Você está na página 1de 6

The 8th INTERNATIONAL SYMPOSIUM ON ADVANCED TOPICS IN ELECTRICAL ENGINEERING

May 23-25, 2013


Bucharest, Romania

Wavelet Solution
of the Time Independent Schrodinger Equation
for a Rectangular Potential Barrier
2

Simona-Mihaela BIBIC1 , Emilia-Simona MALUREANU


1

Department of Applied Mathematics, POLITEHNICA University of Bucharest, Romania


2
Electrical Material Laboratory, POLITEHNICA University of Bucharest, Romania
simona bibic@mathem.pub.ro, emalureanu@elmat.pub.ro

AbstractElectronics industry is a major user of MetalInsulator-Metal (MIM) structures. When the tunneling phenomenon occurs at the metal-insulator contact barrier, as for the
MIM structures, we are dealing with the Fowler-Nordheim field
emission. In MIM structures we have to calculate the currents
densities that go through. Unfortunately, the calculation methods
currently used lead to values that get predictions of emission
current density too low. The current density is calculated based
on the transmission coefficient through the barrier, which is determined by solving the Schrodinger equation. One of the oldest
and most efficient methods of solving the Schrodinger equation
is the WKB method. This paper proposes the wavelet method
for solving the 1D Schrodinger equation, that is: determining
the wave function using the harmonic multiresolution analysis.
KeywordsSchrodinger equation, dyadic wavelets, potential
barriers

I. I NTRODUCTION
As the size of the structures used in electronics industry goes
towards nanaoscopic scale, the importance of quantum effects
must be seriously considered. Metal-Insulator-Metal (MIM)
structures begin to assert themselves in microelectronics industry against those based on silicon because they allow a
substantial reduction in production costs and achieve higher
speeds compared to the latter [1].
In the case of MIM structures with a dielectric thickness of
a few nanometers, conduction can occur in two ways.
At high temperatures and electric field strengths smaller
than 108 V/m, the electrons, due to increased thermal agitation,
gain enough energy to climb the potential barrier that occurs at
metal-insulator interface. This conduction mechanism is called
Schottky injection [2] [3].
At normal temperatures and electric field intensity that
exceeds 108 V/m, the potential barrier, corresponding to the
metal-insulator contact becomes very thin and there will
appear a quantum type phenomenon called tunneling, when
electrons, although they do not have enough energy to climb
the potential barrier, they manage to pass through the barrier.
When the tunneling phenomenon occurs in metal-insulator
contact barrier, as in the MIM structures, this is a Fowler Nordheim injection [2] [3].
In the case of devices that are based on the phenomenon of
Fowler - Nordheim injection [1] [4] [5], the emission density

current must be determined. In MIM structures where the


conduction is achieved by tunneling the barrier, the waves
associated to the electrons suffer reflection from passing
through the barrier so, in this case, the current density is
calculated according to the barrier transmission coefficient,
meaning the probability that the incident electrons to pass
through the barrier. The quantum mechanical reflection can
limit the current gain for these structures, so it is important for
the optimal device design to know the extent of this reflection.
The transmission coefficient is determined by solving the
Schrodinger equation. The exact calculation of the current
density depends on the form of the potential barrier.
In the ideal case, when the image forces or the existence of
the electric double layer on the surface of the metal are not
considered, the representation of the metal - insulator contact
is made by a rectangular potential barrier. For this type of
barrier, the Schrodinger equation can be analytically solved.
But in reality, the barrier is different from this rectangular type
and the more realistic we represent the barrier, the more closer
we are to the experimental density current value. For these
types of barrier there is no analytical solution but numerical
ones. Unfortunately, the calculation methods based on simple
solvable models of the tunneling barrier lead to values that get
predictions of emission current density too low by a factor of
100 or more [9].
In spite of the variety of approaches to this problem over
the last years, one aspect of the calculations that remains
common to all is the WKB approximation to the solution of
the Schrodinger equation. The general validity of this method
is questionable, sice no wave function interference is taken
into account [10].
This paper proposes the wavelet method for solving the
1D Schrodinger equation, that is: determining the wave function using the harmonic multiresolution analysis, respectively
harmonic and Shannon wavelet on Hilbert space L2 (R). The
wavelet method shall be compared to the analytical one for
the rectangular barrier. Wavelets were before applied for
numerically solving the Schrodinger equation. Griebel and
Hamaekers [6] use the one-dimensional Meyer wavelets in
order to solve the electronic Schrodinger equation. Diaz et
al. [7] use the interpolatory wavelets and Goedecker [8] the

ii) for x = a
C (a) = R (a),
0

C (a) = R (a) .

