Sc5Ni2Te2: Synthesis, structure, and bonding of a metal-metal-bonded chain phase, a relative of Gd3MnI3

Article abstract of DOI:10.1021/ic981073a

Sc₅Ni₂Te₂ has been prepared by high-temperature solid-state techniques and the structure determined at 23 °C by single crystal and powder X-ray diffraction methods to be orthorhombic, Pnma (No. 62) with Z = 4, a = 17.862-(1) A?, b = 3.9533(3) A?, c = 10.6398(6) A?. The structure contains pairs of eclipsed zigzag chains of nickel atoms that are sheathed by scandium atoms and demarcated from other chains by tellurium atoms. The structure is isotypic with that of Hf₅Co_1+xP_3-x, but shifted atomic positions and a different ordering of the main group and late transition elements give it a clearly ID character. The differences in dimensionality, ordering, and bonding are discussed, and comparisons are made with GdsMnls and rare-earth-metal cluster halides in general.

Full text of DOI:10.1021/ic981073a

Inorg. Chem. 1999, 38, 1945-1950
1945
Sc5Ni2Te2: Synthesis, Structure, and Bonding of a Metal-Metal-Bonded Chain Phase,
a Relative of Gd3MnI3
Paul A. Maggard and John D. Corbett*
Department of Chemistry, Iowa State University, Ames, Iowa 50011
ReceiVed September 9, 1998
Sc5Ni2Te2 has been prepared by high-temperature solid-state techniques and the structure determined at 23 °C by
single crystal and powder X-ray diffraction methods to be orthorhombic, Pnma (No. 62) with Z ) 4, a ) 17.862-
(1) Å, b ) 3.9533(3) Å, c ) 10.6398(6) Å. The structure contains pairs of eclipsed zigzag chains of nickel atoms
that are sheathed by scandium atoms and demarcated from other chains by tellurium atoms. The structure is
isotypic with that of Hf5Co1+xP3-x, but shifted atomic positions and a different ordering of the main group and
late transition elements give it a clearly 1D character. The differences in dimensionality, ordering, and bonding
are discussed, and comparisons are made with Gd3MnI3 and rare-earth-metal cluster halides in general.
Introduction
Metal-rich chalcogenides of scandium and yttrium also show
some notable contrasts with parallel structures and stoichiom-
The plethora of new metal-rich chalcogenide phases among
the early transition metals have been important for understanding
the expression of and interrelationships between metal-metal
bonding features among these many compounds. Incorporation
of late transition metals has long been known to stabilize both
metal-rich halides and chalcogenides that are otherwise unstable
with respect to electron count and other binary phases. Recently
reported ternary chalcogenides and phosphides of this type
etries of their most reduced halides. The latter are known only
with proportionately more nonmetal atoms. Twice as many
halogen atoms (X) per chalcogen would be expected for the
same electron count per metal atom, and in fact somewhat more
(2 < X/R < 3) are observed in isolated cluster halides.
Condensed chains or tetramers built of recognizable octahedra
span a range of 1 < X/R < 2. The halides structurally serve to
sheath the metal cores, in all cases leading to clearer definition
of the building blocks. Furthermore, with few exceptions, the
known reduced halides are so electron-poor that they also require
interstitial heteroatoms (Sc2Cl12C, Y4I5C, etc.) which afford
1
2
3
include Ta9M2S6, Ta11M2Se8 (M ) Fe, Co, Ni), Ta8NiSe8 (M
4
5
)
Co, Ni), Hf8MTe6 (M ) Co, Ni, Ru), Zr9M2P4 (M ) Co,
6
7
8
9
Ni), Hf5Co1+xP3-x, (0 < x < 0.5), Hf2NiP, and ScNiP.
The study of bonding features in metal-rich chalcogenides
of the early transition metals has only recently been extended
1
4
central bonding and additional bonding electrons.
to group 3 (R) examples, namely, to Sc2Te10 and Sc8Te3.11 Their
In analogy with studies on later transition-metal-chalcogen
systems, this article presents the first results of the expansion
of this chemistry to ternary systems of scandium, in this case
with the incorporation of nickel. The early-late transition metal
bonding involved appears to reflect the extra stability of such
polar interactions that were first noted by Brewer and Wengert.15
Mixed-metal features in chalcogenides and phosphides are
largely multicapped trigonal prisms of the earlier transition metal
centered by a late transition metal. The nonmetals in these
generally prefer a similar environment, a tricapped trigonal prism
structural and bonding relationships to those of electron-richer
analogues allow one to assess the importance of atom sizes,
valence electron concentrations, and metal-to-nonmetal propor-
tions in the structure and bonding. The smaller number of metal-
based electrons for the earlier transition metals appears to force
a reduction in the metal-metal framework dimensionality, as
shown in particular for Sc8Te3 and Y8Te3, relatives of Ti8Ch3,
12,13
Ch ) S, Se.
