Solution of the one-dimensional Dirac equation with a linear scalar potential

John R. Hiller

Citation: American Journal of Physics 70, 522 (2002); doi: 10.1119/1.1456074

View online: http://dx.doi.org/10.1119/1.1456074
View Table of Contents: http://scitation.aip.org/content/aapt/journal/ajp/70/5?ver=pdfcov
Published by the American Association of Physics Teachers

Articles you may be interested in

Simple bound state energy spectra solutions of Dirac equation for Hulthén and Eckart potentials
J. Math. Phys. 54, 102105 (2013); 10.1063/1.4826358

Analytical solutions of the Dirac equation with a scalar linear potential

AIP Advances 3, 082135 (2013); 10.1063/1.4820388

An approximate state solutions of the Dirac equation for the generalized Morse potential under spin and
pseudospin symmetry
J. Math. Phys. 52, 052303 (2011); 10.1063/1.3583553

Addendum to: “The one-dimensional harmonic oscillator in the presence of a dipole-like interaction” [Am. J.
Phys. 71 (3), 247–249 (2003)]
Am. J. Phys. 71, 956 (2003); 10.1119/1.1592514

Introducing supersymmetric quantum mechanics via the Dirac equation

Am. J. Phys. 65, 1112 (1997); 10.1119/1.18730

This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.97.178.73 On: Sat, 27 Sep 2014 13:24:05
 Solution of the one-dimensional Dirac equation with a linear
scalar potential
John R. Hillera)
Department of Physics, University of Minnesota–Duluth, Duluth, Minnesota 55812
共Received 2 November 2001; accepted 8 January 2002兲
We solve the Dirac equation in one spatial dimension for the case of a linear, Lorentz-scalar
potential. This extends earlier work of Bhalerao and Ram 关Am. J. Phys. 69 共7兲, 817– 818 共2001兲兴 by
eliminating unnecessary constraints. The spectrum is shown to match smoothly to the nonrelativistic
spectrum in a weak-coupling limit. © 2002 American Association of Physics Teachers.
关DOI: 10.1119/1.1456074兴

I. INTRODUCTION stant. The continuity of ũ and ũ ⬘ at x⫽0 requires that

The linear potential V(x)⫽g 兩 x 兩 is a natural choice for a
confining potential in one spatial dimension. The nonrelativ-
istic Schrödinger equation admits a nearly analytical solution
for this potential in terms of the Airy function Ai and the
zeros of this function and its derivative. The Dirac equation,
on the other hand, appears to be problematic for this poten-
tial. If V is introduced as the time component of a Lorentz
two-vector, no bound-state solutions exist.1,2 If it is intro-
duced as a Lorentz scalar, Bhalerao and Ram3 have found
only a very limited set of solutions, with no obvious corre-
spondence to the nonrelativistic solution. Such an outcome in
the scalar case is unexpected because the Klein paradox4 is
not problematic; positive and negative energy particles both
see a confining potential. A nonrelativistic limit for the
positive-energy solutions should reproduce the known non-
relativistic spectrum.
This inconsistency in the scalar case can be resolved.5 The
solution found by Bhalerao and Ram3 turns out to be over-
constrained. Here we will construct a more general solution
and show that the nonrelativistic results are recovered in an
appropriate limit.
To see that the Dirac equation solution should match the
nonrelativistic solution, consider the equation 共with ប⫽1
⫽c兲
either Ai⬘ (⫺(2mg) 1/3˜␧ /g)⫽0 for even solutions or
Ai(⫺(2mg) 1/3˜␧ /g)⫽0 for odd solutions. Let ⫺ ␳ n and ⫺ ␳ n⬘
denote the nth zeros of Ai and Ai⬘ , respectively. Then the
nonrelativistic eigenenergies are ˜␧ n⬘ ⫽ ␳ ⬘n (g 2 /2m) 1/3 for even
solutions and ˜␧ n ⫽ ␳ n (g 2 /2m) 1/3 for odd. The values of ␳ n
and ␳ ⬘n can be obtained from tables in Ref. 7. The first four
are ␳ n ⫽2.3381, 4.0879, 5.5206, 6.7867 and ␳ ⬘n ⫽1.0188,
3.2482, 4.8201, 6.1633.
We would expect the Dirac equation to yield these same
results in the limit of weak coupling. To see that negative
energy solutions do not cause any difficulties, we can use the
methods of Coutinho, Nogami, and Toyama8 to prove the
following extension of their theorem B: For a scalar potential
that is everywhere nonnegative, the positive energy solutions
have energy E⭓m and the negative energy solutions have
E⭐⫺m. The proof depends on the freedom to pick as real
the solutions ũ and ṽ to Eq. 共3兲. The inner products of ũ and
ṽ with the terms of these equations yield
冕 ṽ ũ ⬘ dx⫹ 冕 ũ 共 m⫹V 兲 ũ dx⫽E 冕 ũ 2 dx, 共4a兲
冕 ṽ ũ ⬘ dx⫺ 冕 ṽ共 m⫹V 兲ṽ dx⫽E 冕 ṽ 2 dx, 共4b兲

