Molecular architectures for trimetallic d/f/d complexes: Structural and magnetic properties of a LnNi2 core

Article abstract of DOI:10.1021/ic701612e

A series of cationic, trimetallic d/f/d complexes have been prepared which use a multidentate, macrocyclic amine phenol ligand to coordinate divalent first row d-block transition metal ions (TM) and lanthanides (Ln) ions in close proximity, desirable for magnetic studies. Isolable complexes of the d/f/d cluster compounds with the formula [Ln(TM)₂(bcn)₂]ClO ₄·nH₂O, where H₃bcn is tris-N,N′,N″-(2-hydroxybenzyl)-1,4,7-triazacyclononane, TM = Zn(II) and Ni(II) and Ln = La(III), Nd(III), Gd(III), Dy(III), and Yb(III), were synthesized by a one-pot sequential reaction of stoichiometric amounts of H ₃bcn with the TM(II) and Ln(III) metal ions. The spontaneously formed cationic complexes were characterized by a variety of analytical techniques including IR, NMR, +ESI-MS, and EA. The [TM(HbCn)]·nH₂O and [TM₃(bcn)₂]·nH₂O complexes were also synthesized to probe the building blocks of the d/f/d coaggregated species. The solid-state X-ray crystal structures of [GdNi₂(bcn) ₂(CH₃CN)₂]ClO₄·CH ₃CN and [GdZn₂(bcn)₂(CH₃CN) ₂]ClO₄·CH₃CN were determined to be nearly identical with each TM(II) encapsulated in an octahedral geometry by the N₃O₃ binding pocket of the bcn^3- ligand. The eight coordinate Gd(III) was bicapped by two [TM(bcn)]^- moieties and coordinated by two solvent molecules. Because of the isostructurality of the [LnZn₂(bcn)₂]ClO₄·nH₂O and [LnNi₂(bcn)₂]ClO₄·nH₂O complexes, an empirical approach using the LnZn₂ magnetic data was utilized to remove first-order anisotropic contributions from the LnNi ₂ species. Ferromagnetic spin interactions were determined for the [LnNi₂(bcn)₂]ClO₄·nH₂O complexes, where Ln = Gd(III), Dy(III), and Yb(III), while an antiferromagnetic exchange was observed for Ln = Nd(III).

Full text of DOI:10.1021/ic701612e

Inorg. Chem. 2008, 47, 2280-2293
Molecular Architectures for Trimetallic d/f/d Complexes: Structural and
Magnetic Properties of a LnNi2 Core
Cheri A. Barta, Simon R. Bayly, Paul W. Read, Brian O. Patrick, Robert C. Thompson,*
and Chris Orvig*
Department of Chemistry, UniVersity of British Columbia, 2036 Main Mall,
VancouVer, British Columbia V6T 1Z1, Canada
Received August 13, 2007
A series of cationic, trimetallic d/f/d complexes have been prepared which use a multidentate, macrocyclic amine
phenol ligand to coordinate divalent first row d-block transition metal ions (TM) and lanthanides (Ln) ions in close
proximity, desirable for magnetic studies. Isolable complexes of the d/f/d cluster compounds with the formula
[Ln(TM)2(bcn)2]ClO4 · nH2O, where H3bcn is tris-N,N′,N′′-(2-hydroxybenzyl)-1,4,7-triazacyclononane, TM ) Zn(II)
and Ni(II) and Ln ) La(III), Nd(III), Gd(III), Dy(III), and Yb(III), were synthesized by a one-pot sequential reaction
of stoichiometric amounts of H3bcn with the TM(II) and Ln(III) metal ions. The spontaneously formed cationic
complexes were characterized by a variety of analytical techniques including IR, NMR, +ESI-MS, and EA. The
[TM(Hbcn)] ·nH2O and [TM3(bcn)2]·nH2O complexes were also synthesized to probe the building blocks of the
d/f/d coaggregated species. The solid-state X-ray crystal structures of [GdNi2(bcn)2(CH3CN)2]ClO4 ·CH3CN and
[GdZn2(bcn)2(CH3CN)2]ClO4 ·CH3CN were determined to be nearly identical with each TM(II) encapsulated in an
octahedral geometry by the N3O3 binding pocket of the bcn3- ligand. The eight coordinate Gd(III) was bicapped by
two [TM(bcn)]- moieties and coordinated by two solvent molecules. Because of the isostructurality of the
[LnZn2(bcn)2]ClO4 · nH2O and [LnNi2(bcn)2]ClO4 · nH2O complexes, an empirical approach using the LnZn2 magnetic
data was utilized to remove first-order anisotropic contributions from the LnNi2 species. Ferromagnetic spin interactions
were determined for the [LnNi2(bcn)2]ClO4 · nH2O complexes, where Ln ) Gd(III), Dy(III), and Yb(III), while an
antiferromagnetic exchange was observed for Ln ) Nd(III).
Introduction
Nd(III) core;9–11 however, the magnetic exchange between
transition metal (TM) and lanthanide (Ln) ions is not well
understood. Many obstacles, including the weak d/f interac-
tions, the large anisotropic effects of Ln (except for La(III),
Gd(III), and Lu(III)), and the challenge of designing a ligand
that coordinates both d- and f-block transition metal ions in
close proximity have contributed to the complexity of
understanding this important magnetic exchange.
The existing magnetic studies on heterometallic d-f
complexes have been mainly confined to Schiff base Cu-Gd
systems12–22 first studied by Gatteschi in 1985.14 Only a
handful of examples of Gd-Cu species in a range of
Currently, lanthanides are being explored in a myriad of
areas including in medicine as MRI contrast agents,1,2
radiotherapeutic drugs,3,4 and fluorescent probes,5 in catalysis
for homogeneous carbon-carbon bond formation,6 and in
materials for fluorescent thin films and light-emitting di-
odes.7,8 Lanthanides are also part of the most powerful
permanent magnet known to date consisting of an Fe(II)/
* To whom correspondence should be addressed. E-mail: thompson@
chem.ubc.ca (R.C.T.); orvig@chem.ubc.ca (C.O.).
(1) Lauffer, R. B. Chem. ReV. 1987, 87, 901.
(2) Caravan, P.; Ellison, J. J.; McMurry, T. J.; Lauffer, R. B. Chem. ReV.
1999, 99, 2293.
(3) Cutler, C. S.; Smith, C. J.; Ehrhardt, G. J.; Tyler, T. T.; Jurisson, S. S.;
Deutsch, E. Cancer Biother. Radiopharm. 2000, 15, 531.
(4) Fricker, S. P. Chem. Soc. ReV. 2006, 35, 524.
(5) Thunus, L.; Lejeune, R. Coord. Chem. ReV. 1999, 184, 125.
(6) Mikami, K.; Terada, M.; Matsuzawa, H. Angew. Chem., Int. Ed. 2002,
41, 3554.
(9) Rauluszkiewicz, J.; Szymczak, H.; Lachowicz, H. K. Physics of
Magnetic Material; World Scientific: Singapore, 1985.
(10) Coey, J. M. D. EndeaVour 1995, 19, 146.
(11) Muller, K. H.; Krabbes, G.; Fink, J.; Gruss, S.; Kirchner, A.; Fuchs,
G.; Schultz, L. J. J. Magn. Magn. Mater. 2001, 226–230, 1370.
(12) Costes, J.-P.; Dahan, F.; Dupuis, A. Inorg. Chem. 2000, 39, 5994.
(13) Ramade, I.; Kahn, O.; Jeannin, Y.; Robert, F. Inorg. Chem. 1997, 36,
930.
(7) Bünzli, J.-C. G.; Piguet, C. Chem. Soc. ReV. 2005, 34, 1048.
(8) Kido, J.; Okamoto, Y. Chem. ReV. 2002, 102, 2357.
2280 Inorganic Chemistry, Vol. 47, No. 7, 2008
10.1021/ic701612e CCC: $40.75  2008 American Chemical Society
Published on Web 03/05/2008

