Cooperative thermodynamic control of selectivity in the self-assembly of rare earth metal-ligand helices

Article abstract of DOI:10.1021/ja409882k

Metal-selective self-assembly with rare-earth cations is possible with suitable rigid, symmetrical bis-tridentate ligands. Kinetically controlled formation is initially observed, with smaller cations preferentially incorporated. Over time, the more thermodynamically favorable complexes with larger metals are formed. This thermodynamic control is a cooperative supramolecular phenomenon and only occurs upon multiple-metal-based self-assembly: single-metal ML₃ analogues do not show reversible selectivity. The selectivity is dependent on small variations in lanthanide ionic radius and occurs despite identical coordination-ligand coordination geometries and minor size differences in the rare-earth metals.

Full text of DOI:10.1021/ja409882k

Communication
pubs.acs.org/JACS
Cooperative Thermodynamic Control of Selectivity in the Self-
Assembly of Rare Earth Metal−Ligand Helices
Amber M. Johnson, Michael C. Young, Xing Zhang, Ryan R. Julian, and Richard J. Hooley*
Department of Chemistry, University of CaliforniaRiverside, Riverside, California 92521, United States
S
* Supporting Information
ABSTRACT: Metal-selective self-assembly with rare-
earth cations is possible with suitable rigid, symmetrical
bis-tridentate ligands. Kinetically controlled formation is
initially observed, with smaller cations preferentially
incorporated. Over time, the more thermodynamically
favorable complexes with larger metals are formed. This
thermodynamic control is a cooperative supramolecular
phenomenon and only occurs upon multiple-metal-based
self-assembly: single-metal ML3 analogues do not show
reversible selectivity. The selectivity is dependent on small
variations in lanthanide ionic radius and occurs despite
identical coordination-ligand coordination geometries and
minor size differences in the rare-earth metals.
he synthesis of supramolecular cage complexes through
Tcoordination chemistry is a rapidly expanding field. By
combining metals (M) and ligands (L) with complementary
Figure 1. Synthesis of self-assembling ligand L1, control ligand L2, and
geometry, many types of self-assembled structures can be
bimetallic helices [13Ln2]6− (minimized structure, SPARTAN, AM1
designed and synthesized.1 Three-dimensional cages are
force field).
frequently targeted because their internal cavity can be exploited
in host/guest chemistry, including stabilization of reactive
species2 and biomimetic catalysis.3 With well-defined and
makes selective binding and extraction of Ln metals (e.g., from
spent nuclear waste streams) quite challenging.8 Controlled,
predictable coordination geometries, transition metals are most
often utilized in metal-mediated supramolecular self-assembly.4
metal-selective self-assembly of RE M−L complexes would be
Non-transition metals such as main-group and rare-earth (RE)
metals have a number of advantages (such as oxidative stability
and, in certain cases, diamagnetism) over transition metals in
forming self-assembled systems, but the field of rare-earth-
mediated self-assembly is far less explored. The roadblock is the
variable coordination geometry displayed by lanthanide (Ln)
metals, which makes predictable, controlled self-assembly rather
challenging. By using tridentate chelating ligands, nine-
coordinate complexes can be formed. Lanthanides have been
used most often in the synthesis of M2L3 helices,5 although
tetrahedra6 and other structures7 have been reported. In these
cases, the particular type of RE metal used in the self-assembly
process is often overlooked: each RE is treated the same in terms
of its assembly characteristics. Whereas transition metals display
different coordination numbers and geometries, the coordination
chemistry of the Ln series is relatively unchanged across the
series. Each RE metal strongly favors the +3 oxidation state and
commonly displays nine coordination sites. When attempting to
selectively bind different REs, one can only exploit the Ln
contraction and variations in effective ionic radius (EIR). The
differences are not large: La3+ and Yb3+ are only separated by 0.2
Å in size, and adjacent REs are exceptionally similar in EIR. This
applicable to both construction of functionalized cages and hosts
and remediation of nuclear waste effluents. Here we describe the
effect of self-assembly and cooperativity on the selective formation
of self-assembled Ln helices based on a rigid, symmetrical bis-
tridentate ligand.