Stoichiometry and efficient packing apparently
dictate that some metal atom pairs may be in close proximity
even though theory indicates that there are relatively few or no
electrons involved in their bonding, i.e., a classical result of
matrix effects.
(tetrakaidecahedron). In some ternary phases, the late transition
metal and nonmetal (e.g., Co and P, or Ni and S) may exhibit
unusual mixed metal/nonmetal occupancies of the same sites,
7
as in (Hf5Co1+xP3-x), evidently because of their similar sizes
(
1) Corbett, J. D. J. Alloys Compd. 1995, 229, 10.
and site preferences. In the title compound, Sc5Ni2Te2, the late-
transition metal and the nonmetal have markedly different sizes,
and mixed occupancy is not a factor. Furthermore, the relative
electron deficiency of the host metal and the larger anion ensure
a cooperative reduction in dimensionality of the metal-metal
bonded framework as compared with that in Hf5Co1+xP3-x, etc.
The new Sc5Ni2Te2 is significant in that it represents the
extension of early-late transition metal chemistry to ternary
chalcogenides of the electron-poorer scandium.
(2) Harbrecht, B.; Franzen, H. F. J. Less-Common Met. 1985, 113, 349;
Harbrecht, B. J. Less-Common Met. 1986, 124, 125.
(3) Harbrecht, B. J. Less-Common Met. 1988, 182, 118.
(4) Conard, M.; Harbrecht, B. J. Alloys Compd. 1993, 197, 57.
(5) Abdon, R. L.; Hughbanks, T. Chem. Mater. 1994, 6, 424.
(6) Kleinke, H.; Franzen, H. F. Inorg. Chem. 1996, 35, 5272.
(7) Kleinke, H.; Franzen, H. F. J. Alloys Compd. 1996, 238, 68; Kleinke,
H.; Franzen, H. F. J. Alloys Compd. 1997, 255, 110.
(
8) Kleinke, H.; Franzen, H. F. J. Angew. Chem., Int. Ed. Engl. 1997, 36,
513.
(9) Kleinke, H.; Franzen, H. F. J. Solid State Chem. 1988, 137, 218.
(
10) Maggard, P. A.; Corbett, J. D. Angew. Chem., Int. Ed. Engl. 1997,
18, 336.
(14) Corbett, J. D. In Modern PerspectiVes in Inorganic Crystal Chemistry,
Parth e´ , E., Ed.; NATO ASI Series C; Kluwer Academic Publishers:
Dordrecht, The Netherlands, 1992; pp 27-56.
(11) Maggard, P. A.; Corbett, J. D. Inorg. Chem. 1998, 37, 814.
(12) Owens, J. P.; Franzen, H. F. Acta Crystallogr. 1994, B30, 427.
(13) Weirich, T. E.; P o¨ ttgen R.; Simon, A. Z. Kristallogr. 1996, 211, 927.
(15) Brewer, L.; Wengert, P. R. Metallurg. Trans. 1973, 4, 2674.
1
0.1021/ic981073a CCC: $18.00 © 1999 American Chemical Society
Published on Web 03/31/1999

1946 Inorganic Chemistry, Vol. 38, No. 8, 1999
Maggard and Corbett
Table 1. Some Data Collection and Refinement Parameters for
Table 2. Positional and Isotropic-Equivalent Thermal Parameters
Sc Ni Te
5
2
2
5 2 2
for Sc Ni Te
formula weight
space group, Z
lattice parameters
a (Å)
597.42
Pnma (No. 62), 4
atom
x
z
B
eq (Å2)b
Te1
Te2
Ni1
Ni2
Sc1
Sc2
Sc3
Sc4
Sc5
0.10736(6)
0.73263(6)
0.9159(1)
0.9939(1)
0.1526(2)
0.2730(2)
0.8663(2)
0.5740(2)
0.9808(2)
0.5103(1)
0.3098(1)
0.1038(2)
0.8261(2)
0.7819(3)
0.4786(3)
0.6963(3)
0.4410(3)
0.3281(3)
0.71(4)
0.71(4)
0.83(7)
0.84(8)
0.7(1)
0.8(1)
0.8(1)
0.8(1)
0.8(1)
a
17.862(1)
3.9533(3)
10.6398(6)
751.3(1)
5.281
167.35
3.2, 3.2
b (Å)
c (Å)
3
V (Å )
3
d
calc, g/cm
-
1
µ, cm (Mo KR)
R; R
b
w
(%)
a
b
Guinier data, Cu KR, 23 °C, 58 lines. R ) ∑||F
o
| - |F
c
||/∑|F |;
o
a
All atoms on m, y ) 1/
. b
2
2
-2
F
4
B
eq ) (8π /3)∑
i
∑
j
U
ij
a
i
*a
j
i j
* ba ba .