Trimetallic d/f/d Complexes
coordination environments20,23–28 or other Cu-Ln complexes
where Ln * Gd, are known.21,29–35 Systematic studies on
the structure and magnetic properties of TM-Ln species,
including Ni-Ln complexes, are even rarer.36–43
WehavepreviouslysynthesizedavarietyofNi(II)-Ln(III)-Ni(II)
complexes using tripodal amine phenol ligands, H3tam [1,1,1-
tris(((2′-hydroxybenzyl)amino)methyl)ethane]36 and H3trn,
[tris(2′-hydroxybenzylaminoethyl)amine]41 to probe the mag-
netic interactions between d and f metal ions (Figure 1).
Despite the similarities in the ligand structure, very different
magnetic properties were observed between the two series
of Ni-Ln-Ni species.36,41 The [LnNi2(trn)2]+ complexes
exhibited a ferromagnetic exchange for Ln ) Gd(III), Tb(III),
Dy(III), Ho(III), and Er(III) (and possibly Tm(III) and
Yb(III)), following the trend of other documented Ni-Ln
species (a Ni3Ln2 ladder complex44 and one Ni2Ln clus-
Figure 1. Ligands of interest and proposed binding motifs for the d/f/d
(14) Bencini, A.; Benelli, C.; Caneschi, A.; Carlin, R. L.; Dei, A.; Gatteschi,
D. J. J. Am. Chem. Soc. 1985, 107, 8128.
heterometallic complexes.
(15) Costes, J.-P.; Novitchi, G.; Shova, S.; Dahan, F.; Donnadieu, B.;
Tuchagues, J.-P. Inorg. Chem. 2004, 43, 7792.
(16) Costes, J.-P.; Dahan, F.; Dupuis, A.; Laurent, J.-P. Inorg. Chem. 2000,
39, 169.
ter).36,41,45 These results suggest that Kahn’s prediction
(stating that ferromagnetic couplings are observed for Ln(III)
ions with f7-f13 configurations in Cu(II)-Ln(III) sys-
tems)44,46 may also be relevant for other TM(II)-Ln(III)
species. Alternatively, however, antiferromagnetic interac-
tions were observed for the [LnNi2(tam)2]+ species where
Ln ) Dy(III) and Yb(III).36 To probe these contrasting
results, it was deemed necessary to design new ligands that
cannot only afford discrete d-f mixed-metal complexes to
better understand the effects of the geometric constraints on
the observable magnetic exchange but also to synthesize
systems where the TM can be varied to produce a series of
diamagnetic and paramagnetic complexes in which the first-
orderorbitalcontributionsoftheLn(III)canberemoved.21,29–35,47
Costes et al. were the first to synthesize a Ni-Ln system
isostructural with a Cu-Ln species of interest,31 where the
Ni(II) was held in a square planar environment giving rise
to a low-spin, diamagnetic state. When identical crystal-field
effects are assumed for the two complexes, the magnetic
behavior between the d-f metals for the Cu-Ln species
could be unmasked by removal of the first-order orbital
contributions of the Ln(III) determined from the Ni-Ln
magnetic data. A similar method was used by Matsumoto,35
(17) Costes, J.-P.; Dahan, F.; Dupuis, A.; Laurent, J.-P. Inorg. Chem. 1996,
35, 2400.
(18) Andruh, M.; Ramade, I.; Codjovi, E.; Guillou, O.; Kahn, O.; Trombe,
J. C. J. Am. Chem. Soc. 1993, 115, 1822.
(19) Akine, S.; Taniguchi, T.; Nabeshima, T. Angew. Chem., Int. Ed. 2002,
41, 4670.
(20) Ghoerghe, R.; Andruh, M.; Costes, J.-P.; Donnadieu, B. Chem.
Commun. 2003, 2778.
(21) Guillou, O.; Oushoorn, R. L.; Kahn, O.; Baubekeur, K.; Batail, P.
Angew. Chem., Int. Ed. Eng. 1992, 31, 626.
(22) Costes, J.-P.; Dahan, F.; Dupuis, A.; Laurent, J.-P. Inorg. Chem. 1997,
36, 3429.
(23) Gao, S.; Yang, Y.; Xu, G. J. Alloys Compd. 1995, 225, 234.
(24) Winpenny, R. E. P. Chem. Soc. ReV. 1998, 27, 447.
j
(25) Sakamoto, M.; Manseki, K.; Okawa; H., Coord. Chem. ReV. 2001,
219–221, 379.
(26) Benelli, C.; Gatteschi, D. Chem. ReV. 2002, 102, 2369.
j
(27) Shiga, T.; Ohba, M.; Okawa, H. Inorg. Chem. Commun. 2003, 6, 15.
(28) Benelli, G.; Caneschi, A.; Gatteschi, D.; Guillou, O.; Pardi, L. Inorg.
Chem. 1990, 29, 1750.
(29) Stemmler, A. J.; Kampt, J. W.; Kirk, M. L.; Atasi, B. H.; Pecoraro,
V. L. Inorg. Chem. 1999, 38, 2807.
(30) He, F.; Tong, M.-L.; Chen, X.-M. Inorg. Chem. 2005, 44, 559.
(31) Costes, J.-P.; Dahan, F.; Dupuis, A.; Laurent, J.-P. Chem.sEur. J.
1998, 4, 1616.
(32) Kahn, M. L.; Mathoniere, C.; Kahn, O. Inorg. Chem. 1999, 38, 3692.
(33) Costes, J.-P.; Auchel, M.; Dahan, F.; Peyrou, V.; Shova, S.; Werns-
dorfer, W. Inorg. Chem. 2006, 45, 924.
(34) Osa, S.; Kido, T.; Matsumoto, N.; Re, N.; Pochaba, A.; Mrozinski, J.
J. Am. Chem. Soc. 2003, 126, 420.
(35) Kido, T.; Ikuta, Y.; Suantsuki, Y.; Ogawa, Y.; Matsumoto, N. Inorg.
Chem. 2003, 42, 398.
45,48
Kahn,32,34 Okawa
and co-workers to reduce the misin-
j
(36) Xu, Z.; Read, P. W.; Hibbs, D. E.; Hursthouse, M. B.; Malik, K. M. A.;
Patrick, B. O.; Rettig, S. J.; Seid, M.; Summers, D. A.; Pink, M.;
Thompson, R. C.; Orvig, C. Inorg. Chem. 2000, 39, 508.
(37) Yamaguchi, T.; Sunatsuki, Y.; Kojima, M.; Akashi, H.; Tsuchimoto,
M.; Re, N.; Osa, S.; Matsumoto, N. Chem. Commun. 2004, 1048.
(38) Yue, Q.; Yang, J.; Li, G.-H.; Li, G.-D.; Xu, W.; Chen, J.-S.; Wang,
S.-N. Inorg. Chem. 2005, 44, 5241.
terpretation of the magnetic exchange between the d- and
f-block metals.
Using a similar concept, we have synthesized a series of
heterometallic d/f/d cluster compounds using a slightly more
geometrically demanding trisubstituted macrocyclic amine
phenol ligand, H3bcn (tris-N,N′,N′′-(2-hydroxybenzyl)-1,4,7-
(39) Lisowski, J.; Starynowicz, P. Inorg. Chem. 1999, 38, 1351.
(40) Costes, J.-P.; Dahan, F.; Dupuis, A.; Laurent, J.-P. Inorg. Chem. 1997,
36, 4284.
(41) Bayly, S. R.; Xu, Z.; Patrick, B. O.; Rettig, S. J.; Pink, M.; Thompson,
R. C.; Orvig, C. Inorg. Chem. 2003, 42, 1576.
j
(42) Chen, Q.-Y.; Luo, Q.-H.; Zheng, L.-M.; Wang, Z.-L.; Chen, J.-T. Inorg.
Chem. 2002, 41, 605.
(45) Shiga, T.; Ito, N.; Hidaka, A.; Okawa, H.; Kitagawa, S.; Ohba, M.
Inorg. Chem. 2007, 46, 3492.
(43) Pointillart, F.; Bernot, F.; Sessoli, R.; Gatteschi, D. Chem.sEur. J.
2007, 13, 1602.
(46) Kahn, O.; Guillou, O. Research Frontiers in Magnetism; O’Connor,
C. J., Ed.; World Scientific: London, 1993.
(44) Kahn, M. L.; Lecante, P.; Verelst, M.; Mathoniere, C.; Kahn, O. Chem.
Mater. 2000, 12, 3073.
(47) Li, Y.-T.; Yan, C.-W.; Zhang, J. Cryst. Growth Des. 2002, 3, 481.
j
(48) Shiga, T.; Ohba, M.; Okawa, H. Inorg. Chem. 2004, 43, 4435.
Inorganic Chemistry, Vol. 47, No. 7, 2008 2281

Barta et al.
Scheme 1. Synthesis of H3bcna
a
Ts ) p-CH3C6H4SO2. Reaction conditions: (a) NaOH, H2O/ether, RT, 70%; (b) pyridine, O°C, Ar, 50%; (c) anhydrous DMF, 60% NaH, 70°C, Ar,
94%; (d) conc. H2SO4, 160°C, Ar; (e) H2O/NaOH, CH3Cl, RT, 72%; (f) dry EtOH, salicylaldehyde, NaBH(OAc)3, RT, Ar, 82%.
triazacyclononane)49 to probe the effects of the geometric
were calibrated with listed deuterated solvents. Infrared spectra were
recorded as nujol mulls (KBr windows) in the range 4000–500 cm-1
constraints on the magnetic properties of d/f/d complexes
on a Mattson Galaxy Series FTIR-5000 spectrophotometer and were
(Figure 1). H3bcn has three moderately hard amine donors
referenced to polystyrene (1601 cm-1). Elemental analyses of C,
that can preferentially accommodate divalent d-block transi-
H, and N were performed by Mr. M. Lakha at the University of
tion metals and three very hard phenolate donors used to
British Columbia (UBC). UV–vis spectra were recorded in methanol
chelate harder trivalent f-block metals.
using a Hewlett-Packard 8543 diode array spectrophotometer and
The synthesis and characterization of the geometrically
a 1 cm cuvette. Mass spectra were obtained on a Bruker Esquire
more demanding H3bcn ligand and its subsequent
ion trap (electrospray ionization mass spectrometry, ESI-MS)
[TM(Hbcn)] ·nH2O, [TM3(bcn)2]·nH2O, and [LnTM2(bcn)2]-
ClO4 · nH2O complexes are discussed herein, where TM )
Zn(II) and Ni(II) and Ln ) La(III), Nd(III), Gd(III), Dy(III),
and Yb(III). An empirical determination of the magnetic
interactions between the different metal species can be
elucidated by using the analogous diamagnetic La(III) or
Zn(II) species, similar to the analyses by Costes,31 Kahn,32,44
spectrophotometer. Approximately 1 mg of sample was dissolved
in 1 mL of methanol or acetonitrile for +ESI-MS analysis. Room-
temperature magnetic susceptibilities on powdered samples of
H3[Ni(bcn)]2ClO4, [Ni(Hbcn)] ·H2O, and [Ni3(bcn)2]·2H2O com-
plexes were measured on a Johnson Matthey MSB-1 balance
standardized with CoCl2 ·6H2O and CuSO4 ·5H2O. Variable-tem-
perature and applied-field magnetic susceptibility measurements for
all other powdered samples were performed on a Quantum Design
MPMS XL SQUID magnetometer over the temperature range of
2-300 K at 10 000 and 500 G at UBC. Diamagnetic corrections
for the constituent atoms used Pascal’s constants51 (6.10 × 10-4
cm3 mol-1 per mole of complex, plus 3.89 × 10-5 cm3 mol-1 per
water). A specially designed sample holder, as described previ-
ously,52 was used to minimize the background signal. Magnetic
susceptibilities were corrected for the background signal generated
by the holder.
Okawa,45,48 Matsumoto35 and co-workers. The examination
j
of the magnetic properties of these species allows for a
comprehensive study on the effects of the varied geometric,
spin, and orbital contributions on the overall magnetic
behavior between d and f metals using similar amine phenol
ligands. In addition, the procedures used herein can be
translated to the magnetic analysis of other d/f/d systems
including the [LnCu2(bcn)2]ClO4 · nH2O species fully char-
acterized in the following paper.50
Ligand Synthesis. 1,4,7-Triazacyclononane was synthesized
using a significant modification of a published procedure.53
Diethylene-1,4,7-triamine Tritosylate. Diethylenetriamine (8.26
g, 0.08 mol) and NaOH (9.6 g, 0.24 mol) were dissolved in 80 mL
of water. The solution was added dropwise to p-toluenesulfonyl
chloride (TsCl, 45.6 g, 0.24 mol) dissolved in 240 mL diethyl ether.
After the mixture was continuously stirred for 2 h at RT, the off-
white product was filtered and washed with water and diethyl ether.
The resulting white solid was vaccuum-dried without further
purification to yield 30 g, 70%. 1H NMR (300 MHz, RT,
Experimental Section
Materials. Reagents were purchased from commercial sources
and were used without purification unless otherwise stated. Hydrated
Ln(III) perchlorate salts were obtained from Alfa Aesar as 50%
(w/w) solutions in water and evaporated to dryness. Triethylamine
was purchased from Fisher Scientific and used as received. NMR
solvents were obtained from Sigma-Aldrich and were used as
received. Methanol, ethanol, and acetonitrile were dried over
activated 4 Å molecular sieves prior to use. Water was deionized
(Barnstead D8902 and D8904 Cartridges) and distilled (Hytrex II
GX50-9-7/8 and GX100-9-7/8) before use. Yields for the analyti-
cally pure compounds were calculated on the basis of the respective
metal ion starting material.
3
(CD3)2CO): δ 7.75 (d, 4H, J ) 8.2 Hz, aromatic CH), 7.64 (d,
2H, aromatic CH, 3J ) 8.3 Hz), 7.41 (d, 4H, aromatic CH,3J ) 9.2
Hz), 7.39 (d, 2H, aromatic CH, 3J ) 8.01 Hz), 6.56 (t, 2H, NH, 3J
(49) Auerbach, U.; Eckert, U.; Wieghardt, K.; Nuber, B.; Weiss, J. Inorg.
Chem. 1990, 29, 938.
Caution!. Perchlorate salts are potentially explosiVe and should
be handled with extreme care and used only in small quantities.
(50) Barta, C. A.; Bayly, S. R.; Read, P. W.; Patrick, B. O.; Thompson,
R. C.; Orvig, C. Inorg. Chem. 2008, 47, 2294–2302.
(51) Pascal, P. Chimie Generale; Masson et Cie: Paris, 1949.
(52) Ehlert, M. K.; Rettig, S. J.; Storr, A.; Thompson, R. C.; Trotter, J.
Can. J. Chem. 1991, 69, 432.
1
Physical Measurements. H NMR spectra were recorded on a
Bruker Avance 300 spectrometer (300 MHz) or a Bruker Avance
400 inverse spectrometer (400 MHz) at room temperature (RT) and
(53) Richman, J. E.; Atkins, T. J. J. J. Am. Chem. Soc. 1974, 96, 2268.
2282 Inorganic Chemistry, Vol. 47, No. 7, 2008