The required tridentate coordination motif was provided by
salicylhydrazone-based ligands (Figure 1). Bis-tridentate L1 is
capable of assembly with two metals into supramolecular
aggregates, whereas L2 acts as a control, capable of coordinating
only a single metal. L1 was synthesized by refluxing known9
dihydrazide 3 with salicylaldehyde in ethanol and catalytic acetic
acid. L2 was synthesized from ethyl 4-bromobenzoate in two
steps (see Supporting Information (SI)). In their neutral forms,
neither L1 nor L2 showed any affinity for Ln(OTf)3 salts in
DMSO-d6. Slight, very weak coordination was observed between
L1 and Ln(OTf)3 salts in CD3CN. To increase its coordinative
capabilities, L1 was exhaustively deprotonated by treatment with
NaH. Tetraanionic 1 readily dissolved in DMSO-d6 and
displayed greatly enhanced coordinative properties. Simply
Received: September 24, 2013
Published: November 8, 2013
© 2013 American Chemical Society
17723
dx.doi.org/10.1021/ja409882k | J. Am. Chem. Soc. 2013, 135, 17723−17726

Journal of the American Chemical Society
Communication
Figure 3. Mass spectra and minimized structures (SPARTAN, AM1
force field) of (a) [13Y2]6− and (b) [12Yb2]2−.
complexes that were sufficiently soluble in THF for ESI-MS
analysis. For each of the tested [13Ln2]6− assemblies, the parent
[M2L3-4H]2+ ions could be obtained under ESI analysis in
positive mode. The acidic ionization conditions caused
protonation of the complexes, and singly and doubly charged
cations were observed. While the parent ion was observed in all
cases, the species were susceptible to significant fragmentation.
Loss of ligand was the most common fragmentation, and the
[M2L2-4H]2+ and [M2L2-5H]+ ions were commonly observed.
No higher stoichiometry (e.g., M4L6) aggregates were present,
and the experimental isotope patterns agreed with the predicted
M2L3 stoichiometry (see Figure 3 and SI). The [L13Ln2-4H]2+
cation from [13Y2]6− and [13Sm2]6− was subjected to MS/MS
analysis, resulting in fragmentation to the singly and doubly
charged M2L2 cations. This suggests that the M2L2 complexes
observed are a result of ligand loss under ionizing conditions and
not from incomplete assembly. Analysis of [12Yb2]2− gave no
ions arising from the M2L3 complex, consistent with its NMR
assignment. Despite extensive experimentation, single crystals of
the complexes could not be obtained. The charged species in
DMSO solution were susceptible to protonation with adventi-
tious water, and crystallization from less favorable solvents such
as THF complexes led to metal extrusion.
Figure 2. 1H NMR spectra of self-assembled lanthanide helical
complexes (400 MHz, DMSO-d6).
titrating a DMSO solution of Sm(OTf)3 into a solution of 1
assisted in determining the stoichiometry of the self-assembled
complex. After addition of 0.67 mol equiv of metal, no peaks for
ligand were observable by 1H NMR, and a single, discrete M2L3
complex [13Sm2]6− was observed (see Figure 1 for a minimized
structure).
The ligands can form self-assembled complexes with a range of
RE metals. The chosen metals (La, Pr, Sm, Yb, and lanthanide
surrogate Y) are spread across the Ln series and provide a range
of ionic radii (Y is similar in size to Ho). Highly paramagnetic
1
metals such as Gd and Dy were avoided to allow H NMR
While each of the tested REs was capable of forming self-
assembled helices, the ligand showed significant selectivity
between differently sized metals. Two types of experiments were
performed to illustrate the selectivity: initial kinetic preference
via a displacement test and thermodynamic preference after
equilibration. The initial kinetic selectivity was tested by titrating
a DMSO-d6 solution of LnA(OTf)3 into a solution of preformed
analysis. Titration data were consistent for all the larger metals
(La, Pr, Sm, Y) tested. The only deviation from the M2L3
stoichiometry was observed with the smaller REs such as Yb.