2
1/2
R
w
) [∑w(|F
o
| - |F
c
|) /∑w(F
o
) ] ; w ) σ
.
Experimental Section
All materials were handled in He-filled or N
2
-filled gloveboxes to
reduce contamination by adventitious impurities. The elements were
used as received: Sc turnings, Aldrich 99.7%; Te powder, Aldrich
9
9.99%; Ni powder, Alfa 99.95%. The synthesis of Sc
5
Ni
2
Te
2
began
Te
1
0
with the preparation of Sc
2
Te
3
as described previously. The Sc
2
3
,
Ni, and Sc to give a 3:1:1 (Sc:Ni:Te) stoichiometry were first loaded
into a tantalum tube welded at one end. The other end of the tube was
then crimped shut inside the glovebox and transferred to an arc-welder.
The tube was sealed after the welder had been evacuated and backfilled
with argon. Such containers were then sealed inside evacuated and well-
flamed fused silica jackets, heated at 1000 °C for 24 h, cooled to 700
°
C at 5 °C/hr, and then allowed to cool in air. Guinier powder diffraction
on the product of the first reaction showed what was subsequently found
to be Sc Ni Te had been obtained in g80% yield, plus ScTe. Further
reactions with the indicated 5:2:2 stoichiometry at higher or lower
5
2
2
5 2 2
Figure 1. Near-[010] section of the chain structure of Sc Ni Te (99.9%
16
temperatures only yielded a neighboring ternary phase Sc
ScTe and ScNi. Also, arc-melting reactions at the 5:2:2 composition
did not yield Sc Ni Te . However, reactions loaded off-stoichiometry
NiTe) yielded higher quantities of Sc Ni Te , evidently because
6
NiTe
2
or
probability ellipsoids). The Ni atoms are black, Sc, white, and Te, gray
ellipsoids. The shortest interchain distance is marked. The a axis is
horizontal.
5
2
2
(
∼Sc
3
5
2
2
some Ni had been lost into the container in the former reactions. The
same synthetic techniques with Fe, Co, and Cu (M) as the late transition
displacement parameters, and a complete distance list are in the
Supporting Information. These as well as the F /F listing are also
o c
available from J.D.C.
metal did not yield any of the analogous Sc
Powder X-ray Diffraction. The powder diffraction patterns of Sc
Te were obtained with the aid of an Enraf-Nonius Guinier powder
camera and monochromatic Cu KR radiation. The samples were
5 2 2
M Te phases.
5
-
Band Calculations. Extended H u¨ ckel calculations were carried out
within the tight-binding approximation19 for the full structure of Sc
Ni Te at 48 k-points spread out over the irreducible wedge. Hii
parameters employed were the values iterated to charge consistency
Ni
2
2
5
-
1
2
2
powdered, mixed with standard silicon (NIST), and placed between
two strips of cellophane tape on a frame that mounted on the sample
rotation motor. Lattice parameters were obtained with the aid of least-
squares refinement of 58 indexed lines with 2θ values calibrated by a
nonlinear fit to the positions of the standard Si lines (Table 1).
Single-Crystal Diffraction. Several irregularly shaped, silvery
crystals were mounted inside 0.3-mm i.d. glass capillaries that were
sealed off and mounted on metal pins. Their quality was checked by
means of Laue photographs. A diffraction data set for the best crystal
was measured on a Rigaku AFC6R diffractometer (monochromated
16
for Sc from Sc
4
-
2
Te,10 and for Ni from Sc
6
NiTe
2
(eV): Sc 4s, -6.75;
p, -3.38; 3d, -6.12; Ni 4s, -5.58; 4p, -2.41; 3d, -7.82; Te 6s,
21.20; 6p, -12.00. Very similar energies were also obtained for the
first two from charge iteration on ScNi (CsCl type). The charge
interation for Ni gave much more suitable results for this polar
compound than those from density functional theory (-8.13, -4.18,
20
-
12.40 eV, respectively ).