关 ␣ p⫹ ␤ 共 m⫹g 兩 x 兩 兲兴 ␺ ⫽E ␺ , 共1兲 where g 兩 x 兩 has been replaced by a generic scalar potential V

6 and an integration by parts has been performed in the first
in a representation where
term of Eq. 共4a兲. For positive E and V, Eq. 共4b兲 implies that

␣ → ˜␣ ⬅ ␴ y ⫽ 冉 0
i
⫺i
0
冊 , ␤ → ˜␤ ⬅ ␴ z ⫽ 冉 1
0
0
⫺1
冊 . 共2兲
兰 ṽ ũ ⬘ dx⭓0 and then Eq. 共4a兲 yields E⭓m. Analogous steps
for negative E yield 兰 ṽ ũ ⬘ dx⭐0 and E⭐⫺m. Thus the two
parts of the spectrum are completely separate, and we are
As usual, let ␺ ⫽ ˜␺ ⬅( ũṽ ) and decompose the matrix equation allowed to focus on the positive energy solutions only.
into two coupled equations,
⫺ ṽ ⬘ ⫹ 共 m⫹g 兩 x 兩 兲 ũ⫽Eũ, 共3a兲
II. SOLUTION OF THE DIRAC EQUATION
ũ ⬘ ⫺ 共 m⫹g 兩 x 兩 兲ṽ ⫽E ṽ . 共3b兲
To solve the Dirac equation directly, we use the same rep-
For energies E⫽m⫹␧ ˜ near the rest mass m, with ˜␧ Ⰶm, and
resentation as in Ref. 3, that is,
for weak coupling gⰆm 2 , Eq. 共3b兲 yields ṽ ⯝ũ ⬘ /2m. The
substitution of this result into Eq. 共3a兲 yields ⫺ũ ⬙ /2m
⫹g 兩 x 兩 ũ⯝␧
˜ ũ, which is immediately recognized as the non-
relativistic Schrödinger equation.
␣⬅␴ y⫽ 冉 0
i
⫺i
0
冊 , ␤ ⬅ ␴ x⫽ 冉 冊
0
1
1
0
. 共5兲

The solution to the Schrödinger equation is obtained by For ␺ ⫽( uv ) they obtain the coupled equations
noting that ũ ⬙ ⫽2m(g 兩 x 兩 ⫺␧ ˜ )ũ is the differential equation
u ⬘ ⫹ 共 m⫹g 兩 x 兩 兲 u⫽E v , 共6a兲
for a shifted Airy function,7 which yields the solution ũ(x)
⫽NAi((2mg) 1/3关 兩 x 兩 ⫺␧ ˜ /g 兴 ), with N a normalization con- ⫺ v ⬘ ⫹ 共 m⫹g 兩 x 兩 兲v ⫽Eu. 共6b兲

522 Am. J. Phys. 70 共5兲, May 2002 http://ojps.aip.org/ajp/ © 2002 American Association of Physics Teachers 522
This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.97.178.73 On: Sat, 27 Sep 2014 13:24:05
 These equations decouple in terms of the variable ␰
⫽ 冑g(m/g⫹ 兩 x 兩 ), such that for x⬎0, H ␯ 共 ␰ 兲 ⬃2 ␯ e ␰ /2⌫
2
冉 冊冉 冊
␯ ⫹1
2
t␯
z 2␯ ⫺1
1/4
Ai共 t ␯ 兲 , 共12兲

冉 ⫺
d2
d␰2 冊
⫹ ␰ 2 u⫽ 共 E 2 /g⫹1 兲 u, 共7a兲 where z ␯ ⫽ ␰ / 冑2 ␯ ⫹1 and, for z ␯ ⭐1,

t ␯ ⫽⫺ 共 43 共 2 ␯ ⫹1 兲关 cos⫺1 z ␯ ⫺z ␯ 冑1⫺z ␯2 兴 兲 2/3.

