ENSEÑANZA REVISTA MEXICANA DE FÍSICA 48 (1) 76–87 FEBRERO 2002

A simple and effective approach to calculate the energy of complex atoms

Héctor O. Di Rocco
Instituto de Fı́sica Arroyo Seco, Facultad de Ciencias Exactas, Universidad Nacional del Centro
Pinto 399, 7000 Tandil, Argentina
Recibido el 03 de abril de 2001; aceptado el 24 de octubre de 2001

It is shown in this paper that, using only common concepts of well known modern physics and quantum mechanics textbooks (as one- and
two-electron atoms, perturbation theory), we can develop a simple and powerful method to calculate the binding energies of complex electron
configurations, as well as ionization energies, X-ray levels, etc.

Keywords: Atomic structure; Z-expansion; relativistic corrections

En este artı́culo se muestra que usando solamente conceptos comunes bien conocidos que aparecen en los libros de fı́sica moderna y de
mecánica cuántica (átomos con uno y dos electrones, teorı́a de perturbaciones), podemos desarrollar un método simple y poderoso para
calcular las energı́as de ligadura de configuraciones electrónicas complejas, energı́as de ionización, niveles de rayos X, etc.

Descriptores: Estructura átomica; desarrollo en Z; correcciones relativistas

PACS: 31.10; 31.15

1. Introduction charge (e ≈ 4.8 × 10−10 esu), the electron rest mass

(m ≈ 9.1 × 10−28 g) and the reduced Planck constant
It is well known that exact solutions of the Schrödinger equa- (~ ≡ h/2π ≈ 1 × 10−27 erg s). In this system the a.u.
tion can be found in a few cases [1–5]. With relation to ato- of length is the radius of the first Bohr orbit
mic physics, are treated in general hydrogenic atoms and the
ground configuration of He but, in this case, not all necessary ~2
a0 = ≈ 5.29 × 10−9 cm,
calculations are presented in detail [6–8]. me2
In this work it is showed that, provided only with the the time unit is
knowledge about atoms with one and two electrons, we can
treat a general theory that permit us to obtain the energy for ~3
τ0 = ≈ 2.42 × 10−17 s,
electron configurations of arbitrary complexity. The method me4
is based on the Z −1 expansion due to Layzer [9] and origi- and the energy unit can be, indistinctly, the Rydberg or the
nated in works of Hylleraas on the ground state of He. Hartree:
On one hand, this is an interesting exercise for non-
graduate students. We show as to attain data even for complex me4 e2
1 Ry = 2
= ≡ 13.6058 eV
atoms that are well compared with the experiment in diverse 2~ 2a0
cases: binding energies, energy of internal (sub) shells and 1 Ht =2Ry.
ionization energies of atoms when three or more of the va-
lence electrons are missing. For neutral or few times ionized In the following paragraphs ri = |ri | the distance of
atoms the results are encouraging (better than 75% for neu- the i-th electron from the nucleus, rij = |ri − rj | is the dis-
trals, in the more stringent case). tance between the i-th and the j-th electrons, li and si are
In this work we will use only common concepts ap- the orbital and spin angular-momentum operators, in units
pearing in quantum mechanics books: one- and two-electron of ~ and ξi (ri ) is the spin-orbit operator, measured in energy
atoms and the Rayleigh-Schrödinger perturbation theo- units. In this units system, the hamiltonian for N electrons
ry [1–5]. But, in order to give an idea of the usefulness of moving in the field of a nucleus of charge Z looks as
this approach, we gives modern references to research pa-
X X 2Z XX 2
pers. In the first part of this paper, we use a non-relativistic H=− ∇2i − +
approach whereas in the second part we use an approximate i i
ri i>j
rij
relativistic treatment. In order to simplify the notation, we in- X
dicates with m (not ml ) the component of the orbital angular + ξi (ri )(li · si ), (1)
momentum whereas we use µ (not ms ) the component of the i
spin momentum. if energies [and ξi (ri )] are measured in Ry or
n µ ¶ X
2. Atomic units 1X 2Z 1 X
H =− ∇2i + + + ξi (ri )(li · si ), (2)
2 i=1 ri r
i>j ij i
In atomic physics it is useful to use the so called ato-
mic units (a.u.), based in the elementary electron if energies [and ξi (ri )] are measured in Ht.
 A SIMPLE AND EFFECTIVE APPROACH TO CALCULATE THE ENERGY OF COMPLEX ATOMS 77

3. A brief refreshment about electron The subtle questions of the fundamental importance of
configurations and coupling schemes these schemes are out of the scope of this paper (see the book
by Messiah [4]). An specialized although elementary account
To each electron of a complex atom we can assign a pair of can be found in the books by Eisberg [11] and Woodgate [12];
quantum numbers (ni li ) in the non-relativistic approach or a a text about the importance of jj coupling in nuclear struc-
triplet of quantum numbers (ni li ji ) in the relativistic one. A ture is the one by Talmi and de-Shalit [13].
complex electron configuration is denoted, respectively, as As well as the numbers n, l (or n, l, j) indicates an elec-
tron configuration, the numbers L, S give the so called terms.
(n1 l1 )w1 (n2 l2 )w2 . . . (nq lq )wq Denonting by Eav the configuration-average energy
or P
states Ek
Eav = ,
(n1 l1 j1 )w1 (n2 l2 j2 )w2 . . . (nq lq jq )wq , number of states