Results and Discussion
1
Mo KR radiation) at room temperature. Twenty-five centered reflec-
tions gathered from a random search were used to determine provisional
lattice constants and the crystal system. Two octants of data were
collected (h, k, (l) to 2Θmax ) 60° and corrected for Lorentz and
polarization effects. The data were further corrected for absorption with
the aid of two ψ-scans. Of 4627 measured reflections, 1938 had I >
Structural Description. A near-[010] section of the Sc5Ni2-
Te2 structure viewed along the short 3.95 Å axis is given in
Figure 1. The atom distribution can be viewed as pairs of
extended zigzag chains of nickel (black) that are sheathed by
scandium (open) and are in turn well-separated by single
tellurium atoms (gray). The shortest distance between separate
metal chains, d(Sc2-Sc3) ) 3.69 Å (marked), is at best a weak
interaction (below). Figure 2 illustrates the repeat unit in one
chain along with atom labels and distances that have appreciable
overlap populations (vide infra). To help understand the
structure, Figure 3 shows a side view of one-half of the
composite chain as viewed more or less along [301], Figure 1.
The repeat units here are rectangular scandium pyramids [Sc2,
3
σ(I) and 653 of these were unique. Extinction conditions and statistical
evidence for centricity indicated one possible space group, Pnma. The
1
7
structure was solved by direct methods (SHELXS ) and refined with
18
the package TEXSAN in this space group. After isotropic refinement,
the data averaged with Rav ) 9.2%, and the final anisotropic refinement
converged at R(F)/R
w
5 2 2
) 3.2/3.2% for the composition Sc Ni Te . Some
data for these processes are listed in Table 1, and the atomic positions
and isotropic-equivalent temperature factors are given in Table 2.
Additional data collection and refinement parameters, the anisotropic
(16) Maggard, P. A.; Corbett, J. D., unpublished.
(17) Sheldrick, M. SHELXS-86. Universit a¨ t G o¨ ttingen, Germany, 1986.
(18) TEXSAN, Version 6.0, Molecular Structure Corp., The Woodlands,
Texas, 1990.
(19) Hoffman, R. J. Chem. Phys. 1963, 39, 1397; Whangbo, M.; Hoffman,
R. J. Am. Chem. Soc. 1978, 100, 6093.
(20) Vela, A.; G a´ zquez, J. L. J. Phys. Chem. 1988, 92, 5688.

Sc5Ni2Te2: Synthesis, Structure, and Bonding
Inorganic Chemistry, Vol. 38, No. 8, 1999 1947
Table 3. Selected Metal-Metal Distances (Å) and Overlap
5 2 2
Populations in Sc Ni Te
atom 1
atom 2
distance
overlap pop. per pair
Sc1
Sc1
Sc4
Sc2
Sc3
Sc3
Sc1
Sc4
Sc2
Sc4
Sc2
Sc1
Sc1
Sc1
Sc5
Ni1
Ni1
Ni2
Ni2
Ni2
Ni1
Ni1
Ni1
Ni1
Ni2
Ni1
Ni2
Sc2
Sc4
Sc5
Sc4
Sc4
Sc5
Sc5
Sc4
Sc2
Sc4
Sc3
Sc3
Sc2
Sc1
Sc5
Ni2
Ni2
Sc3
Sc4
Sc5
Sc1
Sc4
Sc2
Sc5
Sc4
Sc4
Sc1
3.17
3.27
3.31
3.48
3.44
3.38
3.32
3.53
3.95a
3.95a
3.69b
×2
0.199
0.194
0.172
0.102
0.108
0.099
0.093
0.082
0.045
0.037
0.036
0.028
0.018
0.015
0.009
0.020
-0.007
0.209
0.191
0.192
0.184
0.178
0.174
0.174
0.107
0.104
0.098
×2
×2
×2
×2
×2
×2
×2
×2
b
3.83
×2
×2
×2
×2
×2
3.88b
3.95a
3.95a
2.66c
3.27
2.67
2.62
2.61
2.62
2.64
2.70
2.65
2.86
2.86
2.87
×2
×2
×2
×2
5 2
Figure 2. The centrosymmetric repeat unit in the Sc Ni chain, with
the numbering scheme and independent distances marked. Nickel atoms
and Ni-Ni bonds are darker.
a
Unit cell repeat in the chain. b Interchain distance. c Shortest
distance between two nickel chains.
Figure 2 consists of Sc1 (×2) and Sc4 (×2) as two of the side
edges, with the third edge Ni2-Ni1 pair which also centers the
next interpenetrating trigonal prism. The 2:5 ratio of nickel to
1
4
scandium, relative to that in Sc6NiTe2, predicates such
condensation. Another description of this one-dimensional array
is in terms of zigzag chains. If the Sc-Ni connectivity of the
structure is neglected for the moment, it can be seen in Figure
2 that the outside shell of the chain is composed entirely of
zigzag Sc-Sc chains along Bb that share vertexes, the comparable
Ni-Ni chains being added internally. Although the later
interpretation is not the most useful in terms of understanding
the local bonding, it is visually helpful.