冉 冊
共13兲
d2
⫺ 2 ⫹ ␰ 2 v ⫽ 共 E 2 /g⫺1 兲v , 共7b兲 Equation 共13兲 looks promising because the desired Airy
d␰
function is present. We next expand in powers of ␣ ⫺1 and ␧,
and for x⬍0, using cos⫺1(1⫺y)⯝冑2y⫹ 冑y 3 /72, to obtain

冉 ⫺
d2
d␰2 冊
⫹ ␰ 2 u⫽ 共 E 2 /g⫺1 兲 u, 共8a兲 z ␯⫽
1⫹ 冑g 兩 x 兩 / ␣
冑1⫹2␧⫹␧ 2 ⫺1/␣ 2
⯝1⫺y ⫺ , z ␯ ⫹1 ⯝1⫺y ⫹ 共14兲

冉 ⫺
d2
d␰2 冊
⫹ ␰ 2 v ⫽ 共 E 2 /g⫹1 兲v . 共8b兲
t ␯ ⯝⫺ 冉 冊
2m 4
g2
1/3
y⫺ , t ␯ ⫹1 ⯝⫺ 冉 冊
2m 4
g2
1/3
y⫹ , 共15兲
Obviously these are harmonic oscillator-type equations. The
normalizable solutions can be constructed from the Hermite with
functions9 of order ␯ and ␯ ⫹1, where E 2 ⫽2( ␯ ⫹1)g, as3 冑g 兩 x 兩

再
1
y ⫾ ⬅␧⫺ ⫾ . 共16兲
Ce ⫺ ␰ 2 /2
H ␯ ⫹1 共 ␰ 兲 , x⬎0 ␣ 2␣2

u⫽ E 2
Note that 冑g 兩 x 兩 / ␣ is of order ␧ at the classical turning point,
C⬘ e ⫺ ␰ /2H ␯ 共 ␰ 兲 , x⬍0, which sets a natural scale for x, and that the nonrelativistic
冑g

再
correspondence implies that ␧ is of order (g 2 /m 4 ) 1/3
共9兲 ⬃ ␣ ⫺4/3. Thus the ␧ 2 terms can be dropped relative to ␣ ⫺2 .
E ⫺ ␰ 2 /2
C e H ␯共 ␰ 兲 , x⬎0 To lowest order, the expansions in Eqs. 共14兲 and 共15兲 im-
v⫽ 冑g ply that
2
C ⬘ e ⫺ ␰ /2H ␯ ⫹1 共 ␰ 兲 , x⬍0. 2 2
e ⫺ ␰ /2H ␯ 共 ␰ 兲 , e ⫺ ␰ /2H ␯ ⫹1 共 ␰ 兲
However, because ␰ is always positive, ␯ is not restricted to ⬃Ai共共 2m 4 /g 2 兲 1/3关 兩 x 兩 / ␣ ⫺␧ 兴 兲 . 共17兲
being an integer.10
Continuity at x⫽0 requires that CH ␯ ⫹1 ( ␣ ) To compare with the original nonrelativistic reduction, we
⫽C ⬘ EH ␯ ( ␣ )/ 冑g and C ⬘ H ␯ ⫹1 ( ␣ )⫽CEH ␯ ( ␣ )/ 冑g, where must relate the two representations of the Dirac matrices.
␣ ⬅m/ 冑g is defined as in Ref. 3. These conditions yield They are related by a unitary transformation

