Covalent bonding and the trans influence in lanthanide compounds

Article abstract of DOI:10.1021/ic901571m

A pair of mer-octahedral lanthanide chalcogenolate coordination complexes [(THF)₃Ln(EC₆F₅)₃ (Ln = Er, E = Se; Ln = Yb, E = S)] have been isolated and structurally characterized. Both compounds show geometry-depe

Full text of DOI:10.1021/ic901571m

552 Inorg. Chem. 2010, 49, 552–560
DOI: 10.1021/ic901571m
Covalent Bonding and the Trans Influence in Lanthanide Compounds
Karsten Krogh-Jespersen,* Michael D. Romanelli, Jonathan H. Melman, Thomas J. Emge, and John G. Brennan*
Department of Chemistry and Chemical Biology, Rutgers, The State University of New Jersey, New Brunswick,
New Jersey 08903
Received August 14, 2009
A pair of mer-octahedral lanthanide chalcogenolate coordination complexes [(THF)3Ln(EC6F5)3 (Ln = Er, E = Se; Ln =
Yb, E = S)] have been isolated and structurally characterized. Both compounds show geometry-dependent bond
lengths, with the Ln-E bonds trans to the neutral donor tetrahydrofuran (THF) significantly shorter than the Ln-E
bonds that are trans to negatively charged EC6F5 ligands. Density functional theory calculations indicate that the
structural trans influence evidenced by the differences in these bond lengths results from a covalent Ln-E interaction
involving ligand p and Ln 5d orbitals.
Introduction
extent that the electronic spectra of solid-state ErF3 and
molecular (DME)2Er(SC6F5)3 are virtually identical.10 Struc-
tural features, until recently, could always be rationalized by
ionic bonding models, with bond lengths predicted by the
summation of ionic radii, taking into account the metal
oxidation state and coordination number. In the past decade,
however, exceptions to these rules have appeared.
In transition-metal and main-group chemistry, covalent
bonding interactions often produce a trans effect, in which a
metal-ligand bond length has been influenced by the identity
of the ligand trans to the bond in question.11 We originally
noted years ago, in the description of the mer-octahedral lanth-
anide thiolate coordination compound (py)3Yb(SPh)3,12 that
there appeared to be a structural trans influence, with the
Yb-S bond trans to SPh being significantly longer than the
Yb-S bond trans to the neutral pyridine donor. The same
nonclassical bond-length distributions have since been ob-
served in a series of related compounds, including chalcogen-
ido clusters13,14 and, most recently, a fluorinated phenolate.15
A fundamental understanding of f-block chemistry remains
a seminal challenge, thwarted by the complex angular proper-
ties of the f orbitals and the varying relative extensions of
(n)f vs (n þ 1)d and (n þ 2)s,p orbitals. Molecular actinide
chemistry has examples of clearly defined covalent bonding
interactions involving overlap with 5f orbitals, as illustrated in
the coordination of carbon monoxide to uranium(III)1,2 or the
variable-energy photoelectron spectra of uranocene.3 Numer-
ous structural pairs4-7 exist for which actinide-ligand bond
lengths do not obey the simple radius summation rules8,9 that
for decades dominated the interpretation of bond lengths in
actinide complexes.
In contrast, structure/property relationships in lanthanide
(Ln) chemistry have always appeared to be less complicated.
Ln magnetic properties can almost always be treated as
classical isolated spin systems, with 4f-4f interactions rarely
measurable above 20 K. Ln electronic properties are also
notoriously independent of the coordination sphere, to the
(10) Kornienko, A.; Emge, T. J.; Kumar, G. A.; Riman, R. E.; Brennan,
J. G. J. Am. Chem. Soc. 2005, 127, 3501.
*To whom correspondence should be addressed. E-mail: kroghjes@
rci.rutgers.edu (K.K.-J.), bren@rci.rutgers.edu (J.G.B.).
(1) (a) Parry, J.; Carmona, E.; Coles, S.; Hursthouse, M. J. Am. Chem.
Soc. 1995, 117, 2649. (b) Brennan, J. G.; Andersen, R. A.; Robbins, J. L. J. Am.
Chem. Soc. 1986, 108, 335.
(2) Evans, W. J.; Kozimor, S. A.; Nyce, G. W.; Ziller, J. W. J. Am. Chem.
Soc. 2003, 125, 12821.
(3) Brennan, J. G.; Cooper, G.; Green, J. C.; Payne, M. P.; Redfern, C. M.
J. Am. Chem. Soc. 1989, 111, 2373.
(4) Brennan, J. G.; Stults, S. D.; Andersen, R. A.; Zalkin, A. Organo-
metallics 1988, 7, 1329.
(5) Brennan, J. G.; Stults, S. D.; Andersen, R. A.; Zalkin, A. Inorg. Chim.
Acta 1987, 139, 201.
(6) Mehdoui, T.; Berthet, J. C.; Thuery, P.; Ephritikhine, M. Dalton
Trans. 2004, 579.
(7) Mazzanti, M.; Wietzke, R. L.; Pecaut, J.; Latour, J. M.; Maldivi, P.;
Remy, M. Inorg. Chem. 2002, 41, 2389.
(8) Raymond, K. N.; Eigenbrot, C. W. Acc. Chem. Res. 1980, 13, 276.
(9) Shannon, R. D. Acta Crystallogr., Sect. A 1976, 32, 751.
(11) (a) Appleton, T. G.; Clark, H. C.; Manzer, L. E. Coord. Chem. Rev.
1973, 10, 335. (b) Nakajima, K.; Kojima, M.; Fujita, J.; Ishii, T.; Ohba, S.; Ito, M.;
Saito, Y. Inorg. Chim. Acta 1985, 99, 143. (c) Roecker, L.; Dickman, M. H.;
Nosco, D. L.; Doedens, R. J.; Deutsch, E. Inorg. Chem. 1983, 22, 2022.
(d) Kastner, M. E.; Smith, D. A.; Kuzmission, A. G.; Cooper, J. N.; Tyree, T.;
Yearick, M. Inorg. Chim. Acta 1989, 158, 185. (e) Elder, R. C.; Trkula, M. Inorg.
Chem. 1977, 16, 1048. (f) Stein, C. A.; Ellis, P. E., Jr.; Elder, R. C.; Deutsch, E.
Inorg. Chem. 1976, 15, 1618. (g) Elder, R. C.; Trkula, M. J. Am. Chem. Soc.
1974, 96, 2635.
(12) Lee, J.; Brewer, M.; Berardini, M.; Brennan, J. Inorg. Chem. 1995, 34,
3215.
(13) Freedman, D.; Melman, J. H.; Emge, T. J.; Brennan, J. G. Inorg.
Chem. 1998, 37, 4162.
(14) Kornienko, A.; Emge, T.; Hall, G.; Brennan, J. G. Inorg. Chem. 2002,
41, 121.
(15) Norton, K.; Kumar, G. A.; Dilks, J. L.; Emge, T. J.; Riman, R. E.;
Brik, M. G.; Brennan, J. G. Inorg. Chem. 2009, 48, 3573.
r
pubs.acs.org/IC
Published on Web 12/21/2009
2009 American Chemical Society

Article
Inorganic Chemistry, Vol. 49, No. 2, 2010 553
Trans influences have also been noted in pairs of chemically
related compounds,16 but interpretations of the different bond
lengths are complicated by the difficulty of comparing com-
pounds for which different steric properties or different
crystal-packing forces may also be influencing the length of
the Ln-ligand bond.
While theoretical descriptions of covalent actinide-ligand
bonding abound, i.e., U(C8H8)217,18 or Cp3U(L) (L = CO,
PR3),19-21 examinations of Ln systems with clearly defined
covalent characteristics have yet to reveal any physical
characteristics that could clearly be attributed to a degree
of covalent bonding. The bis-arene lanthanides have been
described22-24 in terms of significant M-L π donation
between the Ln d orbitals and the arene π* orbitals. A density
functional theory (DFT) analysis of eight-coordinate Ln
compounds with S-based anions concluded that Ln-S bonds
are, in fact, “ionocovalent”, but a clear illustration of how
covalent bonding might impact the physical properties could
not be discerned, given the difficulty of visualizing orbital
overlap in these eight-coordinate structures.25 Finally, a
theoretical investigation into the bonding in Ln(N(ER)2)3
suggested that the Ln-E bonds were essentially ionic in
either the octahedral (all E) or nine-coordinate (6E, 3N)
geometries.26 The coordination environments in all of these
Ln compounds were either too symmetric or too asymmetric
to reveal any directional bonding effects.
