Vous êtes sur la page 1sur 9

Home Search Collections Journals About Contact us My IOPscience

Controlling orbital collapse from inside and outside a transition element

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

1998 J. Phys. B: At. Mol. Opt. Phys. 31 3557

(http://iopscience.iop.org/0953-4075/31/16/009)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 128.193.15.53
The article was downloaded on 01/05/2011 at 07:09

Please note that terms and conditions apply.


J. Phys. B: At. Mol. Opt. Phys. 31 (1998) 3557–3564. Printed in the UK PII: S0953-4075(98)92554-8

Controlling orbital collapse from inside and outside a


transition element

J P Connerade† and V K Dolmatov‡


† The Blackett Laboratory, Imperial College, Prince Consort Road, London SW7 2BZ, UK
‡ S V Starodubtsev Physical-Technical Institute, G Mavlyanova Str. 2, 700084 Tashkent,
Uzbekistan

Received 18 March 1998

Abstract. We report a study of bimodal behaviour of atomic wavefunctions for a transition


element (Cr) and how this behaviour can be controlled and even turned into single-mode character
either by (i) varying the effective nuclear charge for nonintegral values, or (ii) placing the atom
in a spherical cavity of adjustable radius. Our conclusions have relevance to the emergence of
valence instabilities for transition metals in the solid state. Also it is shown that the very existence
of the bimodal wavefunctions provides an explanation for the frequently experienced instability
of self-consistent field algorithms when a double-valley potential occurs, and suggestions are
made on how to improve the numerical procedure for such calculations.

1. Introduction

The phenomenon of orbital collapse in atomic species, arising when the atomic potential is
of a double-well form, has important consequences for many properties of atoms as well as
molecules, clusters and solids, and therefore the phenomenon has attracted much attention
for some time (Griffin et al 1969, 1971, Connerade 1978a, 1997, Karaziya 1981). The
expectation is normally to find orbitals entirely on one side or on the other of the potential
barrier once a double-well potential has developed.
However, another possibility exists which is much rarer, namely that the eigenfunctions
should possess two maxima of not too dissimilar amplitudes, one on each side of the barrier.
We refer to this as a bimodal solution, one mode being considered as an eigenfunction of
the inner well and the other, of the outer well. Bimodal solutions were first found to
occur naturally in the nf series of Ba+ by Connerade and Mansfield (1975). They were
shown to account for an anomalous variation of the quantum defect which had initially
been described by Saunders et al (1934) as a perturbation of a novel type, but had never
properly been accounted for. Another important aspect of bimodal solutions is the ease
with which the self-consistent field procedure may ‘flip’ from one solution to another,
leading to ambiguities in the interpretation of Koopman’s theorem. In critical cases, it may
become impossible to decide which is the ‘correct’ solution. Thus Band and Fomichev
(1980) raised the question whether there might not be a coexistence of collapsed and
‘anticollapsed’ states for rare-earth elements, as a result of their Dirac–Fock calculations.
Similarly, Connerade and Mansfield (1982) found two different solutions satisfying the usual
tests for Hartree–Fock wavefunctions in Ba+ . It was suggested by Connerade (1982, 1983)
that, while the coexistence of solutions is unlikely for free atoms, both solutions might
possess physical significance for cases of mixed valence in solids, thereby suggesting

0953-4075/98/163557+08$19.50
c 1998 IOP Publishing Ltd 3557
3558 J P Connerade and V K Dolmatov

an alternative theory to parametrized approaches based on the Anderson impurity model.