Trimetallic d/f/d Complexes
Table 1. List of the [LnTM2(bcn)2]ClO4 ·nH2O Complexes Prepared and Their Analytical Data
%C
%H
%N
complex
found
calcd
found
calcd
found
calcd
+ESI-MS (m/z)
[LaZn2(bcn)2]ClO4 ·4H2O
[NdZn2(bcn)2]ClO4 ·3H2O
[GdZn2(bcn)2]ClO4 ·H2O
[DyZn2(bcn)2]ClO4 ·H2O
[YbZn2(bcn)2]ClO4 ·H2O
[LaNi2(bcn)2]ClO4 ·3H2O
[NdNi2(bcn)2]ClO4 ·2H2O
[GdNi2(bcn)2]ClO4 ·H2O
[DyNi2(bcn)2]ClO4 ·H2O
[YbNi2(bcn)2]ClO4
48.90
49.08
50.06
50.79
50.58
49.71
49.14
50.41
50.04
50.76
48.76
49.23
50.10
50.60
50.19
49.93
49.28
50.62
49.84
50.71
4.82
5.11
5.20
5.12
4.87
5.25
4.79
4.91
5.02
4.79
5.15
5.05
4.83
4.72
4.68
5.12
4.75
4.88
4.72
4.73
6.81
6.52
6.68
6.96
6.45
6.51
6.18
6.89
6.32
6.71
6.32
6.38
6.49
6.56
6.50
6.47
6.39
6.56
6.46
6.57
1157
1163
1176
1181
1192
1145
1150
1163
1169
1179
3
2
) 5.82 Hz), 3.20 (tt, 4H, CH2, J ) 7.53 Hz, J ) 3.18 Hz), 3.06
(tt, 4H, CH2, 3J ) 7.53 Hz, 2J ) 3.09 Hz), 2.43 (s, 6H, CH3), 2.42
(s, 3H, CH3). 13C NMR (75.5 MHz, RT, (CD3)2CO): δ 144.07 (2C),
138.87 (2C), 130.76, 130.61, 128.18, 127.94, 50.28, 43.18, 21.48
(2C). MS (+ESI): m/z 588 ([L + Na]+).
RT, D2O): δ 3.43 (s, 12H, CH2). 13C NMR (75.5 MHz, RT, D2O/
CD3OD): δ 45.78.
1,4,7-Triazacyclononane (tacn). Tacn ·3HCl (9.2 g, 0.0388 mol)
was dissolved in 20 mL of deionized H2O. The pH was adjusted
to ∼13 by addition of NaOH pellets (∼4.3 g, 2.8 equiv), and the
product was extracted into chloroform by continuous extraction for
48 h. The chloroform layer was removed, and the aqueous layer
was further extracted with chloroform (3 × 30 mL). The combined
chloroform layers were dried over NaSO4 and filtered. The solvent
was removed under reduced pressure, affording a pale yellow oil
Ethyleneglycol Ditosylate. Over 30 min, small portions of TsCl
(114 g, 0.6 mol) were slowly added to ethylene glycol (18.6 g, 0.3
mol) dissolved in 500 mL of pyridine under Ar and cooled to 0 °C
using an ice bath. After it was continuously stirred at 0 °C under
Ar for 5 h, the solution was poured onto crushed ice (∼500 mL)
and acidified using concentrated HCl until pH 1 (∼700 mL of HCl).
The white precipitate was washed with cold water, recrystallized
from hot methanol, and dried for 48 h in vacuo to yield 50 g, 50%.
1
to yield 3.6 g, 72%. H NMR (300 MHz, RT, D2O): δ 2.78 (s,
12H, CH2). 13C NMR (75.5 MHz, RT, D2O/CD3OD): δ 46.92. MS
(+ESI): m/z 130 ([L + H]+). MS (+ESI, high-res.) calcd for
C6H16N3: 130.1344. Found: 130.1347.
1H NMR (300 MHz, RT, CD3Cl): δ 7.71 (d, 4H, aromatic CH, J
3
3
) 6.6 Hz), 7.32 (d, 4H, aromatic CH, J ) 8.0 Hz), 4.17 (s, 4H,
N,N′,N′′-Tris(2-hydroxybenzyl)-1,4,7-triazacyclononane
(H3bcn). To a solution of tacn (3.1 g, 0.0279 mol) in dry ethanol
was added salicylaldehyde (10.22 g, 0.0837 mol) dropwise. The
solution was allowed to stir for 2 h under Ar before the solvent
was removed under reduced pressure, leaving an orange oil. The
oil was dissolved in dry dichloromethane under Ar, followed by
the slow addition of NaBH(OAc)3 (35.48 g, 0.1674 mol) over a
period of 3 h. The white cloudy solution was allowed to stir for an
additional 3 h under Ar after which excess NaBH(OAc)3 was
quenched by the addition of a saturated solution of sodium
bicarbonate. The product was extracted from the aqueous layer with
chloroform (3 × 50 mL) and dried over MgSO4, and the volatiles
were removed in vacuo. The product was recrystallized in MeOH/
H2O yielding an off white solid that was obtained in excellent yield
with respect to tacn, 10.23 g, 82%. 1H NMR (300 MHz, RT,
CH2), 2.43 (s, 6H, CH3). 13C NMR (75.5 MHz, RT, CDCl3): δ
145.46, 132.40, 130.13, 128.08, 66.92, 21.81. MS (+ESI): m/z 393
([L + Na]+).
1,4,7-Triazacyclononane Tristosylate. Diethylene-1,4,7-triamin-
etritosylate (24.1 g, 0.0427 mol) was dissolved in 100 mL of
anhydrous DMF dried over 4 Å sieves. Small portions of 60% NaH
in oil (previously rinsed with petroleum ether, 6 g, 0.25 mol) were
slowly added. The reaction was heated to 70 °C under Ar. After
the mixture had been continuously stirred for 30 min, a solution of
ethyleneglycol ditosylate (15.8 g, 0.0427 mol) in dry DMF (200
mL) was added dropwise over a 5 h period. The solution was stirred
for 20 h under Ar at 70 °C, after which the reaction volume was
reduced to ∼200 mL under reduced pressure and slowly added to
∼1000 mL ice and water affording a yellow precipitate. The
solution was stored at 4 °C for 5 h before filtration; the filtrate was
washed with water, ethanol, and ether, dried in vacuo overnight,
and used without further purification to yield 24.1 g, 94%. 1H NMR
3
CDCl3): δ 7.19 (t, 3H, aromatic CH, J ) 7.8 Hz), 6.93 (d, 3H,
3
3
aromatic CH, J ) 7.2 Hz), 6.90 (d, 3H, aromatic CH, J ) 7.2
Hz), 6.76 (t, 3H, aromatic CH, 3J ) 7.5 Hz), 3.77 (s, 6H, BnCH2),
2.43 (s, 12H, CH2). H NMR (300 MHz, RT, CD3OD): δ 7.22 (t,
3H, aromatic CH, J ) 8.1 Hz), 7.16 (d, 3H, aromatic CH, J )
3
1
(300 MHz, RT, CD3Cl): δ 7.69 (d, 6H, aromatic CH, J ) 8.1
3
3
3
Hz), 7.32 (d, 6H, aromatic CH, J ) 8.1 Hz), 3.42 (s, 12H, CH2),
2.43 (s, 9H, CH3). 13C NMR (75.5 MHz, RT, CDCl3): δ 143.88,
132.57, 129.86, 127.47, 51.86, 21.51. MS (+ESI): m/z 614 ([L +
Na]+).
7.2 Hz), 6.87 (t, 3H, aromatic CH, J ) 8.1 Hz), 6.84 (d, 3H,
3
3
aromatic CH, J ) 7.2 Hz), 3.86 (s, 6H, BnCH2), 2.85 (s, 12H,
CH2). 13C NMR (75.5 MHz, RT, CD3OD): δ 164.18, 133.15,
131.07, 123.33, 120.79, 117.10, 60.50, 51.89. MS (+ESI): m/z 448
([L + H]+). Anal. Calcd (found) for C27H33N3O3: C, 72.46 (72.34);
H, 7.43 (7.34); N, 9.39 (9.64).
1,4,7-Triazacyclononane ·Trihydrochlorate (tacn ·3HCl). 1,4,7-
Triazacyclonone tritosylate (24.1 g, 0.0407 mol) was added to 40
mL of concentrated H2SO4. The brown slurry was heated at
160 °C for 30 min under Ar and allowed to cool to RT before
being added dropwise to cold EtOH (200 mL), followed by 300
mL of diethyl ether. After storage at 4 °C overnight, the brown
precipitate was filtered, dissolved in 150 mL of H2O, and decol-
orized by boiling the solution for 15 min in the presence of Norit
and Celite. After the solution was clarified by filtration, the solvent
was removed under reduced pressure to afford a light brown oil.
An off-white precipitate immediately formed upon the addition of
25 mL of concentrated HCl, followed by 150 mL of ethanol. The
solid was filtered, washed with ethanol and ether, and recrystallized
Synthesis of Complexes. See Table 1 for a list of the
[LnTM2(bcn)2]ClO4 ·nH2O species prepared and their characteriza-
tion data.
[Ni(Hbcn)] ·H2O. To a solution of H3bcn (27.9 mg, 0.062 mmol)
in 20 mL methanol was added Ni(ClO4)2 ·6H2O (22.9 mg, 0.062
mmol). The pale yellow solution turned cloudy immediately after
the addition. After the solution was stirred for 1 h, the yellow
powder was filtered, washed with cold methanol, and air-dried to
1
yield 16.9 mg, 54%. H NMR (300 MHz, RT, CD3OD): δ 7.14
(m, 6H, o-benzyl H and m-phenol H), 6.83 (m, 6H, o-phenol H
and p-phenol H), 3.82 (s, br, 6H, benzyl CH2, V1/2 ) 20 Hz,), 2.79
1
from H2O/EtOH (1:8) to yield 9.2 g, 95%. H NMR (300 MHz,
Inorganic Chemistry, Vol. 47, No. 7, 2008 2283

