Solid State Sciences 7 (2005) 1464–1474

www.elsevier.com/locate/ssscie

Effects of relativistic motion of electrons on the chemistry of gold and

platinum
Martin Jansen ∗
Max-Plank-Institut für Festkörperforschung, Heisenbergstrasse 1, D-70569 Stuttgart, Germany
Received 24 March 2005; accepted 7 June 2005
Available online 25 October 2005
Dedicated to my esteemed colleague C.N.R. Rao on the occasion of his 70th birthday

Abstract
Experimental evidence proving the unique stabilization of the 6s orbital in platinum and gold is presented. The conclusions are drawn from the
chemical reactivities, of both elements, as well as from structural and spectroscopic features of selected compounds. In particular, the opening of
a band gap in transparent CsAu and Cs2 Pt, backed by band structure calculations, are regarded conclusive indications of Au− and Pt2− to exist
as closed shell species in these compounds.
 2005 Elsevier SAS. Open access under CC BY-NC-ND license.

Keywords: Aurides; Platinides; Relativistic effects; Band structure calculation

1. Introduction the relativistic motion for the chemical and physical properties
of the elements.
When in 1926 Erwin Schrödinger formulated his famous Qualitatively, Dirac was quite aware of the differences of
equation, he disregarded the fact that the effective mass of an electronic structure calculations based on his theory in compar-
electron depends on the velocity by which it moves. However, ison to non-relativistic treatments, however, he underestimated
already within the terms of the old quantum theory, Sommer- seriously their importance for chemistry [5]. Meanwhile, with
feld had explained the fine structure of hydrogen as an rel- relativistic calculations of the electron structures of all rele-
ativistic effect [1,2]. The same formula for the fine structure vant atoms available, it has been shown that the differences
of hydrogen-like atoms results when applying P.A.M. Dirac’s are clearly significant, and by no means negligible. The con-
equation [3–5], published in 1928. Since then, a vast amount sequences for the chemical properties of the elements have
of literature concerning the proper theoretical treatment of rel- been discussed extensively by Grant [10], and Pyykkö and De-
ativistic effects on the atomic structure and chemical bonding sclaux [12]. In particular, P. Pyykkö has shown up the horizon
has been published. The early period was characterized on one of expectations, in a series of reviews [13–16], also considering
hand by some disputes about the different ways the electron recent experimental results. In general, the achievements of the-
spin was treated in Schrödinger’s and Dirac’s equations, on the ory are well documented, and thoroughly reviewed in a number
other by the development of computational techniques for solv- of essays [17–22] and monographs [23–28].
ing Dirac’s equation [6–9]. The first relativistic calculations of Being experimentalists, we felt attracted by this subject be-
atomic structures by I.P. Grant [10], and, more comprehensive, cause of the strong effects as predicted. Moreover, it appeared
by J.P. Desclaux [11] mark a mile stone, in as much as they pro- necessary to us to provide a broader basis for an experimental
vided a first quantitative basis for studying the significance of validation of the theoretical results. The calculations are ex-
tremely demanding, still nowadays, and a number of regrettable
approximations have to be accepted. The relativistic effects as
* Tel.: +49 711 6891500; fax: +49 711 6891502. discussed in the literature result from a comparison of two dif-
E-mail address: [email protected] (M. Jansen). ferent ways of calculation, and in principle one or both of them
1293-2558  2005 Elsevier SAS. Open access under CC BY-NC-ND license.
doi:10.1016/j.solidstatesciences.2005.06.015
 M. Jansen / Solid State Sciences 7 (2005) 1464–1474 1465

may be numerically unreliable. That’s why particularly in this 2. Evidence for the stabilization of the 6s orbital of gold
case experimental verifications are needed.
Qualitatively, the effects resulting from taking the depen- 2.1. Atomic data
dence of the electron mass on its velocity into account were
addressed already by Sommerfeld [1,2] and Dirac [4]. Since The purely atomic quantities “ionisation energy” and “elec-
then this picture has been refined, and extended by the so called tron affinity” most significantly and unequivocally document
indirect effects (see Table 1). In particular, the numerical re- a pronounced lowering of the 6s orbital energy in the region
sults by Grant [10] and Desclaux [11] have allowed for making around gold, in the periodic system. Here, where the 4f and
quantitative comparisons. Most illustrative is the graph of the 5d contractions superimpose, we find a maximum in ionisation
6s orbital contraction over the atomic number as chosen by energies over the atomic numbers: Pt (9.0 eV), Au (9.23 eV),
Hg (10.44 eV), Tl (6.11 eV), and Pb (7.42 eV), all data from
Pyykkö and Desclaux [12] which is given in Fig. 1. Here also
[29]. As a particularily prominent feature, gold shows the high-
the 5d shell is included, showing an orbital expansion due to
est electron affinity of all metals (cf. Fig. 2), approaching the
an indirect (relativistic) effect. In this representation gold and
values as found for the halogens. Although not an atomic prop-
platinum appear conspicuously unique with respect to the rel-
erty, the lattice constant of elemental gold (a = 407.8 pm [30])
ativistic 6s orbital contraction. For this reason, we have started
falling below the one of isostructural silver (a = 408.6 pm [30])
our work directed towards gaining experimental evidence for provides a rather firm argument, in the context of this discus-
the effects calculated with these elements. However, it has to sion.
be admitted already at this point that such an effort will be se-
riously biased since a significant share of the stabilisation of 2.2. The case of CsAu
the 6s orbital results from conventional orbital contraction and
shielding effects (cf. final paragraph). First evidence for an easy formation of a compound CsAu
was deduced by W. Biltz and F. Weibke from the volume
Table 1 shrinkage at reacting gold and (among other alkali metals) ce-
Relativistic effects sium [31]. The composition was confirmed by A. Sommer [32],
1. Direct Spin-orbit splitting and the CsCl type of structure was assigned [33,34]. From a se-
Instead of the orbital angular momentum l and the spin angular ries of physical investigations, such as optical absorption spec-
momentum s their vector sum j = l + s is used (e.g., splitting of tra [33], X-ray photoemission and Mößbauer spectra [35], it
a p orbital to the p1/2 and p3/2 subshells).
was concluded to be a semiconductor exhibiting a considerable
2. Direct Contraction of s and p orbitals
The contraction is due to the relativistic mass increase as amount of ionic bonding, and containing gold with one extra
followed by the decrease of the Bohr radius. The contraction of localized electron as a negatively charged auride ion. This ideal
the spherical p1/2 subshell is comparable to the contraction of picture has been disturbed significantly by some further litera-
s1/2 . The p3/2 subshell contracts much less. ture reports, stating a significant mutual replacement of Cs and
3. Indirect Expansion of d and f orbitals
Au to take place [34], the colour to be black [36], or a positive
The s and p orbitals contracting inside and outside the d and f
orbitals screen the nuclear attraction of the latter thus leading to temperature coefficient of the electron resistivity [37].
their radial expansion. Unambiguously assigning effective charges to atoms consti-
tuting a collective solid is principally impossible. When trying
to achieve this goal, employing theoretical tools, the arbitrari-
ness results from the necessary partitioning of space and at-
tributing the appropriate volume increments to a given atom.
While experimental probes sensitive to the effective charges
of atoms strongly vary in the spatial range they cover. How-
ever, integral overall charges of complex or ionic entities unveil
themselves in experiments involving ion transport, be it in solid
or liquid electrolytes.

