .
2ρ 1 1 2π
2 3
3ne R
∫
G = ------------------- ξ dξ ∫∫ρ 1–ρ
2
Applying the standard method to Eq. (12), we obtain mv F L∆
0 0 0 (13)
2ρξµ cos α – D ( q – 1 ) exp ( – ∆ϕ/µ )
ϕ = 2 arctan ----------------------------------------
- , × ------------------------------------------------ + 1 dρdα.
µ (ξ – 1)
2 2
1 – q exp ( – ∆ϕ*/µ )

where In the absence of the external magnetic field (µ = 0)

in the case of a dc electric field (when Ω = 0 and ∆ =
λ = R/Λ) and the diffuse reflection of electrons from the
D = 4ρ ( ρ – 2ρξµ sin α + ξ µ )
2 2 2 2
surface of the wire (q = 0), formula (13) coincides with
– [ ξ µ + 2ρ – 2ρξµ sin α – µ ] .
2 2 2 2 2 the result of [2]:
1 1π
2 3
The electron trajectory intersects the boundary of the 6ne R
G = ------------------- ξ dξ ∫ ∫∫ρ 1–ρ
2
wire when D > 0 (Eq. (12) has real roots). In this case, mv F L∆
the traveling time of electrons along the trajectory 0 00
(parameter t') is finite and positive.
⎧ ∆ 2 ⎫
× ⎨ 1 – exp – --- ( ξ cos α + 1 – ξ sin α ) ⎬dρdα.
2
When D < 0 (Eq. (12) has no real roots), the electron
trajectory does not intersect the boundary of the wire ⎩ ρ ⎭
(t' ∞). In this case, one should take into account
that the term containing the exponential function Let us represent the integral conductivity (13) (note
in (10) vanishes. that it is a complex quantity) as
The period of the electron, i.e., the time interval
between two successive collisions of the electron with G = G 0 P ( λ, Ω, µ, q ), (14)

JOURNAL OF EXPERIMENTAL AND THEORETICAL PHYSICS Vol. 103 No. 5 2006

 DEPENDENCE OF THE ELECTRIC CONDUCTIVITY OF A THIN CYLINDRICAL WIRE 773

where S, mΩ

2 3 8 1
3ne R
G 0 = -----------------, 2
mv F L 7

1 1 2π 6
1
P ( λ, Ω, µ, q ) = --- ξ dξ ∫ ∫∫ρ 1–ρ
2
3
∆ 5
0 0 0 (15) 4
4
( q – 1 ) exp ( – ϕ∆/µ )
× ------------------------------------------------ + 1 dρdα.
1 – q exp ( – ϕ*∆/µ ) 3 5

The computed values of the modulus M(λ, Ω, µ, q) 2

and the argument A(λ, Ω, µ, q) (phase) of the dimen- 6
sionless integral conductivity P(λ, Ω, µ, q) of a long 1
µ
0 1 2
cylindrical wire are shown in Fig. 3.
Fig. 2. Theoretical (curves) and experimental (symbols)
dimensional integral magnetoresistance S of a thin metal
3. DISCUSSION OF THE RESULTS wire in the case of a dc electric field (Ω = 0) as a function of
the dimensionless external magnetic field induction µ:
Figure 2 illustrates the comparison of the dimen- (1) λ = 1.06, q = 0.3; (2) λ = 0.88, q = 0; (3) λ = 0.65, q =
sional integral magnetoresistance S of a thin cylindrical 0.3; (4) λ = 0.45, q = 0; (5) λ = 0.35, q = 0.3; and (6) λ =
wire measured in [4] in the case of a dc electric field 0.14, q = 0.
(Ω = 0) and the integral magnetoresistance calculated
for the same case by formula (13) for various values of
the dimensionless induction µ of the external magnetic The results presented above imply that the measure-
field. In the experiment we used a 4.5-cm long sodium ment of the longitudinal magnetoresistance provides an
wire 30 µm in diameter. The magnetoresistance was efficient means for determining the specular reflection
measured in the range of temperatures from 4.2 to coefficient of metals. Note that the parameter q is an
15.5 K. This is equivalent to the variation of the mean important kinetic characteristic of a metal, but measuring
free path of electrons in the wire since the variation of this parameter faces serious technical difficulties [11].
temperature leads to the variation of the collision rate of It follows immediately from the aforesaid that the
electrons in the metal. application of the method described to determine the
Curves (1), (3), and (5) correspond to the case of specular reflection coefficient of the surface of metal
specular–diffuse reflection of electrons from the samples whose linear dimensions are comparable with
boundary of the wire. To show that the consideration of the electron mean free path is especially important.
the specular character of reflection is essential, we also In the limit of purely specular reflection of electrons
present curves (2), (4), and (6), which correspond to the from the boundary of the metal wire (q = 1), we obtain
case of purely diffuse reflection of electrons. The exper- the following expression for the dimensionless integral
imental data do not contain the frequency of bulk elec- conductivity P(λ, Ω):
tron collisions in the wire. However, comparing the the-
oretical and experimental results, we can determine this
π 1 π λ + iΩ
quantity for a given specular reflection coefficient q of P ( λ, Ω ) = --- ------------------- = --- ---------------
-, (16)
the metal. Accordingly, Fig. 2 presents two types of 3 ∆ ( λ, Ω ) 6 λ2 + Ω
curves: solid curves for (q = 0.3) and dashed curves (for
q = 0). It follows from the aforesaid that, for small val- this formula (with regard to (14)) corresponds to the
ues of the dimensionless induction µ of the external classical result for a cylindrical wire. This is associated
magnetic field, the theoretical results (1), (3), and (5) with the fact that, for q = 1, the boundary of the wire
must coincide with (2), (4), and (6), respectively. does not influence the electron distribution function
A comparison of the experimental and theoretical f(r⊥, v⊥, vz). Therefore, high-frequency current inside a
curves shows that the integral conductivity G(λ, Ω, µ, q) specularly reflecting metal wire satisfies a local Ohm’s
of a long cylindrical wire in a longitudinal magnetic law for any relation between the wire radius R and the
field calculated by the kinetic method agrees with electron mean free path Λ, as well as for any dimen-
experimental data to within 5% provided that the reflec- sionless induction µ of the external magnetic field.
tion of electrons from the metal surface is specular. Thus, there are no nonlocal (surface) phenomena in the
These investigations allowed us to determine the spec- case of specular reflection, and the integral conductivity
ular reflection coefficient q. For the sodium wire of the wire does not depend on the longitudinal mag-
described in [4], the value of q equals 0.3. netic field.