5 2
Figure 3. Side view of one-half of the Sc Ni metal chain (Figure 2),
with the short b ) 3.95 Å repeat horizontal. The Ni atoms are darkened
with the gray one further away. Note the construction of Ni-based
rectangular pyramids of Sc; one is marked by a heavier outline.
Sc1 (×2), Sc4 (×2) and Sc3, Sc4 (×2), Sc5 (×2)] (one is
highlighted) that share Sc4-Sc4 edges in pairs and Sc1-Sc4-
Sc5 edges infinitely along Bb . Each rectangular pyramid has a
nickel atom 0.54 Å outside of the base. Two of these composites
are then assembled base-to-base with a relative displacement
of b/2 to generate the full chains seen in Figures 1 and 2. This
assembly generates additional Sc1-Sc5 (top and bottom) and
Sc4-Sc4 bonds across Figure 1 plus the pair of parallel zigzag
chains of nickel that run down the central channel of the chain.
The Ni-Ni distance within the zigzag chains is 2.66 Å (the
The description of all of the pairwise atom-atom distances
within the Sc5Ni2 chain is fairly involved. Some of these are
marked in Figure 2, and they are also listed in order of
decreasing pairwise overlap populations in Table 3. (The
complete distance listing is given in the Supporting Information.)
It should be noted that the chain is centrosymmetric and only
one-half of the interactions need to be considered. While Sc1
is bonded twice to Ni1 (2.62 Å) and once to Ni2 (2.87 Å, across
the double Ni chain), Sc5 is connected in parallel but more
tightly, twice to Ni2 (2.61 Å) and once to Ni1 (2.65 Å). Sc1
and Sc5 also form a zigzag pattern atop the double nickel chains
with Sc1-Sc5 ) 3.32 Å (×2). Internally, Sc4 occupies a special
position more inside of the chain, and the farthest from any
tellurium atom. The Sc4 has two short contacts to each of Ni1
and Ni2, 2.64 and 2.62 Å, respectively, and long diagonals
across the chain center to both Ni (2.86 Å). Interestingly, there
are also short distances between Sc4 and both Sc1 and Sc5,
edges in the square pyramids (3.27 and 3.31 Å), perhaps because
of their common nickel neighbors. Finally, the comparable Sc2
and Sc3 atoms, the apexes of the rectangular pyramids described
earlier, are also bonded to their basal Ni1 (2.70 Å) and Ni2
21
single bond distance is given as 2.31 Å ) whereas the closest
separation between the eclipsed Ni chains is 3.27 Å. The
scandium atoms about the two nickel strings exhibit Sc-Ni
distances over 2.62-2.87 Å, whereas Sc-Sc distances around
the outside of the chain span 3.17-3.48 Å.
Numerous older structures have demonstrated that a preferred
environment for the late transition metal is in the center of a
2
-8
multicapped trigonal prism of the early transition metal.
In
the present structure, four such chains of nickel centered within
confacial scandium trigonal prisms can be viewed as having
been condensed together, but the relationship is not as clear
and direct. One such Sc2-capped member in the upper right of
(2.67 Å), respectively, and also twice each to Sc4 and Sc1 or
(21) Pauling, L. The Nature of the Chemical Bond; Cornell University
Press: Ithaca, NY, 1960; p 400.
Sc5 at 3.17-3.48 Å. The two opposed double chains of

1948 Inorganic Chemistry, Vol. 38, No. 8, 1999
Maggard and Corbett
2
2
Figure 4. A [010] view of the metal chain in Gd
3
MnI
3
along the
Ni chain
short repeat. Compare with the top or bottom half of the Sc
in Figure 3.
5
2
rectangular pyramids, Sc2(Sc6/2)Ni2, left and right in Figure 2,
are then interconnected through Sc1-Sc5, Sc4-Sc4, and Ni1-
Ni2 bonding. (This description is helpful later in understanding
the overlap populations.) The nonequivalent nickel atoms are
reflected in the unequal but generally similar distances about
them, the most disparate of which being Ni1-Sc5, 2.65 Å and
Ni2-Sc1, 2.87 Å (not drawn) which reflect the skewness of
the centrosymmetric chain. Fourier difference maps do not show
any extra electron density in the central channel of the chain.
The chains are not significantly interbonded, as will be shown
for the shortest separation of this kind marked in Figure 1, 3.69
Å for Sc2‚‚‚Sc3. The Te-Te distances are all g3.95 Å. These
structural motifs and atomic distances support the 1D chain
assignment to Sc5Ni2Te2.
Figure 5. The densities-of-states from the EHTB band calculation for
Sc Ni Te . The separate Sc and Ni contributions are projected out.