Fig. 1. The relativistic contraction of the 6s orbital and expansion of the 5d and
the 5d∗ orbitals for the elements with Z from 55 to 100 based on calculations
by Desclaux [11]. Fig. 2. Electron affinities of selected elements. The data are taken from [71].
1466 M. Jansen / Solid State Sciences 7 (2005) 1464–1474

2.3. Diffusion of Au− in ionic solids According to Eq. (1), two rather wide band gap educts
(Cs2 O: 1.9 eV [41], CsAu: 2.6 eV [33]) have undergone an all
In order to check for the mobility of Au− as an integral en- solid state reaction forming another insulating (ionic) product
tity in ionic solids, we have reacted solid Cs2 O and CsAu in Cs3 AuO, with a band gap of 2.35 eV [42]. This observation
a temperature regime of 300 ◦ C, which is far below the de- provides strong evidence for all valence electrons being local-
composition or melting temperatures of these reactants. Indeed, ized, allowing for assigning the most probable and reasonable
according to Eq. (1) a novel cesium auride oxide was obtained charge distribution of (Cs+ )3 Au− O2− . Besides structural fea-
in quantitative yields [38]. tures, like bond lengths, also Au LIII and Au LI XANES give
Cs2 O + CsAu = Cs3 AuO, (1) unambiguous support to this view (Fig. 4(a) and (b), respec-
tively) [43]. The positions of absorption edges, as measured by
Cs3 AuO, adopting the ideal hexagonal perovskite type of struc-
ture (Fig. 3(b)), is of light orange to yellow colour, and ex- XANES, correlate with the valence states. In the case of the Au
tremely sensitive to moisture and carbon dioxide of the at- LIII spectra, however, the exact position of the edge is hidden by
mosphere. Single crystals grow if Cs2 O and CsAu are reacted the pre-edge feature, corresponding to an excitation of an s elec-
in an excess of elemental cesium as a flux which is removed by tron into empty d states. The areas of those pre-edge features
evaporation, after the reaction is finished. Along similar routes, correspond to the number of empty d states. Fig. 4(a) shows a
also K3 AuO and Rb3 AuO are accessible, while no indications nice correlation with the intuitively assigned oxidation states,
have been obtained of analogous lithium or sodium compounds more importantly, the results are not in conflict with assigning
to exist [39,40]. In contrast to Cs3 AuO, the potassium and ru- a negative valence state to gold, since LIII XANES features no
bidium auride oxides are metallic (opaque) and are crystallizing pre-edge peaks for the respective compounds. Au LI XANES
as ideal cubic anti-perovskites (Fig. 3(a)). produces chemical shifts that are significantly lower for CsAu

Fig. 3. Crystal structures of A3 MO ((a), A = K, Rb; (b), A = Cs) with the coordination polyhedra around Au− and O2− emphasized (blue spheres: alkali metal
atoms; yellow spheres: gold atoms; red spheres: oxygen atoms; orange lines: unit cell edges).