where wq denote the occupation number of the electron sub- it is important to know that the energy of the terms, relative
shell. For example, for neutral Ne, the electron configuration to Eav can be found using the vector model of the atom [6,11]
can be written as: and can be calculated in terms of the so called Slater integrals
(see below). Although the deduction is not trivial and it is apt
(1s)2 (2s)2 (2p)6 for advanced courses, they can be obtained using a computer
program presented in Ref. 14. For example, for the configu-
or ration p2 (or p4 ) the followings terms appear:
(1s1/2 )2 (2s1/2 )2 (2p1/2 )2 (2p3/2 )4 , 3 2
E(3 P ) = Eav − F (pp),
25
respectively, when using the non-relativistic or the relativistic
3
solutions of the central field problem. E(1 D) = Eav + F 2 (pp),
When we consider the non-relativistic Hamiltonian 25
[Eqs. (1) or (2)], the LS (or Rusell-Saunders) coupling is 12
E(1 S) = Eav + F 2 (pp). (6)
valid when the electrostatic interactions are stronger than 25
the spin-orbit whereas jj coupling is applied in the re-
verse case [4]. When considering the approximate, relativis- 4. One- and two-electron atoms
tic Hamiltonian, the notion of jj coupling is the natural one
(remember that the spin-orbit interaction is relativistic in ori- 4.1. One-electron atoms
gin).
We denote the LS coupling by the sequence Few things are necessary to remind from the non-relativistic
©£ ¤ hydrogenic atoms:
{(L1 , L2 )L2 }L3 , . . . Lq ,
£ ¤ ª i) The discrete state energies, relative to the ionization
{(S1 , S2 )S2 , S3 }S3 , . . . Sq Jq Mq , (3) limit (taken as zero), are given by

where the script letters indicates the various intermediate and Z2

En = − .
final quantum numbers. Analogously, the jj coupling implies n2
the sequence
© ª ii) The non-relativistic wavefunction for one-electron
[(l1 , s1 )j1 , (l2 , s2 )j2 ]J2 , . . . JM. (4) atoms is of the form called spin-orbital
There are a biunivocal correspondence between the cou- Ψ(r, θ, ϕ, µ) = r−1 Pnl (r)Ylm (θ, ϕ)χ(µ), (7)
pled LS and jj states. For example:
where P (r) and Ylm (θ, ϕ) are normalized according to
s2 1 S0 ←→ (s2 )0,
the relations
2p2 3 P0 ←→ (2p1/2 )2 0, Z ∞
|Pnl (r)|2 dr = 1;
2p2 1 S0 ←→ (2p3/2 )2 0, 0
ZZ
2p2 3 P1 ←→ (2p1/2 2p3/2 )1, |Ylm (θ, ϕ)|2 sin θ dθdϕ = 1. (8)
2 3
2p P2 ←→ (2p1/2 2p3/2 )2,
Explicitly, when r is measured in units of a0
2p2 1 D2 ←→ (2p3/2 )2 2, n−l−1
X
.. Pnl (r) = Cnl rl+1 e−Zr/n ak rk , (9)
. (5) k=0

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

78 HÉCTOR O. DI ROCCO

with The difficulty in the treatment of many-electron atoms is

k in the term 2/rij of the Hamiltonian (1 or 2) that, without the
(−2Z/n)
ak = , (10) spin-orbit term, can be re-written using conventional sym-
k!(2l + 1 + k)!(n − l − 1 − k)!
bols as
µ 2 ¶
and ∇i ∇2j Z Z 1
½ ¾1/2 H(in Ht) = − + + + +
22l+2 Z 2l+3 (n − l − 1)!(n + l)! 2 2 ri rj rij
Cnl = . (11)
n2l+4 2
X
Abundant theoretical material about the spherical har- ≡ H0 (i) + H12 (i, j). (12)
monics can be found in quantum mechanics as well as in elec- i=1
tromagnetism books (see the above mentioned references, for To proceed with such term, that impedes the variable se-
example the book by Jackson [15]. paration, we introduce the normalized spherical harmonics,
according to [15]
4.2. Two-electron atoms µ ¶1/2
4π
This topic is included in some texts about modern physics [6], Cq(k) = Ykq , (13)
2k + 1
therefore we will be shyntetic in this section. A more explicit (k)
such that the matrix element, denoted hlm|Cq |l0 m0 i, results
treatment can be found in the book by de la Peña [5] as well as ZZ
in the one by Borowitz [1]. More detailed treatments are pre- hlm|Cq(k) |l0 m0 i ≡ ∗
Ylm Ykq Yl0 m0 sin θ dθdφ; (14)
sented in some books about quantum chemistry (See Refs. 7
or 8). and can be calculated directly, although it is laborious. The
We must evaluate hΨ|H|Ψi where |Ψi is a determinan- result can be expressed in terms of the 3j symbols due to
tal (antisymmetric) wavefunction and H is the Hamiltonian Wigner, that can be calculated in closed form in terms of fac-
operator. torials (this material can be found in the books of Landau-
Lifshitz or Messiah [3,4]:
ZZ
∗
£
Ylm Ykq Yl0 m0 sin θ dθdφ = (−1)−m (2l + 1)(2l0 + 1)]1/2 S3j (l, k, l0 ; 0, 0, 0)S3j (l, k, l0 ; −m, q, m0 )

= δ(q, m − m0 )ck (lm, l0 m0 ). (15)

The above equation also defines the coefficients ck (lm, l0 m0 ).

It is known from the courses about electrostatics that the following expansion is valid [15]:
∞
X k k
X
1 r< (k)
= k+1
(−1)q C−q (θ1 , φ1 )Cq(k) (θ2 , φ2 ). (16)
r12 r>
k=0 q=−k

It can be shown [7] that, if each spin-orbital is written as in Eq. (7), the average value h1/r12 i ≡ hΨ|1/r12 |Ψi is evaluated as
¿ ¯ ¯ À
¯ 1 ¯ X
i(1)j(2)¯¯ ¯t(1)u(2) = δ(µi , µt )δ(µj , µu )
¯ Rk (ij, tu)
r12
k
X
× δ(q, mt − mi )δ(q, mj − mu )(−1)q ck (li mi , lt mt )ck (lj mj , lu mu ), (17)
q

where the Rk (ij, tu) are the generalized Slater integrals (see
below in this same paragraph). The summation over k involve
only a small number of non-zero terms [see below Eqs. (18) indicates that since the electrostatic interaction does not ope-
and (19)]. Using the fundamental property of de Dirac δ func- rate on the electrons spins, not only is the total spin con-
tion, we note that the matrix elements 17 are zero unless served, but so also is the spin of each electron separately.
q = mt − mi = mj − mu , therefore mi + mj = mt + mu . In particular, from Eq. (17), it is introduced the Coulomb
This is a reflection of the conservation of angular mo- or direct integral J(ij):
mentum: the electrostatic interaction betweeen two electrons ¿ ¯ ¯ À
¯ 1 ¯
cannot change the total orbital angular momentum of the two J(ij) ≡ ij ¯ ¯ ¯ij
electrons, nor the z-component. The situation for the spins is r12 ¯
X
even more restrictive, the δ-factors giving = ck (li mi , li mi )ck (lj mj , lj mj )F k (ij), (18)
µi = µt , µj = µu , k