Independently, Schlüter and Varma (1983) reached similar conclusions on the basis of a
Thomas–Fermi approximation. Band et al (1988) developed a quasiatomic Dirac–Fock
model for intermediate valence in solids by using Wigner–Seitz boundary conditions. The
particular model of Schlüter and Varma (1983) was criticized by Bringer (1983), but the
question of the applicability of such quasiatomic models to atoms in the condensed phase
remains open. They provide a natural explanation for the occurrence of valence instabilities
in specific regions of the periodic table, and they also do not rely so heavily on adjustable
parameters as alternative methods.
In the light of the important consequences of the collapsed or bimodal-collapsed
character of eigenfunctions for certain species, the possibility of controlling orbital collapse
acquires significance. It was first found by Connerade (1978b, c) that the bimodal character
of the nf series of Cs or Ba+ could be controlled by exciting a valence electron, and that,
in favourable cases, the variation might be such as to make bimodal solutions appear and
to enable the relative amplitudes of the inner and outer maxima to be controlled. Some
experimental progress along these lines was made (Maiste et al 1980a, b, Lucatorto et al
1981).
Until recently, the nf series of Ba+ seemed to be the only natural occurrence of the
bimodal behaviour of eigenfunctions. Relatively recently, however, Dolmatov (1993a, b)
(see also Dohrmann et al 1996) demonstrated the bimodal character of an excited 3d∗
(anti)collapsed orbital (arising from a 3p → 3d∗ transition) in free neutral Cr and also
satisfactory control of such a behaviour by slightly varying the nuclear charge Z in self-
consistent-field Hartree–Fock equations, to account for the unique properties of the 3p
absorption spectrum of free Cr. Actually, the theoretical basis for this procedure can be
traced back to the foundations of the g-Hartree method. Dietz has shown (see Connerade
and Dietz (1987) for details) that there is a family of central potentials represented as
gV direct +(1−g)V exchange (V direct and V exchange are the direct and exchange parts of the atomic
potential, respectively) which makes the action defined on the Lagrangian of quantum field
theory (Itzykson and Zuber 1980) stationary. It turns out that solutions to this problem can
be approximated by introducing a slight, fractional additional effective nuclear charge in
the Hartree–Fock equations. The physical interpretation of this approximation is that the g-
Hartree potential interpolates between the N-electron V N and (N −1)-electron V N−1 atomic
potentials in an optimized manner (Connerade et al 1984, 1985) and therefore, roughly the
same effect can also be achieved by a small variation of the nuclear charge Z.
In this paper, we demonstrate one more way of controlling the bimodal character of
collapsed eigenfunctions of the atom, namely by placing the atom inside a spherical cavity
of adjustable radius R0 , with finite or infinite potential wells given by a potential step V0 ;
this is achieved by altering the external boundary conditions for the atomic wavefunctions
in self-consistent-field Hartree–Fock equations. This time the atom chosen for control is
not a rare-earth element, but belongs to the 3d transition sequence.

2. Theoretical method

Our calculations are performed in a spin-dependent self-consistent field scheme, namely, the
spin-polarized Hartree–Fock approximation (SPHF) (Slater 1974), applied to calculations of
Cr. Both free Cr and Cr confined in a spherical cavity are considered. For the latter case,
the SPHF approximation is generalized and used to study confined atoms for the first time.
The ground state of Cr is characterized by two half-filled 3d5 ↑ and 4s1 ↑ subshells
(↑ (↓) denotes the upward (downward) spin direction), whose electrons all have a co-
Controlling orbital collapse 3559

directed spin orientation (chosen explicitly as the upward direction), according to Hund’s
rule. In the frame of SPHF, all otherwise closed subshells nl 2(2l+1) are divided into two
subshells with opposite spin orientations, nl↑2l+1 and nl↓2l+1 , because of differences in
the exchange interaction experienced by nl↑ and nl↓ electrons, caused by the imbalance
in overall number of nl↑ and nl↓ electrons in the atom. Thus, the SPHF ground-state
configuration of Cr looks as follows
. . . 3p3 ↑ 3p3 ↓ 3d5 ↑ 4s1 ↑.
The inner-shell excited configuration of Cr we consider in this paper is due to a
3p↓ → 3d↓ transition:
. . . 3p3 ↑ 3p2 ↓ 3d5 ↑ 3d∗ ↓ 4s↑,
in which a 3p↓ electron is excited to an initially unoccupied 3d∗ ↓ orbital; the 3d↑ and 4s↑
electrons are considered as just spectators to the inner-shell transitions.
As follows from the SPHF approximation adopted in this paper (see also Amusia et al
1983), the single-electron radial wavefunctions (Pnlµ (r)) as well as single-electron energies
(Enlµ ) acquire their spin dependence from the explicit dependence on z-projection (µ) of
the electron spin, and can be obtained by solving the SPHF radial equation (Slater et al
1969)
!
eff
−1 d2 Znlµ l(l + 1)
− + Pnlµ = Enlµ Pnlµ (r). (1)
2 dr 2 r 2r 2
eff
The operator Znlµ , which can be regarded as the operator of an effective nuclear charge
seen by an electron in the orbital nlµ, is defined by
(  
eff
Znlµ Z X 1 n0 l 0 µ0
− Pnlµ (r) = − + V (r) + (Nn0 l 0 µ0 − δnlµ,n0 l 0 µ0 ) Y0 n0 l 0 µ0 ;
r r n0 l 0 µ0
r r
  )
Nnlµ − 1 X k 1 nlµ
− c (l0; l0) Yk nlµ; Pnlµ (r)
2l k>0
r r
X  
Nn0 l 0 µ0 ck (l0; l 0 0) n0 l 0 µ0 Pn0 l 0 µ0 (r)
−2δµ,µ0 Y k nlµ; . (2)
k,n0 l 0 6=nl
[(4l + 2)(4l 0 + 2)]1/2 r r
Here atomic units are used; Nnlµ is the number of electrons in the subshell nlµ; the
summations extend over all occupied states n0 l 0 µ0 in the atom; the coefficients ck represent
products of three spherical harmonics; the functions Yk are defined as
  Z ∞ k
1 n0 l 0 µ0 r<
Yk nlµ; = Pnlµ (r 0 )Pn0 l 0 µ0 (r 0 ) dr 0 (3)
r r 0 r>k+1
where r< and r> represent the smaller and the greater of r 0 and r, respectively. We have
also inserted an additional radial potential V (r) in (2), to simulate the situation of the atom
confined inside a spherical cavity.
It follows from (1) and (2) that radial wavefunctions (and also the phenomenon of
eff
orbital collapse) can be controlled by changing the effective nuclear charge Znlµ seen by
an electron.
eff
One way to change Znlµ ‘from the inside’, is to run a set of calculations of orbitals by
inserting different values for the nuclear charge Z into SPHF equations (1) and (2). The
other, opposite, way we are going to consider in detail later in this paper, is a perturbation
‘from outside’, i.e. applying a spherically symmetric pressure to the outermost boundary of
3560 J P Connerade and V K Dolmatov