Fig. 4. Au LIII (a) and Au LI (b) XANES of selected gold compounds, normalized to their maximum intensity, and dependence between the energy position of the
Au LI absorption edges and the oxidation state of gold (c).
 M. Jansen / Solid State Sciences 7 (2005) 1464–1474 1467

Fig. 5. Crystal structures of Cs7 Au5 O2 (top) and Rb5 Au3 O2 (bottom). For colour code see Fig. 3.

and the A3 AuO-family, as compared to elemental gold, again are best described as intergrowths of slabs corresponding to
documenting their exceptional character (Fig. 4(b) and (c)). the binary aurides (CsAu, RbAu) and to the ternary aurates(I)
(Cs3 AuO2 , Rb3 AuO2 ), see Fig. 5. Seemingly, these building
2.4. Disproportionation of gold units can be combined in varying compositions; this potential,
however, has not yet been fully explored. The Mößbauer spec-
It is well known that the halogens Cl2 , Br2 and I2 dispropor- tra [45] clearly confirm the presence of gold in two strongly
tionate in basic aqueous solutions according to Eq. (2). different valence states (Fig. 6), with the chemical shifts corre-
sponding to Au+ and Au− , respectively.
Cl±0 − +1 −
2 + 2OH = [Cl O] + Cl
−1 + H O.
2 (2)
Most amazingly, the notoriously “noble” metal gold easily re- 2.5. Parallels between aurides and halides
acts in a similar way, forming well crystalline solids containing
disproportionated gold [44,45], Eqs. (3)–(5). If we regard Au− a closed shell species, one would expect a
number of qualities, e.g., in its crystal chemistry, to be similarly
2Cs2 O + 5Au±0 + 3Cs = [CsAu−1 ]4 [Cs3 Au+1 O2 ], (3)
found with the halides. Indeed, one could merge the two latter
2Rb2 O + 5Au±0 + 3Rb = [RbAu−1 ]4 [Rb3 Au+1 O2 ], (4) paragraphs using this relationship as a head line.
The structural equivalence of halides and aurides, as show-
2Rb2 O + 3Au±0 + Rb = [RbAu−1 ]2 [Rb3 Au+1 O2 ]. (5)
ing up in isostructural CsAu, CsI, CsBr and CsCl, or mixed
In both cases, the halogens and gold, the driving forces behind CsAu/CsBr solid solutions [46,47], have been worked out by us
these reactions are the same, in principle. In the first place, it is more extensively: the halides Cl− , Br− and I− form the com-
the intrinsic stability (positive electron affinity) of the monoan- plete series of anti-perovskites A3 XO (A = K, Rb, Cs; X = Cl,
ions, secondly the stabilization of the cations by bond formation Br, I), in all possible combinations of the constituents [48].
(complexation) to oxygen. The crystal structures of the auride From Table 2 one can deduce that oxide aurides and halides
aurates(I) feature properties that are without precedent. They are forming a homogeneous family of perovskites, obeying the
1468 M. Jansen / Solid State Sciences 7 (2005) 1464–1474

Table 2
Selected crystallographic parameters of M3 XO (M = K, Rb, Cs; X = Br, Au, I)
Compound Volume (cm3 mol−1 ) K–X (Å) Compound Volume (cm3 mol−1 ) Rb–X (Å) Compound Volume (cm3 mol−1 ) Cs–X (Å)
K3 BrOa 85.6 3.69 Rb3 BrOa 98.3 3.86 Cs3 BrOc 112.8 4.09
K3 AuOa 86.6 3.71 Rb3 AuOa 100.3 3.89 Cs3 AuOc 112.9 4.08
K3 IOa 88.8 3.73 Rb3 IOb 106.2 3.99 Cs3 IOc 121.5 4.19
a Compound crystallizes in the cubic anti-perovskite (P m3̄m) CaTiO structure type. b Compound crystallizes in the mixed cubic-hexagonal anti-perovskite
3
(P 63 /mmc) BaTiO3 structure type. c Compound crystallizes in the hexagonal anti-perovskite (P 63 /mmc) CsNiCl3 structure type.

Fig. 7. (a) Solution of CsAu in liquid ammonia and (b) yellow solid CsAu
recrystallized from liquid ammonia together with dark-blue CsAu·NH3 [51].

super-alkaliion, with respect to size, here Au− experiences the

Fig. 6. 197 Au(Pt) Mößbauer spectrum of Cs7 Au5 O2 [51].
weakest bonding interactions, as compared to all other aurides
known.
classical rules of crystal chemistry. In particular, these studies When slowly evaporating the solvent, CsAu is not recov-
have allowed us to identify the halide most similar to Au− . ered directly. Instead, a deep blue crystalline precipitate forms
Without any doubt, this is bromide since the sizes of the unit in a first step, which transforms to bright yellow CsAu above a
cells and the bond lengths are virtually identical within the pairs temperature of 225 ± 5 K. It should be mentioned that recrystal-
Cs3 AuO/Cs3 BrO, Rb3 AuO/Rb3 BrO and K3 AuO/K3 BrO. lizing CsAu from liquid ammonia is yielding CsAu of highest
purity, to our experience. This observation of the true colour of
2.6. Electrolytic dissociation of CsAu in liquid ammonia stoichiometric CsAu (see Fig. 7(b)), by the way, is suited to re-
move all the discrepancies on CsAu in the literature, mentioned
Pure CsAu easily dissolves in liquid ammonia. These solu- above: they all seem to be related to varying purities of the sam-
tions are transparent and of very pale yellow color (Fig. 7(a)), ples studied.
they stay stable at least for several days, if, as an indispensable Investigating the blue crystalline precipitate was a demand-
requirement, vessels, solvent and gas atmosphere used are kept ing task, since the crystals had to be kept at a temperature well
painstakingly clean. Since these solutions show a high electrical below −50 ◦ C, all the time, and at the same time were ex-
conductivity, CsAu must have undergone an electrolytic disso- tremely sensitive to traces of oxygen and water. According to a
ciation. Similar solutions are accessible by dissolving gold in a single crystal X-ray structure analysis, the blue crystals have the
solution of cesium in liquid ammonia [49]. This observation is composition CsAu·NH3 [51,52]. There is a simple way of de-
proving again that Au− is an intrinsically stable entity. scribing the crystal structure by formally deriving it from CsAu:
Such an electrolyte offers all the opportunities of solution the cubic parent structure is cut into slabs, oriented parallel to
chemistry for further exploring the properties of Au− . Here, we (110), and of half the face diagonal thickness. These slabs are
present two results, a metathesis reaction and the formation of intercalated by the solvent molecules ammonia (cf. Fig. 8). Be-
a solid solvate. sides the changes in the coordination of cesium, the drop at
[NMe4 ]Au forms through metathesis by adding an ion ex- the monoatomar Au–Au separations from 426 to 302 pm are
change raisin (IE), loaded with [NMe4 ]+ , to the solution of the most striking changes in the local structural features. The
CsAu in liquid ammonia [50], Eq. (6). characteristic blue color, which is reminiscent of the color of
electrons solvated by ammonia, is clearly indicating changes
Cs+ (solv.) + Au− (solv.) + [NMe4 ]+ − IE
= [NMe4 ]Au + Cs+ − IE. (6) in the bonding situation. Indeed, in the Mößbauer experiment
(Table 3), a significantly reduced chemical shift was observed,
[NMe4 ]Au is the first colorless auride, showing all prototypic as compared to CsAu or Cs3 AuO. Thus, in CsAu·NH3 , gold
properties of a salt like solid. It is isostructural to the analogous is at a less negative valence state than −1. In this context,
bromide, and is exhibiting the largest Au− –Au− separation ever Cs solid state NMR spectra have yielded rather illustrative re-
found in the solid state. Since [NMe4 ]+ may be regarded a sults [53]. As can be seen from Fig. 9, the resulting signal
 M. Jansen / Solid State Sciences 7 (2005) 1464–1474 1469

