Journal of the Less-Common Metals, 90 (1983) 83-88 83

MAGNETISM OR BONDING: A NEARLY PERIODIC TABLE OF

TRANSITION ELEMENTS

J. L. SMITH and E. A. KMETKO*

Los Alamos National Laboratory, Los Alamos, NM 87545(U.S.A.)
(Received June 26,1982)

Summary

A table of the transition metal elements is constructed to show their

dominant electronic features. The strictly periodic table of the elements is
misleading for the study of many useful metallic materials. The actinide series is
central to this new view.

1. Introduction

The important properties of a large number of technical and research

materials occur because their d or f electrons are in a state of delicate balance
between magnetic and non-magnetic behavior. Itinerant electron magnetism,
hydrogen storage, catalysis, intermediate valence, structural instabilities and
many other phenomena tend to occur in transition metal materials near this
cross-over from localized electron behavior (magnetic moments) to itinerant
electron behavior (metallic bonds). The common source and systematic occur-
rence of these phenomena are not generally appreciated. It is the ideas expressed
by our figures that are new and bring together many well-known arguments.
We illustrate this cross-over with the elements, but its importance rests in
the extension to intermetallic compounds and alloys. Quantitative models for
bonding and for stability of compounds do not work well in the cross-over
region. We do not attempt to improve this situation but merely point out its
generality and offer empirical guidelines for dealing with its effects. Finally,
since the actinide elements are central to our systematics, we discuss their
behavior in some detail.
Free-atom d and f electrons usually possess magnetic moments. In the
majority of condensed phases, these moments disappear [l] because the orbitals
overlap and bond (itinerant behavior); bonding is inconsistent with the

* Deceased.

Elsevier Sequoia/Printed in The Netherlands

84

existence of magnetic moments (localized behavior) for the same electrons [a].
The most sensitive measure of the onset of this bonding is a decrease in the
atomic volume. The cross-over from magnetic moments to bonding shows up as a
sudden decrease in the lattice constant in studies of a series of pure elements, of
a series of isostructural compounds or of the compression of individual
materials.

2. The new table

The long periods of the d and f transition metals show a smooth trend
toward localized d or f electrons, i.e. the ratio of the position of the maximum of
the atomic wavefunctions to the metallic radii decreases monotonically [2]. If
the five long periods are arranged in order of decreasing localization, Fig. 1 is
obtained. This suggests that the cross-over to localized behavior can be marked
by the diagonal shaded region and that the d and f metals can be considered on
an equal footing. The precise position of the shaded region is somewhat
arbitrary and depends on which property is used as a criterion, Crossing the
shaded region in the figure from left to right can be viewed as a Mott transition
from a metallic state to an insulating magnetic state for the relevant d or f
electrons. In general the Mott model describes a metal becoming a magnetic
insulator, but here the magnetic elements remain metallic because of the
presence of other electrons. Johansson [3] originated this idea for the case of
cerium, but Fig. 1 suggests greater generality. For a compound or an alloy, if one
element is near the cross-over while the others are not, the position of the former
largely determines the electronic behavior. (Of course if one of the constituents
comes from the magnetic moment region, that moment is unaffected unless its
configuration energy happens to match that of a band state.) For mixtures
containing more than one element near the diagonal, the exact behavior is quite
sensitive to many parameters and no quantitative predictions can be made
from the figure or by any other method. However, if the behavior of such a
material is already known modifications can be made predictably by shifting
compositions with respect to the diagonal region, and irrespective of whether d
or f electrons are involved in this situation the strict arrangement of the
columns and the periods of the periodic table offer little guidance to the
tailoring of properties. Figure 1 provides such guidance.
It is implicit in Fig. 1 that properties result from the overlap of the relevant
wavefunctions and hence from the interatomic spacing of elements near the
diagonal. This has been quantitatively worked out for the actinides and their
compounds by Hill [4]. Thus behavior is shifted diagonally toward magnetic
moments in alloys or compounds by dilution with elements far from the
diagonal. Similarly, static high pressure moves the properties of materials
toward bonding because the wavefunction overlap is increased.
Since the behavior of materials is suggested by the position of its constitu-
ents with respect to the diagonal, a close examination of the shaded region is
appropriate. Materials close to this delocalization-localization transition
region can have their properties modified appreciably by small perturbations.
 85

Mognefic
MOnmfS full
partially filled shell

Zn

Cd

5d Bol Lu Hf Ta W Re OS Ir Pt 1Au Hg
/
Bonding
Fig. 1. A nearly periodic table of transition metals showing the cross-over between electron
bonding behavior and magnetic moment formation.