the atomic species by placing it in the centre of a spherical cavity of radius R0 with infinite
potential step V0 :
(
0 if r < R0
V (r) = (4)
V0 if r > R0

and letting Pnlµ satisfy the boundary condition Pnlµ |r=R0 = 0.


In this work, the SPHF equations were solved within spheres of different radii R0 and
a potential step V0 of −10 au was applied at r = R0 . The consistency of our method
was tested (i) by checking that our solutions reproduce the exact hydrogenic results of
Sommerfeld and Welker (1934), (ii) by checking that, for R0 → ∞, our solutions tend to
those of the free atom and (iii) by checking that, for a given value of R0 , the total energy
is reasonably stable for further increases in the height of the potential step.

3. Results and discussion

3.1. Controlled collapse as a function of Z


In figure 1, we show the binding energies of the 3d↑, 4s↑ and 3d∗ ↓ electrons, together with
the total energy of the system, plotted as a function of Z, the atomic number, considered
as a nonintegral quantity (see the previous section). Notice the variations in binding energy
(especially for the excited 3d∗ ↓ orbital) which are indicative of orbital collapse, and the
sensitive dependence on even very slight changes in Z. Notice how the 3d5 ↑ and 4s↑
binding energies vary as a function of Z, and how this variation is counteracted by the
variation for 3d∗ ↓. This behaviour will be further commented on below.
Next, we show, in figure 2, the behaviour of the 3d∗ ↓ excited orbital. This is a separate
orbital for the excited electron, because the 3d5 ↑ group is a spectator group already present
in the ground-state configuration before the 3p↓ electron is excited.

Figure 1. Binding energies for the 3d5 ↑, 4s1 ↑ and 3d∗ ↓ electrons as well as the total energy
E total of free Cr as a function of the nuclear charge Z.
Controlling orbital collapse 3561

Figure 2. A set of radial wavefunctions P3d∗ ↓ (R) of the excited 3d∗ ↓ orbital (arising from the
3p↓ → 3d↓ transition) of free Cr for different values of the nuclear charge Z, as marked in the
figure. Zones X and Y are circled for the sake of clarity.

The most noteworthy feature of figure 2 is the fact that there exist two classes of
solutions to the SPHF equations, characterized by two different zones (marked X and Y )
in the figure, through which, to a good approximation, all the curves of a given family
must pass. These two zones allow us to distinguish between the two classes of solution.
In a double-well potential, the complete solution is made up from eigenfunctions of each
individual well joined smoothly at some boundary between the two wells. The curves may
thus be joined either to the right or left of the maximum in the eigenfunction of the outer-
well. Since, for high Z, the outer electron tends to be attracted inwards, the wavefunction
obtained by matching the curves to the right of the maximum (type 1 solution) are favoured,
whereas the curves obtained the other way (type 2 solution) are favoured at lower Z.
An interesting question, however, is what happens in between these two situations, i.e.
under circumstances in which the two eigenfunctions are joined close to the maximum of
the outermost one. It is very difficult to obtain SCF solutions in this parameter range, as
the algorithm used to optimize the wavefunction becomes unstable around this zone. The
reason for this is that the trial function can ‘flip’ from type 1 to type 2 between iterations.
Indeed, we conclude from our study that current self-consistent field algorithms are
inadequate in such a situation, and that new procedures are required for critical double-well
potentials. Rather than allow the solutions to flip between type 1 and type 2, one should
construct a complete map for each type of solution independently. A future possibility
might be, in order to inhibit the instability, to construct a new algorithm by making small
changes in the trial functions close to either of the X or Y , and larger changes elsewhere, so
that it remains either a type 1 or a type 2 solution throughout the calculation. A convenient
factor for this purpose would be
(r − a)2
ξ=
(r + a)2
which tends to 1 when r → 0 and when r → ∞ but is zero when r = a, and a is taken as
the radius rX or rY . If part of the correction applied to the trial function is multiplied by ξ ,
the effect should be to inhibit ‘flipping’ in the unstable range. However, such a procedure
would be unsuitable in this paper, because it would effectively force the wavefunction
to be of type 1 or type 2, whereas our purpose is to show without biasing them in any
3562 J P Connerade and V K Dolmatov