Fig. 8. Perspective representations of the non-hydrogen atom structures of the Cm [51] (left) and C2/c [52] (right) modifications of CsAu·NH3 (blue spheres:
cesium atoms; yellow spheres: gold atoms; green spheres: nitrogen atoms; orange lines: unit cell edges).

Table 3
Mößbauer parameters of aurides, and of elemental gold [51]
Compound Chemical shift δ (mm−1 )
CsAu 7.00(2)
RbAu 6.70(2)
CsAu·NH3 5.96(1)
Au −1.23

shows a pronounced tailing towards stronger shielding, how-

ever, with the peak maximum still at the same shift as found
for CsAu. At inspecting the T1 relaxation time across the sig-
nal, it turns out that it varies significantly. This is a clear proof
for a inhomogeneous line broadening encountered. From these
experimental observations a rather conclusive picture may be
derived. The partial coordination of cesium by ammonium re-
duces the coulomb field experienced by Au− , thus destabilizing
the latter. On the other hand, the ammonia partial structure of-
fers sites for electrons where to go. The response is a partial Fig. 9. 133 Cs-NMR spectrum (top panel) of CsAu·NH3 at 86 K (solid line).
charge transfer which makes gold less negative, and an open The dotted line is a fit to the model described in [53]. The bottom panel shows
the frequency shift dependence of spin-lattice relaxation rate (T1−1 ) of 133 Cs
shell species, resulting in gold–gold homoatomar bonds. The in CsAu·NH3 at 86 K.
electrons, moved to the neighborhood of some of the cesium
atoms, influence the shielding of their cores and, at the same
3. Evidence for the stabilization of the 6s orbital of
time, the relaxation times. The spatial distribution of the elec-
platinum
trons in the Cs/NH3 sublattice is incommensurate, as can be
seen from the inhomogeneous broadening (no resolvable split-
ting) of the 133 Cs-NMR signal. This consistent interpretation is 3.1. Cesium platinide
getting support from comparing the Wannier orbitals of gold in
CsAu and CsAu·NH3 [52]. While in the former six well local- Platinum exhibits the second highest electron affinity among
ized orbitals result, there are only five localized in the latter, and all metals, cf. Fig. 2. While gold is best compared to the halo-
a rather diffuse one, allowing for sufficient overlap to produce gens which also need one more electron to achieve a closed
gold–gold bonding. shell, or subshell, respectively, the best main group references
1470 M. Jansen / Solid State Sciences 7 (2005) 1464–1474