When Yb(OTf)3 was titrated into a solution of 1, a full equivalent
of metal was required for complete formation of complex,
suggesting [12Yb2]2− is formed in this case. The titration data fit
with the predicted structure of M2Lx helices. Triple-stranded
helices are formed with the larger metals, with nine coordination
sites of the Ln being filled. The smaller Yb is less disposed to fit
three tridentate ligands in its coordination sphere and instead
forms a double-stranded M2L2 complex. Figure 2 shows the 1H
NMR spectra of the Ln complexes. Diamagnetic Y and La and
weakly paramagnetic Sm appear as expected in the aromatic
region, while paramagnetic Pr and Yb shift the complex
resonances strongly downfield and upfield, respectively.
Further characterization was provided by diffusion NMR.
DOSY spectra were taken for [13Y2]6− and [13Sm2]6− in DMSO-
d6. The observed diffusion coefficient for [13Y2]6− was 7.24 ×
10−11 m2/s, and that of [13Sm2]6− was 8.13 × 10−11 m2/s,
indicating that the two complexes display almost identical overall
size and charge properties (for spectra see SI). No change in the
1H NMR spectra was observed upon addition of different
counteranions. Use of different LnX3 salts (such as chloride or
nitrate) also gave identical complexes, indicating that the Ln
centers are saturated by 1, with no interaction with the
counterions.
[13LnB ]6− and determining the extent of displacement of the
2
1
first metal by the second from H NMR integration. This
displacement assay showed a clear correlation between EIR of
the metal and the preference for displacement, with smaller
radius cations being favored (for full spectra and tabulation, see
SI). Almost complete displacement of La3+ was observed upon
titrating 0.67 mol equiv of Y(OTf)3 (EIR = 1.02 Å) to [13La2]6−
(see SI). When La(OTf)3 (EIR = 1.18 Å) was added to [13Y2]6−,
no displacement was observed. Interestingly, no heterometallic
[13YLa]6− was observed in this case. With other metal
combinations that had smaller variance in EIR, lower overall
selectivity and the presence of heterometallic complexes were
observed. While overlapping signals plagued the analysis of
heterometallic complexes of two diamagnetic metals (Y, La, Sm),
paramagnetic Pr3+ and Yb3+ acted as shift reagents, separating the
signals and aiding identification. Figure 4 shows the correspond-
ing NMR spectra, as well as the selectivity and EIR data.
To determine whether this kinetic selectivity was a supra-
molecular effect or merely a characteristic of the coordinating
ligand, control experiments were performed with 2. When 0.33
mol equiv of Ln(OTf3) was added to a DMSO-d6 solution of 2, a
clean ML3 complex [23Ln]3− was formed. ESI-MS analysis
While the complexes formed with 1 were only soluble in
DMSO, treatment of L1 with potassium tert-butoxide gave
17724
dx.doi.org/10.1021/ja409882k | J. Am. Chem. Soc. 2013, 135, 17723−17726

Journal of the American Chemical Society
Communication
product was that of the larger metal complex. As Δ(EIR)
increased, the thermodynamic favorability for incorporation of
the larger metal increased. The ligand was again strongly selective
for the Y/La combination, which displays a Δ(EIR) between the
metals of only 0.16 Å. Smaller Δ(EIR) led to observation of
heterometallic complexes, and the concentration of
[13LnALnB]6− was also dependent on Δ(EIR), with a greater
size difference resulting in a smaller proportion of heterometallic
complex. The observed selectivity with our rigid anionic
coordinators is much greater than that observed with flexible,
neutral bis-tridentate Ln chelators.11 In that case, very similar
proportions of (LnA)2, (LnB)2, and (LnAB) complexes were
formed upon addition of equimolar amounts of different
metals.11
The inverted selectivity upon equilibration phenomenon is
truly a supramolecular effect and requires a self-assembling ligand
to occur. When ML3 complexes were formed from monometallic
control ligand 2, essentially no equilibration was observed, even
after extensive periods of time. In this case, both the kinetic and
thermodynamic preferences for metal complexation are
determined by the electronic coordination properties of the
ligand alone. This favors small metals with a larger charge:size
ratio (see Figure 5d−f).