Barta et al.
a
Scheme 2. Synthesis of [LnTM2(bcn)2]+
a
TM ) Ni(II) or Zn(II) and Ln ) La(III), Nd(III), Gd(III), Dy(III), or Yb(III).
(m, 12H, CH2). MS (+ESI): m/z 505 ([TM(HL) + H]+). µeff:
diamagnetic. Anal. Calcd. (found) for C27H33N3NiO4: C, 62.09
(62.34); H, 6.37 (6.26); N, 8.05 (8.30).
7.7), 6.78 (t, 6H, p-phenol H,3J ) 7.8 Hz), 6.13 (d, 6H, o-phenol
H, 3J ) 8.2 Hz), 4.23 (d, 6H, benzyl CH2, 2J ) 10.8 Hz), 3.39 (d,
6H, benzyl CH2, 2J ) 12.0 Hz), 3.07 (ddd, 6H, CH2, 2J ) 16.2 Hz,
3
2
H3[Ni(bcn)]2ClO4. To a solution of H3bcn (27.9 mg, 0.062
mmol) in 20 mL acetonitrile was added Ni(ClO4)2 ·6H2O (27.9 mg,
0.062 mmol). The solution turned a golden yellow color im-
mediately after the addition and was stirred for 1 h at RT. The
solution was filtered through glass wool and allowed to stand at
RT, upon which purple needle crystals deposited. The crystals were
filtered off, washed with cold water, and air-dried to yield 6.6 mg,
3J(A,B) ) 15.0 Hz, J ) 2.7 Hz), 2.65 (dd, 6H, CH2, J ) 15.06
Hz, 3J ) 3.9 Hz), 2.34 (dd, 6H, CH2, 2J ) 12.3 Hz, 3J ) 3.6 Hz),
2
3
3
1.96 (ddd, 6H, CH2, J ) 13.5 Hz, J ) 12.3 Hz, J ) 5.4 Hz).
MS (+ESI): m/z 1085 ([TM2L3 + H]+). Anal. Calcd (found) for
C54H60N6O6Zn3: C, 59.76 (59.80); H, 5.57 (5.72); N, 7.74 (7.89).
General Procedure for [LnNi2(bcn)2]ClO4 ·nH2O, Ln(III)
) La, Nd, Gd, Dy, Yb (Scheme 2). To a solution of
Ni(ClO4)2 ·6H2O (91.4 mg, 0.25 mmol) in acetonitrile was added
H3bcn (111.9 mg, 0.25 mmol). A blue-green precipitate formed
immediately in the pale yellow solution. The reaction was refluxed
and allowed to stir for 1 h. Hydrated Ln(ClO4)3 ·6H2O (68.2–72.4
mg, 0.125 mmol) was added after the solution was cooled to RT,
affording a pale purple solution. The solution was stirred for an
additional 30 min at RT before NEt3 (75.6 mg, 0.75 mmol) diluted
in 5 mL acetonitrile was added dropwise. After it was stirred for 5
min, the solution turned yellow. Pale purple powder deposited upon
slow evaporation of the solvent (except in the case of
[GdNi2(bcn)2]ClO4 and [YbNi2(bcn)2]ClO4 where prismatic purple
crystals formed). The solid was filtered, washed with cold aceto-
nitrile, and dried in air yielding 45–85%. 1H NMR (300 MHz, RT,
CD3OD) was uninformative.
1
21%. H NMR (300 MHz, RT, CD3OD) is uninformative. MS
(+ESI): m/z 505 ([TM(HL) + H]+). µeff: 2.77 µB. Anal. Calcd
(found) for C54H63ClN6Ni2O10: C, 58.49 (58.14); H, 5.81 (5.89);
N, 7.46 (7.67).
[Zn(Hbcn)] ·H2O. To a solution of H3bcn (27.9 mg, 0.062 mmol)
in methanol was added Zn(ClO4)2 ·6H2O (23.1 mg, 0.062 mmol).
The pale yellow solution became clear immediately after the
addition and was continuously stirred for 1 h at RT. Evaporation
of the solvent to ∼2 mL afforded a white precipitate which was
filtered, washed with cold methanol, and air-dried to yield 25.9
mg, 82%. 1H NMR (300 MHz, RT CDCl3): δ 7.14 (d, 3H, o-benzyl
3
3
H, J ) 7.2 Hz), 6.98 (t, 3H, m-phenol H, J ) 7.2 Hz), 6.78 (t,
3
3
3H, p-phenol H, J ) 7.2 Hz), 6.14 (d, 3H, o-phenol H, J ) 7.2
2
Hz), 4.23 (d, 3H, benzyl CH2, J ) 11.4 Hz), 3.39 (d, 3H, benzyl
CH2, 2J ) 11.4 Hz), 2.96 (ddd, 3H, CH2, 2J ) 15.0 Hz, 3J ) 14.1
Hz, 3J ) 3.9 Hz), 2.66 (dd, 3H, CH2, 2J ) 15.3 Hz, 3J ) 5.4 Hz),
General Procedure for [LnZn2(bcn)2]ClO4 ·nH2O, Ln(III)
) La, Nd, Gd, Dy, Yb. To a solution of H3bcn (111.9 mg, 0.25
mmol) in methanol was added Zn(ClO4)2 ·6H2O (93.1 mg, 0.25
mmol). The pale yellow solution immediately turned clear upon
addition of the metal salt. The solution was refluxed and allowed
to stir for 1 h. The hydrated Ln(ClO4)3 ·6H2O (68.2 – 72.4 mg,
0.125 mmol) was added to the solution after it had cooled to RT.
The clear solution was stirred for 30 min at RT and NEt3 (75.6
mg, 0.75 mmol) diluted in 5 mL methanol was added dropwise.
The solution turned slightly yellow after stirring for an additional
five minutes. An off-white perchlorate salt deposited upon slow
evaporation of the solvent (except in the case of [GdZn2(bcn)2]ClO4
which was redissolved in acetonitrile and formed prismatic colorless
crystals upon slow evaporation of the solvent) and was filtered,
2
3
2.31 (dd, 3H, CH2, J ) 12.5 Hz, J ) 3.9 Hz), 1.97 (ddd, 3H,
CH2, J ) 12.8 Hz,3J ) 12.3 Hz, J ) 6.0 Hz). MS (+ESI): m/z
511 ([TM(HL) + H]+). Anal. Calcd (found) for C27H33N3O4Zn:
C, 61.31 (61.02); H, 6.29 (6.30); N, 7.94 (8.20).
2
3
[Ni3(bcn)2]·2H2O. To a solution of H3bcn (27.9 mg, 0.062
mmol) in methanol was added Ni(ClO4)2 ·6H2O (22.9 mg, 0.062
mmol). The pale yellow solution immediately turned clear following
the addition. The solution was refluxed for 1 h and allowed to cool
before addition of the third equivalent of Ni(ClO4)2 ·6H2O (11.4
mg, 0.031 mmol). The solution was allowed to stir at RT for 30
min before NEt3 (18.8 mg, 0.186 mmol) was added dropwise. The
solvent was evaporated to ∼2 mL. A light purple precipitate formed
after additional solvent was slowly evaporated. The powder was
filtered, washed with cold methanol, and air-dried to yield 28.4
mg, 43%. MS (+ESI): m/z 1065 ([TM3L2 + H]+). µeff : 5.54 µB.
Anal. Calcd (found) for C54H64N6Ni3O8: C, 58.90 (59.29); H, 5.86
(5.84); N, 7.63 (7.83).
1
washed with cold ethanol, and dried in air yielding 60–85%. H
NMR for [LaZn2(bcn)2]ClO4 (400 MHz, RT, CDCl3, Supporting
Information): δ 7.12 (d, 6H, o-benzyl H, 3J ) 6.2 Hz), 6.96 (t, 6H,
m-phenol H, 3J ) 7.1 Hz), 6.75 (t, 6H, m-phenol H, 3J ) 7.3 Hz),
6.12 (d, 6H, o-phenol H, 3J ) 8.1 Hz), 4.20 (d, 6H, benzyl CH2, 2J
2
[Zn3(bcn)2]. To a solution of H3bcn (27.9 mg, 0.062 mmol) in
methanol was added Zn(ClO4)2 ·6H2O (23.1 mg, 0.062 mmol). The
pale yellow solution immediately turned clear following the
addition. The solution was refluxed for 1 h and allowed to cool
before addition of the third equivalent of Zn(ClO4)2 ·6H2O (11.6
mg, 0.031 mmol). The solution was allowed to stir at RT for 30
min before NEt3 (18.8 mg, 0.186 mmol) was added dropwise. A
white precipitate formed after the solvent was evaporated to ∼2
mL which was filtered, washed with cold methanol, and air-dried
) 11.7 Hz), 3.37 (d, 6H, benzyl CH2, J ) 11.7 Hz), 2.95 (ddd,
2
3
3
6H, CH2, J ) 16.0 Hz, J ) 14 Hz, J ) 3.9 Hz), 2.64 (dd, 6H,
2
3
2
CH2, J ) 15.5 Hz, J ) 5.32 Hz), 2.30 (dd, 6H, CH2, J ) 12.5
3
2
3
Hz, J ) 4.3 Hz), 1.95 (ddd, 6H, CH2, J ) 12.8 Hz, J ) 12.0
Hz, J ) 5.5 Hz). 13C NMR (75 MHz, RT, CD3OD): δ 167.02
3
(ring C-OH), 132.41 (m-C), 131.16 (o-benzyl C), 125.05 (benzyl
C), 121.92 (p-phenol C), 116.83 (o-phenol C), 63.48 (benzyl CH2),
57.64 (CH2).
X-ray Crystallographic Analyses. Single crystals of the per-
chlorate salts of H3[Ni(bcn)]2ClO4 · 3CH3CN, [YbNi2(bcn)2]-
ClO4 ·CH3CN, and [GdZn2(bcn)2(CH3CN)2]ClO4 ·CH3CN, isolated by
1
to yield 37.7 mg, 56%. H NMR (300 MHz, RT, CDCl3): δ 7.13
3
(d, 6H, o-benzyl H, J ) 6.4 Hz), 6.98 (t, 6H, m-phenol H,3J )
2284 Inorganic Chemistry, Vol. 47, No. 7, 2008