Because Ln ions are extensively used in catalysis, an
appreciation of the extent to which covalent bonding might
impact Ln systems is of practical importance, and the
development of proper methodologies for computational
modeling of such processes is highly advantageous. Here
we outline the synthesis and structural characterization of
two new mer-octahedral Ln coordination complexes with
S- and Se-based anions, and we present results from DFT
calculations that suggest that the observed bond-length
distributions originate from covalent bonding involving the
overlap of ligand-based p orbitals with the Ln 5d orbitals.
Figure 1. ORTEP diagram of 1 with green indicating F atoms and gray
C atoms. The H atoms were removed for clarity.
according to a modified procedure27 of the initial synthesis,28 and
(SeC6F5)2 was prepared according to the literature. Er (chips)
29
and Yb (powder) metals were purchased from Aldrich and used as
received. Melting points were recorded in sealed capillaries and are
uncorrected. IR spectra were recorded on a Thermo Nicolet
Avatar 360 FTIR spectrometer from 4000 to 450 cm-1 as Nujol
mulls on CsI plates. Visible absorbance spectra were recorded on
either a Varian DMS 100S spectrometer or a Cary 50 Bio with the
samples dissolved in THF, placed in either a 1.0 mm ꢀ 1.0 cm
Spectrosil quartz cell or a 1.0 cm2 special optical glass cuvette, and
scanned between 190 and 1000 nm, with ranges depending on the
metal ion present in the sample. Elemental analyses were per-
formed by Quantitative Technologies, Inc. (Whitehouse, NJ).
Synthesis of (THF)3Yb(SC6F5)3 (1). Yb (0.083 g, 0.48 mmol)
and Hg(SC6F5)2 (0.447 g, 0.747 mmol) were combined in THF
(20 mL), and the mixture was stirred until Yb was consumed and
elemental Hg was visible in the bottom of the flask (overnight).
The pale-yellow solution was filtered away from the Hg (0.11 g,
73%), reduced in volume under vacuum to ca. 15 mL, and
layered with hexane (15 mL) to give yellow crystals (0.098 g,
21%) that melt at 79 °C, give a yellow-orange solid at 135 °C,
and continue to darken up to 350 °C. IR: 2950 (s), 2855 (s), 2724
(w), 1625 (w), 1508 (s), 1462 (s), 1377 (s), 1261 (w), 1172 (w),
1074 (m), 1002 (m), 971 (m), 857 (m), 722 (w) cm-1. UV-vis:
For a 6.1 mM THF solution with a 1.00 cm path length, this
compound shows a broad absorption maximum centered at
394 nm (ε = 208 L mol-1 cm-1) and a shoulder at 486 nm (ε =
Experimental Section
General Methods. All syntheses were carried out under ultra-
pure nitrogen (Welco Praxair), using conventional drybox or
Schlenk techniques. Tetrahydrofuran (THF; Aldrich) was puri-
fied with a dual-column Solv-Tek solvent purification system
and collected immediately prior to use. Hg(SC6F5)2 was prepared
3
3
88 L mol-1 cm-1). Anal. Calcd for C30H24YbF15O3S3: C, 36.5;
3
3
H, 2.45. Found: C, 36.5; H, 2.56. Single-crystal X-ray diffraction
˚
data (233 K): space group P21/c; a = 9.581(2) A; b = 15.885(2)
3
˚
˚
˚
A; c = 25.035(3) A; β = 100.03(1)°; V = 3452.3(9) A ; Z = 4;
D
calcd = 1.898 g cm-3
.
Synthesis of (THF)3Er(SeC6F5)3 (2). Er (0.084 g, 0.50 mmol),
(16) (a) Deacon, G. B.; Feng, T.; Skelton, B. W.; White, A. H. Aust.
J. Chem. 1995, 48, 741. (b) Panda, T. K.; Trambitas, A. G.; Bannenberg, T.; Hrib,
C. G.; Randoll, S.; Jones, P. G.; Tamm, M. Inorg. Chem 2009, 48, 5462.
(17) Moritz, A.; Dolg, M. Chem. Phys. 2007, 337, 48.
(18) Rosch, N.; Streitweiser, A. J. Am. Chem. Soc. 1983, 105, 7237.
(19) Vetere, V.; Maldivi, P.; Adamo, C. J. Comput. Chem. 2003, 24, 850.
(20) Bursten, B. E.; Rhodes, L. F.; Strittmatter, R. J. J. Am. Chem. Soc.
1989, 111, 2758.
(21) Maron, L.; Eisenstein, O.; Andersen, R. A. Organometallics 2009, 28, 3629.
(22) DiBella, S.; Lanza, G.; Fragala, I. L.; Marks, T. J. Organometallics
1996, 15, 3985.
(23) Hong, G. Y.; Schautz, F.; Dolg, M. J. Am. Chem. Soc. 1999, 121,
1502.
(24) King, W. A.; DiBella, S.; Lanza, G.; Khan, K.; Duncalf, D. J.; Cloke,
F. G. N.; Fragala, I. L.; Marks, T. J. J. Am. Chem. Soc. 1996, 118, 627.
(25) Roger, M.; Barros, N.; Arliguie, T.; Thuery, P.; Maron, L.;
Ephritikhine, M. J. Am. Chem. Soc. 2006, 128, 8790.
(26) Gaunt, A. J.; Reilly, S. D.; Enriquez, A. E.; Scott, B. L.; Ibers, J. A.;
Sekar, P.; Ingram, K. I. M.; Kaltsoyannis, N.; Neu, M. P. Inorg. Chem. 2008,
47, 29.
(27) Melman, J. H.; Emge, T. J.; Brennan, J. G. Inorg. Chem. 2001, 40,
1078.
(28) Peach, M. E. J. Inorg. Nucl. Chem. 1973, 35, 1046.
3
(SeC6F5)2 (0.37 g, 0.75 mmol), and Hg (0.026 g) were combined
in THF (ca. 20 mL), and the mixture was stirred until the metal
flakes were completely consumed (8 days). The pale-pink solu-
tion was filtered and concentrated to ∼5 mL. The solution was
held at -5 °C for 2 days, brought to room temperature, and
layered 2:1 with hexanes to give pale-pink lathes (0.42 g, ∼75%)
that melt at 153 °C. IR: 2958 (s), 1634 (w), 1605 (w), 1531 (w),
1507 (s), 1474 (s), 1254 (m), 1070 (s), 960 (s), 809 (s), 669 (m), 604
(w) cm-1. UV-vis: For a 0.104 M THF solution with a 1.00 mm
path length, this compound shows absorption maxima at 655
(2.04 L mol-1 cm-1), 547 (ε = 1.3 L mol-1 cm-1), 523 (ε =
3
3
3
3
27 L mol-1 cm-1), 490 (ε = 3.3 L mol-1 cm-1), 453 (ε =
3
3
3
3
3.0 L mol-1 cm-1), 409 (ε = 8.6 L mol-1 cm-1), and 380 (ε =
3
3
3
3
3
3
75 L mol-1 cm-1) nm. These absorption peaks correspond
4
4
to the F9/2, , , , ,
4S3/2 2H11/2 4F7/2 4F5/2,3/2 2G9/2, and G11/2
€
(29) Klapotke, T. M. Eur. J. Inorg. Chem. 1999, 1359.

554 Inorganic Chemistry, Vol. 49, No. 2, 2010
Krogh-Jespersen et al.
Figure 2. ORTEP diagrams for the four independent molecules within the unit cell of2 with greenindicatingF atoms and gray C atoms. The H atoms were
removed for clarity.
4
transitions typically seen from the I15/2 ground state of Er3þ
.