冑 冉 冊
C ⬘ ⫽⫾C and i 1 1
U⫽ , 共18兲
E 2 ⫺1 1
H ␯ ⫹1 共 ␣ 兲 ⫽⫾ H ␯共 ␣ 兲 , 共10兲
冑g such that
the latter being the eigenvalue condition. This condition has
a rich set of solutions when free of the restriction to integer
␯; there are infinitely many solutions for any positive value
˜␺ ⫽ 冉冊 ũ
ṽ
⫽U ␺ ⫽U
u
v
. 冉冊 共19兲

of ␣. Therefore, we have ũ⫽ (i/&) (u⫹ v ) and ṽ ⫽ (i/&) ( v

The sign that appears in the eigenvalue condition 共10兲 cor-
responds to the parity of the solution. The parity operator11 is
reflection in x combined with multiplication by the Dirac
matrix ␤. Because ␰ is independent of the sign of x, we find
that
⫺u), with u and v given by Eq. 共9兲. Thus for large ␣, ũ does
indeed reduce to an Airy function with the correct argument,
given ␧⫽␧ ˜ /m and ␣ ⫽m/ 冑g.
To the next order in ␣ ⫺1 we obtain, with the aid of
Stirling’s formula for ⌫(( ␯ ⫹2)/2), 7

␤ 冉 u 共 ⫺x 兲
v共 ⫺x 兲
⫽冊冉
v共 ⫺x 兲
u 共 ⫺x 兲
⫽⫾
u共 x 兲
v共 x 兲
. 冊 冉 冊 共11兲 H ␯ 共 ␰ 兲 ⬃2 ␯ ⫺1/4 ␰ 2 /2
e ⌫ 冉 冊冉 冊
␯ ⫹1
2
2m 4
g2
1/12

The presence of such a symmetry is, of course, necessary for ⫻ 共 1⫹y ⫺ /4兲 Ai共 ⫺ 共 2m 4 /g 2 兲 1/3y ⫺ 兲 , 共20兲
the match to the nonrelativistic solution.
and

III. NONRELATIVISTIC LIMIT H ␯ ⫹1 共 ␰ 兲 ⬃2 ␯ ⫹3/4e ␰

2 /2
冑 ␯ ⫹1 ␯ ⫹1
2
⌫
2 冉 冊冉 冊 2m 4
g2
1/12

To recover the nonrelativistic solution, we must take the ⫻ 共 1⫹y ⫹ /4兲 Ai共 ⫺ 共 2m 4 /g 2 兲 1/3y ⫹ 兲 , 共21兲
appropriate limit. We write E⫽m(1⫹␧) and consider small
where y ⫾ is defined by Eq. 共16兲. The combination that ap-
␧ as well as small g. The latter corresponds to large ␣ and pears in the eigenvalue condition
large ␯. In the limit of large ␯, we find from Ref. 7 that H ␯
has the asymptotic form 0⫽H ␯ ⫹1 共 ␣ 兲 ⫿ 冑2 ␯ ⫹2H ␯ 共 ␣ 兲

523 Am. J. Phys., Vol. 70, No. 5, May 2002 John R. Hiller 523
This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.97.178.73 On: Sat, 27 Sep 2014 13:24:05
 value conditions. Therefore the eigenvalues will match in the
limit of large ␣ and small ␧.

IV. RESULTS AND CONCLUSIONS

Fig. 1. Lowest four eigenvalues ␧⬅E/m⫺1 for the scalar potential as func-
tions of 1/␣ ⬅ 冑g/m for 共a兲 even and 共b兲 odd parity. The solid lines are the
nonrelativistic results, and the points are positive-energy relativistic results.
then reduces to
A comparison of the relativistic and nonrelativistic eigen-
values is made in Fig. 1, where each is plotted as a function
of 1/␣ for the four lowest levels of each parity. The nonrel-
ativistic values were already obtained earlier as explicit func-
tions of ␣, which are simply plotted as lines in the figure.
The relativistic values were computed numerically with the
use of MATHEMATICA to solve the eigenvalue condition 共10兲
at selected values of ␣. The dimensionless ␧ is related to ␯ by
␧⫽ 冑2 ␯ ⫹2/ ␣ ⫺1. For 1/␣ near 0, that is, large ␣, the rela-
tivistic and nonrelativistic results are indistinguishable. As ␣
decreases, they separate smoothly.
From these results we see that the Dirac equation with a
scalar linear potential is a well-defined problem in one di-
mension, with a rich set of solutions and a smooth nonrela-
tivistic limit. As an exercise, one could extend this work to
include calculation of the relativistic wave functions and
make direct comparisons with the nonrelativistic Airy func-
tions. They will match in the large ␣ limit. A second inter-
esting exercise is a comparison of the ultrarelativistic,
strong-coupling limit of the spectrum to the ␣ ⫺4/3 behavior
of the nonrelativistic spectrum. The plots in Fig. 1 appear to
imply ␣ ⫺1 behavior for small ␣.
a兲
Electronic mail: jhiller@d.umn.edu
1
H. Galić, ‘‘Fun and frustration with hydrogen in a 1⫹1 dimension,’’ Am.
J. Phys. 56 共4兲, 312–317 共1988兲.
2
A. Z. Capri and R. Ferrari, ‘‘Hydrogenic atoms in one-plus-one dimen-
sions,’’ Can. J. Phys. 63, 1029–1031 共1985兲.
3
R. S. Bhalerao and B. Ram, ‘‘Fun and frustration with quarkonium in a