for platinum are the chalcogens (two electrons missing for a

closed shell). Amazingly, and in contrast to gold, platinum even
surpasses the highest electron affinity found for a chalcogen
(sulfur). This, however, does not mean that the prospect for re-
alizing a negatively charged metal might be more promising
for platinum than for gold, because for the former only a dou-
bly charged anion would represent a closed shell species, and
all divalent anions are known to be intrinsically unstable with
respect to loss of an electron. Therefore, we started our experi-
ments aiming at the synthesis of a platinide by offering cesium
which is the most electropositive species available in the Peri-
odic Table.
By reacting platinum and cesium in welded Ta-ampoules,
Cs2 Pt, the first compound in the Cs/Pt system, was obtained at
the unexpectedly high temperature of 700 ◦ C [54]. The trans-
parent, red crystals are extremely sensitive towards oxidation
and hydrolysis. Cs2 Pt adopts the Ni2 In type of structure. Thus,
platinum is in a nine fold coordination by cesium, forming a
triply capped trigonal prism (see Fig. 10). Since this type of
structure is found for alloys as well as for heteropolar com-
pounds, no direct conclusions concerning the charge distribu-
tion in Cs2 Pt can be drawn from structural arguments. How-
ever, high pressure polymorphs of the alkalimetal monochalco-
genides occur in the Ni2 In structure [55–58], which moreover
is closely related to the PbCl2 structure as realized by Cs2 S,
Cs2 Se and Cs2 Te, at ambient conditions [59–62]. Recalling the
pressure-homologue rule, it is thus justified to regard Pt2− a Fig. 11. TB-LMTO-ASA band structure of Cs2 Pt. The widths of the bands
heavier homologue of S2− , Se2− and Te2− . Much more con- exhibit the contributions of different orbitals: Pt 5d (top), and Pt 6s (bottom).
vincing than the structural evidence for Cs2 Pt being a charge
separated (ionic) compound is the fact that it is transmissive to 3.2. Barium platinides
visible light, with the red color indicating a band gap of 1.5 eV.
Regarding Cs2 Pt a truly ionic compound gets strong support Barium is another electropositive metal, showing at least a
by (scalar) relativistic band structure calculations. As can be low first ionization potential. Therefore, it is to be expected that
seen from Fig. 11 a well pronounced band gap of 1.3–2.1 eV also barium might stabilize platinides. While the platinum rich
(depending on the method of calculation) develops. While the part of the Ba/Pt system was investigated previously [63], there
band of primarily 6s character submerges below the Pt 5d band, is no knowledge documented about the more barium rich re-
demonstrating its strong stabilization, the 5d band is barely gions [64]. Reacting the elements in an equimolar ratio yields
showing any dispersion. This all is in perfect agreement with prismatic opaque crystals of composition BaPt, as confirmed by
assuming Pt2− being realized in Cs2 Pt, for the first time. single crystal X-ray analysis [65]. Although the crystal struc-
ture basically is of the NiAs type (Fig. 12) there are striking
topological deviations from the aristotype. This is reflected
in a rather obvious way by an anomalously low c/a ratio of
1.15, the lowest ever observed for this structure type. This is
caused by homoatomar platinum–platinum bonding to form
linear chains of equidistant platinum atoms running along the
c-axis, which, as a consequence, is shrunk considerably. The
platinum–platinum bonds are as short as 2.70 Å, a value be-
low the corresponding one as was found in Krogmann’s salt
K2 [Pt(CN)4 ]Cl0.32 ·2.6H2 O (2.88 Å) [66,67]. Here again, ana-
lyzing the band structure leads to a deeper understanding of the
bonding. As can be seen from Fig. 13, the bands originating
from Pt 5d orbitals show virtually no dispersion within sections
of the Brillouin zone vertical to the  → A direction, while par-
Fig. 10. Crystal structure of Cs2 Pt with the coordination polyhedra around Pt allel to this direction pronounced dispersion, i.e., strong Pt–Pt
and Cs1 atoms emphasized (blue spheres: cesium atoms; yellow spheres: plat- interaction, is noticeable. These features nicely comply with the
inum atoms; orange lines: unit cell edges). structural anisotropy as discussed above. Exclusively along the
 M. Jansen / Solid State Sciences 7 (2005) 1464–1474 1471

either arrives at a closed shell situation or is using its original

and excessive valence electrons to from polyatomic anionic net-
works. In this sense, BaPt is the first example of a Zintl phase
where the polyanionic structure is established by a transition
element.
Obviously, barium is able to transfer but one of its valence
electrons to platinum. One way of arriving at more negatively
charged platinum might be to offer more barium. Indeed, the
new platinide obtained at a Ba/Pt ratio of 3:2 contains platinum
atoms pairwise connected to each other [68]. In terms of the
Zintl–Klemm concept, in Ba3 Pt2 , platinum has captured more
valence electrons from barium, thus approaching the closed
shell situation of 6s2 5d10 . As a consequence, there is less need
to form homoatomar bonds as compared to BaPt, and the in-
finite 1∞ Pt-chains are broken to isolated dumbbells (Fig. 14).
Fig. 12. Crystal structure of BaPt with the coordination polyhedra around Pt and The same structure type (Er3 Ni2 ) is adopted by a significant
Ba atoms emphasized (blue spheres: barium atoms; yellow spheres: platinum number of comparable intermetallics AE3 T 2 (AE = alkali earth
atoms; orange lines: unit cell edges).
elements, T = transition metal). From the compilation given in
Table 4, one can see a monotonous increase of the cell volumes
directions –K–M–, some bands (Ba 5d, 6s and 6p) cross the with the size of the respective alkaline earth element. However,
Fermi level indicating metallic conductivity, as was observed with the same alkaline earth element, the unit cell volumes of
experimentally. The low dimensional nature of the conductivity the platinides are significantly smaller than those of the respec-
to be expected from the calculations has not been confirmed ex- tive palladium compounds. This is also reflected by the volume
perimentally, yet. Addressing the charge transfer from barium shrinkage at the formation of AE3 T 2 from the elements, which
to platinum by Bader or ELF analyses yields an approximate is highest for the platinides. These data are providing evidence
transfer of one electron per barium atom. The scenario resulting for some peculiar bonding behaviour of the platinum alloys, as
from these considerations might be best described within the compared to the remaining representatives. The variations of
Zintl–Klemm paradigm: the electropositive constituent trans- the T–T bond lengths, on the other hand, are indicating a com-
fers charge to the more electronegative one which on its part plex interplay of geometric and electronic influences. Since the

Fig. 13. (a) Brillouin zone for a hexagonal crystal with special k points as considered in the band structure diagrams. The widths of the bands show the contribution
of different atomic orbitals (b) Pt 5d, (c) Pt 6s, (d) Ba 5d. The orange line marks the Fermi level.
1472 M. Jansen / Solid State Sciences 7 (2005) 1464–1474