(9)

The boundary conditions give the following restrictions on


the coefficients
Ar + Al = Br + Bl ,
ik0 (Ar Al ) = ik(Br Bl ),
Br expika +Bl expika = Cr expik0 a +Cl expik0 a ,
Fig. 1.

ik(Br expika Bl expika ) = ik0 (Cr expik0 a Cl expik0 a ) .


(10)

Rectangular potential barrier.

Daubechies wavelets in order to obtain a numerical solution


of the time independent Schrodinger equation.
II. THE ANALYTICAL SOLUTION OF THE

SCHR ODINGER
EQUATION FOR THE
RECTANGULAR POTENTIAL BARRIER
Let us consider the one-dimensional time independent
Schrodinger equation
~2 d2
+ U (x)(x) = E(x),
2m dx2
where the potential energy U (x) is by the form
(
U0 , 0 x a
U (x) =
, U0 > 0 ,
0
, otherwise

(1)

q
ix 2mE
~2

+k2 exp

ix

2mE
~2

R = |r|2 , T = |t|2

(11)

with
(2)

and ~ represents Plancks constant divided by 2, m represents


the electron rest mass, is the wave function, E is the total
energy of the particle.
For U (x) = 0 the solution of (1) is
(x) = k1 exp

To study the case of quantum tunneling, let us consider


the following situation: a particle incident on the barrier from
the left side (Ar ) (see Fig.1). It may be either reflected (Al )
or transmitted (Cr ). To find the amplitudes for reflection and
transmission for incidence from the left, we put in the above
equations Ar = 1 (incoming particle), Al = r (reflection),
Cl = 0 (no incoming particle from the right), and Cr = t
(transmission). We then eliminate the coefficients Bl , Br from
the equation and solve for r and t.
The reflection and transmission coefficient are

r = Br + Bl 1,
t = Br expia(kk0 ) +Bl expia(k+k0 ) .

(12)

In Fig.2 one can see the analytical solution of the


of the rectSchrodinger wavefunction, for a width a = 1A
angular potential barrier, after and before tunneling.

(3)

and (3) can be detailed


q as following.
Considering k0 = 2mE
~2 , then
L (x) = Ar expik0 x +Al expik0 x ,

if x < 0 ,

(4)

R (x) = Cr expik0 x +Cl expik0 x ,

if x > a .

(5)

and

If U (x) = U0 (E < U0 ), then (1) becomes


d2
k 2 (x) = 0 ,
dx2

(6)
Fig. 2. The wave function after tunneling the rectangular barrier when a = 1.

where k 2 = 2m(U~02E) .
In this case, the solution has the form
C (x) = Br expikx +Bl expikx .

(7)

Applying the boundary conditions to (4),(5) and (7), the


coefficients Ar,l , Br,l , Cr,l can be determined [12]. We shall
use two sets of conditions
i) for x = 0
L (0) = C (0),
0

L (0) = C (0).

(8)

III. DYADIC WAVELETS


Some differential problems such as those corresponding to
certain physical processes have led to the necessity of using
wavelets theory (orthonormal bases of compactly supported
functions which depend on two parameters: the dilation parameter used to change the scale and the translation parameter
used to slide in time) as a scientific working tool for solving
them and more, relying on the obtained information to describe
the phenomenon itself. Therefore, based on the properties

of wavelets like good localization and resolution control the


concept of orthonormal wavelet series and representation of a
function f L2 (R) by such an expression were introduced.
To illustrate this, dyadic wavelet approximation represents
a numerical method used to obtain numerical solutions of
some physical problems. An orthonormal dyadic wavelet is
a function L2 (R) (see [22] [21] [19] [11] [16] [17]),
such that the family {Dn Tk []}n, kZ is an orthonormal basis
for L2 (R), where D and T are the expansion (dilatation) and
translation unitary operators on L2 (R) expressed as

HW


b HW (0) = 0.
a wavelet) and its energy EHW = (x) = 1,
2
The Calderons admissibility constant is
HW
2
Z
b ()
(23)
0 < CHW =
d = ln(2) < .
||
R
More, according to multiresolution analysis the followings are
satisfied
Z
Z
HW
HW
(x)dx = 1 ,
(x)dx = 0 .
(24)
R

D [(x)] = 2 2 (2x) ,
T [(x)] = (x 1) .