way, that these bimodal solutions arise naturally. Consequently, we were unable to obtain
convergence in the range between Z = 24.045 and 24.050 where the solution suddenly flips
from type 2 to type 1.
As the calculations show, altering the nuclear charge turns out to control orbital collapse
in much the same way as exciting a valence electron to a Rydberg state (Connerade 1978b, c).

3.2. Controlled collapse inside a cavity


Another way of controlling orbital collapse is to place the excited atom within a spherical
cavity (Connerade 1997), which is easily achieved by altering the external boundary
conditions and solving the Hartree–Fock equations within a sphere of given radius. In
a sense, altering the nuclear charge and placing the atom in a spherical cavity can be
regarded as two opposite forms of perturbation, one being the most internal and the other,
the most external to the atom.
Atoms in spherical cavities were considered very early in the development of quantum
mechanics. A hydrogen atom placed inside a spherical cavity with impenetrable walls was
considered by Sommerfeld and Welker (1938), who showed that, from the s states of the free
hydrogen atom, exact solutions could be obtained. Even at this early stage, they recognized
the connection this problem has to the theory of condensed matter. More recently, various
applications in solid state physics were reviewed by Jaskólski (1996). Connerade (1997)
used confined atoms in the context of ion storage to treat controlled orbital collapse in
many-electron atoms contained in a spherical cavity, and showed that, because of orbital
collapse, the ground configurations of these atoms depend on the size of the cavity.
In figure 3, we show how the 3d∗ ↓ wavefunction of the 3p↓ → 3d∗ ↓-excited
configuration of Cr behaves when placed in a spherical cavity consisting of a potential
step of height 10 au. We have found by calculation that the solutions are stable in energy
for a given radius as the height of the step is increased, once levels of about 10 au are
reached. The graph shows 3d∗ ↓ wavefunctions for different radii. This figure should be
compared with figure 2 above. Note how the same bimodal character of the wavefunctions

Figure 3. A set of wavefunctions P3d∗ ↓ (R) for the excited 3d∗ state (arising from the
3p↓ → 3d↓ transition) of confined Cr for different values of the radius R0 of a spherical
cavity, as marked in the figure. Zones X and Y are circled for the sake of clarity. Note that, as
was pointed out in the text, the outer zone Y spreads out, simply because of the phase shift of
the outer eigenfunction produced by the presence of the outer cavity wall.
Controlling orbital collapse 3563

Figure 4. Binding energies for the 3d5 ↑, 4s1 ↑ and 3d∗ ↓ electrons as well as the total energy
E total of confined Cr as a function of the radius R0 of a spherical cavity.

emerges, and that orbital collapse can equally well be controlled by using an external cavity
or by varying the nuclear charge. Indeed, much of the character of the functions in figure 2
is preserved in figure 3. Even the inner zone X occurs at approximately the same radius in
both figures. The outer fixed zone Y is not as well preserved in figure 3, simply because of
the phase shift of the outer eigenfunction produced by the presence of the outer cavity wall.
Again, we were unable to obtain convergence in a critical range between R0 = 11 and 14.
In figure 4, we show the binding energies of the 3d and 4s electrons and the total energy
of the configuration. The significant point to note is the ‘atomic swing’ effect (Connerade
1997) in which the binding energy of the spectator spin-up valence electrons is decreasing,
while the binding energy of the excited 3d∗ ↓ orbital at first increases and then decreases as
the cavity radius is reduced. As a consequence of these opposing motions, the total energy
of the configuration remains remarkably steady in the same range. Notice also the faster
change of binding energy in the range 11 < R0 < 14, which happens to be precisely the
range in which computations become very difficult and we failed to obtain convergence,
essentially for the reasons given above.