5
2
2
Remarkably similar structural characteristics are found in the
chains in Gd3MnI3,22 the (Gd3Mn)2 portion of which is shown
along the short 4.13 Å axis in Figure 4.23 This is very similar
to the upper or lower half of the Sc5Ni2 chain (Figures 2,3).
The iodide was likewise described as the base-to-base assembly
of two chains of trans-edge-sharing rectangular pyramids, which
is a very atypical halide structure. This has been classified as a
distortion extreme of a family of compounds originating with
Pr3(Ru)I3, viz., double chains of Z-centered R6 octahedra that
share trans edges and are further condensed side-by-side. Their
progressive distortions in other examples can be described as
the partial fusion of the recognizable twin octahedral chains in
Pr3I3Ru, etc., as two adjoining octahedra begin to merge and
the Z elements approach one another.24 Although the Sc5Ni2-
Te2 and Gd3MnI3 compositions and geometries may be readily
interrelated, the electron-poorer interstitial Mn also exhibits
significant Mn-Mn bonding.
Theoretical Calculations. Intuitively, the character of the
bonding in Sc5Ni2Te2 must be highly delocalized, the bonded
metal neighbors ranging from five about Sc2 and Sc3 through
eight for Sc1 and Sc5, 10 for each Ni, and 14 for Sc4. Band
calculations seemed necessary to clarify the situation. Figure 5
shows the total DOS for Sc5Ni2Te2. Typical for these relatively
electron-poor compounds, the Fermi level resides on the low-
energy side of a prominent conduction band composed of about
Figure 6. Total COOP (overlap-weighted orbital population) curves
for indicated pairwise interactions in Sc Ni Te .
5
2
2
10
2 6
rigorously, as 2 Ni (d ) and 2 Te (s p ), with the 11 electrons
left over in the conduction band.
Figure 6 shows the COOP curves for total Ni-Ni, Sc-Sc,
and Sc-Ni bonding. EF falls largely within bonding regions
for both Sc-Sc and Sc-Ni and 0.4 eV below the onset of Sc-
Sc antibonding states. On the other hand, the Fermi level falls
in a nonbonding region for Ni-Ni interactions, these being
virtually closed shell above ∼-7.0 eV. This is a common
behavior for M-M bonding of late transition metals as they
10
25
achieve a closed shell d configuration, as noted in La2Ni2I.
Other cases of early-late transition metal chalcogenide structures
have been observed to optimize the M-Ni bonding as revealed
4
by COOP curves. It is interesting that this is not the situation
here, albeit in a structure that does not involve the trigonal
prisms in a linear face-sharing motif as before.
A complex structure such as this presents an ample variety
of distances and bond strengths. Comparisons of bond distances
with overlap populations allow one to ascertain where matrix
effects, separations fixed more by geometric relationships, may
be more important than bonding in near-neighbor contacts. For
this purpose, pairwise overlap populations for Sc-Sc, Sc-Ni,
and Ni-Ni are listed in Table 3 in decreasing magnitude along
with the corresponding distances. These change in parallel fairly
well, but there are some significant deviations that assist in
highlighting important bonding details. For one, the Sc1-Sc5
bridging interactions at 3.32 Å lie fairly low in the list (0.093).
The natural assumption is that this distance is moderately to
heavily influenced by matrix effects, in concert with the
9
0% scandium d and s orbitals and 10% nickel contributions.
The nickel d (and Sc d) states comprise all of the lower valence
bands between ∼-8.5 and -7.0 eV, with tellurium states being
the majority at still lower energies (off scale). A near gap at
∼
-6.9 eV occurs with 36 of the 47 total electrons per formula
unit. The band below there can be described simply, but not
(22) Ebihara, M.; Martin, J. D.; Corbett, J. D. Inorg. Chem. 1994, 33, 2078.
(23) This relationship was pointed out to us by a perceptive reviewer.
(24) Payne, M. W.; Dorhout, P. K.; Kim, S.-J.; Hughbanks, T. R.; Corbett,
J. D. Inorg. Chem. 1992, 32, 1389.
(25) Hong, S.-T.; Martin, J. D.; Corbett, J. D. Inorg. Chem. 1998, 37, 3385.

Sc5Ni2Te2: Synthesis, Structure, and Bonding
Inorganic Chemistry, Vol. 38, No. 8, 1999 1949
suggestion made earlier that the structure could be described
as two edge-sharing sheets built from square pyramids and
“
glued” together by Ni-Sc (and Ni-Ni) bonding, Figure 2.
The other cross-distance, Sc4-Sc4, is longer but more central,
and it has only a slightly lower population, 0.082 at 3.53 Å.