(13)

Generalizing, for n, k integer values and taking into account


(13), then
n

Dn Tk [(x)] = n k (x) = 2 2 (2n x k) .

Using the Fourier property




F f(x) = fb() ,

results immediately the Fourier transform for the conjugates


of (15), (16), i.e.,

(14)
HW

By definition [11], the harmonic scaling function (x)


HW
(father wavelet) and the harmonic wavelet function (x)
(mother wavelet) are complex functions of the form
HW

HW

HW

, : R C ,

def

(x) =

exp2ix 1
2ix

(15)

respectively,
def

HW

HW

(x) = exp2ix (x) .

(16)

Hence, the harmonic wavelet function actually represents the


scaling wavelet function modulated.
If we assume as Fourier transform of f (x) L1 L2 the
function fb(), where
Z
def
b
f () =
f (x) expix dx
(17)
R

then, we can easily get Fourier transform for (15) and (16),
i.e.,
b HW () = ( + 2)

(18)
b HW () = () .

The characteristic (or box) function () is defined as

2 4

1,
() =

0,
elsewhere.
HW

(19)

cHW () = ( + 4)

(26)

d
HW () = ( + 6) .

(27)

Remark 1: The harmonic scaling function and the harmonic


wavelet function, respectively their conjugates satisfy
D HW HW E D HW HW E D HW HW E D HW HW E
,

,
=
= ,
=
=1

E
D
D HW
E
D HW
E
D HW
E

HW

, HW = HW ,
HW =
, HW = 0
=
,

HW

HW

HW

HW

HW ,
HW

E
D HW
,
HW = 0

HW
,

HW , HW

HW , HW

E
D HW
HW = 0 .
,

HW

(28)
According to (21), the proof is trivial.
SW
The Shannon scaling function (x) is a particular case of
(15), [21] [17]
h HW  x i
SW
(x) = Re
= sinc(x)
(29)
2
and also, according to the multiresolution analysis, the ShanSW
non wavelet function (x) can be expressed as


SW
SW
SW
1
(x) = 2 (2x 1)
x
.
(30)
2
Their Fourier transforms are

(20)

HW

Also, remark that (x) and (x) are functions with


bonded spectrum.
According to Plancherel-Parseval theorem, if f, g L2 (R)
the Hilbert space, then
Z
1 Db E
hf, gi (x) =
f (x)g(x)dx =
f , gb () ,
(21)
2
R
and in particular

1


kf (x)k2 = fb() .
2
2

(25)

(22)

In this respect, it is easy to prove that the harmonic wavelet


fulfills the admissibility condition (a necessary condition to be

b SW () = ( + 3)

(31)

respectively,
h 

i
b SW () = exp i
2

+ 3 ( + 3) .
2

(32)

By computation (21), (30), (32), it shows that the Shannon


b SW (0) = 0 and the Calderons
wavelet energy is ESW = 1,
admissibility constant is CSW = ln(4). Similarly, are fulfilled
the multiresolution analysis conditions, see (24).
Remark 2: The harmonic scaling function and the harmonic
wavelet function, respectively their conjugates satisfy
D SW SW E D SW SW E
, = , =1
(33)
D SW SW E D SW SW E

= 0.

IV. T HE DYADIC WAVELET SOLUTION OF THE

S CHR ODINGER
EQUATION FOR THE RECTANGULAR

B. The wavelet solution

POTENTIAL BARRIER

A. Problem formulation
Let us consider the time-independent Schrodinger equation
00 (x) = K(x) (x) ,

In this section we will consider the simplest case for


(38), i.e., h = n = k = 0, which represents the wavelet
approximation of level zero (coarse form) of the particular
solution. Therefore, according to harmonic wavelet analysis,
(38) has the following form
HW

(34)

HW

HW

HW

(44)

In this respect, relation (36) becomes

where
2m
K(x) = 2 (U (x) E) .
~

(35)

HW

d2
dx2

+
HW
HW

d
dx

+
HW

+ 2HW HW

y 00 + (y 0 )2 = K(x) .