Table 4
Volumes per formula unit and T –T distances for selected AE3 T 2 compounds
Compound Vcell /Z (Å) V (%)a d(T –T )AE3 T2 (Å)b d (%)c
Ca3 Pd2 129.9 −18.9 2.71 (2.75) −1.45
Ca3 Pt2 124.7 −22.5 2.68 (2.77) −3.24
Sr3 Ag2 177.7 −12.2 2.99 (2.89) +3.46
Sr3 Au2 163.0 −19.3 3.04 (2.88) +5.56
Sr3 Pt2 149.0 −24.9 2.84 (2.77) +2.53
Ba3 Ag2 200.6 −10.7 3.12 (2.89) +7.96
Ba3 Au2 184.2 −17.9 3.05 (2.88) +5.90
Ba3 Pt2 165.8 −24.9 3.03 (2.77) +9.39
a Volume change 3V = 100 × [V (AE T ) − 3V (AE) − 2V (T )]/[3V (AE) + 2V (T )]. b In parentheses d(T –T ) —metal–metal distances in elements are
3 2 T
given [30]. c T –T distance change d = 100 × [d(T –T )AE3 T2 − d(T –T )T ]/[d(T –T )T ].

Fig. 14. View of the Ba3 Pt2 crystal structure along the c-axis emphasizing.
Pt-pairs. For colour code see Fig. 12.

Pt–Pt separations always are shortest, the respective alkaline

earth element given, there is clear evidence for electronic effects
stabilizing the Pt–Pt bonding, compared to Pd–Pd. In particu-
lar, in context with the considerable spread of the Pt–Pt bond
lengths from 268 to 303 pm for Ca3 Pt2 and Ba3 Pt2 , the low-
ering of the 6s bonds makes platinum more flexible to comply
with varying bond lengths, and stabilizes the more ionic atoms Fig. 15. Crystal structure of Ba2 Pt with the coordination polyhedra around Pt
atoms emphasized. For colour code see Fig. 12.
at larger distances. The variation of the Pt–Pt distances over the
rather wide range as mentioned is imposed by the matrix of the
structure type, and correlates with the sizes of the respective
alkaline earth components.
Ba2 Pt, the platinide richest in barium known so far, is the
first example of a solid composed of two metals that adopts
the CdCl2 -type of structure (Fig. 15) which is otherwise found
preferably for metal salts of polarizable halogenides [69]. This
strong indication for Ba2 Pt being principally a heteropolar com-
pound is corroborated by the interatomic distances observed.
The Ba–Pt separation of 315 pm, for instance, is in fair agree-
ment with the sum of the ionic radii (303 pm), and much
shorter than the sum of the metallic radii (361 pm). These
structural features justify an assignment of charges according
to (Ba2+ )2 Pt2− ·2e− , and are suggesting a close relationship to
the alkaline earth subnitrides.
The charge distribution as proposed is in good accordance
with the results of LMTO-TB-ASA band structure calculations, Fig. 16. Total and partial densities of states (DOS; LMTO-TB-ASA) of Ba2 Pt
cf. Fig. 16. Since the Pt 5d and 6s bands are narrow, and clearly (left). Total DOS of Ba2 N (right). The dashed lines mark the respective Fermi
below the Fermi level, we regard platinum a doubly charged levels.
 M. Jansen / Solid State Sciences 7 (2005) 1464–1474 1473

monoanion. The excess electrons which are responsible for the

metallic character of Ba2 Pt are mainly located in the neighbour-
hood of barium, in the interslab area.

4. Stabilization of the 6s orbitals of gold and platinum,

primarily of relativistic origin?