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

 A SIMPLE AND EFFECTIVE APPROACH TO CALCULATE THE ENERGY OF COMPLEX ATOMS 79

5. Average energies for complex configurations

TABLE I. Electron-pair interaction energies for a non-relativistic
system. In the non-relativistic approach, a complex configuration is
hssi 0
F (ss) denoted as
hppi F 0 (pp) − 2F 2 (pp)/25 (n1 l1 )w1 (n2 l2 )w2 . . . (nq lq )wq , (24)
hddi F 0 (dd) − 2F 2 (dd)/63 − 2F 4 (dd)/63
Pq
where N = j=1 wq is the total number of electrons of the
0
hss i F 0 (ss0 ) − G0 (ss0 )/2 atom or ion.
hspi F 0 (sp) − G1 (sp)/6 For the Hamiltonian [Eq. (2)], without the spin-orbit
hsdi F 0 (sd) − G2 (sd)/10 term, the energy
0
hpp i F 0 (pp0 ) − G0 (pp0 )/6 − G2 (pp0 )/15 E = hΨ|H|Ψi
0 1 3
hpdi F (pd) − G (pd)/15 − 3G (pd)/70 can be written as the sum of kinetic, electron-nuclear and
hdd0 i F 0 (dd0 ) − G0 (dd0 )/10 − G2 (dd0 )/35 − G4 (dd0 )/35 electron-electron Coulomb energies:
X X X
Eav = Ek (i) + En (i) + E(i, j)
and the exchange integral K(ij): i i i>j
¿ ¯ ¯ À X X
¯ 1 ¯ = I(i) + E(i, j). (25)
K(ij) ≡ ij ¯ ¯ ¯ji
r12 ¯ i i>j

X For example, for the neutral Be

= δ(µi , µj ) [ck (li mi , lj mj )]2 Gk (ij). (19)
Eav = 2I(1s) + 2I(2s) + E(1s, 1s) + E(2s, 2s)
k

We can see, using the properties of the 3j symbols [3] +4E(1s, 2s).
(and therefore of the ck coefficients, both derived from the This very complex many-body problem can be solved us-
integral over spherical harmonics) that: ing the Hartree-Fock approach (a numerical one), as can be
i) l + k + l0 must be even. briefly viewed in the books by Eisberg [11], Landau [3] or
Messiah [4]. Instead, in this paper we use the following ap-
ii) The triangle relation of the classical vector model must proach.
be satisfied: |l − l0 | ≤ k ≤ l + l0 .
In Eq. (19) F k (i, j) and Gk (i, j) are [6,14] 6. The Z −1 expansion
F k (ij) ≡ Rk (ij, ij)
Layzer’s formulation of the Z-dependent theory of the many-
ZZ k
r< electron atom can be regarded as the starting point [9]. From
= k+1
|Pi (r1 )|2 |Pj (r2 )|2 dr1 dr2 , (20) the approximate Hamiltonian in atomic units (e = me = ~ = 1
r>
and energies measured in Hartrees) given by Eq. (2):
Gk (ij) ≡ Rk (ij, ji) n µ ¶ X
1X 2 2Z 1
ZZ k H(N, Z) = − ∇i + + , (26)
r< 2 i=1 ri r
= P (r )Pj (r1 )Pi (r2 )Pj (r2 ) dr1 dr2 . (21)
k+1 i 1
i>j ij
r>
and introducing the new variable ρ = Zr, Eq. (26) becomes
Now it is necessary to average over the magnetic quantum (the parameter λ is commonly used in place of Z −1 )
numbers. For non-equivalent orbitals:
H(N, Z) → Z 2 H(N, λ) = Z 2 (H0 + λV ), (27)
E(ij) = hij|1/r12 |ijiav − hij|1/r12 |jiiav
µ ¶2 where
1 X li k lj n µ ¶
0
= F (ij) − Gk (ij), (22) 1X 2 2 X 1
2 0 0 0 H(N, λ) = − ∇i + +λ .
k 2 i=1 ρi ρ
i>j ij
whereas for the equivalent ones:
The first term of the right hand side,
µ ¶2
2li + 1 X li k lj n µ ¶
0
E(ii) = F (ii) − F k (ii). (23) 1X 2
4li + 1 0 0 0 H0 = − ∇2i + (28)
k>0 2 i=1 ρi
Slater integrals can be calculated in closed form if hydro- is a sum over non-interacting hydrogenic Hamiltonians, and
genic orbitals are used, but the final result is long and compli- X 1
cate. Electron-pair interaction energies for a non-relativistic V ≡ H1 = . (29)
ρ
i>j ij
system are in Table I.