Computational Details
Anal. Calcd for C30H24F15O3Se3Er: C, 32.1; H, 2.16. Found: C,
31.6; H, 1.97. Single-crystal X-ray diffraction data (100 K):
Electronic structure calculations have been carried out at the
DFT level of theory.34 A large majority of the calculations
made use of the PBE35 combination of exchange and correla-
tion functionals, although the TPSS,36 B3LYP,37 and M0538
functionals were also tested. The SDD-model MWB57 and
MWB59 quasi-relativistic effective core potentials (ECPs)
were used to represent the atomic cores of the Ln atoms Er
and Yb.39 These large-core ECPs incorporate all of the elec-
trons in the 4f shell, leaving only the 5s25p66s25d1 (valence)
electrons explicitly included in the calculations, and are appro-
priate for LnIII oxidation states. Computational justification
for the use of large-core ECPs with Ln compounds has been
provided.40 Inner-shell electrons on Se atoms were replaced by
the 28-electron MWB28 ECP;41 calculations employing BasisI
(see below) also made use of an ECP for S atoms (MWB10).41
˚
˚
space group P21; a = 19.481(2) A; b = 15.625(2) A; c =
3
˚
˚
23.555(2) A; β=105.766(2)°; V=6900(1) A ; Z=8; Dcalcd
=
2.159 g cm-3
X-ray Structure Determination. Data for 1 were collected on
.
3
an Enraf-Nonius CAD4 diffractometer with graphite-mono-
˚
chromatized Mo KR radiation (λ=0.710 73 A). The crystal of 1
was glued to a glass fiber and placed in a cold nitrogen stream.
Data for 2 were collected on a Bruker Smart APEX CCD
diffractometer with graphite-monochromatized Mo KR radia-
˚
tion (λ=0.710 73 A) at 100 K. The crystal of 2 was immersed in
Paratone oil and placed in a cold nitrogen stream. The data for 1
and 2 were corrected for Lorenz effects, polarization, and
absorption, the latter by a multiscan method (SADABS).30
The structures were solved by direct methods (SHELXS86).31
All non-H atoms were refined (SHELXL97)32 based upon Fobs
.
2
(34) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and
Molecules; University Press: Oxford, U.K., 1989.
(35) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
3865.
(36) Tao, J. M.; Perdew, J. P.; Staroverov, V. N.; Scuseria, G. E. Phys.
Rev. Lett. 2003, 91, 146401.
(37) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5468. (b) Lee, C.; Yang, W.;
Parr, R. G. Phys. Rev. B 1988, 37, 785.
All H-atom coordinates were calculated with idealized geome-
tries (SHELXL97). Scattering factors ( f°, f0, f00) are as described
in SHELXL97. ORTEP diagrams33 for 1 and 2 are shown in
Figures 1 and 2, respectively. Significant bond lengths and
angles for 1 and 2 are given in Tables 1 and 2, respectively.
Complete crystallographic details are given in the Supporting
Information.
(38) Zhao, Y.; Schultz, N. E.; Truhlar, D. G. J. Chem. Phys. 2005, 123,
161103.
(39) (a) Dolg, M.; Stoll, H.; Savin, A.; Preuss, H. Theor. Chim. Acta 1989,
75, 173. (b) Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1993, 85, 441.
(c) Yang, J.; Dolg, M. Theor. Chem. Acc. 2005, 113, 212.
(40) (a) Eisenstein, O.; Maron, L. J. Organomet. Chem. 2002, 647, 190.
(b) Maron, L.; Eisenstein, O. New J. Chem. 2001, 25, 255. (c) Maron, L.;
Eisenstein, O. J. Phys. Chem. A 2000, 104, 7140.
(30) SADABS, Bruker Nonius area detector scaling and absorption correc-
tion, version 2.05; Bruker-AXS Inc.: Madison, WI, 2003.
(31) Sheldrick, G. M. SHELXS86, Program for the Solution of Crystal
€
€
Structures; University of Gottingen: Gottingen, Germany, 1986.
(32) Sheldrick, G. M. SHELXL97, Program for Crystal Structure Refine-
€
€
ment; University of Gottingen: Gottingen, Germany, 1997.
(33) Persistence of Vision Raytracer, version 3.6; 2004 [computer software
retrieved from http://www.povray.org/download/].
(41) Bergner, A.; Dolg, M.; Kuechle, W.; Stoll, H.; Preuss, H. Mol. Phys.
1993, 80, 1431.

Article
Inorganic Chemistry, Vol. 49, No. 2, 2010 555
˚
˚
Table 1. Significant Bond Lengths [A] and Angles [deg] for 1
Table 2. Significant Bond Lengths [A] and Angles [deg] for 2
Yb1-O1
Yb1-O2
Yb1-S1
S1-C1
2.271(4)
2.308(5)
2.678(2)
1.766(7)
1.755(7)
Yb1-O3
Yb1-S2
Yb1-S3
S2-C7
2.280(4)
2.642(2)
2.680(2)
1.741(6)
Molecule 1
Er1-O3
Er1-O2
Er1-Se1
Se1-C1
Se3-C13
2.297(4)
2.319(4)
2.8299(7)
1.915(6)
1.906(6)
Er1-O1
2.304(4)
2.8042(7)
2.8550(7)
1.927(6)
Er1-Se2
Er1-Se3
Se2-C7
S3-C13
O1-Yb1-O3
O3-Yb1-O2
O3-Yb1-S2
O1-Yb1-S1
O2-Yb1-S1
O1-Yb1-S3
O2-Yb1-S3
S1-Yb1-S3
C7-S2-Yb1
172.04(16)
82.44(17)
93.45(13)
90.75(12)
81.89(14)
87.56(12)
99.33(14)
177.90(6)
108.2(2)
O1-Yb1-O2
O1-Yb1-S2
O2-Yb1-S2
O3-Yb1-S1
S2-Yb1-S1
O3-Yb1-S3
S2-Yb1-S3
C1-S1-Yb1
C13-S3-Yb1
89.74(17)
94.51(11)
167.45(14)
89.64(12)
86.25(6)
92.21(12)
92.65(6)
107.2(2)
109.1(2)
O3-Er1-O2
O3-Er1-Se2
O2-Er1-Se2
O1-Er1-Se1
Se2-Er1-Se1
O1-Er1-Se3
Se2-Er1-Se3
C1-Se1-Er1
C13-Se3-Er1
85.39(14)
90.59(10)
167.16(11)
89.27(12)
87.916(19)
83.68(12)
91.511(19)
105.55(17)
101.51(17)
O3-Er1-O1
O1-Er1-O2
O1-Er1-Se2
O3-Er1-Se1
O2-Er1-Se1
O3-Er1-Se3
O2-Er1-Se3
Se1-Er1-Se3
C7-Se2-Er1
172.89(15)
87.66(16)
96.52(12)
91.02(10)
79.98(11)
96.14(10)
101.04(11)
172.82(2)
99.30(16)
Two collections of atomic basis sets were assembled. A
large number of exploratory calculations were carried out on
1, 2, and related species with basis sets of modest size (BasisI).
BasisI consists of the small valence basis sets provided with
the MWB ECPs: [5s4p3d] for Yb and Er and [2s3p] for S and
Se.39,41 In addition, S and Se atoms received a single set of
d-type polarization functions.42 The outermost p function in
the [2s3p] S and Se valence basis sets effectively serves as a
diffuse function. O atoms carried a Dunning-Huzinaga
[3s,2p] basis set, augmented by a set of d-type polarization
functions and a set of diffuse p functions,43 whereas valence
double-ζ-quality Dunning-Huzinaga basis sets were placed
onall C, F, and H atoms ([3s,2p] forC and F and [2s] for H).43
Selected calculations were made with modified Er or Yb basis
sets as described in the text.
A much larger basis set, BasisII, was formed for the final
calculations. BasisII contains for Er and Yb the recently
developed MWB-II [6s5p5d2f2g] basis sets, which include
several sets of polarization functions;44a,b for the Se valence
electrons, we used an augmented triple-ζ-quality basis set
[4s4p3d2f].44c All-electron Dunning-type basis sets were used
for the remaining atoms: aug-cc-pVTZ for O and S atoms,
D95(d) for C and F atoms, and [2s] for H atoms.43,45 Calcula-
tions on the molecular compounds 1 and 2 involved about
600 (BasisI) and 1100 (BasisII) basis functions, respectively.