The experimental results presented on gold and platinum are

providing convincing evidence for a considerable stabilisation
of their 6s orbitals, and of the states or bands related to the 6s
configuration. Beyond the structural arguments and the band
structure calculations, which to a certain amount suffer from
arbitrariness and numerical inaccuracy, respectively, CsAu and
Cs2 Pt being transmissive to visible light give strong support to
regarding them basically ionic solids with full charge separa-
tion. Moreover, this view is getting strong support from the fact Fig. 17. First ionisation potentials of the selected elements; diamonds/normal
that Au− is stable in liquid electrolytes. Thus, a striking lower- labels, circles/italic labels and squares/bold labels correspond to the 2nd, 5th
ing in energy, and a concomitant contraction, of the 6s orbitals and 6th period, respectively. The data are taken from [29].
of gold and platinum has to be regarded a proven fact.
Now, what is the origin of these prominent features? To an- way avoiding to form isolated Tl− species, our present work is
swer this question, one first has to consider the positions of gold focussing at bringing Tl− anions at a larger distance to each
and platinum in the Periodic Table. These are indeed unique, other, e.g., by synthesizing (NMe4 )Tl or Cs3 TlO. If one, or
since here for the fist time f (4f) and d (5d) contractions super- both, of these hypothetical solids shows insulating and diamag-
impose, affecting most strongly the least shielded s electrons. netic properties this would be an exemplar for which even a
Thus, the 6s orbital contraction might be well assigned to con- qualitative understanding of the electron structure would indis-
ventional orbital effects, in the first place. This view is getting pensably require to consider the relativistic dependence of the
some support by the fact that the 6s orbital expands when pro- electron mass on its velocity.
ceeding beyond mercury, where the 6p orbitals are filled in.
This change of the trend has to be attributed exclusively to or- Acknowledgements
bital effects; to be more explicit, to the stronger shielding of the
nuclear charge by p electrons, as compared to d and f electrons. The author gratefully acknowledge Andrey Karpov for valu-
Of course, quantum chemical calculations intended to be able discussion, and assistance in preparing the manuscript,
close to physical reality have to consider the relativistic varia- Jürgen Nuss for his help with the figures.
tion of the electron mass with its velocity. However, neither the-
oretical nor experimental measures seems to exist, that would References
allow to impartially differentiate between the contributions of
conventional orbital effects and of relativistic effects to the 6s [1] A. Sommerfeld, Annalen der Physik 51 (1916) 1.
orbital contraction. [2] A. Sommerfeld, Annalen der Physik 51 (1916) 125.
In order to being able to unambiguously judge whether the [3] P.A.M. Dirac, Proc. R. Soc. London A 117 (1928) 610.
[4] P.A.M. Dirac, Proc. R. Soc. London A 118 (1928) 351.
relativistic variation of the electron mass is of significance in [5] P.A.M. Dirac, Proc. R. Soc. London A 123 (1929) 714.
chemistry, one would need to determine experimentally a qual- [6] G. Breit, Phys. Rev. 36 (1930) 383.
ity that can only be reproduced by using Dirac’s equation. Such [7] G. Breit, Phys. Rev. 39 (1932) 616.
a quality might be a 6p2 configuration producing a diamagnetic [8] B. Swirles, Proc. R. Soc. London A 152 (1935) 625.
ground state. A treatment according to Dirac would lift the de- [9] B. Swirles, Proc. R. Soc. London A 157 (1936) 680.
[10] I.P. Grant, Adv. Phys. 19 (1970) 747.
generacy of the p orbitals, and would split them into two sets, [11] J.-P. Desclaux, At. Data Nucl. Data Tables 12 (1973) 311.
p1/2 and p3/2 , according to Dirac’s notation. The p1/2 subshell [12] P. Pyykkö, J.-P. Desclaux, Acc. Chem. Res. 12 (1979) 276.
is lower in energy and more strongly contracted, and is able to [13] P. Pyykkö, Adv. Quantum Chem. 11 (1978) 353.
accommodate two spin paired electrons yielding a diamagnetic [14] P. Pyykkö, Chem. Rev. 88 (1988) 563.
ground state. Such a scenario can be derived already from the [15] P. Pyykkö, Angew. Chem. 114 (2002) 3723;
P. Pyykkö, Angew. Chem., Int. Ed. 41 (2002) 3573.
atomic spectra and the first ionisation potentials, see Fig. 17. [16] P. Pyykkö, Angew. Chem. 116 (2004) 4512;
Normally, one would expect the 15th group elements to ex- P. Pyykkö, Angew. Chem., Int. Ed. 43 (2004) 4412.
hibit a peaking 1st ionisation potential (half filled p-shell). This [17] K.S. Pitzer, Acc. Chem. Res. 12 (1979) 271.
is true up to the 5th period, however, within the 6th row, the [18] P.A. Christiansen, W.C. Ermler, K.S. Pitzer, Annu. Rev. Phys. Chem. 36
element lead with a 6p2 configuration shows a higher ionisa- (1985) 407.
[19] B.A. Heß, Ber. Bunsen-Ges. Phys. Chem. 101 (1997) 1.
tion energy than bismuth (6p3 ). A good candidate for realizing [20] N. Kaltsoyannis, J. Chem. Soc., Dalton Trans. (1997) 1.
such an electron configuration in a chemical compound might [21] N. Bartlett, Gold Bull. 31 (1) (1998) 22.
be Tl− . Since CsTl is known to contain Tl6 -clusters [70], this [22] P. Schwerdtfeger, Heteroat. Chem. 13 (2002) 578.
1474 M. Jansen / Solid State Sciences 7 (2005) 1464–1474