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

80 HÉCTOR O. DI ROCCO

If Z −1 (= λ) is a small parameter, the Rayleigh-Schrö-

dinger perturbation theory (RSPT) can be applied [4,11]. TABLE II. Some Average Coulomb energies E1 calculated in terms
Layzer showed that, within the framework of the Z −1 -de- of Slater integrals with Z = 1. More results can be found in the
pendent theory, the wave function and the total energy can be Refs. 16 and 17.
written as the expansions He 0.62500
2 Li 1.02281
Ψ(N, λ) = Ψ0 + λΨ1 + λ Ψ2 + . . . (30)
Be 1.57100
and B 2.33445
E(N, λ) = E0 + λE1 + λ2 E2 + . . . (31) C 3.27251
N 4.38518
Then O 5.67245
E(N, Z) = Z 2 E0 + ZE1 − E2 + . . . (32) F 7.13433
Ne 8.77083
where, exactly
n
1X 1
E0 = hψ0 |H0 |ψ0 i = − (33) If we restrict the expansion (32) up to E2 , the energy can
2 i=1 n2i
and be written as
X w (Z − σ )2
nl nl
E1 = hψ0 |H1 |ψ0 i. (34) E=− , (35)
2n2
n,l
According to the Eqs. (22) and (23), E1 is given in terms where wnl is the number of electrons in the n, l shell and
of Slater’s integrals F k and Gk evaluated with hydrogenic σnl is the corresponding screening parameter. Comparing
wavefunctions with Z = 1. In Table II we display these ave- Eqs. (32) and (35), we find that the σnl satisfy
rage Coulomb energies for the ground configurations of He Xw
nl
to Ne, in order to explain our following results. More va- E1 = σ , (36)
n2 nl
lues are exposed in the works of Safronova et al. [16] and the n,l
and
present author [17]. We remark the recent publication of these Xw
nl 2
last cited works in order to indicate how a simple approach E2 = σ . (37)
2n2 nl
can give valuable theoretical data apt to interpret complex n,l
experiments. Then, to second-order approximation in the non-relativistic
context, the average energy of a configuration is given by

µ ¶ · X X ¸
Z2 w2s + w2p w3s + w3p + w3d 1
Eav = − w1s + + + · · · +Z w (w − 1)E1 (ii) + wi wj E1 (ij) − E2 , (38)
2 4 9 2 i i i i,j

where wi is a short notation for wn l , the number of elec-

i i
trons in the ni , li shell.
7. Firsts examples of applications
6.1. About the concept of screening
7.1. Binding energies and relativistic contributions
The concept of screening (and screened orbitals) is of old
data and it is impossible to give a short account in this pa- For the He atom, E1 = F 0 (1s, 1s) = 0.625, therefore
per. In the past, screening parameters were obtained using from Eq. (36), σ1s = 0.3125, a result known from varia-
spectroscopic data, numerical calculations and theoretical ap- tional calculations (see Appendix A). With Z1s = 1.6875,
proaches. A short review can be found in the paper from the E(He) = −2.8477 Ht; better results could be obtained
author [17]. if relativistic corrections are employed. In this case,
In more refined approachs the screening parameters all σ1s = 0.2961 and E = −2.9033 Ht; almost the experimental
have Z expansions of the form value.
For Li I: 1s2 2s, we have E1 = 1.0228 = 2σ1s + σ2s /4.
s = s0 + s1 Z −1 + s2 Z −2 + . . . , Assuming that σ1s = 0.3125 (neglecting the external
screening), σ2s = 1.5912 and E(Li I) = −7.4707 Ht. In-
whereas other authors used electrostatic considerations and troducing the concept of external screening, we write
deduced parameters dependent of both Z, N : σ(Z, N ) [17]. σ1s = 0.3125 + g(1s, 2s), estimate g(1s, 2s) accord-
Looking for simplicity, in this work, we will use only the ing to Appendix B and calculate σ2s . For the Li I
concept of external screening, as shown below and in the Ap- case, g(1s, 2s) = 0.0143, therefore σ2s = 1.4771 and
pendix B. E(Li I) = −7.4361 Ht, instead of the previous value

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

 A SIMPLE AND EFFECTIVE APPROACH TO CALCULATE THE ENERGY OF COMPLEX ATOMS 81

TABLE III. Center of gravity binding energies in Rydbergs for TABLE IV. K-shell binding energies in eV for some representative
ground configurations [including relativistic corrections according neutral atoms.
to Eq. (39)]. The values with a (∗ ) were calculated using Hartree- Atom Ours Experiment
Fock methods because no experimental values are available.
Be 123.8 119.30
Atom This method Sucher [24] Experiment C 297.6 283.8
He 5.695 6.0 5.807 Ne 876.0 870.10
Be 29.245 29.50 29.337 Mg 1316.3 1311.20
C 75.580 75.50 75.712 Si 1856.2 1846.00
Ne 257.460 269.60 258.102 Ar 3211.6 3202.90
Mg 400.960 414.20 400.621 Zn 10392.8 10367.1
Si 580.060 593.80 579.732 32 11131.9 11103.1
Ar 1058.040 1072.80 1058.23 33 11897.6 11866.7
Zn 3585.28 3626.00 3588.46∗ 34 12690.1 12657.8
Kr 5573.66 5570.00 5575.30∗ 35 13509.6 13473.7
Kr 14356.2 14325.6

of −7.4707 Ht. The experimental one is −7.4344 Ht. The

IP is obtained after calculating E(Li II). The He-like Li II this behaviour is easy to understand from Eq. (32). In parti-
has σ1s = 0.3125, therefore E(Li II) = −7.2227 Ht imply- cular, it is a simple exercise to show that
ing that neglecting g(1s, 2s), I = 0.248 Ht ≡ 6.75 eV, Z2
whereas Iexp = 5.39 eV. Considering now the external a= 2,
2nk
screening, I = 0.2134 Ht ≡ 5.807 eV. where nk is the principal quantum number of the removed
Using Z-independent screening parameters we electron [19].
have for the Be atom σ1s = 0.3125 + 2 × 0.0143.
Then, σ1s = 0.3125 + 2 × 0.0143 = 0.3410 and accord- 7.3. K-shell binding energies
ing to Eq. (36) and Table II, σ2s = 1.8349 and thus
E(Be I) = −14.56 Ht. The experimental value is found to In Table IV we present the K-shell binding energies for some
be −14.6685 Ht. The difference between these values lies neutral atoms from Be to Kr. We establish a comparison with
within 0.7%. experimental results tabulated in the Handbook of Chemistry
and Physics [20]. In general, we can see that our simple ap-
Proceeding in this form, in Table III we show binding
proach gives an agreement of the order of 0.2%.
energies for a number of elements in order to compare with
the values reported by the experiment [18]. The most impor-
tant relativistic contribution is due to the more internal elec- 8. Now, the relativity, why?
trons and can be estimated as a simple sum over terms of the We see that the most important relativistic contribution to the
form total energy is due to the more internal electrons and can be
µ ¶ well estimated as a simple sum over terms of the form given
α2 Zef
4
4n
Erel = − − 3 wnl Ht. (39) by the Eq. (39).
8n4 j + 1/2
However, we know from the modern physics courses
7.2. The isoelectronic sequences of neon and argon that in the X-ray spectra appear a fine structure such that,
for example, the 2p1/2 electrons have a notorious different
In Ref. 17 we exhibit, as examples, the ionization poten- energy than the 2p3/2 ones [6,11,21]. We will show that, al-
tials for the Ne I and Ar I isoelectronic sequences cal- though the energy of the valence electrons do not differ ap-
culated by means of our approach. A comparison with preciably between relativistic or non-relativistic treatments,
the experiment indicates that agreement is within 76% we can easily calculate the different subshell energies.
for Ne I and better than 3% for the third member of
the serie. Our IP’s can be fitted by the adjusted em- 8.1. Approximate relativistic wave functions and some of
pirical curve IP(eV) = 128.29 − 45.42Z + 3.4231Z 2 . For their properties
the Ar I isoelectronic sequence, the fit function is The relativistic theory of the H atom is briefly presented in
IP = 285.16 − 43.06Z + 1.5861Z 2 . Now, the IP’s are bet- the books by Merzbacher [2] and also in the one by Mes-
ter than 7% for the 4-th member of the serie. Also the exper- siah [4]. Also brief are the expressions for the hydrogen rela-
imental ionization potentials follow an empirical law of the tivistic radial wave functions that can be found in these clas-
type sic texts.
They have a large and a small component, denoted re-
I = aZ 2 + bZ + c; spectively by Fnlj (r) and Gnlj (r) (other authors use the