Molecular geometrieswerefully optimizedon thepotential
energy surfaces. Normal-mode analysis (PBE/BasisI) verified
that the located stationary points were actual minima on the
PES. Electronic population analysis employed the natural
bond order (NBO) scheme of Weinhold et al.46
Molecule 2
Er2-O6
Er2-O5
Er2-Se6
Se4-C31
Se6-C43
2.286(4)
2.314(4)
2.8438(6)
1.915(6)
1.917(6)
Er2-O4
Er2-Se5
Er2-Se4
Se5-C37
2.298(4)
2.8111(6)
2.8454(6)
1.912(5)
O6-Er2-O5
O6-Er2-Se5
O5-Er2-Se5
O4-Er2-Se6
Se5-Er2-Se6
O4-Er2-Se4
Se5-Er2-Se4
C31-Se4-Er2
C43-Se6-Er2
82.62(14)
94.40(10)
170.61(11)
86.76(10)
87.417(18)
85.72(10)
86.397(19)
103.65(16)
104.13(17)
O6-Er2-O4
O4-Er2-O5
O4-Er2-Se5
O6-Er2-Se6
O5-Er2-Se6
O6-Er2-Se4
O5-Er2-Se4
Se6-Er2-Se4
C37-Se5-Er2
165.27(14)
82.67(14)
100.22(10)
95.72(10)
101.71(10)
93.49(10)
84.91(10)
169.27(2)
104.71(16)
Molecule 3
Er3-O7
Er3-O8
Er3-Se9
Se7-C61
Se9-C73
2.314(4)
2.322(4)
2.8233(7)
1.919(6)
1.894(7)
Er3-O9
Er3-Se8
Er3-Se7
Se8-C67
2.318(4)
2.8042(6)
2.8278(6)
1.917(5)
O7-Er3-O8
O7-Er3-Se8
O8-Er3-Se8
O9-Er3-Se9
Se8-Er3-Se9
O9-Er3-Se7
Se8-Er3-Se7
C61-Se7-Er3
C73-Se9-Er3
85.91(14)
91.06(10)
166.48(11)
92.86(10)
89.05(2)
88.24(10)
95.939(18)
102.30(17)
103.37(17)
O7-Er3-O9
O9-Er3-O8
O9-Er3-Se8
O7-Er3-Se9
O8-Er3-Se9
O7-Er3-Se7
O8-Er3-Se7
Se9-Er3-Se7
C67-Se8-Er3
175.50(14)
90.56(14)
93.01(10)
89.13(10)
77.74(11)
89.43(10)
97.20(11)
174.83(2)
99.03(16)
All computational work was performed using the Gaussian
0347 software package.
Results
mer-Octahedral coordination compounds (THF)3Ln-
(EC6F5)2 [Ln = Yb, E = S (1); Ln = Er, E = Se (2)]
Molecule 4
Er4-O12
Er4-O11
Er4-Se12
Se10-C91
Se12-C103
2.304(4)
2.337(4)
2.8276(6)
1.903(5)
1.903(7)
Er4-O10
Er4-Se11
Er4-Se10
Se11-C97
2.310(4)
2.8019(6)
2.8334(6)
1.913(5)
(42) Check, C. E.; Faust, T. O.; Bailey, J. M.; Wright, B. J.; Gilbert, T. M.;
Sunderlin, I. S. J. Phys. Chem. A 2001, 105, 8111.
(43) Dunning, T. H.; Hay, P. J. In Modern Theoretical Chemistry;
Schaefer, H. F., III, Ed.; Plenum: New York, 1976; Vol. 3, pp 1-28.
(44) (a) Yang, J.; Dolg, M. Theor. Chem. Acc. 2005, 113, 212. (b) Weigand,
A.; Cao, X.; Yang, J.; Dolg, M. Theor. Chem. Acc. 2009, accepted. (c) Martin,
J. M. L.; Sundermann, A. J. Chem. Phys. 2001, 114, 3408. MWB basis sets were
retrieved from http://www.theochem.uni-stuttgart.de/pseudopotentials.
(45) (a) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007. (b) Kendall,
R. A.; Dunning, T. H., Jr.; Harrison, R. J. J. Chem. Phys. 1992, 96, 6796.
(c) Woon, D. E.; Dunning, T. H., Jr. J. Chem. Phys. 1993, 98, 1358.
(46) (a) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88,
899. (b) Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F. QCPE
Bull. 1990, 10, 58.
O12-Er4-O11
O12-Er4-Se11
O11-Er4-Se11
O10-Er4-Se12
Se11-Er4-Se12
O10-Er4-Se10
Se11-Er4-Se10
C91-Se10-Er4
C103-Se12-Er4
82.62(14)
90.76(10)
165.42(10)
91.54(10)
89.22(2)
87.49(10)
95.989(19)
98.48(15)
100.23(17)
O12-Er4-O10
O10-Er4-O11
O10-Er4-Se11
O12-Er4-Se12
O11-Er4-Se12
O12-Er4-Se10
O11-Er4-Se10
Se12-Er4-Se10
C97-Se11-Er4
176.04(14)
94.91(14)
92.34(10)
90.97(10)
77.97(11)
89.74(10)
96.95(11)
174.73(2)
97.21(16)
(47) Frisch, M. J.; et al. Gaussian 03, revision E.01; Gaussian, Inc.:
Wallingford, CT, 2004. See the Supporting Information for the complete reference
for Gaussian 03.
were isolated in high yield and fully characterized.
The thiolate compound 1 is prepared most readily by

556 Inorganic Chemistry, Vol. 49, No. 2, 2010
Krogh-Jespersen et al.
transmetalation reactions of elemental Yb with Hg(SC6F5)2
in THF (eq 1),
Table 3. Comparison of Computed (PBE/BasisII) and Experimental (X-ray)
Metal-Ligand Bond Lengths [A] for 1 and 2 and Computed (PBE/BasisII)
Wiberg Bond Orders at DFT-Optimized and X-ray Geometries
a,b
˚
bond length
bond order
2Yb þ 3HgðSC6F5Þ2 f 2ðTHFÞ3YbðSC6F5Þ3 þ 3Hg ð1Þ
DFT
X-ray
DFT
X-ray
whereas selenolate 2 can be prepared by the direct reduction
of diselenide RSeSeR with elemental Er in THF (eq 2) and a
trace Hg catalyst.
Yb1-O1
Yb1-O2
Yb1-O3
Yb1-S1
Yb1-S2
Yb1-S3
2.339
2.440
2.341
2.708
2.652
2.716
0.100
-0.060
2.271
2.308
2.280
2.678
2.642
2.680
0.033
-0.037
0.129
0.119
0.123
0.500
0.559
0.487
0.143
0.132
0.139
0.512
0.568
0.509
2Er þ 3C6F5Se-SeC6F5 f 2ðTHFÞ3ErðSeC6F5Þ3 ð2Þ
The different approaches reflect different chemistries associated
with the ligands, not with the redox behavior of Ln, given that
attempts to reduce the disulfide with both redox active and
inactive Ln do not appear to proceed at an appreciable rate.