4. Conclusion

The calculations we have described demonstrate that orbital collapse may be controlled
either from the centre, by altering the binding strength of the inner well, or from the outer
reaches of the atom, by confining the outer well.
This conclusion gives some insight into the observed difference between the spectra of
free atoms and of the corresponding solids. Indeed, in a solid, the transfer of an electron
from the inner well to the outer reaches of an individual atom can arise from a combination
of two effects: first, as outer electrons are delocalized into the conduction band, the effective
nuclear charge appears to be altered. Second, since the atom is contained inside a Wigner–
Seitz cell, there is also an external effect due to the confining cavity. It is, therefore, not
3564 J P Connerade and V K Dolmatov

surprising, just by analogy with the effects of an external cavity or of a fractional nuclear
charge as considered in this paper, that, for example, the spectrum of metallic chromium
should bear very little relation to the spectrum of the free atom, although in cases where
complete atomic orbital collapse occurs, giant resonances persist with very similar properties
from the free atom to the solid (Sonntag et al 1969).
Another application of the methods we have described arises when an atom is trapped
inside a spherical fullerene cage. The simplest model for this situation is a sphere of charge,
whose potential can be represented in a similar way to the model we have described. Indeed,
we are currently extending our method to provide a simple description of endohedrally
captured atoms.

Acknowledgments

This research was supported by the Royal Society of London under a Joint Research
Agreement. JPC is grateful to colleagues from the Uzbekistan Academy of Sciences for
hospitality at the Starodubtsev Institute (Tashkent) where this research was carried out.

References

Amusia M Ya, Dolmatov V K and Ivanov V K 1983 Sov. Phys.–JETP 58 67–72


Band I M and Fomichev V I 1980 Phys. Lett. 75A 178
Band I M, Kikoin K A Trzhavskovskaya M B and Khomskii D I 1988 Sov. Phys.–JETP 67 1561
Bringer A 1983 Solid State Commun. 46 591
Connerade J-P 1978a Contemp. Phys. 19 415
——1978b J. Phys. B: At. Mol. Phys. 11 L409
——1978c J. Phys. B: At. Mol. Phys. 11 L381
——1982 J. Phys. C: Solid State Phys. 15 L367
——1983 J. Less-Common Met. 93 171
——1997 J. Alloys Compounds 225 79
Connerade J-P and Dietz K 1987 Comment. At. Mol. Phys. XIX 283
Connerade J-P, Dietz K, Mansfield M W D and Weymans G 1984 J. Phys. B: At. Mol. Phys. 17 1211
Connerade J-P, Dietz K and Weymans G 1985 J. Phys. B: At. Mol. Phys. 18, L309
Connerade J-P and Mansfield M W D 1975 Proc. R. Soc. A 346 565
——1982 Phys. Rev. Lett. 48 131
Dohrmann Th, von dem Borne A, Verweyen A, Sonntag B, Wedowski M, Godehusen K, Zimmermann P and
Dolmatov V K 1996 J. Phys. B: At. Mol. Opt. Phys. 29 4641–58
Dolmatov V K 1993a J. Phys. B: At. Mol. Opt. Phys. 26 L585–8
——1993b J. Phys. B: At. Mol. Opt. Phys. 26 L393–8
Griffin D C, Andrew K L and Cowan R D 1969 Phys. Rev. 177 62
Griffin D C, Cowan R D and Andrew K L 1971 Phys. Rev. A 3 1233
Itzykson C and Zuber J B 1980 Quantum Field Theory (New York: McGraw-Hill)
Jaskólsky W 1996 Phys. Rep. 271 1-66
Karaziya R I 1981 Sov. Phys.–Usp. 24 775
Lucatorto T J, McIlrath T, Sugar J and Younger S M 1981 Phys. Rev. Lett. 47 1124
Maiste A A, Ruus R E, Ruchas S A, Karaziya R I and Elango M A 1980a Sov. Phys.–JETP 51 474
——1980b Sov. Phys.–JETP 52 844
Saunders F A, Schneider E G and Buckingham E 1934 Proc. Nat. Acad. Sci. USA 20 291
Schlüter M and Varma C M 1983 Helv. Phys. Acta 56 147
Slater J C 1974 The Self-consistent Field for Molecules and Solids (New York: McGraw-Hill)
Slater J C, Mann J B, Wilson T M and Wood J H 1969 Phys. Rev. 184 672–94
Sommerfeld A and Welker H 1938 Ann. Phys., Lpz. 32 56
Sonntag B, Haensel R and Kunz C 1969 Solid State Commun. 7 597

Vous aimerez peut-être aussi