Trimetallic d/f/d Complexes
Table 2. Selected Bond Lengths (Å) and Angles (deg) in
H3[Ni(bcn)]2ClO4 ·3CH3CN
Table 3. Selected Bond Lengths (Å) and the Intermetallic Angle (deg)
in the [GdNi2(bcn)2(CH3CN)2]ClO4 ·CH3CN,
[YbNi2(bcn)2]ClO4 ·CH3CN, and [GdZn2(bcn)2(CH3CN)2]ClO4 ·CH3CN
Complexes
N(1)-Ni(1)
N(2)-Ni(1)
N(3)-Ni(1)
N(4)-Ni(2)
N(5)-Ni(2)
N(6)-Ni(2)
O(1)-Ni(1)
O(2)-Ni(1)
2.133(3)
2.125(3)
2.181(3)
2.086(3)
2.210(3)
2.138(3)
2.034(4)
2.075(4)
O(3)-Ni(1)
O(4)-Ni(2)
O(5)-Ni(2)
O(6)-Ni(2)
O(7)-Cl(1)
O(8)-Cl(1)
O(9)-Cl(1)
O(10)-Cl(1)
1.987(4)
2.028(3)
1.991(3)
2.104(3)
1.433(3)
1.423(4)
1.420(3)
1.378(3)
crystal data
GdNi2
YbNi2
GdZn2
Ln-O1
2.342(2)
2.384(2)
2.351(2)
2.354(2)
2.390(2)
2.379(2)
2.082(2)
2.058(2)
2.064(2)
2.094(2)
2.065(2)
2.039(2)
2.083(3)
2.069(3)
2.074(3)
2.068(3)
2.079(3)
2.067(3)
3.1191(5)
3.1182(5)
5.905(3)
142.417(13)
2.203(6)
2.193(6)
2.190(7)
2.184(6)
2.195(6)
2.202(6)
2.083(6)
2.115(6)
2.076(6)
2.100(6)
2.102(6)
2.2084(7)
2.069(8)
2.053(7)
2.064(8)
2.062(7)
2.050(8)
2.048(8)
2.8990(13)
2.8910(13)
5.787(13)
2.385(4)
2.382(4)
2.350(3)
2.341(4)
2.394(4)
2.357(3)
2.093(4)
2.047(4)
2.122(4)
2.108(4)
2.086(4)
2.070(4)
2.150(5)
2.134(4)
2.137(4)
2.152(4)
2.129(4)
2.136(4)
3.1259(7)
3.1254(6)
5.905(7)
Ln-O2
Ln-O3
Ln-O4
Ln-O5
Ln-O6
O(3)-Ni(1)-O(1)
O(3)-Ni(1)-O(2)
O(1)-Ni(1)-O(2)
O(3)-Ni(1)-N(2)
O(1)-Ni(1)-N(2)
O(2)-Ni(1)-N(2)
O(3)-Ni(1)-N(1)
O(1)-Ni(1)-N(1)
O(2)-Ni(1)-N(1)
N(2)-Ni(1)-N(1)
O(3)-Ni(1)-N(3)
O(1)-Ni(1)-N(3)
O(2)-Ni(1)-N(3)
N(2)-Ni(1)-N(3)
N(1)-Ni(1)-N(3)
87.28(17)
84.15(17)
85.29(17)
100.25(17)
170.93(15)
90.42(13)
171.12(15)
89.64(13)
103.90(17)
83.63(13)
93.31(13)
102.33(17)
171.87(15)
82.42(12)
79.21(14)
O(5)-Ni(2)-O(4)
O(5)-Ni(2)-N(4)
O(4)-Ni(2)-N(4)
O(5)-Ni(2)-O(6)
O(4)-Ni(2)-O(6)
N(4)-Ni(2)-O(6)
O(5)-Ni(2)-N(6)
O(4)-Ni(2)-N(6)
N(4)-Ni(2)-N(6)
O(6)-Ni(2)-N(6)
O(5)-Ni(2)-N(5)
O(4)-Ni(2)-N(5)
N(4)-Ni(2)-N(5)
O(6)-Ni(2)-N(5)
N(6)-Ni(2)-N(5)
86.44(17)
168.43(16)
89.13(12)
87.77(15)
85.96(14)
102.58(15)
102.17(17)
169.49(14)
83.50(11)
88.34(11)
88.08(12)
102.73(15)
82.47(11)
170.12(12)
83.77(11)
TM1-O1
TM1-O2
TM1-O3
TM2-O4
TM2-O5
TM2-O6
TM1-N1
TM1-N2
TM1-N3
TM2-N4
TM2-N5
TM2-N6
Ln-TM1
Ln-TM2
TM1-TM2
TM1-Ln-TM2
176.39(4)
141.68(4)
slow evaporation of the solvent, were mounted on a glass fiber, and
measurements were made on a Rigaku/ADSC CCD area detector with
graphite-monochromated Mo KR; radiation (λ ) 0.71073 Å). The
structures were processed and corrected for Lorentz and polarization
effects, and absorption using the d*TREK program.54 The structures
were solved by direct methods55 and expanded using Fourier tech-
niques.56 All calculations were performed using the teXsan57 crystal-
lographic software package of the Molecular Structure Corporation,
and SHELXL-97.58 The non-hydrogen atoms were refined anisotro-
pically. Hydrogen atoms were fixed in calculated positions, but not
refined.
Purple needle crystals of [GdNi2(bcn)2(CH3CN)2]ClO4 ·CH3CN
were obtained via slow evaporation of acetonitrile. The sample was
mounted on a glass fiber and cooled to 173 K. Data sets were
collected on a Bruker X8 APEX CCD area detector with graphite-
monochromated Mo KR radiation (λ ) 0.71073 Å). Data were
collected and integrated using the Bruker SAINT59 software
package and corrected for Lorentz and polarization effects as well
as for absorption (SADABS).60 The structure was solved by direct
methods (SIR92).55 Non-hydrogen atoms were refined anisotropi-
cally, while hydrogen atoms were added but not refined. Final
refinement was completed using SHELXL-97.58 Selected crystal-
lographic data for the complexes are presented in Tables 2 and 3
and in the Supporting Information for this paper.
published53 (Scheme 1). The phenol arms were successfully
attached to the polyamine macrocycle by an in situ reductive
alkylation with sodium triacetoxyborohydride, the only
reducing agent found to prepare the desired H3bcn in
moderate to high yields, with minimal side-products.
[TM(Hbcn)]· nH2O Complexes. The isolation of the
Ni(II)- and Zn(II)-bcn complexes was instrumental in study-
ing the building blocks of the larger coaggregated
[LnTM2(bcn)2]+ complexes (abbreviated as LnTM2). The
binary transition metal complexes were synthesized using
one equivalent of TM(ClO4)2 ·6H2O (TM ) Zn(II) or Ni(II))
and H3bcn in either methanol or acetonitrile to give a parent
peak with 100% relative intensity in the +ESI-MS spectra
corresponding to the proton adducts of the [TM(Hbcn)]
complexes.
A rigid octahedral geometry where the three phenolate
oxygen donors and three amine nitrogens are tightly bound
to the Zn(II) ion, is postulated from the sharp resonances
1
observed in the H NMR data of [Zn(Hbcn)] (Figure 2).
The ring hydrogens, H3–6 are split into four distinct
signals, all significantly shifted. The largest upfield shift
was observed for H3, the hydrogen ortho to the chelating
phenolate oxygen (∆δ ≈ 0.79 ppm). Splitting of the singlet
from the benzylic protons into AB doublets upon com-
plexation additionally confirms the rigidity of the complex.
The diasterotopic hydrogens are further distinguished by
assuming the proton in the pseudoaxial position, H2B, is
more shielded than the benzylic proton in the pseudoequa-
torial environment, H2A, because of the proximity of the
collinear p-orbital from the neighboring benzene ring. The
excess shielding causes the pseudoaxial hydrogen to
resonate at lower frequencies (3.39 vs 4.23 ppm of the
pseudoequatorial proton).61
Results and Discussion
Synthesis of H3bcn. A modified Richmond-Atkins route
was used to synthesize H3bcn in higher yields than previously
(54) D*TREK Area Detector Software,version 7.11; Molecular Structure
Corporation: The Woodlands, TX, 2001.
(55) Altomare, A.; Cascarano, M.; Giaconazzo, C.; Guagliardi, A. J. Appl.
Crystallogr. 1994, 26, 343.
(56) Beurskens, P. T.; Admiraal, G.; Beurskens, G.; Bosman, W. P.; de
Gelder, R.; Israel, R.; Smits, J. M. M. The DIRDIF-94 program system;
Crystallography Laboratory, University of Nijmegen: Nijmegen, The
Netherlands, 1994.
(57) teXsan: Crystal Structure Analysis Package; Molecular Structure
Corporation: The Woodlands, TX, 1985, 1992.
(58) Sheldrick, G. M. SHELXL-97; University of Gottingen: Germany,
1997.
(59) SAINT,version 6.02; Bruker AXS Inc.: Madison, WI, 1999.
(60) SADABS,version 2.05; Bruker AXS Inc.: Madison, WI, 1999.
Inorganic Chemistry, Vol. 47, No. 7, 2008 2285

Barta et al.
Figure 2. 1H NMR spectra of H3bcn (top) and [Zn(Hbcn)] (bottom) (300
MHz, CDCl3, RT, *solvent).
Figure 3. ORTEP diagram of the cation in H3[Ni(bcn)]2(ClO4)·3CH3CN;
thermal ellipsoids are drawn at 50%.
In addition, the conformational constraints on the ethylenic
backbone imposed by the chelation of Zn(II) renders the
ethylenic hydrogens magnetically and chemically inequiva-
lent (Figure 2) resulting in four distinct multiplets assigned
to each of the ethylenic hydrogens in the N-CH2-CH2-N
bridge. Assignments were confirmed by 1H-1H COSY
experiments. The elemental analysis was also consistent with
a stoichiometry of [Zn(Hbcn)] ·H2O, suggesting the formation
of a neutral encapsulated product.
tion sphere around the metal ion difficult. Elevating or
lowering the temperature resulted in no significant sharpen-
ing, broadening, or coalescence of the resonances, implying
a fluxional process in solution with a low energy barrier
between interconverting states. The broadening of the
hydrogen resonances may also be complicated by the
exchange of the coordinated amine nitrogens and phenol
oxygens with surrounding solvent molecules. The 1H NMR
spectrum from the paramagnetic purple Ni(II) species was
uninformative.
A solid-state structure of the purple paramagnetic material
H3[Ni(bcn)]2ClO4 ·3CH3CN (Figure 3 and Table 2) revealed
Ni(II) encapsulated in a distorted octahedral environment by
the bcn3- ligand via three phenolate O groups and three
amine N functionalities. Each of the two independent Ni(II)
moieties is located near an inversion center with approximate
C3 symmetry about an axis perpendicular to the plane of the
bcn3- ligand.
The Ncis-Ni-Ncis bite angle (82.6°) about the Ni(II) center
varies from the 90° idealized for an octahedral structure
because of the constraints imposed by the five-membered
chelate ring of the tacn moiety; however, the Ncis-Ni-Ocis
angle is close to 90° (89.1°), suggesting that the six-
membered ring between the phenolate O and the amine N is
large enough to accommodate Ni(II) in an octahedral
environment (Table 2).
The average Ni-O and Ni-N bond lengths do not differ
between the two [Ni(bcn)]- units (Ni(1) ) 2.044 and 2.141
Å, Ni(II) ) 2.031 and 2.137 Å, respectively); however, the
Ni-N bonds are slightly longer than those observed for
[Ni(tacn)2]2+ (2.139 vs 2.116 Å, respectively),64 and signifi-
cantly longer than optimum values for an octahedral NiN6
structure of 2.08 Å.65 The elongation of the Ni-N bonds
are correlated with the less strongly coordinating tertiary
Interestingly, different stoichiometries of the Ni(II)-bcn
complexes were isolated depending on the solvent used. A
purple paramagnetic material was obtained when synthesized
in acetonitrile with an EA consistent with H3[Ni(bcn)]2ClO4,
whereas a yellow diamagnetic material precipitated using the
analogous reaction in methanol with the stoichiometry of
[Ni(Hbcn)]· H2O. The purplematerial is not soluble in alco-
hols and only sparingly soluble in acetonitrile, whereas the
yellow material is soluble in both solvents. The only
difference detected in the fingerprint region of the IR is a
prominent band at 1061 cm-1 for the purple material,
-
correlating with a ClO4 counterion and confirming the
formation of a cationic complex for the purple Ni(II) species.
The +ESI-MS spectra for both species were nearly identical
with the parent peak corresponding to the proton adduct of
the [Ni(Hbcn)] complex.
1
The shifting of the resonances in the H NMR spectrum,
in addition to the absence of resonances corresponding to
the free ligand, confirm the coordination of Ni(II) to the
polyamine phenol ligand for the yellow diamagnetic species.
The multiplets in the aromatic region, the broad peaks for
the organic framework, and the small chemical shift changes
relative to the free ligand (∆δ < 0.02 ppm) suggest exchange
of the amine nitrogens and phenol oxygens around the metal
on the NMR time scale, making elucidation of the coordina-
(61) Fraser, R. R.; Gurudata; Reyes-Zamora, C.; Swingle, R. B Can.
J. Chem. 1968, 46, 1595.
(62) Cole, E.; Copley, R. C. B.; Howard, J. A. K.; Parker, D.; Ferguson,
G.; Gallagher, J. F.; Kaitner, B.; Harrison, A.; Royle, L. J. Chem.
Soc., Dalton Trans. 1994, 1619.
(64) Zompa, L. J.; Margulis, T. N. Inorg. Chim. Acta 1978, 28, L157.
(65) Graham, B.; Grannas, M. J.; Hearn, M. T. W.; Kepert, C. M.; Spiccia,
L.; Skelton, B. W.; White, A. H. Inorg. Chem. 2000, 39, 1092.
(63) Beattle, J. K.; Elsbernd, H. S. J. Am. Chem. Soc. 1970, 92, 1946.
2286 Inorganic Chemistry, Vol. 47, No. 7, 2008