冉 冊冉 冊
1⫹1 dimension,’’ Am. J. Phys. 69 共7兲, 817– 818 共2001兲.
␯ ⫹1 2m 4 1/12 4
0⯝ 冑2 ␯ 2 ␯ ⫺1/4e ␣ /2⌫
2 The term ‘‘Klein paradox’’ refers to a scenario where transmission through
2 g2 a barrier occurs via transitions to negative-energy states. For additional
discussion and references, see Refs. 1 and 2.
5
⫻ 关共 1⫹y ⫺ /4兲 Ai共 ⫺ 共 2m 4 /g 2 兲 1/3y ⫹ 兲 See also A. S. de Castro, quant-ph/0110178, which became available after
the present work was completed.
⫿ 共 1⫹y ⫹ /4兲 Ai共 ⫺ 共 2m 4 /g 2 兲 1/3y ⫺ 兲兴 x⫽0 . 共22兲
6
The tildes are used to distinguish this representation from the one used in
Ref. 3.
First-order Taylor expansions of the Airy functions about 7
M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions
⫺(2m 4 /g 2 ) 1/3␧ yield 8
共Dover, New York, 1972兲.
F. A. B. Coutinho, Y. Nogami, and F. M. Toyama, ‘‘General aspects of the
0⯝Ai共 ⫺ 共 2m 4 /g 2 兲 1/3␧ 兲 ⫺ 共 2 ␣ 兲 ⫺2/3Ai⬘ 共 ⫺ 共 2m 4 /g 2 兲 1/3␧ 兲 bound-state solutions of the one-dimensional Dirac equation,’’ Am. J.
Phys. 56 共10兲, 904 –907 共1988兲.
⫿ 关 Ai共 ⫺ 共 2m 4 /g 2 兲 1/3␧ 兲 9
N. N. Lebedev, Special Functions and Their Applications 共Dover, New
York, 1972兲, p. 294.
⫹ 共 2 ␣ 兲 ⫺2/3Ai⬘ 共 ⫺ 共 2m 4 /g 2 兲 1/3␧ 兲兴 , 共23兲 10
The usual solution of odd-order Hermite polynomials for the half-
1
where terms of order higher than ␣ have been ⫺2/3 harmonic oscillator, where V⫽ 2 m ␻ 2 x 2 for x⬎0 but is infinite for x⬍0, is
correct because the solutions to H ␯ (0)⫽0 are the odd integers, not be-
dropped. For even parity 共the upper sign兲, we have cause ␯ is first restricted to being an integer by the boundary conditions.
Ai⬘ (⫺(2m 4 /g 2 ) 1/3␧)⯝0 and for odd parity, 11
J. D. Bjorken and S. D. Drell, Relativistic Quantum Mechanics 共McGraw–
Ai(⫺(2m 4 /g 2 ) 1/3␧)⯝0, which are the nonrelativistic eigen- Hill, New York, 1964兲, p. 24.

524 Am. J. Phys., Vol. 70, No. 5, May 2002 John R. Hiller 524
This article is copyrighted as indicated in the article. Reuse of AAPT content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.97.178.73 On: Sat, 27 Sep 2014 13:24:05