(36)

A potential solution to (34) has the following form


(x) = (xC1 + C2 ) expW(x) ,

(37)

(38)

where

hZ

h h (x) +

n k n k (x) ,

n, kZ

and represents the wavelet series expansion of the particular


solution of (34).
Remark 3: If the potential barrier has the form
(
U0 , x1 x x2
U (x) =
, U0 > 0 ,
(39)
0
, otherwise

+ 2
HW HW

constants C1 and C2 have the form

W(x2 )
A expW(x1 )

C1 = B exp

x2 x1

(40)

d2
dx2

d2
dx2
HW !2

!2

d
dx

+ HW

HW

HW

+ HW

+ HW

d2
dx2

!2

HW

2
+ HW

d
dx

+ HW

d
dx

HW

+ 2HW
HW

56HW 52HW 10HW HW + 14


HW HW =

3
3

56HW 5
2HW 10
HW HW + 14HW HW =

3
3

b
K()d
2
2

b
K()d
.
4

(46)

SW

SW

SW

W (x) = SW (x) + SW (x) ,

(47)

and (36) takes the form


SW

d2
dx2
2

+ SW

SW

+ SW
d
dx

d2
dx2

SW

d
dx

+ SW

!2

SW

+ 2SW SW

!2
+

(48)

SW

d d
dx
dx

= K(x) .

SW

In order to illustrate the numerical method, the simplest


form of (38) will be considered and the solution will be
determined in terms of the dyadic wavelets (e.g., harmonic,
Shannon analysis).
One should note that, in Fourier domain,
h
i
s
F f (s) (x) = (i) fb() .
(42)
Also, in the following, some results are obtained using
frequency convolution(see [18]), i.e.,
Z
1
\
f (x)
g(x) =
fb( )b
g ( )d .
(43)
2 R

HW

d d
+
dx
dx

Similarly, the Shannon wavelet expression of (38) is

SW

W(x1 )

x1 B expW(x2 )

C2 = x2 A exp
.
x2 x1

HW
HW
HW
HW
HW dHW
d d
d d
d
+ 2HW HW
+ 2
HW HW
+
dx
dx
dx
dx
dx
dx
HW
HW
HW
HW
d

d
d d
+ 2HW HW
= K(x) .
dx
dx
dx
dx

SW

(41)

!2

(45)
HW
HW
HW ,
HW and
By inner product with functions , ,
according to (21), (28), (42), (43), results the coefficients HW ,
HW and their complex conjugates
HW , HW (for more technical
details of computation see [11] [16] [15] [18]). Finally, they
are determined by solving the following nonlinear system of
equations

Z 2
3
b

8HW 2HW + 6HW


HW + 54HW HW + 22
HW HW = 3
K()d

Z0 0

2
b

K()d
8
HW
HW + 6HW
HW + 54HW HW + 22HW HW = 3

applying the boundary conditions (see the analytical solution,


(8), (9)), we have the following relations for this case
(x1 ) = A , (x2 ) = B ,

HW

HW

HW

According to the WKB method [12], substituting (x) =


expy(x) the following nonlinear ODE is obtained,

W(x) =

HW

(x) + (x) +
(x) .
W (x) = HW (x) +
HW
HW
HW

SW

By inner product with functions , respectively and


according to (21), (33), (42), (43), the coefficients SW and
SW are determined by solving the following nonlinear system
of equations (see [17])