Trimetallic d/f/d Complexes
nitrogens and the steric demands on the ligand upon
accommodation of the Ni(II) ion, similar to that seen in a
Ni4L species, where L ) tetrakis(1,4,7-triazacyclonon-1-yl).65
Alternatively, the average Ni-O bond lengths (2.014 Å) in
the [Ni(bcn)]– species are very similar to other deprotonated
phenoxide species.22,41 It is also interesting to note that the
Ni-N and Ni-O distances in the axial positions are slighter
longer than the comparable bonds in the equatorial positions
(2.050 and 2.136 Å vs 2.021 and 2.132 Å, respectively)
attributed to steric strain of the ligand.
adduct of the monometallic [TM(Hbcn)] complexes). The
only differences observed in the IR of the monometallic and
the trimetallic TM(II) species was the bathochromic shift of
the VC-O bands attributed to decreased strength in the
phenolate oxygen O-TM dative bond from the oxygens
bridging an additional metal.
Only one set of resonances is seen in the 1H NMR spectra
for [Zn3(bcn)2], implying a 3-fold symmetric complex where
the Zn(II) is held in a rigid octahedral geometry. A small
upfield shift of the hydrogen (H3) ortho to the phenolate
oxygen (∆δ ≈ 0.01) compared to the [Zn(Hbcn)] spectra
was observed. Slight upfield shifts and larger couplings
between the protons on the ethylenic bridge are also observed
suggesting an increasingly strained environment upon bridg-
ing the third Zn(II) metal ion (∆δ ≈ 0.11 ppm, ∆3J ≈ 1.2
Hz, ∆2J ≈ 0.8 Hz for H1).
Three bridging hydrogens equidistant to the six phenolic
oxygen atoms were found between the two encapsulated
Ni(II) structures. The dimers have intermolecular O-O
contacts (2.705 Å) consistent with other O-HO hydrogen
bonds in the literature (near 2.8 Å); however, the exact
location of the O-H hydrogen atoms could not be located
because of the three hydrogens being disordered over the
six possible sites. The large thermal ellipsoids on the
oxygens, expected for intermolecular hydrogen bonding,
suggest that the hydrogen atoms reside between the two
halves of the molecule rather than being bound to one-half
of the complex. One perchlorate and three noncoordinating
acetonitrile molecules were also found in the unit cell.
On the basis of the diamagnetic properties, it is postulated
that the yellow [Ni(Hbcn)] material is a square planar d8
Ni(II); this hypothesis is strengthened by analyzing the
UV–vis data of the yellow material. One weak band resolved
at 586 nm, resulting from the 1A1g f 1A2g transition, matches
literature values for Ni(II) in the square planar conforma-
tion.49,66 The purple Ni(II) complex has absorbances char-
acteristic for an octahedral environment, arising at 544 and
[LnTM2(bcn)2]ClO4 · nH2O. Air-stable complexes of the
trinuclear [LnTM2(bcn)2]ClO4 · nH2O unit were produced
using protocols adopted from the synthesis of analogous
[LnNi2(tam)2]+ and [LnNi2(trn)2]+ cationic species (TM )
Ni(II) and Zn(II), Ln ) La(III), Nd(III), Gd(III), Dy(III),
Yb(III)).36,41 Spontaneous, self-assembled bicapped struc-
tures, where the Ln(III) is bridged between the three
deprotonated phenolate oxygens from two [TM(bcn)]- pre-
organized ligands, were readily formed. The amount of base
added is instrumental to the formation of these complexes;
the desired d/f/d complexes were not isolated if a slight
excess or shortage of the base was used. The complexes
precipitated from the respective reaction mixtures affording
light-purpleoroff-whitesolidsfortheNi(II)-orZn(II)-Ln(III)
species, respectively (Scheme 2).
Attempts to recrystallize the complexes (in hot MeOH/
H2O or CH3CN/H2O) or purification by column chromatog-
raphy resulted in decomposition of the product and presumed
hydrolysis of the Ln(III) ions. Washing the precipitated solid
thoroughly with cold solvents to remove unreacted ligand,
additional Ln(ClO4)3 and TM(ClO4)2 salts, or other side
products, was the only way to yield complexes pure by
elemental analysis (if the trimetallic d/f/d species did not
crystallize upon standing in solution). The compounds were
characterized by EA, +ESI-MS, IR, and NMR (only of the
diamagnetic [LaZn2(bcn)2]+). All results are consistent with
the formation of cationic d/f/d species shown in Table 1.
The elemental analyses for all the lanthanide complexes
agreed with the empirical formula [LnTM2(bcn)2]ClO4 ·nH2O
suggesting that labile reaction solvent molecules (either
acetonitrile of methanol) observed in the solid state crystal
structure analysis were replaced with water molecules upon
storage under air. Hydrated species have been observed with
previously synthesized Ln(III)-containing systems where
coordinated solvent molecules are associated with the
metal to fulfill the large coordination preference of the Ln(III)
ions.
3
3
3
3
900 nm from A2g(F) f T1g(F) and A2g(F) f T2g(F)
transitions, respectively.36,41
It is also interesting to note that the yellow material can
readily be converted to the purple material by dissolution
and refluxing in a variety of solvents, including acetonitrile.
The purple material, however, could not be converted into
the yellow diamagnetic material indicating that the yellow
complex is an intermediate. This point is confirmed by
allowing a sample of the yellow Ni(II) material to stand in
methanol for several weeks. During this time, the solution
turns red and purple crystals deposit that solve for the
same stoichiometry as the purple paramagnetic Ni(II)
complex.
[TM3(bcn)2] · nH2O Complexes. Addition of 1 equiv of
TM(ClO4)2 · 6H2O to 2 equiv of deprotonated [TM(bcn)]-
formed trimetallic transition metal species with the expected
stoichiometry confirmed by EA ([Ni3(bcn)2] · 2H2O for the
trimetallic Ni(II) species and [Zn3(bcn)2] for the trimetallic
Zn(II) complex). The peaks in the +ESI-MS spectra that
correlate to the trimetallic species (with the expected isotope
patterns) only resolved for 10% of the relative intensity,
suggesting that the stability of the neutral [TM3(bcn)2]
complexes was significantly reduced compared to that of the
respective [TM(Hbcn)] building block species (the parent
peak with 100% relative intensity corresponded to the proton
Electrospray ionization mass spectrometry in the positive
ion mode was the primary diagnostic tool for the detection
of the cationic d/f/d clusters, giving diagnostic mass spectra
with the expected isotopic patterns. The mass spectra are
devoid of any other significant peaks besides the parent ion
(66) Lever, A. B. P. Inorganic Electronic Spectroscopy; Elsevier: New
York, 1984.
Inorganic Chemistry, Vol. 47, No. 7, 2008 2287

Barta et al.
Figure 4. +ESI spectrum of [LaZn2(bcn)2]+; inset displays the experimental isotope pattern closely matching the calculated.
peak for the cationic complex, [LnTM2(bcn)2]+ (Figure 4).
The isotope patterns were also in good agreement with the
calculated patterns.
between Ln(III) ions and Ni(II) or Zn(II) are attributed to
mass differences among the metal ions. The small differences
between the TM/Ln complexes were also confirmed by the
crystal structure of [GdNi2(bcn)2(CH3CN)2]ClO4 ·CH3CN and
the analogous [GdZn2(bcn)2(CH3CN)2]ClO4 · CH3CN.
The two TM(II) in the [LnTM2(bcn)2]+ cationic complexes
appear to be in comparable coordination environments in
solution, as reflected by the absence of splitting of the
absorbance peaks in the UV–vis. The absorptions at 900 and
544 nm for the octahedral Ni(II) are associated with 3A2g(F)
f 3T2g(F) and 3A2g(F) f 3T1g(F) transitions, respectively.36,41
Both the [GdNi2(bcn)2]+ and [GdZn2(bcn)2]+ cations of
the perchlorate salt (Figure 5, Table 3) revealed a V-shaped
metallic core where the Zn(1)-Gd-Zn(2) angle is slightly
smaller (141.68°) than that observed for the Ni(1)-Ln-Ni(2)
complexes (142.4°). The Gd(III) ion is eight-coordinate in
a square antiprism distorted to C2V bicapped trigonal
prism36,67 geometry, bicapped by two deprotonated tridentate
[Ni(bcn)]- or two [Zn(bcn)]- units, plus two coordinating
acetonitrile molecules. The end capping units are nearly
identical with each TM(II) chelated in a distorted octahedral
geometry via the three amine and three phenolato functions
from the fully deprotonated bcn3- ligand, similar to the
coordination mode seen for the paramagnetic
H3[Ni(bcn)]2ClO4 · 3CH3CN species. The Ni-N and Ni-O
bond distances in [GdNi2(bcn)2(CH3CN)2]ClO4 · CH3CN are
3
3
The A2g(F) f T1g(P) transition is obscured by a strong
charge-transfer band at 303 nm. A π-π* transition of the
K-band of the benzene ring appears at 249 nm.42 The
spectrum of the d/f/d species is very similar to that of the
octahedral [Ni(Hbcn)] system, suggesting that the incorpora-
tion of the Ln(III) does not significantly distort the coordina-
tion environment around the Ni(II) ion.
1
The H NMR spectrum was instrumental in elucidating
the solution structure of the [LaZn2(bcn)2]+ complex. The
relative shifts in the resonances conclusively indicate binding
1
of the ligand to the metal ions. Interestingly, the H NMR
spectrum of [LaZn2(bcn)2]+ is nearly superimposable upon
that of [Zn(Hbcn)], suggesting that the TM(HL) structure is
maintained upon coordination to the La(III) ion. The minor
differences observed between the two spectra include the
ring proton resonances being shifted slightly upfield upon
accommodating the Ln(III) ion (∆δ ≈ 0.02 ppm) compared
to those in the 1H NMR spectra for [Zn(Hbcn)]. The ethylenic
and benzylic proton resonances (which exhibit the same
splitting seen for the [Zn(Hbcn)] and [Zn3(bcn)2] species)
are also slightly shifted upfield from the Zn(II) complexes
(∆δ > 0.03 ppm) possibly because of the smaller donation
of electron density to La(III). The absence of additional
splitting of the resonances further suggests that the 3-fold
symmetry is maintained. In addition, only one set of
resonances in the 13C NMR spectrum was observed; one peak
for each of the carbons in the ring, one resonance for the
benzylic carbon and one peak for the chemically and
magnetically equivalent carbons in the ethylenic bridge.
Regardless of the TM(II) or the Ln(III) used, the IR spectra
for the [LnTM2(bcn)2]ClO4 · nH2O complexes were nearly
superimposable in the 400–3500 cm-1 range, suggesting the
formation of isostructural complexes in the solid state. The
small <10 cm-1 variations in the IR peak frequencies
slightly
longer
than
those
observed
in
H3[Ni(bcn)]2ClO4 · 3CH3CN but very similar to other litera-
ture d/f complexes. In addition, the TM-O, TM-N, and
Gd-O bond lengths in the trimetallic species are nearly
identical regardless of the TM(II). The averaged Zn-Gd
distance is slightly longer (3.125 Å) than the Ni-Gd distance
of 3.119 Å attributed to the larger ionic radius of the Zn(II)
ion; however, the interatomic Zn-Zn distance is nearly
identical with the Ni-Ni distance in the TM-Ln-TM
˚
complexes (5.905 A).
A decrease in the number of coordinated solvent molecules
surrounding the smaller Ln(III) ions is observed for
[YbNi2(bcn)2]ClO4 · CH3CN, where the Yb(III) is 6-coordi-
nated by six bridging phenolate oxygens. The decrease of
the coordination number with the increasing atomic number,
that is, decrease in ionic radius, is commonly observed for
Ln(III) systems, especially in systems containing labile
coordinated solvent molecules. All Ln-O bond lengths
uniformly reflect the periodic decrease in the ionic radius of
the Ln(III) ion (2.367 Å for Gd(III) and 2.177 Å for Yb(III)),
as well as the compressed O-Yb-O angle (average of 77.6°)
(67) Muetterties, E. L.; Guggenberger, L. J. J. Am. Chem. Soc. 1974, 96,
1748.
2288 Inorganic Chemistry, Vol. 47, No. 7, 2008

Trimetallic d/f/d Complexes
Figure 5. ORTEP diagrams of the cations in [GdNi2(bcn)2(CH3CN)2]ClO4 (right) and [GdZn2(bcn)2(CH3CN)2]ClO4 (left) with expanded views of the
Gd(III) coordination sphere; thermal ellipsoids are drawn at 50% probability.
compared to 84.6° for the Gd(III) complex. The Ni-Yb-Ni
angle displays a nearly linear metallic core of 176.4°. The
intermetallic bonds for Ni-Yb are slightly shorter than the
Ni-Gd bonds (2.895 vs 3.119 Å), with a Ni-Ni interatomic
distance of 5.787 Å. Uniquely, these d/f/d complexes do not
contain coordinated counterions. Powder diffraction data was
also collected for the GdNi2 and YbNi2 systems; however,
the data did not match the crystal data because of differences
in solvation. Upon standing in air, these compounds are
known to exchange coordinated reaction solvents (methanol
or acetonitrile) for ambient water observed both in the IR
and EA spectra, and assumed to be the cause of the
differences in the powder and crystal diffraction data.
Magnetic Properties. The magnetic susceptibilities of the
[LnTM2(bcn)2]ClO4 · nH2O species were collected on pow-
dered samples using a SQUID magnetometer over a tem-
perature range of 2-300 K. Unfortunately, regression values
at high temperatures (>100 K) are only adequate for data
taken at the higher magnetic field (10 000 G). Thus, all ꢀMT
values reported at RT will be those measured at a magnetic
field of 10 000 G. To minimize effects from magnetic
saturation, ꢀMT values reported herein at <100 K were those
obtained using the magnetic field of 500 G. All measure-
ments were corrected for diamagnetism using Pascal’s
constants.51
ity, ꢀM, at room temperature (RT) for many rare earth
compounds. Thus, the experimental data will be compared
to the calculated values derived from eq 1
2
ꢀMT ) [2 × gNi SNi(SNi + 1) + gJ2J(J + 1)]/8
(1)
where g is the Landé factor fixed at 2.00, gJ ) 3/2 + [S(S
+ 1) - L(L + 1)](2J(J + 1))-1, L is the total orbital angular
moment quantum number, S is the total spin angular moment
quantum number, and J is the total angular momentum
quantum number.
The RT ꢀMT values for [LnZn2(bcn)2]ClO4 · nH2O, where
Ln ) Nd(III), Gd(III), Dy(III), and Yb(III), were determined
to be ꢀMT ) 1.67, 7.56, 14.16, and 1.95 cm3 K mol-1,
respectively, shown in Figure 6. These values closely
correspond to the calculated RT values (1.69, 7.88, 14.13,
and 2.57 cm3 K mol-1 using eq 1) where the magnetic
contribution is solely originating from the isolated Ln(III)
ion. The ꢀMT decreases slightly as the temperature is lowered
from 300 to 10 K for all LnZn2 species. Below 10 K, the
ꢀMT decreases significantly reaching a minimum at 2 K, an
intrinsic characteristic of Ln(III) attributed to the depopula-
tion of the spin–orbit sublevels similar to that seen in other
d/f and d/f/d heterometallic species.31,36,41,48
The absence of a magnetic exchange between the d/f/d
metals for the LnZn2 systems make these species an attractive
baseline for the analogous paramagnetic Ni(II) systems,
similar to the empirical approach developed by Costes et
The free-ion approximation, where only one level 2S+1LJ
is thermally populated and second-order contributions are
ignored, is a good approximation of the magnetic susceptibil-
Inorganic Chemistry, Vol. 47, No. 7, 2008 2289