Both 1 and 2 adopt mer-octahedral geometries, with
inequivalent chalcogenolate andTHF ligand positions within
the primary coordination sphere. In each structure, there is a
consistent pattern of Ln-E bond lengths, with the Ln-E
bond trans to anionic E significantly longer than the Ln-E
bond trans to neutral THF. In thiolate 1, there is a single
crystallographically unique molecule in the unit cell, and the
c
ΔYb-O
d
ΔYb-S
bond length
bond order
DFT
X-ray
DFT
X-ray
Er4-O12
Er4-O11
Er4-O10
Er4-Se12
Er4-Se11
Er4-Se10
2.373
2.478
2.375
2.858
2.807
2.874
0.104
-0.059
2.304
2.337
2.310
2.828
2.802
2.833
0.030
-0.029
0.128
0.114
0.120
0.585
0.646
0.569
0.155
0.142
0.151
0.622
0.698
0.620
c
˚
˚
ΔEr-O
Yb-S2 bond [2.642(2) A] trans to THF is 0.037 A shorter
d
ΔEr-Se
˚
thaneither of the Yb-S bonds [2.678(2) and2.680(2) A] trans
a Cartesian geometries of 1 and 2 are available as Supporting Infor-
mation. b Comparison is made to molecule 4 in the unit cell of 2 because
the conformation of this molecule shows the strongest similarity to the
to thiolate. The selenolate compound 2 shows the same
directional dependence of the metal-ligand bond lengths,
but in this case, there are four crystallographically indepen-
dent molecules per unit cell and thus a range of bond-length
conformation of 1. c ΔLn-O = (Ln-O bond length when trans to anionic
d
E) - (average Ln-O bond length when trans to THF). ΔLn-E
=
˚
differences. The Er-Se bonds trans to THF are 0.038 A
(Ln-E bond length when trans to THF) - (average Ln-E bond length
when trans to anionic E).
˚
˚
˚
(Er1), 0.034 A (Er2), 0.021 A (Er3), and 0.029 A (Er4) shorter
than the average of the Er-Se bonds trans to SeC6F5. The
bond-length difference in the thiolate compound is slightly
computed Ln-E distances (Yb1-S2 and Er4-Se11) deviate
less than 0.01 A from the X-ray values. The distinct asym-
˚
˚
greater (∼0.007 A) than the average of the four sets of bond-
metry observed inthe metal-ligand bond lengths of 1 and 2 is
clearly expressed, and even enhanced (relative to the crystallo-
graphic values), in the electronic structure calculations. We
define the asymmetry parameter ΔLn-E as the difference
between the Ln-E bond length, when trans to THF, and the
length differences in the selenolate compound.
A similar pattern is noted for the Ln-O(THF) bonds; the
Ln-O bond for the THF ligands trans to the chalcogenolate
ligands is significantly longer than the Ln-O bonds trans
to O(THF). In the thiolate compound 1, the difference is
average of the two Ln-E bond lengths, when trans to
˚
0.032 A. In the selenolate compound 2, the four independent
˚
anionic E, and compute ΔYb-S =-0.060 A and ΔEr-Se
=
˚
molecules have Er-O bonds trans to Se that are 0.019 A
˚
-0.059 A, as compared with observed values of -0.037 and
˚
˚
˚
˚
(Er1), 0.022 A (Er2), 0.006 A (Er3), and 0.030 A (Er4) A
longer than the average of the Er-O bonds trans to THF.
Again, the difference in the bond lengths is greater in the
˚
-0.029 A (Er4), respectively. When ΔLn-O is defined in an
analogous manner [ΔLn-O=(Ln-O bond length when trans
to anionic E) -(average Ln-O bond length when trans
to THF)], the computed asymmetry in Ln-O distances is
˚
thiolate compound (by ∼0.013 A). Attempts to measure the
trans influence exerted by other Lewis bases invariably led to
the formation of seven-coordinate species, either by incor-
poration of an additional base or by dative coordination of
fluoride;12,15,27 only the THF derivatives formed six-coordi-
nate products.
also considerably larger than that observed, viz., ΔYb-O
=
˚
ΔEr-O ∼ 0.10 A, as compared with the experimental value of
˚
approximately 0.03 A.
Computed values for the Ln-E bond lengths are not
strongly sensitive to the type of DFT methodology used. Table
S-1 in the Supporting Information presents optimized metal-
ligand bond lengths for 1 and 2 obtained with BasisII and
exchange-correlation functionals of increasing sophistication
(representatives from different rungs on “Jacob’s Ladder”48):
PBE,35 a pure generalized gradient approximation (GGA)
functional (second rung; these data are also in Table 3);
TPSS,49 a meta-GGA functional (third rung); B3LYP,50 the
The X-ray structure of 1 provided the initial geometry for
unconstrained geometry optimizations of 1 and 2 carried out
with several DFT calculations and ECP/basis set combina-
tions (see the Computational Details section); no significant
changes in the overall conformation occurred during any of
the optimizations. The optimized PBE/BasisII Ln-E bond
lengths are listed in Table 3 (full Cartesian geometries of 1
and 2 are available as Supporting Information). The com-
puted metal-ligand bond lengths are systematically larger
than the observed X-ray values, and the differences are
largest for the weak Ln-O(THF) bonds. The computed
Ln-O bond lengths, when O(THF) is trans to anionic E
(Yb1-O2 and Er4-O11), deviate most strongly from the
(48) Perdew, J. P.; Schmidt, K. In Density Functional Theory and Its
Applications to Materials; Van Doren, V. E., Van Alsenoy, K., Geerlings, P., Eds.;
American Institute of Physics Press: New York, 2001.
(49) Tao, J. M.; Perdew, J. P.; Staroverov, V. N.; Scuseria, G. E. Phys.
Rev. Lett. 2003, 91, 146401.
(50) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5468. (b) Lee, C.; Yang, W.;
Parr, R. G. Phys. Rev. B 1988, 37, 785.
˚
observed values (∼0.14 A). Conversely, the corresponding

Article
Inorganic Chemistry, Vol. 49, No. 2, 2010 557
standard hybrid-GGA functional (fourth rung); and, finally,
M05,51 representing hybrid meta-GGA functionals (also
fourth rung). The bond lengths obtained with the TPSS,
B3LYP, and M05 functionals are overall very similar in
magnitude to those obtained with the PBE functionals; the
largest differences in the stronger Ln-S,Se bonds are about
Table 4. Comparison of Computed (PBE/BasisI) Metal-Ligand Bond Lengths
˚
[A] for 1 and 2 with Modified Yb or Er Basis Sets
Yb basis set
[5s4p3d]
[5s4p]
[5s4p2d]c
Yb1-O1
Yb1-O2
Yb1-O3
Yb1-S1
Yb1-S2
Yb1-S3
2.331
2.411
2.344
2.727
2.681
2.729
0.074
-0.047
2.399
2.410
2.383
2.877
2.867
2.874
0.019
-0.008
2.389
2.429
2.400
2.806
2.780
2.797
0.035
-0.021
˚
0.02 A, whereas the differences in the weaker Ln-O bonds
˚
may approach 0.04 A. With TPSS functionals, the Ln-O
bonds are slightly shorter than those obtained with PBE, but
the Ln-S,Se bonds are longer. The bond lengths obtained
with the two hybrid functionals, B3LYP and M05, are mostly
larger than those obtained with the functionals that do not
contain exact exchange, PBE and TPSS. Importantly, sub-
stantial asymmetry in the metal-ligand bond lengths is
present with all functionals applied. The asymmetry para-
meters (quoting data for 1 only; the results on 2 are similar)
a
ΔYb-O
b
ΔYb-S
Yb basis set
[5s1p3d]d
[2s4p3d]
[5s4p3d3f]
[5s4p3f]
Yb1-O1
Yb1-O2
Yb1-O3
Yb1-S1
Yb1-S2
Yb1-S3
2.322
2.388
2.337
2.717
2.675
2.720
0.059
-0.044
2.288
2.391
2.318
2.727
2.677
2.732
0.088
-0.052
2.322
2.405
2.335
2.719
2.674
2.719
0.077
-0.045
2.372
2.397
2.374
2.864
2.850
2.860
0.024
-0.012
˚
˚
˚
are larger for M05 (ΔYb-S=-0.062 A and ΔYb-O=0.114 A)
˚
than for PBE (ΔYb-S = -0.060 A and ΔYb-O = 0.100 A)
a
˚
ΔYb-O
but slightly smaller for TPSS (ΔYb-S = -0.051 A and
b
ΔYb-S
˚
˚
Δ
Δ
Yb-O = 0.075 A) and B3LYP (ΔYb-S = -0.063 A and
˚
Er basis set
[5s4p3d]
[5s4p]
[5s1p3d]c
[2s4p3d]
Yb-O=0.075 A).