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

82 HÉCTOR O. DI ROCCO

reverse notation!), with the general property that, when With these quantities, and calling for brevity rFnlj ≡ Pnlj ,
(Zα)2 → 0, Fnlj (r) → Rnl (r), and Gnlj → 0. The norma- our hydrogen radial wave-function has the form [2]
lization condition is µX n0 ¶
v+λ
Z ∞
¡ 2 ¢ Pnlj (r) = Cnj fv r e−Zr/N . (41)
Fnlj + G2nlj r2 dr = 1. (40) v=0
0 In Eq. (41)
In our case, we introduce one heuristic approach apt to the (2λ)!(2Z/N )v+λ−1 n0 !
purpose of this paper and also to diverse applications, there- fv = (−1)v (v − n0 + N − κ), (42)
v!(2λ + v)!(n0 − v)!
fore we take Gnlj = 0 and normalize Fnlj (r) according to
Eq. (40). Heuristic approachs are very common when model- for v = 0 up to v = n0 − 1 and
ling complex physical situations, for example when atomic (N − κ)(2λ)!(2Z/N )v+λ−1
and fluid equations are coupled or when plasmas in non- fv = (−1)v , (43)
(2λ + v)!
thermal equilibrium exists [22]. In this form we have a wave
for v = n0 . The normalization constant is found by perform-
function for each sub-shell defined by the individual quan-
ing the integral given by the Eq. (40) with the appropriate
tum numbers (nlj) and we can calculate the corresponding
values of fv . In such manner, we find in place of Eq. (11),
energies for each subshell.
" #1/2
Taking into account that diverse notations exists for the (2Z/N )2λ+1
relativistic functions, we shall summarize the symbols to be Cnj = P2n0 , (44)
α
used. Given the quantum numbers n, l, we construct the fol- α=0 Aα Γ(α + 2λ + 1)(N/2Z)
lowing quantities, with α = 1/137.037: with
α
X
1
j+ = l + , Aα = fβ fα−β . (45)
2
¯ ¯ β=0
¯ 1 ¯¯
¯
j− = ¯l − ¯, A general property of these wave functions is that, denot-
2 +
ing by Pnl the non-relativistic functions, Pnlj the relativistic
·µ ¶2 ¸1/2 −
functions with j = j+ and by Pnlj when j = j−
1
λ+ = j+ + − α2
2 +
Pnlj ' Pnl ,
£ ¤1/2
= (l + 1)2 − α2 → (l + 1), and
·µ ¶2 ¸1/2 −
1 Pnlj ' crl e−Zr/N + Pnl .
λ− = j− + − α2
2
8.2. jj coupling and the calculation of the Slater integrals
= (l2 − α2 )1/2 → l,
We give here an heuristic point of view, indicating that a
n0+ = n − j+ − 1/2 = n − l − 1,
correct relativistic Hamiltonian for many-electron atoms was
n0− = n − j− − 1/2 = n − l, derived by Breit. The electrostatic part of the relativistic en-
· µ ¶¸1/2 ergy of an atom is the straightforward generalization of the
2 0 1 nonrelativistic energy
N+ = n − 2n+ j+ + − λ+ → n,
2 X X
· µ ¶¸1/2 E= I(i) + [J(i, j) − K(i, j)], (46)
1 i i,j
N− = n2 − 2n0− j− + − λ− → n,
2 where the energies must be calculated usingrelativistic wave-
µ ¶ functions.
1
κ+ = − j+ + = −(l + 1), In a analogous way to the non-relativistic theory, the en-
2 ergy of an atomic configuration may be expressed in terms
µ ¶
1 of Slater integrals, nowin the jj coupling scheme. For a
κ− = + j− + = +l. two-electron system and from Eq. (46), the electron-electron
2
Coulomb energy can be written (neglecting here magnetic
and retardation effects) as
¿ ¯ ¯ À X
¯ 1 ¯ £ ¤
¯
na la ja , nb lb jb ¯ ¯na la ja , nb lb jb = fk (a, b)F k (a, b) − (−1)ja +jb +J gk (a, b)Gk (a, b) . (47)
r12 ¯
k

Now the diverse summations, corresponding 3 − nj symbols and coefficients fk and gk depend on quantum numbers l0 s
and j 0 s and not on l0 s, L0 s and S 0 s. In the Appendix, we give explicit expressions for these coefficients and the calculated ones

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

 A SIMPLE AND EFFECTIVE APPROACH TO CALCULATE THE ENERGY OF COMPLEX ATOMS 83

TABLE V. Electron-pair interaction energies for a relativistic system.