Barta et al.
Figure 6. Temperature dependence of ꢀMT for [LnZn2(bcn)2]ClO4 ·nH2O
complexes at 10 000 G: (O) NdZn2; (4) GdZn2; (]) DyZn2; (0) YbZn2.
Figure 7. ꢀMT vs T plot of [LaNi2(bcn)2]ClO4 ·3H2O at 500 G from 100-2
K (O) or at 10 000 G from 300-2 K (insert, 2). The solid and the dotted
lines represent the theoretical curve (text) fit from 100 to 4 K or 100 to 2
K, respectively.
al.31 and used by Kahn,32,44 Okawa,
and Matsumoto35
45,48
j
and co-workers. Costes et al. removed the unquenched orbital
angular momentum contributions from the Ln(III) by cal-
culating the difference, ∆(ꢀMT), between the ꢀMT values for
the Cu-Ln species and the structurally homogeneous Ni-Ln
complexes, where the diamagnetic Ni(II) was held in a square
planar geometry. The magnetic exchange between the d/f/d
metal ions for the complexes herein were elucidated using a
similar empirical approach defined by eq 2.
parameters, invariably gave vanishingly small values for TIP.
Accordingly, we set TIP to zero and employed gNi and J′ as
variable parameters to get the best fit shown as the solid
line in Figure 7. The best fit gave gNi ) 2.151 (0.0001), J′
) -0.030 cm-1 (0.003), and F ) 1.63 × 10-5. F is the
goodness of fit value given by eq 3 where N is the number
of data points.
[
[
(
)
calc
)
(
)
2]]1⁄2
2 ⁄ Σ ꢀ
(3)
∆ꢀMT ) (ꢀMT)LnNi - (ꢀMT)LnZn ) 2(ꢀMT)Ni + JNiLn(T)(2)
F ) 1/N Σ ꢀ
- ꢀ
(
2
2
obsd
obsd
Ln ) La(III), Nd(III), Gd(III), Dy(III), and Yb(III) and
JNiLn(T) is related to the nature of the overall exchange
interaction between the Ni(II) and Ln(III), such that positive
or negative values indicate a ferromagnetic or antiferromag-
netic interaction, respectively. Similar crystal field effects
are assumed to be operative in both the Ni(II) and Zn(II)
complexes due to similarities in their core structures.
Intermolecular Ni(II)-Ni(II) magnetic interactions are as-
sumed to be negligible for the d/f/d systems.
The ꢀMT plot of [LaNi2(bcn)2]ClO4 · 3H2O corresponds
well to the expected curve for isolated Ni(II) ions (Figure
7). LaNi2 exhibits a ꢀMT value of 2.42 cm3 mol-1 K at RT,
corresponding to two Ni(II) ions with µeff per Ni(II) ) 3.11µꢁ
and g ) 2.10. Below 10 K, the ꢀMT curve for LaNi2 decreases
rapidly due to zero-field splitting of the octahedral Ni(II)
transition metal ion or weak intra and/or intermolecular
antiferromagnetic interactions, also seen in [LaNi2(tam)2]+
and [LaNi2(trn)2]+. Nonetheless, the µeff values determined
for the two Ni(II) ions are in the range of previously reported
literature values.68
Our inability to fit the lowest temperature data to the model
likely arises from the fact that the strong temperature
dependence of ꢀT in this region arises primarily from the
effects of zero field splitting of Ni(II), an effect not included
in this model. This analysis, which avoids this region, can
be considered to give a valid measure of the strength of the
interaction between the Ni centers and the clear conclusion
from this is that the coupling is extremely small. The J′ value
of -0.03 cm-1 is, for example, 20 times smaller than that
observed by Shiga et al. for their Ni-La-Ni compound (J
) -0.63 cm-1).45 This result prompted us to attempt to fit
our data to a ZFS model69 in which exchange coupling is
ignored. Setting J′ ) 0 and TIP ) 0, as above, and employing
gNi and D, the ZFS parameter, as variable parameters, the
best fit to the data over the full range from 2-100 K is shown
as the dotted line in Figure 7. This fit yielded g ) 2.154
(0.003), D ) 2.60 (0.07), and F ) 6.52 × 10-4. The
goodness of this fit confirms that the abrupt drop observed
in ꢀMT at the lowest temperatures is due primarily to ZFS
and that the exchange coupling between the nickel centers
(ignored in this particular model) is very weak in this
compound.
In order to confirm the extent of interaction between the
two terminal Ni(II) ions, we examined fits of the LaNi2 data
to the theoretical model described by Shiga et al.45 Attempts
to fit the data to this model over the full 2-100 K
temperature range were unsuccessful unless the 2 and 3 K
data were ignored. Attempts to fit the data over the 4-100
K range employing gNi, the exchange parameter J′ and
temperature independent paramagnetism, TIP, as variable
The ꢀMT value for the [LnNi2(bcn)2]ClO4 · nH2O species
followed the isolated-ion approximation of three isolated
metal ions at RT. The ꢀMT for NdNi2 was determined to be
3.70 cm3 K mol-1 at RT, closely matching the value (3.64
cm3 K mol-1 calculated using eq 1) expected for two
magnetically isolated Ni(II) ions (SNi ) 1) and one Nd(III)
(68) Earnshaw, A. Introduction to Magnetochemistry; Academic Press: New
(69) O’Connor, C. J. In Progress in Inorganic Chemistry; Lippard, S. J.,
Ed.; Interscience: New York, 1982; Vol. 29, pp 203.
York, 1968.
2290 Inorganic Chemistry, Vol. 47, No. 7, 2008

Trimetallic d/f/d Complexes
Table 4. Observed and Calculated Magnetic Susceptibilities for [LnNi2(bcn)2]ClO4 ·nH2O Complexes
Ln(III)
ground state
gJ
ꢀMTa (calcd)
ꢀMTobsd (RT)
ꢀMTobsd (2 K)
ꢀMTmax or min (T K)
∆(ꢀMT)b
sign of JNiLn(T)c
La
1S0
2.85
3.64
9.88
16.17
4.57
2.00
3.70
10.80
17.48
5.05
2.55
1.02
16.06
14.73
4.26
2.55 (2)
1.02 (2)
17.55 (13)
14.85 (3)
4.26 (2)
Nd
Gd
Dy
Yb
4I9/2
8/11
2
4/3
8/7
0.44
8.86
6.08
3.56
-
+
+
+
8S7/2
6H15/2
2S7/2
a ꢀMT calculated using eq 1. b ∆(ꢀMT) values were obtained using eq 2 at 2 K. c The sign of JNiLn(T) was determined after removing both Ln(III) and
Ni(II) contributions (eq 5).
Figure 8. Temperature dependence of ꢀMT for [LnNi2(bcn)2]ClO4 ·nH2O
Figure 9. M vs H curve for [GdNi2(bcn)2]ClO4 ·H2O at 10 K. The solid
complexes at 500 G: (2) LaNi2; (O) NdNi2; (4) GdNi2; (]) DyNi2; (0)
line indicates the theoretical curve for St ) 11/2 (g ) 2).
YbNi2.
of other Ni-Gd systems reported where a ferromagnetic
exchange was determined between the Ni(II) and Gd(III)
metal ions.
ion (J ) 9/2, gJ ) 8/11, Table 4, Figure 8). As the
temperature decreased, the ꢀMT values reached a minimum
of 1.02 cm3 K mol-1 at 2 K, well below the calculated value
of 3.64 cm3 K mol-1.
We examined fits of the magnetic data for GdNi2 to the
model described by Shiga et al.45 in which exchange integrals
J, between Ni(II) and Gd(III) centers, and J′, between Ni(II)
centers, are considered. To minimize the number of potential
variables, we examined fits where gGd was fixed at 2.00, gNi
was fixed at 2.15, the value determined by analysis above
of the LaNi2 data, and TIP was fixed at 0.000 320 cm3 mol-1,
the same value of TIP used in the data analysis by Shiga et
al.45 Fixing J′ at -0.03 cm-1, the value determined above
for the exchange between Ni(II) centers in LaNi2 complex,
and allowing only J to vary, did not lead to acceptable fits.
To illustrate this, we have shown the best “fit” which was
obtained for J ) 4.5 (0.6) cm-1 as the solid black line in
Figure 10. Allowing for intermolecular coupling by incor-
porating a θ term (as T-θ) in the theoretical expression,
we were able to reproduce the maximum in ꢀMT near 13 K
as seen by the dotted black line in Figure 10. This curve,
which is still a very poor fit to the data, was the best fit
obtained where J and θ were allowed to vary. The values of
J and θ for this fit are 5.15 (0.3) cm-1 and -0.63 (0.07) K,
respectively.
Better modeling of the data for the GdNi2 complex was
obtained when we allowed both J and J′ to vary, as well as
θ. This modeling gave the best fit curve shown as the red
dotted line in the Figure 10 with J ) 11.0 (1) cm-1, J′ )
-16 (2) cm-1, θ ) -0.49 (0.04) K, and F ) 3.8 × 10-3.
While the theoretical curve closely models the data above
the maximum, the detail below the maximum is poorly
modeled. Setting θ to zero and fitting only the data above
the maximum (13-100 K) yielded the solid red line shown
The ꢀMTvalues at RT for [GdNi2(bcn)2]ClO4 · H2O,
[DyNi2(bcn)2]ClO4 · H2O, and [YbNi2(bcn)2]ClO4 were de-
termined to be 10.80, 17.48, and 5.05 cm3 K mol-1,
respectively, closely matching the calculated values (9.88,
16.17, and 4.57 cm3 K mol-1, respectively, using eq 1 and
Table 4) for three magnetically isolated metal ions. The ꢀMT
data for GdNi2 rise subtly reaching a maximum of 17.55
cm3 K mol-1 at 13 K and decrease slightly to reach a value
of 16.06 cm3 K mol-1 at 3 K (Figure 8). Although the
absolute reason for the observed maximum at 13 K is not
known, one reviewer suggested it may be due to intermo-
lecular antiferromagnetic interactions (further observed in
the modeling of the ꢀMT curve as discussed below) caused
by dipolar effects. The maximum value at 13 K is signifi-
cantly greater than the calculated value for the three isolated
metal ions of 9.88 cm3 K mol-1 and closely matches the
ꢀMT for a ferromagnetic coupling of the spins (ꢀMT ) 17.88
µB) calculated using eq 4.
ꢀ T ) (g S (S + 1) )1/2/2.828)2
(4)
[
]
M
T
T
ST
)
2SNi
+
SLn (SNi
)
1, SGd
)
7/2 for
[GdNi2(bcn)2]ClO4 · H2O), and g is the Landé factor (g )
2).
In addition, the M vs H curve follows the Brillouin
function for a ST ) 11/2 (SNi ) 1 and SGd ) 7/2), suggesting
that the ground state (ST ) 11/2), arising from a ferromag-
netic interaction, is significantly populated at 10 K (Figure
9). The [GdNi2(bcn)2]ClO4 · H2O system follows the trend
Inorganic Chemistry, Vol. 47, No. 7, 2008 2291