This is because electrons in a material that is near the transition have a larger
number of states of such properties as valence, bonding, crystal structure and
magnetic moments that are almost degenerate than is the case when the
material is far from the transition. For example, the three pure elements in Fig. 1
that possess the largest number of allotropic crystal structures are cerium,
plutonium and manganese which are all on the diagonal. This shows the
sensitivity to perturbations and the enhanced number of states. Similarly small
pressure or composition changes cause large changes in physical properties
such as those which occur in the formation of stainless steel from iron. Such
structural changes signal the importance of the precise phonon spectrum and
the possibility of martensitic transformations, valence fluctuations, charge
density waves, solitons and shape memory effects [5] in materials near this
cross-over. These materials also have a high density of electronic states at the
Fermi energy and are sensitive to the occurrence of many-body effects. All of the
above properties correlate well with interesting materials, and we suggest that
the large number of almost degenerate physical states for materials at this cross-
over, and not any other single aspect such as a high density of electronic states,
is the fundamental cause.
The multiplicity of electronic structures is what is important in chemical
processes such as hydrogen storage and catalysis. The effectiveness of materials
for such processes correlates with the cross-over in Fig. 1. The figure also
suggests low temperature measurements for studies of materials near the cross-
over. Materials from the magnetic moments region usually order magnetically,
while those from the bonding region tend to be superconductors. Materials from
the cross-over region tend to be itinerant electron magnets (easily perturbed) or
to have no long-range order. Since superconductivity and magnetism are
relatively well understood theoretically, systems of materials under study
because of other important properties should have their superconducting or
magnetic properties measured. Then such useful information as densities of
states, electron-phonon coupling or coherence lengths can be extracted.
86

It is clear that Fig. 1 is useful as an empirical guide to understanding the

roles of various elements (and the importance of their atomic volumes and
magnetic moments) in materials studied for a wide variety of reasons. Even
without quantitative understanding, such guidance can be important. For
example, the so-called Matthias rules [S] and the Bardeen-Cooper-Schrieffer
theory of superconductivity [7] have both been available for 25 years, but it is
the empirical approach which has led to the development of useful
superconductors.

3. Examples

We can discuss some specific examples in terms of Fig. 1. When nickel and
chromium are added to iron to make stainless steel the magnetism is lost, the
crystal structure is changed, the strength is decreased and the corrosion
resistance is increased. Yet the conversion back to the undesirable form (toward
the magnet moments region) occurs all too readily in welding, loading and
cooling and remains a significant area of research. The compounds ZrZn,, Sc,In
and TiBe, represent an interesting group of weakly magnetic materials [8].
They turn out to be the magnets with the magnetic element the farthest from the
diagonal region in Fig. 1. They are of course magnetic because the other atoms
have decreased the relevant wavefunction overlap. It is then clear that
materials whose behavior most departs from the pattern of Fig. 1 should be
studied with the most care. The u phase of uranium has a large reverse isotope
effect in its superconducting transition temperature [9] which is sufficiently
unusual that the results are not generally accepted. It is of course due again to
the extreme sensitivity to perturbation of materials at the cross-over, and in this
case a small change in the phonon spectrum is responsible. In contrast, the
cubic y phase of uranium has a normal isotope effect because it is out of the
cross-over region as the interatomic uranium distance is smaller [4]. There must
be other similarly large isotope effects to be found. Finally there are now
materials known such as ErRh4B, that are both magnetic and superconducting
[lo] which would seem to contradict Fig. 1. Rather, these ternary compounds
are sufficiently complex that the lattice can accommodate two types of
microscopic interaction simultaneously. Quite simply, different atoms are
involved. The rhodium atoms are bonding and provide the superconductivity
while the erbium atoms are well into the magnetic moments region and
maintain their magnetic moments as mentioned earlier.
The actinide elements provide the clearest example of the full cross-over of
behavior in Fig. 1. Figure 2 shows connected binary alloy phase diagrams for the
light actinides [ll]. The shaded areas are our guesses, but the systematics are
clear. The diagram begins and ends with nothing unusual for a transition metal.
However, there is a gradual decrease in the symmetry of the crystal structures
which then abruptly disappears just past plutonium showing the transit of the 5f
band. In the center there is a minimum in the melting points, and the many low
symmetry crystal structures all have densities greater than when the elements
 87

temperature PC)

1500
c

AC Th Pa u Np Pu Am ‘m
C..
Fig. 2. Connected binary alloy phase diagrams for the light actinides: b.c.c., body-centered cubic;
ortho, orthorhombic; mono, monoclinic; d.h.c.p., double hexagonal close packed. The shaded areas
are uncertain. The solid areas are two-phase regions.