hs+ s+ i F 0 (s+ s+ )
hp− p− i F 0 (p− p− )
hp+ p+ i F 0 (p+ p+ ) − F 2 (p+ p+ )/15
hd− d− i F 0 (d− d− ) − F 2 (d− d− )/15
hd+ d+ i F 0 (d+ d+ ) − 24F 2 (d+ d+ )/525 − 10F 4 (d+ d+ )/525
hs+ s0+ i F 0 (s+ s0+ ) − G0 (s+ s0+ )/2
hs+ p∓ i F 0 (s+ p∓ ) − G1 (s+ p∓ )/6
hs+ d∓ i F 0 (s+ d∓ ) − G2 (s+ d∓ )/10
hp− p0− i F 0 (p− p0− ) − G0 (p− p0− )/2
hp− p+ i F 0 (p− p+ ) − G2 (p− p+ )/10
hp− d− i F 0 (p− d− ) − G1 (p− d− )/6
hp− d+ i F 0 (p− d+ ) − G3 (p− d+ )/14
hp+ p+ 0 i F 0 (p+ p0+ ) − G0 (p+ p0+ )/4 − G2 (p+ p0+ )/20
hp+ d− i F 0 (p+ d− ) − G1 (p+ d− )/60 − 9G3 (p+ d− )/140
hp+ d+ i F 0 (p+ d+ ) − G1 (p+ d+ )/10 − G3 (p+ d+ )/35
hd− d0− i F 0 (d− d0− ) − G0 (d− d0− )/4 − G2 (d− d0− )/20
hd− d+ i F 0 (d− d+ ) − G2 (d− d+ )/70 − G4 (d− d+ )/21
hd+ d0+ i F 0 (d+ d0+ ) − G0 (d+ d0+ )/6 − 4G2 (d+ d0+ )/105 − G4 (d+ d0+ )/63

ones are presented in Table V. For some examples,

TABLE VI. Relativistic Slater’s integrals (in Ht).
hs+ s+ i = F 0 (s+ s+ );
Integral This work Asaad [23]
0 F 2 (p+ p+ ) F 0 (1s, 1s) 0.624926 0.625003
hp+ p+ i = F (p+ p+ ) − .
15 F 0 (1s, 2s) 0.209844 0.209891
The Slater parameters F k , Gk for hydrogenic wavefunc- F 0 (1s, 2p− ) 0.242852 0.242822
tions can be calculated analytically, but the expressions are F 0 (1s, 2p+ ) 0.242800 0.242817
tedious and not very illuminating as to be presented here. The F 0 (2s, 2s) 0.150367 0.150389
results shown small differences with the non-relativistic ones F 0 (2s, 2p− ) 0.162117 0.162113
but the jj approach permit us to calculate the subshell bind- F 0 (2s, 2p+ ) 0.162107 0.162112
ing energies (and X-ray ones, if we like). Some examples can
F 0 (2p− , 2p− ) 0.181667 0.181629
be viewed in Table VI and compared with the numbers of
F 0 (2p− , 2p+ ) 0.181652 0.181627
Asaad [23]. So, the term energies E0 and E1 of a given con-
figuration can be calculated. For the neutral Ne, for example, F 0 (2p+ , 2p+ ) 0.181637 0.181625
F 2 (2p+ , 2p+ ) 0.087712 0.087909
E0 (Ne I) = 2I(1s) + 2I(2s) + 2I(2p− ) + 4I(2p+ ), G0 (1s, 2s) 0.021921 0.021942
G1 (1s, 2p− ) 0.051225 0.051195
where
G1 (1s, 2p+ ) 0.051180 0.051194
I(nl± ) ≡ Inj G1 (2s, 2p− ) 0.087955 0.087910
· µ ¶¸ G1 (2s, 2p+ ) 0.087968 0.087910
1 (Zα)2 4n
=− 1 + − 3 , (48) G2 (2p− , 2p+ ) 0.087714 0.087909
2n2 4n2 ji + 1/2
and
Average Coulomb energies E1 calculated in term of
E1 (Ne I) = h1s1si + h2s2si + 4h1s2si + 4h1s2p− i
Slater integrals with Z = 1 are presented in Table VII.
+8h1s2p+ i + 4h2s2p− i + 8h2s2p+ i + h2p− 2p− i With respect to the terms, we must consider the relations
between the Slater integrals F k (ii), Gk (ij) and their rela-
+6h2p+ 2p+ i + 8h2p− 2p+ i. tivistic counterparts. We develop here one case; other ones

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

84 HÉCTOR O. DI ROCCO

TABLE VII. Electron pair energies (in Ht). TABLE IX. LII − LIII shell binding energies (in eV) for some
representative atoms.
h1s+ 1s+ i 0.624926
h1s+ 2s+ i 0.198833 Z Element Our values (eV) Experiment (eV) [20]
h1s+ 2p− i 0.234314 18 Ar 247.0 247.3
h1s+ 2p+ i 0.234270 245.2 245.2
h2s+ 2s+ i 0.150367 31 Ga 1135.07 1142.3
h2s+ 2p− i 0.147458 1112.22 1115.4
h2s+ 2p+ i 0.147445 32 Ge 1239.22 1247.3
1212.82 1216.7
h2p− 2p− i 0.181667
33 As 1350.06 1358.6
h2p− 2p+ i 0.172881
1319.70 1323.1
h2p+ 2p+ i 0.175790 34 Se 1464.62 1476.2
1429.88 1435.8
35 Br 1583.97 1596.0
TABLE VIII. Some relationships between the non-relativistic and 1544.40 1549.9
the relativistic expressions for the Slater integrals. 36 Kr 1708.15 1727.2
1663.27 1674.9
F 2 (pp) [2G2 (−+) + F 2 (++)]/3
2
F (dd) [7F 2 (−−) + 6G2 (−+) + 12F 2 (++)]/25
F 4 (dd) [4G4 (−+) + F 4 (++)]/5
· µ ¶¸
0
G (ss) G0 (++) Z 2 X wi (Zα)2 4ni
Eav = − 1+ −3
G1 (sp) [G1 (+−) + 2G1 (++)]/3 2 i n2i 4n2i ji + 1/2
2
G (sd) [2G2 (+−) + 3G2 (++)]/5 · X X ¸
1
+Z w (w −1)E1 (ii)+ wi wj E1 (ij) −E2 , (49)
2 i i i i,j