The Sc and Ni orbital energies are close enough that they do
mix well and give substantial overlap populations, whereas this
is not true for the Ni-Ni bonding when the atoms are this well-
reduced and have virtually closed shells. In this case, the
observed distance in the zigzag chain, which corresponds to a
Pauling bond order of 0.26, is quite misleading regarding the
actual bonding magnitude.
Other noteworthy differences are the relatively large overlaps
for Sc2-Sc2 (0.045) and Sc4-Sc4 (0.037) (but not for Sc1-
Sc1, Sc3-Sc3, and Sc5-Sc5) along the 3.95 Å chain repeat.
This result is not surprising for Sc4, which resides centrally
within the chain and shows effects that parallel those noted for
Sc2Te and Sc8Te3 where electrons appear to be concentrated
within metal-rich cluster chains that are isolated by nonmetals
5
Figure 7. Near-[010] projection of the structure of Hf Co1+xP3-x (0
7
e x e 0.5) with Co gray, P black. The chain unit evident in Sc
5 2
Ni -
2
Te (Figure 1) is outlined.
This condensation of the building units expresses the greater
number of metal-based electrons for hafnium compared with
scandium and, most certainly, the smaller size of phosphorus
compared with tellurium. Once again, as was the case for the
M8Ch3 phases (M ) Sc, Y; Ch ) Se, Te vs Te8Ch3, Ch ) S,
9
,10
neighbors.
But strong bonding along the chain for Sc2 as
well seems unusual as these atoms reside more on the periphery
of the chains, with three tellurium near neighbors. The larger
population here may mean additional bonding takes place for
scandium that otherwise has fewer (five) near metal neighbors.
Scandium 3 has the largest number of close Te neighbors (4)
and thence little Sc3-Sc3 bonding.
9
Se), there is a cooperative effect of increased anion size and
decreased valence electron concentration that acts to reduce the
dimensionality of the scandium interactions.
The interchain Sc-Sc distances lie near the bottom of the
overlap population list. The shortest distance between the chains
In addition to the reduction in dimensionality in Sc5Ni2Te2,
there is also a substantial difference in ordering of the nonmetal
(Sc2-Sc3, 3.69 Å) has an overlap population of 0.036, followed
7
and transition metal atoms. In Hf5Co1+xP3-x, a phase width (0
by smaller values for analogous but longer distances (Sc1-
Sc3, 0.028; Sc1-Sc2, 0.018). There are evidently only meager
amounts of electron density between the chains (<∼15% of
the larger internal populations), but whether these are remarkable
or substantial is doubtful.
e x e 0.5) arises from the varying occupancy of one intrachain
position by cobalt (gray) and phosphorus (black). The authors
attributed this in part to the similar atomic sizes and their
distances to hafnium. For Sc5Ni2Te2, the intrachain zigzag chains
are composed solely of nickel atoms, and the interchain cavities
are filled only by tellurium, which emphasize the lowered
dimensionality. This new ordering rests with the fact that
tellurium and nickel have disparate atomic sizes and very
different distances to scandium. Furthermore, a large phase
width is no longer a structural feature. The resultant ordering
still generates only fairly weak Ni-Ni bonding. The variability
between these structures is remarkable.
The notion was posited before that Sc-Ni bonding holds the
chain together, an idea that goes back to the original studies of
1
5
early-late transition metal bonding by Brewer and Wengert.
These overlaps are given at the bottom of Table 3. The seven
shorter types of Sc-Ni contacts around 2.61-2.70 Å have
overlap populations of about 0.17 to 0.21, while the three longer
contacts of 2.86 Å involving Sc4 or a long diagonal to Sc1
have lower values, ∼0.10. One important detail is that the Ni2-
Sc3 distance to the apex of one rectangular pyramid has a
somewhat larger overlap population than for the analogous Ni1-
Sc2 (0.209 vs 0.174), in parallel with the generally different
distances about these two-pseudo-equivalent atoms (i.e., Sc2-
Sc1 vs Sc3-Sc5). Similarly the “equivalent” Sc1-Ni2 and
Sc5-Ni1 diagonals differ by 0.22 Å and in parallel, so do the
overlap populations. The causes of these distortions are complex
and perhaps tied up with the factors behind optimization of
overall bonding.