Z
2
2 + 3 2
7 + 15 2
1

b
SW SW +
SW =
K()d

SW +
2
3

7 2
3
1
2SW =
3 SW
4
2

Z

b
K()
exp
2

i
2

d +

b
K()
exp

i
2


d

(49)
b
In the relaions above was denoted by K()
the Fourier
transform of (35).
C. Results
This paper considers three cases for the rectangular potential
barrier, when the ends of the barrier are in x1 = 0, x2 = a = 1,
a second one, when x1 = 0 and x2 = a = 3 and when x1 = 0
and x2 = a = 10. The wave function was calculated, with

the analytical method and with the wavelet methods, for these
values of x. The absolute error was calculated for each value
of in x1 andx2 as xnum xAN , with xnum representing
the numerical solution and xAN being the exact solution. The
results are listed in the tables below.
From Figures 3, 4, 5, 6, one can observe that inside as
well as outside the potential barrier, the harmonic wavelet
solution and the Shannon wavelet solution represent a good
approximation of the analytical solution.
The transmission coefficient
2)
Tr = (x2 ) (x

(50)

will be calculated for the analytical and the wavelet numerical


approximation, using (50) for these three cases of the potential
barrier.
TABLE I
VALUES OF FOR x1 = 0 AND x2 = a = 1
x
AN
HW

SW

x1 = 0
1.642 0.742i
1.642 0.742i
1.642 0.742i
1.642 0.742i
1.642 0.742i
1.642 0.742i
1.642 0.742i

x2 = a = 1
0.144 + 0.125i
0.144 + 0.125i
0.144 + 0.125i
0.144 + 0.125i
0.144 + 0.125i
0.144 + 0.125i
0.144 + 0.125i

Fig. 3. Real part of analytical solution by dash line and real part of harmonic
solution by solid line for a = 3.

TABLE II
A BSOLUTE ERROR
Absolute error for x1 = 0 and x2 = 1

Method
HW
SW

7 1010 + 1 1010 i
1 109 3 1010 i

0
1 1010 1 1010 i

TABLE III
VALUES OF FOR x1 = 0 AND x2 = a = 3
x
AN
HW

SW

x1 = 0
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i

x2 = a = 3
0.002 + 0.002i
0.002 + 0.002i
0.002 + 0.002i
0.002 + 0.002i
0.002 + 0.002i
0.002 + 0.002i
0.002 + 0.002i

Fig. 4. Real part of analytical solution by dash line and real part of harmonic
solution by solid line for a = 10.

TABLE IV
A BSOLUTE ERROR
Method
HW
SW

Absolute error for x1 = 0 and x2 = 3


2 1010 1 1010 i
2 109 1 109 i

1 1011 + 7 1012 i
5 1011 i

TABLE V
VALUES OF FOR x1 = 0 AND x2 = a = 10
x
AN
HW

SW

x1 = 0
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i
1.666 0.745i

x2 = a = 10
1.36 109 + 1.22 109 i
1.37 109 + 1.23 109 i
1.37 109 + 1.23 109 i
1.4 109 + 1.23 109 i
1.4 109 + 1.225 109 i
1.368 109 + 1.229 109 i
1.369 109 + 1.225 109 i

Fig. 5. Real part of analytical solution by dash line and real part of Shannon
solution by solid line for a = 3.

TABLE VI
A BSOLUTE ERROR
Method
HW
SW

Absolute error for x1 = 0 and x2 = 10


1 1010 i
3 109 + 1 109 i

1 1011 + 3 1012 i
9 1013 + 2 1012 i

TABLE VII
T RANSMISSION COEFFICIENT FOR x2 = 1, x2 = 3 AND x2 = 10
Method
AN
HW
SW

x2 = 1
0.036
0.036
0.036

x2 = 3
1 105
1 105
1 105

x2 = 10
3 1018
3 1018
3 1018

Further results, regarding wavelet based wave function approximation for other potential barrier types, will be reported
elsewhere.
ACKNOWLEDGMENT
This work has been supported by the Sectorial Operational
Programme Human Resources Development 2007-2013 of the
Romanian Ministry of Labour, Family and Social Protection
through the Financial Agreement POSDRU/107/1.5/S/76813.
R EFERENCES

Fig. 6. Real part of analytical solution by dash line and real part of Shannon
solution by solid line for a = 10.