are in Table VIII. From Tables I and V with E1 and E2 satisfying the relations (36) and (37). Ioni-
0 2 zation energies for the valence electrons do not differ ap-
hppi = F (pp) − F 2 (pp), preciably from the non-relativistic case. As examples, LII
25
and LIII shell bindings energies are in Table IX. Agreement
hp− p− i = F 0 (p− p− ),
with experimental values are better than 1% [20].
1 2
hp+ p+ i = F 0 (p+ p+ ) − F (p+ p+ ),
15
10. Comparison with other simple approachs
1
hp− p+ i = F 0 (p− p+ ) − F 2 (p− p+ ).
10 Some simple approachs for the calculation of atomic struc-
tures were published in the last two decades. In 1978,
In the respective complete shells there are 15 pairs pp, 1
Sucher [24] presented a simplified version of the atomic
pair p− p− , 6 pairs p+ p+ and 8 pairs p− p+ . Multiplying ade-
shell model and calculated the ground-state energy of any
quately and equalizying, results in
atom. The agreement with Hartree-Fock calculations was
F 0 (−−) + 6F 0 (++) + 8F 0 (−+) within 5% for He and Ne and better than 0.5% for Z > 45.
F 0 (pp) = No calculations for ions or for shell binding energies were
15
presented, nor for excited
P states. On the other hand, based
and in the use of the virial ri Fi as the model potential energy
F 2 (++) + 2G2 (−+) operator, Kregar [25] calculated screening parameters for any
F 2 (pp) = . configuration. These works of Kregar were generalized by the
3
present author [26].
When comparing the above mentioned papers with the
present approach, we can conclude that
9. Further applications and results i) Our results for ground configurations binding energies
are clearly better than the values presented by Sucher
We base our heuristic approach in the expansion given by (see Table III).
Eq. (32) which contains itself a contribution proportional
to Z 4 , because E0 is now given by Eq. (49). To second or- ii) Our values for binding energies and ionization poten-
der approximation tials are very similiar to those calculated by Kregar.

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

 A SIMPLE AND EFFECTIVE APPROACH TO CALCULATE THE ENERGY OF COMPLEX ATOMS 85

iii) We can calculate sub-shell binding energies (and X-ray Appendix B

spectra) whereas the works of Sucher and Kregar are,
essentially, non-relativistic. A simple approach to the screening
As was said above, there are many works about screening
11. Conclussions
parameters and screened functions. Here, we gives a simple
The heuristic point of view presented in Ref. 17 using the point of view apt to calculate external screenings [27]. Total
Z-expansion theory supplemented with the inclusion of the screenings are deduced according to Eq. (36) (see the exam-
external screening concept is generalized to the relativistic ples given above).
case in the jj coupling. We use the normalized large com- Schrödinger equation is non-separable, due to the term
ponent relativistic wave-function such that we can calculate
1
the binding energies for each orbital indexed by the n, l, j = (ri2 + rj2 − 2ri rj cos ωij )−1/2 .
quantum numbers. From the analysis of the tables we can rij
obtain the following conclusions. The center-of-gravity bind-
ing energies calculated in the Z −1 approach are in very good Making u = 1/ri , v = 1/rj , f = 1/rij and developing
agreement with experimental values. As can be viewed in a two-variable function in a Taylor series to a first order
Ref. 17 the ionization potentials for the (Ne I) and (Ar I) · ¸
∂f ∂f
isoelectronic sequence are within 3% for three times ionized f (u + h, v + k) ≈ f (u, v) + h+ k ,
atoms and better for higher ionization degrees: 1% for five ∂u ∂v
times ionized atoms and 0.5% for the 15-th spectra of the se- it results
quence. Results are worse for neutrals (within 25%) yet better
than other screening approaches; for example, the values of ¡ −2 ¢−3/2
1 ¡ −2 ¢
−2 −1/2 u−3
0 u0 + v0−2
Safronova et al. [16] are negative for low ionization degrees. = u 0 + v0 +
rij ri
K- and L- shell binding energies are in very good agreement
with experiment [20]. An important aspect to be taken into ¡ ¢
−2 −3/2
−2
¡ −2 ¢
−2 −3/2 v0−3 u−2
0 + v0
account is that, in our approach there are no adjustable pa- −u0 u0 + v0 +
rameters. Better results can be attained with small efforts in rj
the calculation of external screening parameters but exten- ¡ ¢
−2 −3/2
−v0−2 u−2
0 + v0 .
sive use of research papers must be made. We think that the
numbers obtained with this simple approach will give to the
Because the sum of the 1st, 3rd and 5th terms are zero,
student a clear idea of the power of the approximation meth-
the second one (and mutatis mutandis) the fourth can be put
ods applied to complex many-body problems.
in the form
½ · µ ¶−2 ¸¾−3/2 · µ ¶−2 ¸−3/2
Appendix A v0 v0
u−3
0 u−2
0 1 + 1 +
u0 u0
The variational calculation of the He ground = ;
ri ri
configuration energy
therefore
This topic can be found in the books by Messiah [4], Karplus-
Porter [7] and Levine [8]. If we assign to each electron of 1 Aij Aji
= + ,
the 1s2 configuration an hydrogenic radial wave-function rij ri rj
with effective charge Ze
with
P1s = 2Ze3/2 re−Ze r · µ ¶−2 ¸−3/2
v0
and using the Hamiltonian (2) without the spin-orbit term, Aij = 1 + .
u0
then
hHi = −2Ze2 + 4Ze (Ze − 2) + 1.25Ze . Using for the mean values

The values of hHi and Ze corresponding to the minimum 2Z

u0 = hni li |r|ni li i−1 =
are determined by differentiation; that is, from 3n2i − li (li + 1)
dhHi we have, calling xij to
= −4Ze + 8Ze − 8 + 1.25 = 0;
dZe
v0 3n2j − lj (lj + 1)
therefore Ze = 2 − 5/16 = 1.6875 and hHi = 5.6953 xij = =
Ry = 77.49 eV. u0 3n2i − li (li + 1)