Calculations with the smaller basis set (BasisI) and these
Er4-O12
Er4-O11
Er4-O10
Er4-Se12
Er4-Se11
Er4-Se10
2.368
2.443
2.370
2.876
2.843
2.892
0.074
-0.041
2.416
2.440
2.430
3.035
3.034
3.041
0.017
-0.004
2.362
2.419
2.363
2.871
2.845
2.891
0.057
-0.036
2.336
2.414
2.351
2.878
2.834
2.896
0.071
-0.053
same four DFT functionals (Table S-2 in the Supporting
Information) generally result in slightly shorter bonds and
diminished asymmetry parameters, relative to those obtained
with BasisII. For example, the asymmetry parameters ob-
tained for 1 using PBE functionals and BasisI are ΔYb-S
=
a
˚
˚
ΔEr-O
-0.048 A and ΔYb-O=0.073 A. The additional calculations
just summarized (variations in the applied DFT functionals
and basis sets) support the conclusion that the computed
bond lengths shown in Table 3, and the trends they display,
are part of a general pattern and not fortuitous results arising
from the selection of one particular computational model.
The computed charge distributions for 1 and 2 were
subjected to NBO analysis.46 Wiberg bond indices52 for the
metal-ligand bonds, obtained at the DFT equilibrium geo-
metries and at the experimental X-ray geometries (PBE/
BasisII), support the presence of covalent Ln-S,Se bonding
(Table 3). At the DFT-optimized geometries, computed
Yb-S bond orders are 0.49 and 0.50 for the two nearly
equivalent Yb-S bonds but are larger at 0.56 for the Yb-S
bond trans to THF. Similarly, the Er-Se bond orders for the
nearly equivalent bonds are 0.64 and 0.66, whereas the
unique Er-Se bond trans to THF shows a larger bond order
of 0.69. The Ln-O bonds, on the other hand, are over-
whelmingly ionic in character with uniformly small bond
orders (∼0.12); the Ln-O bond order is smallest when the
bond is trans to an anionic group. Within a particular type of
bond (Yb-S, Er-Se, or Ln-O), the magnitude of the bond
order correlates qualitatively with the length of the bond. The
X-ray-derived geometries feature shorter metal-ligand
bonds, and hence the bond orders obtained at these geo-
metries are slightly larger than those at the computed
equilibrium geometries (Table 3). However, the patterns
exhibited by the bond orders are identical at the two sets of
geometries. The distinctly nonzero Ln-E Wiberg bond
orders support the notion that covalency contributes to
Yb-S and Er-Se bonding; furthermore, the variations
computed in these bond orders suggest that the asymmetry,
b
ΔEr-Se
a ΔLn-O = (Ln-O bond length when trans to anionic E) - (average
Ln-O bond length when trans to THF). b ΔLn-E = (Ln-E bond length
when trans to THF) - (averageLn-E bond length when trans to anionic
E). c Innermost d-type basis function deleted. d Innermost p-type basis
function kept.
computed and observed, in the length of the Ln-E bonds is
associated with the differences in the magnitudes of the
covalent interaction.
As mentioned earlier, the electronic configuration for
neutral Yb and Er considered in the calculations is formally
5s25p65d16s2; hence, a metal electronic configuration ap-
proaching 5s25p65d06s0 would be representative of a fully
formed LnIII cation. However, the natural electronic config-
urations (NECs) for Yb and Er in 1 and 2 are 5s25p5.99
-
5d0.916s0.276d0.13 and 5s25p5.985d1.166s0.296p0.016d0.13, respec-
tively, resulting in formal charges of only 1.68þ for Yb and
1.41þ for Er. This analysis indicates that the positive charge
on the Ln atom is associated exclusively with a loss of electron
density from the 6s orbitals; Yb (or Er) actually has a slightly
higher occupancy of d electrons in the valence shell in
complex 1 (or 2) than in the free atom. Virtually no changes
occur in the Ln p orbital occupancies from the free atom to
either complex. The low occupancy of metal 6s orbitals does
not appear capable of inducing metal-ligand bond length
asymmetry and, consequently, the NECs suggest that partial
occupancy of d orbitals is the principal source of covalency
and trans influence in Ln complexes 1 and 2.
To further investigate this issue, additional geometry opti-
mizations on complexes 1 and 2 were carried out with
selectively modified Ln basis sets. Computed metal-ligand
bond lengths from all of the calculations are available in Table
S-3 in the Supporting Information; a condensed version is
presented as Table 4. The Ln basis set in BasisI is of the
[5s4p3d] type; i.e., there are three sets of d-type basis functions
(51) Zhao, Y.; Schultz, N. E.; Truhlar, D. G. J. Chem. Phys. 2005, 123,
161103.
(52) Wiberg, K. B. Tetrahedron 1968, 24, 1083.

558 Inorganic Chemistry, Vol. 49, No. 2, 2010
Krogh-Jespersen et al.
˚
covering the 5d (and higher) atomic orbitals, four p-type basis
functions for the 5p (and higher) atomic orbitals, etc. This type
of Yb basis set led to computed asymmetry parameters of
Table 5. Summary of the Ln-Ligand Bond Lengths (A) in Octahedral MX3Y3
Coordination Compounds
Ln-X
Ln-X
˚
˚
a
Δ
Yb-S=-0.048 A and ΔYb-O=0.073 A in 1 (Table 4 and
molecule
trans to X trans to Y Δx-y
*
ref
Table S-3 in the Supporting Information). Upon deletion of
the most diffuse d function from the Yb basis set, absolute
(OPR3)NdCl3
(THF)3YbCl3
(THF)3YbBr3
(THF)3YbI3
(HMPA)3PrCl3
(HMPA)3DyCl3
(HMPA)3YbCl3
(py)3Yb(SPh)3
2.787(3)
2.532(3)
2.708(1)
2.954(2)
2.729(3)
2.636(3)
2.591(4)
2.666(2)
2.709(3)
2.513(4)
2.665(1)
2.915(2)
2.706(3)
2.626(4)
2.582(5)
2.609(4)
2.720(3)
2.811(1)
2.718(1)
2.084(2)
2.642(2)
2.805(2)
0.078
64
55
56
57
63
65
66
12
53
67
67
changes in all Ln-E bond lengths are less than 10-3 A and no
0.019
0.043
0.039
0.023
0.010
0.009
0.057
0.030
0.015
0.016
0.027
0.039
0.031
˚
changes occur to the asymmetry parameters. When the two
outermost d functions are deleted, the Yb-S bonds lengthen
˚
˚
by ∼0.01 A and the Yb-O bonds shorten by ∼0.03 A,
resulting in small changes in the asymmetry parameters to
Yb-S =-0.045 A and ΔYb-O =0.061 A (Table S-3 in the
˚
˚
Δ
(py)2(THF)Sm(SR)3 2.750(3)
Supporting Information). However, when all d-type functions
are removed from the Yb basis set, all Ln-E bonds lengthen
further as expected, but, importantly, the asymmetry in the
(HMPA)3Sm(SPh)3
(HMPA)3Yb(SPh)3
(THF)3Yb(OC6F5)3
(THF)3Yb(SC6F5)3
(THF)3Er(SeC6F5)3
2.826(2)
2.734(1)
2.111(2)
2.679(2)
2.836(2)
15
this work
this work
˚
Ln-E bonds essentially vanishes, ΔYb-S = -0.008 A and
˚
Δ
Yb-O = 0.019 A ([5s4p]; Table 4). Furthermore, it is the
a Δx-y* = (average Ln-X trans to X) - (average Ln-X trans to Y),
where X is the anionic ligand and Y is the neutral donor ligand. If there is
more than one crystallographically independent molecule/cell, the aver-
age distances for all of the molecules in the cell are used.