There are interesting parallels and contrasts between the
metal-rich chalcogenide compounds of the rare-earth elements
(R) and the longer known halides. Twice as many halogen as
chalcogen atoms would be needed to achieve the same electron
count per metal, and this anion preponderance alone would
produce better separation of the metal-bonded units and reduc-
tion of the dimensionality of the halides, as observed. Isolated
halide clusters are thus much more common, and chains of edge-
sharing metal octahedra are relatively fewer (Y2Cl3, Sc5Cl8C,
Y4I5C, etc.). The halides are, in fact, generally so electron-poor
that interstitial heteroatoms are nearly always required for
stability. Including the electron count of the interstitial gives
Structural Comparisons. Sc5Ni2Te2 is nominally isotypic
with Hf5Co1+xP3- x, but there are in detail many differences in
the bonding. These involve the ordering pattern, the sizes of
the nonmetals and the transition metals, and presumably, the
-
7
2.3-2.8 electrons per cluster metal, 2.8-3.2 e in chains, and
electron counts. Figure 7 shows the Hf5Co1+xP3-x structure
X/R values of roughly 1.0 (condensed clusters) to 2.0 (iso-
nearly along [010], with the bonds in the double zigzag chain
unit that are comparable to those in Figure 1 outlined in black,
and the additional metal-metal interactions about it as open
bonds. The Hf-Hf and Hf-Co bonding in this occurs in
essentially a 3D arrangement, as judged by distances and,
1
4
lated). (The odd binary Y2Cl3 and Sc7Cl10 remain exceptions
-
at 1.8 e /R.) The binary chalcogenides Sc2Te and Sc8Te3 achieve
only slightly lower electron counts than cluster halides, 2.0-
2.25 per metal, but with disproportionately lower Ch/R values,
0.5-0.38. In other words, the grossest analysis suggests that
both classes of the metal-richest group 3 compounds achieve
similar electron counts per metal, ca. 2.0-2.5, but at greatly
7
especially, overlap populations. The few Hf-Hf distances
displayed on the structure show that the equivalent “interchain”
distances are approximately equal to or less than those within.

1950 Inorganic Chemistry, Vol. 38, No. 8, 1999
Maggard and Corbett
different stoichiometries: 1.3-1.7 X/R vs e 0.5 Ch/R.26 The
latter nicely correlates with the notably greater aggregation found
in the metal-richest chalcogenides. As with later transition metal
chalcogenides, these too would appear to be driven by M-M
bonding when the anion number is insufficient to give good
sheathing of the metal aggregate. It’s harder to compare these
two families of compounds electronically when they contain
late transition metal components (Gd3MnI3 vs Sc5Ni2Te2), but
if we ignore the contributions of the latter the difference in VEC
is similarly low (2.0 vs 2.2).
environments. One would in fact expect the chalcogenide-metal
interactions to be somewhat more covalent.
Conclusions. Sc5Ni2Te2 is significant because it represents
the first extension of early-late transition metal bonding in
chalcogenides to the earliest, and most electron-impoverished
transition metals. The structure is built up from double nickel
zigzag chains sheathered by scandium atoms and separated from
other Sc5Ni2 chains by tellurium atoms. The metal substructure
can be envisaged as trans-edge-shared double square pyramid
chains “glued” together at their bases mainly by Sc-Ni bonding.
Reduction in dimensionality compared with Hf5Co1+xP3-x results
from a new ordering of the nonmetal and late transition metal
and the different anionic nature of tellurium. A decreased
number of metal-based electrons is in concert with the absence
of interchain bonds via tellurium in this new phase. These
structural interrelationships aid in the understanding of this new
ternary phase.
Finally, the roles of the anions are rather different, evidently
because of differences in their quantity. Halides (Cl, Br, I)
generally exhibit several very discrete functions on these clusters
or chains, being two to four coordinate as they bridge all exposed
edges and bond exo at all vertexes of the cluster units.
14
Exceptions are rare. The greatly reduced telluride numbers
lead to much higher coordination numbers and less regular
geometries, usually some form of an augmented trigonal prism.
In the present compound, two different tellurides each have 6-7
neighbors at 2.91-3.06 Å, plus two more contacts at 3.16 Å.
These marked differences have generally led us to include the
halides in illustrations but to omit the less specific chalcogenide
Acknowledgment. This work was supported by the National
Science Foundation, Solid State Chemistry, via grants DMR-
9
510278 and -9809850 and was carried out in the facilities of
the Ames Laboratory, U.S. Department of Energy.
Supporting Information Available: Tables of additional crystal-
lographic and refinement parameters, anisotropic thermal parameters,
and a complete listing of nearest-neighbor distances (PDF). This
material is available free of charge via the Internet at http://pubs.acs.org.
(
26) The lowest X/R in individual closed-shell clusters is about 1.7 [Sc-
(
Sc6Cl12N), Y6I10Os]. Oxidation to give larger X/R values and less
interbridged cluster arrays, up to X/R ) 3 (R6X18), are compensated
by incorporation of electron-richer Z or additional alkali metals, or
-
both, the e /R remaining close to 2.3 throughout.
IC981073A