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

86 HÉCTOR O. DI ROCCO

that tions [25]:

½ · 2 ¸ ¾ µ ¶(2nj +1) " 2nX
i −1 µ ¶k #
3nj − lj (lj + 1) 2 −3/2 1 (2nj + 1)! yij
Aij = 1 + . gij = .
3n2i − li (li + 1) 1 + yij k!(2nj )! yij + 1
k=0

When j > i we are calculating the external screening para- Appendix C

meters.
In principle, a best recipe is given below, but the deduc- Calculation of the coefficients fk and gk for the
tion is longer. In the spirit of this paper, the present point relativistic case
of view is sufficient. Since the application of this formula is
straightforward, is not necessary to give a table of these pa- In place of the non relativistic expressions presented in Ta-
rameters. ble I, now we have [28]
Observe that for an orbital pair i, j we call P ¡ ¢
¡ ¢ (2Ji + 1)fk la ja , lb jb ; J
xij = h1/ri i/h1/rj i, and yij = (ni /nj )xij . We propose fk la ja , lb jb = P ,
the following expression, based in electrostatic considera- (2Ji + 1)
where
¡ ¢
fk la ja , lb jb ; J = (−1)J+la +lb +1 (2ja + 1)(2jb + 1)
­ ° ° ®­ ° ° ®
×S6j (J, ja , jb ; k, jb , ja )S6j (1/2, ja , la ; k, la , ja )S6j (1/2, jb , lb ; k, lb , jb ) la °C (k) °la lb °C (k) °lb

and
P
(2Ji + 1)gk (la ja , lb jb ; J)
gk (la ja , lb jb ) = P
(2Ji + 1) 12. Acknowledgments
where The major part of this work was prepared when the author
gk (la ja , lb jb ; J) = (−1)J+jb −ja +1+k (2ja + 1)(2jb + 1) was at the International Centre for Theoretical Physics, as
2
an Regular Associate Member, the Associateship privilege
×S6j (J, ja , jb ; k, ja , jb )S6j (1/2, ja , la ; k, lb , jb ) is strongly acknowledged. Supports of the Consejo Nacional
­ ° ° ®2 de Investigaciones Cientı́ficas y Técnicas (CONICET, Ar-
× la °C (k) °lb .
gentina) and the Universidad Nacional del Centro, are ac-
The so called 6j symbols as well as the expressions for knowledged. The help of Dr. Juan Pomarico in the writing
the matrix elements can be found in the book by Messiah [4]. of this paper is also acknowledged.

1. S. Borowitz, Fundamentals of Quantum Mechanics, (Ben- 10. B. Edlén, “Atomic Spectra”, edited by S. Flugge, Handbuch
jamin, New York, 1967). der Physik, (Springer-Verlag, Berlin, 1964).
2. E. Merzbacher, Quantum Mechanics, 2nd edition, (John Wiley, 11. R.M. Eisberg, Fundamentals of Modern Physics, (John Wiley,
New York, 1970). New York, 1961).
3. L.D. Landau y E.M. Lifshitz, Mecánica Cuántica no Rela- 12. G.K. Woodgate, Elementary Atomic Structure, 2nd edition,
tivista, (Reverté, Barcelona, 1967). (Clarendon Press, Oxford, 1980).
4. A. Messiah, Quantum Mechanics, (Interscience, New York, 13. A. de Shalit and I. Talmi, Nuclear Shell Theory, (Ac. Press,
1961). New York, 1963).
5. L. De la Peña, Introducción a la Mecánica Cuántica, (CECSA,
14. G. Fernández, R. Solanilla, and H.O. Di Rocco, Rev. Mex. Fı́s.
Mexico, 1979).
44 (1998) 92.
6. M. Alonso and E. Finn, Fı́sica III: Fundamentos Cuánticos y
Estadı́sticos, (Fondo Educativo Interamericano, Bogotá, 1971). 15. J.D. Jackson, Classical Electrodynamics, 2nd edition, (John
Wiley, New York, 1975).
7. M. Karplus and R.H. Porter, Atoms and Molecules, (Benjamin,
Menlo Park, 1970). 16. U.I. Safronova et al., Phys. Scripta 47 (1993) 364.
8. I.N. Levine, Quı́mica Cuántica, (Editorial AC, Madrid, 1977). 17. H.O. Di Rocco, Nuovo Cimento D 20 (1998) 131.
9. D. Layzer, Ann. Phys. 8 (1959) 271; Ann. Phys. 29 (1964) 101; 18. C.E. Moore, Atomic Energy Levels, (U.S. Govt. Printing Off.,
Int. J. Quant. Chem. 1S, (1967) 45. Washington, 1971).

Rev. Mex. Fı́s. 48 (1) (2002) 76–87

 A SIMPLE AND EFFECTIVE APPROACH TO CALCULATE THE ENERGY OF COMPLEX ATOMS 87

19. R.J.S. Crossley and C.A. Coulson, Proc. Phys. Soc. 81 (1963) 24. J. Sucher, J. Phys. B: Atom. Molec. Phys. 11 (1978) 1515.
211.
25. M. Kregar, Phys. Scripta 29 (1984) 438; Phys. Scripta 31
20. “CRC Handbook of Chemistry and Physics,” 80th edition,
(1985) 246.
(CRC Press, Boca Raton, FL, 1999), p. 10.
21. B.K. Agarwal, X-Ray Spectroscopy, 2nd edition, (Springer- 26. H.O. DI Di Rocco, Brazilian Journal of Physics 22 (1992) 227.
Verlag, Berlin, 1991).
27. N. Bessis and G. Bessis, J. Chem. Phys. 74 (1981) 3628.
22. D. Salzmann, Atomic Physics in Hot Plasmas, (Oxford Univ.
Press, New York, 1998). 28. B.W. Shore and D.H. Menzel, Principles of Atomic Spectra,
23. W.N. Asaad, Z. Phys. 203 (1967) 362. (Wiley, New York, 1968).

Rev. Mex. Fı́s. 48 (1) (2002) 76–87