Barta et al.
Figure 11. ∆ꢀMT for [LnNi2(bcn)2]ClO4 ·nH2O complexes: (O) NdNi2;
(4) GdNi2; (]) DyNi2; (0) YbNi2.
Figure 10. Attempted fittings (text) of the ꢀMT vs T data for the
[GdNi2(bcn)2]ClO4 ·H2O complex (O).
in Figure 10 for which J ) 11.0 (0.4) cm-1, J′ ) -17 (1)
cm-1, and F ) 4.3 × 10-4. Importantly, both of these
analyses yield large positive values of J(11 cm-1) confirming
the presence of a relatively strong ferromagnetic exchange
between the Ni(II) and Gd(III) centers much larger than
previously observed. For comparison, Shiga et al. observed
J ) 0.79 cm-1 for their system.45 More surprisingly, perhaps,
is the large negative value of J′ obtained from the fits (J′ )
-16 to -17 cm-1).
In view of the difficulties encountered here in analyzing
the data for the GdNi2 complex and in obtaining satisfactory
experimental data at temperatures above 100 K, caution
should be applied in interpreting these results; particularly
the numerical values presented. Nonetheless the results do
point to some interesting general conclusions. There is clearly
evidence for significant ferromagnetic coupling between the
Gd(III) and Ni(II) centers. Moreover, it seems difficult to
ignore the significant antiferromagnetic exchange seen, likely
between the two Ni(II) centers. In view of the fact that the
La(III) analogue studied here clearly gives no evidence for
significant coupling between the Ni(II) centers, this latter
finding suggests that it is not reasonable to always assume
that coupling between two d-metal centers separated by an
f-metal is necessarily unaffected by what metal occupies the
central site. Furthermore, the presence of measurable anti-
ferromagnetic Ni-Ni interactions in combination with
significant ferromagnetic Ni-Gd coupling in such a
Ni-Gd-Ni system would represent a unique form of spin-
frustration. Future studies on this and related Ni/Gd systems
are clearly warranted by this work.
for two Ni(II) ions and the Dy(III) or the Yb(III)
ions.
The differences between the ꢀMT plots of
[LnNi2(bcn)2]ClO4 · nH2O and the analogous [LnZn2(bcn)2]-
ClO4 · nH2O species are shown in Figure 11. Compounds
LnNi2, where Ln ) Gd(III), Dy(III), Yb(III) displayed
∆(ꢀMT) values between 2 and 3 cm3 K mol-1 at 300 K which
correspond to the magnetic contributions of the two isolated
Ni(II) ions. Values of 8.86, 6.08, and 3.56 cm3 K mol-1 were
observed, respectively, after removing the first-order orbital
contributions of the Ln(III) at 2 K. These values are all larger
than ꢀMT ) 2 cm3 K mol-1 calculated for two Ni(II) ions (g
) 2.0), suggesting a ferromagnetic exchange. Alternatively,
after removing the Nd(III) contributions, NdNi2 shows a
minimum ∆(ꢀMT) value of 0.441 cm3 K mol-1, significantly
smaller than the contribution from two isolated Ni(II) ions
at 2 K (ꢀMT ) 2 cm3 K mol-1) confirming an antiferromag-
netic interaction between the TM(II) and Ln(III) ions.
Furthermore, the Ni(II) contribution can be removed by
using the analogous [LaNi2(bcn)2]ClO4 ·3H2O magnetic data
to minimize any second-order effects of the Ni(II) ion that
may complicate the analysis of the magnetic exchange using
eq 5.
JNiLn(T) ) (ꢀMT)LnNi - (ꢀMT)LnZn - (ꢀMT)LaNi
2
(5)
2
2
This empirical approach helps confirm the type of spin
interaction between the d/f/d metal ions, leaving only the
temperature dependent JNiLn(T) value where a negative value
suggests an antiferromagnetic exchange between metal ions
and a positive value suggests a ferromagnetic exchange.
Positive JNiLn(T) values were observed after removing Ni(II)
contributions from the ∆(ꢀMT) plot at 2 K for Gd(III),
Dy(III), and Yb(III) ions (5.70, 2.93, and 0.406 cm3 K mol-1,
respectively), further confirming a ferromagnetic exchange.
Alternatively, NdNi2 has a negative JNiLn(T) value at 2 K of
-2.72 cm3 K mol-1, further suggesting an antiferromagnetic
interaction; however, we would like to note that the conclud-
ing magnetic exchanges of the YbNi2 and NdNi2 complexes
are assigned with caution due to the slight increase or
decrease of the ∆(ꢀMT) plots at low temperatures.
The curve for [DyNi2(bcn)2]ClO4 · H2O decreases slightly
until 30 K where ꢀMT begins to decrease reaching a
minimum value of 14.55 cm3 K mol-1 at 5 K and a
maximum value of 14.85 cm3 K mol-1 at 3 K (Table 4).
A maximum ꢀMT value of 4.26 cm3 K mol-1 at 2 K is
determined for [YbNi2(bcn)2]ClO4. The ꢀMT values for
DyNi2 and YbNi2 (14.85 cm3 K mol-1 at 3 K; 4.26 cm3 K
mol-1 at 2 K, respectively) are below the calculated values
of 16.17 cm3 K mol-1 and 4.57 cm3 K mol-1, respectively,
2292 Inorganic Chemistry, Vol. 47, No. 7, 2008

Trimetallic d/f/d Complexes
Only four other trimetallic LnNi2 systems have been
reported;45 two of the systems were previously synthesized
in our group,36,41 and only the Ni-Gd-Ni species of the
other trimetallic LnNi2 complexes were analyzed.37 The
magnetic data for the [LnNi2(bcn)2]ClO4 · nH2O complexes
was maintained regardless of the TM(II) or Ln(III) ion.
[TM(Hbcn)]· nH2O, where the transition metal is encapsu-
lated by the three amine donors and three phenolate donors,
became a preorganized moiety used to sandwich the Ln(III)
ion between the six donating phenolate oxygens, three from
each [TM(bcn)]- ligand. Solvent molecules completed the
large coordination spheres intrinsic to lanthanides, forming
stable d/f/d complexes where the d- and the f-block metal
ions are in close proximity, allowing the study of their
magnetic exchange. Magnetic interactions in the
[LnNi2(bcn)2]ClO4 · nH2O complexes were determined by
removing the first-order orbital contributions of the Ln(III)
determined by the analogous [LnZn2(bcn)2]ClO4 ·nH2O spe-
cies and analyzing their resulting ꢀMT values using an
empirical approach similar to the data analysis of Costes et
al.31 The [LnNi2(bcn)2]ClO4 · nH2O complexes displayed
magnetic properties similar to those of other literature d- and
f-block species despite different geometric demands, where
an antiferromagnetic exchange was observed for the lighter
Ln(III) (Ln ) Nd(III)), while the heavier Ln(III) ions (Ln
) Gd(III), Dy(III), and Yb(III)) exhibited ferromagnetic
behavior. These conclusions are consistent with the predic-
tions of Kahn et al.44,46 where a ferromagnetic exchange is
observed with 4fn Ln(III) metal ions where n > 7 and an
antiferromagnetic interaction arises if n < 7. This analysis
is also used for the [LnCu2(bcn)2]ClO4 · nH2O species pub-
lished in the following paper50 allowing for the investigation
of the magnetic behavior upon varying both TM(II) and
Ln(III) while maintaining the same core structures. We
remain hopeful that more trimetallic d/f/d literature examples
with varied TM(II), Ln(III) and geometric demands will soon
be synthesized and magnetically characterized to further
understand the intrinsic d/f hetermetallic magnetic interac-
tions.
matched Okawa and co-workers LnNi2 system45 and that of
j
Bayly et al.41 where a ferromagnetic exchange was observed
when Ln ) Gd(III), Dy(III), and Yb(III) and an antiferro-
magnetic interaction when Ln ) Nd(III); however, this data
conflicts with the antiferromagnetic interaction determined
for similar amine phenol [LnNi2(tam)2]ClO4 · nH2O species,
where Ln ) Dy(III) or Yb(III).36
It has been proposed41 that the observation of the anti-
ferromagnetic interactions in the [LnNi2(tam)2]ClO4 · nH2O
species were due to a combination of diminishing paramag-
netic contributions of the Ln(III) ion at low temperatures, in
addition to saturation effects arising from the large applied
magnetic field (10 000 G) that could mask ferromagnetic
interactions between the d/f metal ions; however, the same
results as previously published where the ꢀMT measurements
decreased at lower temperatures for Ln ) Dy(III) and Yb(III)
were confirmed after repeating the ꢀMT measurements at a
lower magnetic field (500 G). The repeated experiments rule
out the probability of the unprecedented interactions being
a function of magnetic saturation. Thus, the cause of the
antiferromagnetic behavior is still undetermined. This odd
magnetic behavior of the [LnNi2(tam)2]ClO4 · nH2O com-
plexes is puzzling due to the similarities of the H3tam, H3trn,
and H3bcn amine phenol ligands.
The only major difference between the trimetallic
Ni-Ln-Ni structures is the variation in the intermetallic
bond lengths. Shorter Ni-Ln and Ni-Ni intermetallic
distances are observed for [LnNi2(bcn)2]ClO4 · nH2O com-
pared to Bayly’s Ni-Ln-Ni reduced Schiff base com-
plexes41 and Okawa’s 2,6-di(acetoacetyl)pyridine systems
j
(∼3.119 vs >3.53 Å for Ni-Ln distances; 5.98 vs >7.21 Å
for Ni-Ni distances, respectively).45 In addition, the Ni-Ni
intermetallic distances for the [LnNi2(tam)2]ClO4 · nH2O
complex were even shorter (∼5.4 Å). The shorter Ni-Ni
distances for the [LnNi2(tam)2]ClO4 · nH2O complexes may
give rise to an antiferromagnetic exchange between the two
terminal Ni(II) species masking any weak ferromagnetic
interaction between the Ln(III) and the terminal Ni(II) ions,
observed between the terminal Ni(II) ions in the
[GdNi2(bcn)2]+ complex discussed herein. Unfortunately,
attempts to remove the Ln(III) contributions of the
[LnNi2(tam)2]ClO4 · nH2O have been unsuccessful due to
failed efforts at synthesizing and crystallizing the analogous
[LnZn2(tam)2]ClO4 · nH2O species.
Acknowledgment. We thank Dr. Cecilia Stevens for her
expertise on magnetics and with the UBC SQUID. We
acknowledge Discovery Grant support from the Natural
Sciences and Engineering Research Council of Canada
(R.C.T. and C.O.), a University Graduate Fellowship (C.A.B.)
from the University of British Columbia, and a Royal Society
Postdoctoral Fellowship (S.R.B.). We also acknowledge one
of our reviewers for many helpful comments.
Supporting Information Available: Complete tables of
crystal parameters, an ORTEP diagram of the cation in
[YbNi2(bcn)2]ClO4 ·CH3CN, selected H1 NMR spectra, selected IR
data, and the temperature dependence plots of µeff for [LnNi2(tam)2]-
ClO4 · nH2O at 500 G, Ln ) Gd(III), Dy(III), and Yb(III).
This material is available free of charge via the Internet at
http://pubs.acs.org.
Conclusion
A series of TM(II)-Ln(III)-TM(II) complexes were
synthesized where a similar core structure of the complex
IC701612E
Inorganic Chemistry, Vol. 47, No. 7, 2008 2293