[23] G.L. Malli (Ed.), Relativistic Effects in Atoms, Molecules, and Solids, in: [48] C. Feldmann, M. Jansen, Z. Anorg. Allg. Chem. 621 (1995) 1907.
NATO ASI Series, New York, 1983. [49] W.J. Peer, J.J. Lagowski, J. Am. Chem. Soc. 100 (1978) 6260.
[24] P. Pyykkö, Relativistic Theory of Atoms and Molecules: Part 1, Springer, [50] P.D.C. Dietzel, M. Jansen, Chem. Commun. (2001) 2208.
Berlin, 1986. [51] A.-V. Mudring, M. Jansen, Ph.D. Thesis A.-V. Mudring, Rheinische
[25] P. Pyykkö, Relativistic Theory of Atoms and Molecules: Part 2, Springer, Friedrich-Wilhelms-Universität, Bonn, 2001.
Berlin, 1993. [52] A.-V. Mudring, M. Jansen, J. Daniels, S. Krämer, M. Mehring, J.P.P. Ra-
[26] P. Pyykkö, Relativistic Theory of Atoms and Molecules: Part 3, Springer, malho, A.H. Romero, M. Parinello, Angew. Chem. 114 (2002) 128;
Berlin, 2000. A.-V. Mudring, M. Jansen, J. Daniels, S. Krämer, M. Mehring, J.P.P. Ra-
[27] P. Schwerdtfeger (Ed.), Relativistic Electronic Structure Theory: Part 1— malho, A.H. Romero, M. Parinello, Angew. Chem., Int. Ed. 41 (2002) 120.
Fundamentals, Elsevier Science B.V., Amsterdam, 2002. [53] S. Krämer, M. Mehring, A.-V. Mudring, M. Jansen, J. Phys. Chem. B 107
[28] P. Schwerdtfeger (Ed.), Relativistic Electronic Structure Theory: Part 2— (2003) 4922.
Applications, Elsevier Science B.V., Amsterdam, 2004. [54] A. Karpov, J. Nuss, U. Wedig, M. Jansen, Angew. Chem. 115 (2003) 4966;
[29] K.M. Mackay, R.A. Mackay, W. Henderson, Introduction to Modern Inor- A. Karpov, J. Nuss, U. Wedig, M. Jansen, Angew. Chem., Int. Ed. 42
ganic Chemistry, fifth ed., Stanley Thornes (Publishers) Ltd, Cheltenham, (2003) 4818.
2000, p. 27. [55] A. Vegas, A. Grzechnik, M. Hanfland, C. Mühle, M. Jansen, Solid State
[30] P. Villars, L.D. Calvert, Pearson’s Handbook of Crystallographic Data for Sci. 4 (2002) 1077.
Intermetallic Phases, second ed., ASM International, Ohio, 1991. [56] A. Vegas, A. Grzechnik, K. Syassen, I. Loa, M. Hanfland, M. Jansen, Acta
[31] W. Biltz, F. Weibke, Z. Anorg. Allg. Chem. 236 (1938) 12. Crystallogr. B 57 (2001) 151.
[32] A. Sommer, Nature 152 (1943) 215. [57] A. Grzechnik, A. Vegas, K. Syassen, I. Loa, M. Hanfland, M. Jansen,
[33] W.E. Spicer, A.H. Sommer, J.G. White, Phys. Rev. 115 (1959) 57. J. Solid State Chem. 154 (2000) 603.
[34] G. Kienast, J. Verma, Z. Anorg. Allg. Chem. 310 (1961) 143. [58] K. Stöwe, Z. Kristallogr. 219 (2004) 359.
[35] G.K. Wertheim, R.L. Cohen, G. Crecelius, K.W. West, J.H. Wernick, Phys. [59] A. Bergmann, Z. Anorg. Allg. Chem. 231 (1937) 269.
Rev. B 20 (1979) 860. [60] W. Klemm, H. Sodomann, P. Langmesser, Z. Anorg. Allg. Chem. 241
[36] U. Zachwieja, Z. Anorg. Allg. Chem. 619 (1993) 1095. (1939) 281.
[37] F. Wooten, G.A. Condas, Phys. Rev. 131 (1963) 657. [61] H. Sommer, R. Hoppe, Z. Anorg. Allg. Chem. 429 (1977) 118.
[38] C. Feldmann, M. Jansen, Angew. Chem. 105 (1993) 1107; [62] I. Miller-Schewe, P. Böttcher, Z. Kristallogr. 196 (1991) 137.
C. Feldmann, M. Jansen, Angew. Chem., Int. Ed. Engl. 32 (1993) 1049. [63] H. Schulz, K. Ritapal, W. Bronger, W. Klemm, Z. Anorg. Allg. Chem. 357
[39] C. Feldmann, M. Jansen, J. Chem. Soc., Chem. Commun. (1994) 1045. (1968) 299.
[40] C. Feldmann, M. Jansen, Z. Anorg. Allg. Chem. 621 (1995) 201. [64] P. Villars, Pauling File, Inorganic Materials Database and Design System
[41] A. Band, A. Albu-Yaron, T. Livneh, H. Cohen, Y. Feldman, L. Shimon, V1.0, Tokyo (Japan) and Vitznau (Switzerland) 2002.
R. Popovitz-Biro, V. Lyahovitskaya, R. Tenne, J. Phys. Chem. B 108 [65] A. Karpov, J. Nuss, U. Wedig, M. Jansen, J. Am. Chem. Soc. 126 (2004)
(2004) 12360. 14123.
[42] C. Feldmann, M. Jansen, Z. Naturforsch. B 51 (1996) 607. [66] K. Krogmann, Angew. Chem. 81 (1969) 10.
[43] A. Pantelouris, G. Küper, J. Hormes, C. Feldmann, M. Jansen, J. Am. [67] K. Krogmann, H.-D. Hausen, Z. Anorg. Allg. Chem. 358 (1968) 67.
Chem. Soc. 117 (1995) 11749. [68] A. Karpov, U. Wedig, M. Jansen, Z. Naturforsch. B 59 (2004) 1387.
[44] A.-V. Mudring, M. Jansen, Angew. Chem. 112 (2000) 3194; [69] A. Karpov, U. Wedig, R.E. Dinnebier, M. Jansen, Angew. Chem. 117
A.-V. Mudring, M. Jansen, Angew. Chem., Int. Ed. 39 (2000) 3066. (2005) 780;
[45] A.-V. Mudring, J. Nuss, U. Wedig, M. Jansen, J. Solid State Chem. 155 A. Karpov, U. Wedig, R.E. Dinnebier, M. Jansen, Angew. Chem., Int.
(2000) 29. Ed. 44 (2005) 770.
[46] R. Wormuth, R.W. Schmutzler, Thermochim. Acta 160 (1990) 97. [70] Z.-C. Dong, J.D. Corbett, Inorg. Chem. 35 (1996) 2301.
[47] R. Erdmann, R. Wormuth, R.W. Schmutzler, Mater. Sci. Forum 73 (1991) [71] T. Andersen, H.K. Haugen, H. Hotop, J. Phys. Chem. Ref. Data 28 (1999)
45. 1511.