possess one of the traditional “close-packed structures” such as face-centered

cubic (f.c.c.). These features are unique to materials that have an f electron
energy band and result from the symmetry of the f electron ligands (similar to p
electrons). The ligands do not favor high symmetry structures but prefer the
greater density obtainable in more perturbed structures or liquids [12]. The
actinides begin as superconductors with thorium, protactinium and uranium.
The next two elements, neptunium and plutonium, are almost magnetic which
inhibits superconductivity. Finally, approximately where the f.c.c. phase
reappears, the f electron band drops below the Fermi energy showing that the
diagonal intermediate region of Fig. 1 has passed. The small region of f.c.c.
structure that occurs in pure plutonium as its lowest density phase can be seen
as the tip of an f.c.c. “iceberg”. This phase in plutonium disappears under only
about 1 kbar of pressure, illustrating the importance of pressure as a parameter
for studying actinides [13] (or for the entire cross-over region of Fig. 1) where
the pressure shifts the details of the diagram to the right in Fig. 2. For example,
americium at 150 kbar and room temperature has the a-U orthorhombic
structure because its localized f electrons are squeezed into a bonding
configuration [14]. It is also clear from the figure that alloying and temperature
are crucial to actinide properties in the transition region (as in the diagonal
region of Fig. 1).
One of the longest standing current questions concerning the electronic
structure of the actinides [13] is emphasized by Fig. 1: how does the f electron
band change over to localized electron behavior? Figure 2 shows a compelling
feature: the smooth line separating the low symmetry structures from the f.c.c.
phase between plutonium and americium must mark the precise cross-over. Its
curvature to the left as the temperature increases simply shows that the lattice
vibrations overtake the last remaining low symmetry f bonding. The difficulties
are that there is not a dramatic volume change when the f electrons delocalize at
room temperature in americium under pressure 113, 141 and that there is no
magnetic moment in f.c.c. plutonium. In both cases it is likely that temperature
effects are crucial and must be dealt with properly both experimentally and
theoretically. This simply points out the continuing question regarding all of
the materials near the shaded area in Fig. 1. The figures we present here are
deceptively simple but should prove thought provoking.

Acknowledgments

We thank C. Y. Huang, A. C. Lawson and J. H. Wood for many discussions.

This work was performed under the auspices of the U.S. Department of Energy.

References

J. S. Schilling, in J. S. Schilling and R. N. Shelton (eds.), Physics of Solids Under

High Pressure, North-Holland, Amsterdam, 1981,p. 345.
J. L. Smith, Physica B, 102(1980) 22.
B. Johansson, Philos. Mag., 30 (1974) 469.
H. H. Hill, in W. N. Miner (ed.), Plutonium 1970 and Other Actinides, American
Institute of Mechanical Engineers, New York, 1970,p. 2.
T. A. Sandenaw, Ser. Metall., 15 (1980) 397.
B. T. Matthias, Prog. Low Temp. Phys., 2(1957) 138.
J. Bardeen, L. N. Cooper and J. R. Schrieffer, Phys. Rev., lOS(1957) 1175.
B. T. Matthias, A. L. Giorgi, V. 0. Struebing and J. L. Smith, J. Phys. (Paris), 39
(1978) L-441.
H. H. Hill, R. W. White, J. D. G. Lindsay, R. D. Fowler and B. T. Matthias, Phys. Rev.,
163 (1967) 356.
R. D. Fowler, J. D. G. Lindsay, R. W. White, H. H. Hill and B. T. Matthias, Phys. Rev.
Lett., 19 (1967) 892.
10 W. Fertig, D. C. Johnston, L. E. Delong, R. W. McCallum, M. B. Maple and B. T.
Matthias, Phys. Rev. Lett., 38(1977) 987.
11 W. G. Moffatt, The Handbook of Binary Phase Diagrams, General Electric Co.,
Schenectady, 1981,and references cited therein.
R. L. Rose, R. E. Kelley and D. R. Lesuer, J. Nucl. Mater., 79 (1979) 414.
12 E. A. Kmetko and H. H. Hill, J. Phys. F, 6(1976) 1025.
13 B. Johansson, H. L. Skriver and 0. K. Andersen, in J. S. Schilling and R. N. Shelton
(eds.), Physics of Solids under High Pressure, North-Holland, Amsterdam, 1981, p.
245.
14 R. B. Roof, R. G. Haire, D. Schiferl, L. A. Schwalbe, E. A. Kmetko and J. L. Smith,
Science, 207 (1980) 1353.