The presence of the second coordination site in a rigid ligand
such as 1 adds another dimension to the selectivity of self-
assembly. Coulombic interactions are no longer the sole
determining factor in coordination: the added strain upon the
rigid coordinating ligand must be considered. The flexible
Figure 4. Identification of heterometallic complexes: (a) [13Pr2]6−; (b)
[13Sm2]6−; and (c) mixture of [13Pr2]6− (red), [13Sm2]6− (blue), and
[13PrSm]6− (green). Below: percentage of complexes after displace-
ment titrations and effective ionic radii of Ln3+ ions.
showed the presence of the [LnL23-2H]+ parent ion, along with
fragmentation into the [LnL22-2H]+ ion, analogous to that
observed with 1. MS/MS analysis of the [LnL23-2H]+ parent ion
showed fragmentation of the ML3 complex into the [LnL22-
2H]+ ion. The monometal [23Ln]3− complexes showed the same
kinetic preference for metals with smaller ionic radii as the
bimetallic helices [13Ln2]6− according to displacement experi-
ments. The relationship between the difference in EIR of the
added metals and the observed concentration of [13Ln-
(small)2]6− upon displacement showed a strongly linear
correlation (see SI). As Δ(EIR) increased, so did the selectivity
for the smaller metal. The kinetic selectivity for smaller metals is
evidently a charge-based phenomenon. As the metal EIR
decreases, the charge:size ratio increases, as does the affinity of
the anionic ligand for the smaller metal.
While these displacement experiments described the kinetic
preference of the Ln−L complexes, far more unusual behavior
was observed upon equilibration of the self-assembled Ln helices.
To test equilibration, an equimolar mixture of metals was added
1
to a solution of 1, and H NMR spectra were periodically
acquired over 30 h. Figure 5 shows the percentage of each
complex (both homometallic [13LnA ]6− and [13LnB ]6−, and
2
2
heterometallic [13LnALnB]6−) vs time with mixtures of Y3+, La3+,
and Sm3+ (for complete analysis see SI). In all samples, the
initially formed kinetic complex was that favored in the
displacement experiments; i.e., the smaller cation was preferen-
tially bound. Over time, the selection preference inverted: for the
highly selective Y/La mixture, the kinetically disfavored
[13La2]6− was observed after 3 h equilibration. Equilibration
was complete after 24 h, and only the larger [13La2]6− complex
was present. The selectivity of self-assembly after equilibration
again depended on Δ(EIR) of the cations, but the
thermodynamically favored complexes are formed with the
larger metal. The Y3+/La3+ mixture (Figure 5a) displays the
largest Δ(EIR), and the initial 96:4:0 ratio of [13Y2]6−:[13La2]6−:
[13LaY]6− complexes was inverted to 6:94:0 after equilibration,
preferentially forming the larger [13La2]6− complex. In the Y3+/
Sm3+ mixture (Figure 5b), with a smaller Δ(EIR), the initial ratio
was 50:9:4, and after equilibration it was 17:37:46. The selection
process after equilibration also showed a direct, linear depend-
ence on Δ(EIR) (for graphs, see SI). Whereas the initial
selectivity favored smaller metal binding, the thermodynamic
Figure 5. Equilibration of lanthanide complexes: (a) 13Y2 vs 13La2; (b)
13Y2 vs 13Sm2; (c) 13Sm2 vs 13La2; (d) 23Y vs 23La; (e) 23Y vs 23Sm; and
(f) 23Sm vs 23La (DMSO-d6, 298 K).
coordination sphere of RE metals allows ligands to shift into
more favorable conformations around the metal to minimize
strain and steric congestion between ligands.12 In the [13La2]6−
complexes, the two coordination sites are connected by rigid
aromatic rings, restricting conformational flexibility and adding
strain to the complex. Larger metals provide more space for a
favorable conformation to be reached and so are preferentially
coordinated, if equilibration can be reached. The use of rigid,
17725
dx.doi.org/10.1021/ja409882k | J. Am. Chem. Soc. 2013, 135, 17723−17726

Journal of the American Chemical Society
Communication
AUTHOR INFORMATION
Corresponding Author
richard.hooley@ucr.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
The authors thank the NSF (CHE-1151773 to R.J.H., CHE-
0747481 to R.R.J.) and UC Riverside for funding.