V. CONCLUSION
This paper considers a numerical wavelet based method to
solve the time independent Schrodinger equation, considering
a rectangular potential barrier. The numerical solution is
compared to the analytical one for four values of x describing
the barrier. These values are important for determining the
transmission coefficient and the density current.
While the analytical method offers solutions only for the
rectangular potential barrier, the numerical wavelet based
method allows solutions for more complex potential barrier
types, where no analytical method is available. From the
results and considering the small obtained error (see Tables
II,IV,VI) it is obvious that the dyadic wavelet based method
offers a very good approximation of the solution compared to
the analytical method. Also, considering the representations
from Figures 3, 4, 5, 6 one can conclude that the wavelet
solution offers a precise approximation for the wavefunction,
inside and outside the potential barrier compared to the analytical one.

[1] E. William Cowell III, Nasir Alimardani, Christopher C. Knutson, John


F. Conley Jr.,Douglas A. Keszler, Brady J. Gibbons and John F. Wager,
Advancing MIM Electronics: Amorphous Metal. Electrodes, Advanced
Materials, 23, pp. 74-78, 2011
[2] Petru V. Notingher, Materiale pentru Electrotehnica, Politehnica Press,
Bucharest, 2005.
[3] I. Munteanu, Fizica Solidului, University of Bucharest Press, 2003.
[4] Amitabh Chandra and Lester F. Eastman, Quantum mechanical reflection
at triangular planar-doped potential barriers for transistors, Journal of
Applied Physics, 53, pp. 9165-9169, 1982.
[5] Leo Esaki, New Phenomenon in Narrow Germanium p-n Junctions,
Physical Review, 109,pp. 603-604, 1958.
[6] M. Griebel and J. Hamaekers, A wavelet based sparse grid method for
the electronic Schrodinger equation, Proceedings of the International
Congress of Mathematicians, 3,pp. 1473-1506, 2006.
[7] Lilliam Alvarez Diaz, Rodrigo Morante Blanco and Ailin Ruiz de Zarate,
The interpolatory wavelets for the numerical solution of the Schrodinger
equation, Revista Investigation Operacional, 22,pp. 184-192, 2001.
[8] Stefan Goedecker, Wavelets and Their Application for the Solution of
Poissons and Schrodingers Equation, Multiscale Simulation Methods in
Molecular Sciences, 42,pp. 507-534, 2009.
[9] R.H.Fowler and L. Nordheim, Electron Emission in Intense Electric
Fields, Proceedings of the Royal Society of London 119,pp. 173-181,
1928.
[10] Andreas Gehring, Simulation of Tunneling in Semiconductor Devices,
Dissertation at Vienna University of Technology, 2003.
[11] D.E. Newland, Harmonic wavelet analysis, Proceedings of the Royal
Society of London, Series A, 443, 203-225, 1993.
[12] I. Licea, Fizica Metalelor, Scientific and Encyclopedic Press, Bucharest,
1986.
[13] X. Dai and D. Larson, Wandering Vectors for Unitary Systems and
Orthogonal Wavelets, American Mathematical Society, Series A, 134,
no. 640, 1998.
[14] E. Weber, On the translation invariance of wavelet subspaces, Journal
Fourier Anal. Appl., 6, 551-558, 2000.
[15] C. Cattani, Harmonic Wavelet Solutions of the Schrodinger Equation,
International Journal of Fluid Mechanics Research, 5, 1-10, 2003.
[16] C. Cattani , Harmonic wavelet solution of Poissons problem with a
localized source, AAPP, Physical, Mathematical, and Natural Sciences,
86, no.2, 1-14, 2009.
[17] C. Cattani, Shannon Wavelet Theory, Mathematical Problems in Engineering, 1-24, 2008.
[18] S.M. Bibic, Harmonic wavelet analysis-connection coefficients for
nonlinear PDE, UPB, Sci. Bull, Series A, 73, no.1, 12-35, 2011.
[19] I. Daubechies, Ten lectures on wavelets, CBMS-NSF Regional Conference Series in Applied Mathematics, SIAM, Philadelphia, 1992.
[20] D. Stefanoiu, O. Stanasila and D. Popescu, Wavelets - theory and
applications, Romanian Academy Press, Bucharest, Romania, 2010.
[21] S. Mallat, A wavelet tour of signal processing, Academic Press, 2nd
edition, 1999.
[22] Y. Meyer, Wavelets and operators, Cambridge University Press, 1993.

Você também pode gostar