Yb-S bond situated trans to the THF molecule (Yb1-S2)
that undergoes the largest change in the bond length and hence
is most sensitive to the presence of d functions in the Yb basis
set. Reduction of the 4p basis set to minimal size (to essen-
tially just cover the 5p6 electrons), while keeping all of the
d functions, induces only a small change in the asym-
covalency and the trans influence are expressed in com-
pounds 1 and 2, substantiating the NEC results presented
above.
metry parameters (relative to their original values), ΔYb-S
=
˚
˚
-0.044 A and ΔYb-O=0.059 A ([5s1p3d]; Table 4); hence, the
more diffuse p functions do not appear to play a significant
role in promoting the trans influence. As expected, removal of
the two most diffuse s functions from the Yb basis set does not
change the bond-length asymmetry parameters either (Table S-3
in the Supporting Information). However, if the s function that
Discussion
Both 1 and 2 have directionally dependent Ln-ligand
bond lengths, with Ln-ligand bonds trans to anions that
are consistently longer than the same Ln-ligand bonds that
are trans to neutral donor ligands. While this concept of a
trans influence is well established in transition-metal and
main-group chemistry, until now there has never, to our
knowledge, been an experimental-theoretical investigation
into the possibility of Ln molecules showing the same
effect.
These bond-length distributions have been measured fre-
quently, but noted rarely, for a wide range of mer-octahedral
LnX3Y3 (X = anion and Y = neutral donor) coordination
complexes. Table 5 contains a list of Ln-X bond lengths for
the known mer-LnX3Y3 compounds that have been char-
acterized by X-ray diffraction. In every case, anions trans to
anions are further from Ln than are anions trans to neutral
donors. The same effect is also noted in cubane cluster
derivatives and a group of octahedral LnX4Y2 coordination
compounds. Table 6 shows the range of Ln compounds with
octahedral coordination environments and their associated
trans influences, ranging from charge-neutral cesium(IV)
alkoxides to anionic (DME)Yb(SePh)4-. Again, in every
case, there is a directional component to the bonding, with
bonds trans to anions significantly longer than bonds trans
to neutral donor ligands. The trans influence appears to
be a general, fundamental property of Ln coordination
compounds.
covers mostly 6s orbital space is also deleted, the asymmetry
˚
parameters actually increase (ΔYb-S=-0.052 A and ΔYb-O
=
˚
0.088 A); population of nondirectional metal s orbitals tends
to diminish the asymmetry. When the basis set on Er is
modified in 2, the metal-ligand bond lengths change in similar
patterns (Table 4). Whereas the initial asymmetry parameters
˚
˚
were ΔEr-Se=-0.041 A and ΔEr-O=0.074 A, deletion of all
d-type basis functions on Er leads to a dramatic reduction to
Er-Se =-0.004 A and ΔEr-O = 0.017 A, but bringing the
p-type basis set down to minimal size produces little change
(ΔEr-Se=-0.036 A and ΔEr-O=0.057 A). Furthermore, the
addition of up to three sets of f-type polarization functions to
the Yb basis set44a ([5s4p3d3f]) leads to a shrinkage of only
˚
˚
Δ
˚
˚
˚
∼0.01 A in all Ln-E bonds and virtually no change in the
˚
asymmetry parameters, ΔYb-S = -0.045 A and ΔYb-O
=
˚
0.077 A. On the other hand, substituting the f-type functions
for the d-type functions ([5s4p3f] basis set) again removes most
˚
˚
of the asymmetry (ΔYb-S=-0.012 A and ΔYb-O=0.024 A).
Thus, the asymmetry in the metal-ligand bond lengths is
brought about primarily by the presence of d-type basis
functions in the Ln basis set. The magnitude of the asym-
metry depends largely on the spatial extent of the innermost
d-type function in the basis set, the one that corresponds most
closely to what we commonly would refer to as the 5d orbital;
deletion of just this single basis function from the Yb basis
Both electropositive (i.e., SePh) and electronegative (i.e.,
OC6F5) ligands display a trans influence. The directiona-
lity was first noted in the description of (py)3Yb(SPh)3,12
˚
˚
set leads to ΔYb-S = -0.021 A and ΔYb-O = 0.035 A
([5s4p2d]; Table 4). The calculations with modified Ln basis
sets establish that it is primarily through Ln 5d orbitals that
(55) (a) Deacon, G. B.; Feng, T.; Junk, P. C.; Skelton, B. W.; Sobolev,
A. N.; White, A. H. Aust. J. Chem. 1998, 51, 75. (b) Qian, C.; Wang, B.; Deng,
D.; Xu, C.; Sun, X.; Ling, R. Jiegou Huaxue 1993, 12, 18.
(56) Deacon, G. B.; Feng, T.; Junk, P. C.; Meyer, G.; Scott, N. M.;
Skelton, B. W.; White, A. H. Aust. J. Chem. 2000, 53, 853.
(57) Emge, T. J.; Kornienko, A.; Brennan, J. G. Acta Crystallogr. 2009,
C65, m422.
(53) Mashima, K.; Nakayama, Y.; Fukumoto, H.; Kanehisa, N.; Kai, Y.;
Nakamura, A. J. Chem. Soc., Chem. Commun. 1994, 2523.
(54) (a) Shotwell, J. B.; Sealy, J. M.; Flowers, R. A. J. Org. Chem. 1999,
64, 5251. (b) Kramer, G. M.; Mass, E. T.; Dines, M. B. Inorg. Chem. 1980, 20,
1415. (c) Kramer, G. M.; Mass, E. T.; Dines, M. B. Inorg. Chem. 1980, 20, 1418.

Article
Inorganic Chemistry, Vol. 49, No. 2, 2010 559
˚
Table 6. Summary of the Ln-Ligand Bond Lengths (A) in Octahedral
Third, there is no fixed period for which trans influences are
maximum. For the set of (THF)3Ln(EC6F5)3 (E=O, S, Se),
the effect is at a maximum value for S, with S > Se > O, and in
the halide derivatives (THF)3LnX3 (X=Cl,55 Br,56 I57), there
is a maximum value for Br, with Br > I > Cl . This is not
surprising, given that any trans influence reflects both the
ability of an ion to donate electron density and the ease with
which a given ion-metal bond can be distorted. While it is
tempting to note that most, but not all, of the largest trans
influences are found in compounds that contain more covalent/
less electronegative ligands, these are also the least stable
complexes with the longest M-L bond lengths under con-
sideration, and as such, these contain the bonds that should
be distorted most easily.
Compounds
Ln-X
trans to X
Ln-X
trans to Y
a
*
Δx-y
ref
[(BIPY)Yb(SBu)3]2
Ce(OR)4(4-NMe2py)2
Ce(hfip)4(TMEDA)
2.627(3)
2.127(3)
2.152(6)
2.791(2)
3.307(1)
2.803(2)
2.669(6)
2.785(2)
2.620(3)
2.112(5)
2.115(5)
2.781(2)
3.280(1)
2.770(2)
2.646(6)
2.742(2)
0.007
0.015
0.037
0.010
0.027
0.033
0.023
0.037
68
69
70
71
-
(DME)Yb(SePh)4
Sm2I4(N-MeIm)4
(py)8Yb4Se4(SePh)4
(py)10Yb6S6(SPh)6
(THF)10Yb6Se6I6
72
13b
13c
14d
a Δx-y* = (average Ln-X trans to X) - (average Ln-X trans to Y),
where X is the anionic ligand and Y is the neutral donor ligand. If there is
more than one crystallographically independent molecule/cell, the aver-
age distances for all of the molecules in the cell are used. b Yb-μ3Se2-
bonds trans to SePh or py. c Yb-μ3S2- bonds trans to SPh or py.
d Yb-μ3Se2- bonds trans to I or THF.
Bonds between Ln and the neutral donors also tend to
follow the same bond-length distribution patterns, with
bonds trans to anions longer than bonds trans to neutral
donors. In 1, the Yb-O bonds trans to O are 2.271(4) and
˚
where it was pointed out that the related compound
2.280(4) A, and the Yb-O bond trans to S is significantly
i
53
˚
(py)3Sm(SC6H3 Pr3)3 had a similar range of geometry-
dependent Ln-S bond lengths. Subsequent studies on the
NIR emission behavior of Ln(OC6F5)3 revealed that (THF)3-
longer, at 2.308(5) A. A similar pattern is noted for the four
independent molecules in 2, but the differences change
significantly within the independent molecules. Because these
dative interactions are significantly weaker than bonds be-
tween charged species, they should be particularly susceptible
to influences such as crystal-packing effects. An example of
this behavior was noted in the structures of [(DME)3-
Ln(SC6F5)2]þ (Ln=La-Gd), where Ln-anion bond lengths
varied consistently with the Ln ionic radius, while the dative
Ln-F separations initially followed the expected trend but
then increased with decreasing Ln as ligand-ligand repul-
sions increased.58 For this reason, we have chosen not to
focus attention on variations in the bond lengths between Ln
and neutral donor ligands, even though the measured dis-
tances in 1 and 2 are consistent with theoretical predictions
(cf. Table 3).