JOURNAL OF EXPERIMENTAL AND THEORETICAL PHYSICS Vol. 103 No. 5 2006

774 ZAVITAEV, YUSHKANOV

M A
8 0.8
(b)
(a)
7 3
0.7

6 0.6
2
5 0.5
3

4 0.4
2
3 0.3
1
1
2 0.2

1 0.1
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
q q
Fig. 3. (a) Modulus M and (b) argument A (phase) of the dimensionless integral conductivity of a wire as a function of the reflection
coefficient q of electrons: (1) λ = Ω = µ = 0.1; (2) λ = Ω = 0.1, µ = 1; and (3) λ = Ω = 0.1, µ = 2.

Irrespective of the character of electron reflection REFERENCES

from the boundary (for any q) and of the longitudinal 1. Yu. I. Petrov, Physics of Small Particles (Nauka, Mos-
magnetic field (for any µ), the macroscopic asymptot- cow, 1982) [in Russian].
ics (16) holds as the wire radius increases, i.e., for 2. R. B. Dingle, Proc. R. Soc. London, Ser. A 201, 545
λ
1 (in this case, one can neglect the rapidly decreas- (1950).
ing exponential term in (15)). 3. D. K. C. MacDonald, Nature 163, 637 (1949).
Figure 3 shows the effect of the reflection coefficient 4. R. G. Chambers, Proc. R. Soc. London, Ser. A 202, 378
q of electrons on the modulus M and the argument A (1950).
(phase) of the dimensionless integral conductivity P. 5. D. K. C. MacDonald, Proc. Phys. Soc. London, Sect. A
Consider the cases of small dimensionless frequency Ω 63, 290 (1950).
(Ω = 0.1) of the electric field and small dimensionless 6. R. G. Chambers, Proc. Phys. Soc. London, Sect. A 65,
inverse mean free path λ (λ = 0.1) of electrons (the 458 (1952).
parameters Ω and λ are the same for all the curves in 7. G. K. White and S. B. Woods, Philos. Mag. 1, 846
Fig. 3) for three values of the dimensionless induction (1956).
µ of the external magnetic field. One can see that the 8. É. A. Manykin, P. P. Poluéktov, and Yu. G. Rubezhnyœ,
modulus and the argument (phase) of the dimensionless Zh. Éksp. Teor. Fiz. 70, 2117 (1976) [Sov. Phys. JETP
integral conductivity P of the wire smoothly vary as the 43, 1105 (1976)].
reflection coefficient q of electrons increases. This vari- 9. W. A. Harrison, Solid State Theory (McGraw-Hill, New
ation is the most noticeable when the wire is in the mag- York, 1970; Mir, Moscow, 1972).
netic field with maximal dimensionless induction µ. 10. J. M. Ziman, Electrons and Phonons (Clarendon,
The variation of these quantities reaches significant val- Oxford, 1960; Inostrannaya Literatura, Moscow, 1962).
ues (for the parameters Ω, λ, and µ chosen for numeri- 11. I. M. Lifshits, M. Ya. Azbel’, and M. I. Kaganov, Elec-
cal calculations) in the case of purely diffuse reflection tron Theory of Metals (Nauka, Moscow, 1971; Consult-
of electrons from the internal surface of the wire ants Bureau, New York, 1973).
(q = 0). For large q (q ≈ 1), these functions merge 12. R. Courant, Partielle Differentialgleichungen (Göttin-
together because the calculated value of the electric con- gen, 1932; Mir, Moscow, 1962).
ductivity coincides with the classical expression (16) for
the conductivity of a cylindrical metal wire. Translated by I. Nikitin

JOURNAL OF EXPERIMENTAL AND THEORETICAL PHYSICS Vol. 103 No. 5 2006