15
Yb(OC6F5)3 displayed equally distinctive bond-length
patterns. This phenoxide compound was actually the inspira-
tion for the present work, because while the initial cursory
examination of the bond lengths revealed the expected
pattern, a closer examination of the structure suggested that
interligand π-π interactions were also capable of distorting
Ln-X bond lengths to a similar degree.
Compounds 1 and 2 were subsequently targeted in an
attempt to evaluate the various steric and electronic factors
that might be important in influencing the bond lengths, but
given the broad range of differences in the Er-Se and Er-O
bond lengths for the four crystallographically independent
molecules in 2, it is clearly impossible to confidently separate
contributions from lattice/steric/electronic influences or to
state unequivocally that one ligand has a stronger covalent
trans influence than does another. This is also similar to
findings in main-group and transition-metal systems,11 where
it has been noted that intermolecular interactions are often
greater than, or equal to, the effect that a trans influence
might have on a bond length.
There are, however, a number of trends in the available
data. First, it appears that the basicity of the neutral donor
ligand has a measurable impact on the bond length trans to it.
There are two molecular pairs in Table 5 with different
neutral donors: (L)3Sm(SPh)3 (L = HMPA, py/THF) and
L3Yb(SPh)3 (L = HMPA, pyridine). In both cases, the
compound with the neutral donor that is the strongest donor
ligand54 also has the greater trans influence.
Fluorination of the arene group appears to increase the
length of the Ln-Se bond and decrease the length of the
Er-O(THF) bond, which is consistent with an electrostatic
component to the bonding, because the polarizing influence
of the electronegative F atoms lessen the charge density at the
Se atom. The average Er-Se bond length in the fac-octa-
hedral selenolate (THF)3Er(SePh)3 is significantly shorter, at
˚
2.7766(6) A, than either the average (for the four independent
molecules) of the crystallographically unique Er-Se bond
˚
trans to THF in 2 [2.8042(7) A] or the average of all of
˚
the Er-Se bonds in 2 (2.8256 A). Given trans influences, the
former comparison seems more appropriate. In contrast, the
˚
average Er-O(THF) distances for 2 (2.304 A) are signifi-
cantly shorter than the three crystallographically equivalent
Er-O(THF) bonds in (THF)3Er(SePh)3 [2.347(3) A].59 This
˚
Second, the differences in the Ln-X bond lengths are
significantly greater when steric interactions are minimal, i.e.,
in the first half of the Ln series. Of the five chloride structures
listed in Table 5, the two with the greatest trans influence
are Pr and Nd compounds. This can be rationalized by
noting that for six-coordinate structures the larger Ln com-
pounds are impacted less by steric repulsions that might
tend to overshadow any trans influence, but it does run
contrary to the assumption that any sort of metal-ligand
orbital overlap will increase as the d orbitals contract across
the Ln series. Alternatively, it could be argued that the early
Ln-X bonds are weaker and thus more susceptible to a trans
influence.
is again consistent with an electrostatic bonding model, in
which fluorination of the selenolate withdraws the electron
density from the ErIII ion, which then compensates by
bonding more strongly with the THF ligands. A similar
effect was noted in comparisons between Ln-SC6F5 and
Ln-SC6H5 compounds.60
While a trans influence is noted here, Ln chemistry is still
best described in ionic terms, and this is evident from the data
(58) Banerjee, S.; Emge, T. J.; Brennan, J. G. Inorg. Chem. 2004, 43, 6307.
(59) Lee, J.; Freedman, D.; Melman, J. H.; Brewer, M.; Sun, L.; Emge,
T. J.; Long, F. H.; Brennan, J. G. Inorg. Chem. 1998, 37, 2512.
(60) Melman, J.; Rhode, C.; Emge, T. J.; Brennan, J. G. Inorg. Chem.
2002, 41, 28.

560 Inorganic Chemistry, Vol. 49, No. 2, 2010
Krogh-Jespersen et al.
in Tables 5 and 6. In particular, the uniformity of the trans
influence can be interpreted as an artifact of predominantly
ionic behavior. In every Ln compound, the negatively
charged ligand lengthens the bond trans to it and thus has
the stronger trans influence, whereas in transition-metal
systems, trans influences occur but relative bond lengthening
can be found trans to either anions or neutral donors.
Compare, for example, GaCl3(terpy),61 where the Ga-Cl
Conclusion
Covalent bonding can have a measurable impact on the
structures of Ln coordination complexes. DFT calculations
indicate that the inequivalent bond lengths found in octa-
hedral Ln coordination complexes result from a covalent
bonding interaction between ligand-based p orbitals and the
Ln 5d orbitals. Arguably, the present work is the first
combined experimental-theoretical investigation demon-
strating that covalent bonding is responsible for the direc-
tionality noted in Ln-X bond lengths.
˚
bond trans to N is 0.13 A shorter than the average of the
Ga-Cl bonds trans to Cl, with IrCl3(PMe2Ph)3,62 where Cl
˚
trans to P is 0.084 A longer than the Ir-Cl bonds trans to Cl.
It would be interesting to determine the structure of a
LnX3Y3 compound with a phosphine donor ligand.
The magnitudes of the trans influences also reflect the
essentially ionic character of the Ln-X bond. The range of
bond-length differences noted in these Ln compounds is
significantly smaller than the range of trans influence found
in main-group or transition-metal structures,11 even though
the Ln-X bonds are considerably longer. This would be
consistent with a greater covalent character for the latter
metals relative to the Ln compounds.
Acknowledgment. We acknowledge support of the NSF
(Grant CHE-0747165). We thank anonymous reviewers
for constructive comments.
Supporting Information Available: X-ray crystallographic
files in CIF format for the crystal structures of 1 and 2; complete
ref 47, optimized geometries of 1 and 2, Tables S-1-S-3, and a
summary of the crystallographic details. This material is avai-
lable free of charge via the Internet at http://pubs.acs.org.
(67) Mashima, K.; Nakayama, Y.; Shibahara, T.; Fukumoto, H.;
Nakamura, A. Inorg. Chem. 1996, 35, 93.
(68) Aspinall, H. C.; Cunningham, S. A.; Maestro, P.; Macaudiere, P.
Inorg. Chem. 1998, 37, 5396.
(69) Suh, S.; Guan, J.; Miinea, L. A.; Lehn, J. S.; Hoffman, D. M. Chem.
Mater. 2004, 16, 1667.
(70) Evans, W. J.; Rabe, G. W.; Ziller, J. W. Inorg. Chem. 1994, 33, 3072.
(71) Geissinger, M.; Magull, J. Z. Anorg. Allg. Chem. 1995, 621, 2043.
(72) Daniele, S.; Hubert-Pfalzgraf, L. G.; Perrin, M. Polyhedron 2002, 21,
1985.
(61) Beran, G.; Carty, A. J.; Patel, H. A.; Palenik, G. J. Chem. Commun.
1970, 222.
(62) Aslanov, A.; Mason, R.; Wheeler, A. G.; Whimp, P. O. Chem.
Commun. 1970, 30.
(63) Spichal, Z.; Necas, M.; Pinkas, J. Inorg. Chem. 2005, 44, 2074.
(64) Radonovich, L. J.; Glick, M. D. J. Inorg. Nucl. Chem. 1973, 35, 2745.
(65) Xing-Wang, Z.; Xing-Wang, L.; Benetollo, F.; Bombieri, G. Inorg.
Chim. Acta 1987, 139, 103.
(66) Hou, Z.; Kobayashi, K.; Yamakazi, H. Chem. Lett. 1991, 265.