■
Figure 6. Plots of percent complex vs time: (a) [13Y2]6− and [13La2]6−
in samples containing 50 mM water (dry) and 150 mM water (wet); (b)
23Y and 23La in samples containing 100 mM water (dry) and 300 mM
water (wet).
REFERENCES
■
(1) Fujita, M.; Umemoto, K.; Yoshizawa, M.; Fujita, N.; Kusukawa, T.;
Biradha, K. Chem. Commun. 2001, 509.
(2) Fiedler, D.; Bergman, R. G.; Raymond, K. N. Angew. Chem., Int. Ed.
2006, 45, 745.
(3) Brown, C. J.; Bergman, R. G.; Raymond, K. N. J. Am. Chem. Soc.
2009, 131, 17530.
strained ligands to control assembly in transition metal complexes
is precedented,1,13 but it is rarely observed for coordinatively
flexible RE metal complexes. 1 is a departure from the more
typical Ln-coordinating ligands that contain flexible methylene
spacers and show little to no selectivity between metals unless
“built in” by the incorporation of different coordinating motifs.14
Interestingly, the presence of water in the sample has a large
effect on the equilibration process. Water is well-known to
reversibly coordinate to Gd-based MRI contrast agents and is
strongly correlated with their contrast and relaxivity.15 For the
[13La2]6− complexes, increased [H2O] caused faster equilibra-
tion to the thermodynamic product. The data are most striking
for the Y3+/La3+ system: Figure 6 shows the change in proportion
of [13Y2]6− and [13La2]6− vs time, along with the control of
[23Y]3−: [23La]3− in samples containing varying amounts of
water. In the presence of 50 mM water, very slow equilibration of
[13Y2]6− to [13La2]6− occurs and the equilibration is not
complete after multiple days. Additional water ([H2O] = 150
mM) caused a much faster change in the equilibrium, and the
graph is similar to Figure 5a. It should be noted that [H2O] in all
equilibration experiments in Figure 5 and the SI was identical.
As would be expected, monometallic control complex [23Y]3−
showed no effect from the presence of additional water. There
was a slight shift in selectivity upon initial complexation, but no
significant equilibration was observed, lending additional
evidence for cooperativity in equilibration. The favorable,
reversible coordination of water to the Ln centers allows more
rapid on/off exchange between the ligands and speeds up the
equilibration process. In the case of the nonequilibrating ligand 2,
this has essentially no effect but is an important component of the
cooperative selectivity of 1 for different RE metals.
(4) Young, N. J.; Hay, B. P. Chem. Commun. 2013, 49, 1354.
(5) (a) Piguet, C.; Bernardinelli, G.; Hopfgartners, G. Chem. Rev. 1997,
97, 2005. (b) Stomeo, F.; Lincheneau, C.; Leonard, J. P.; O’Brien, J. E.;
Peacock, R. D.; McCoy, C. P.; Gunnlaugsson, T. J. Am. Chem. Soc. 2009,
131, 9636. (c) Shi, J.; Hou, Y.; Chu, W.; Shi, X.; Gu, H.; Wang, B.; Sun,
Z. Inorg. Chem. 2013, 52, 5013. (d) Jensen, T. B.; Scopelliti, R.; Bunzli, J.-
̈
C. G. Chem.Eur. J. 2007, 13, 8404. (e) Piguet, C.; Bunzli, J.-C. G. In
̈
Handbook on the Physics and Chemistry of Rare Earths; Gschneider, K. A.,
Jr., Bunzli, J.-C. G., Pecharsky, V. K., Eds.; Elsevier: Amsterdam, 2010;
̈
Vol. 40, pp 301−553. (f) Albrecht, M.; Osetska, O.; Frolich, R.; Bunzli,
̈
̈
J.-C. G.; Aebischer, A.; Gumy, F.; Hamacek, J. J. Am. Chem. Soc. 2007,
129, 14178. (g) Ryan, P. E.; Guenee, L.; Piguet, C. Dalton Trans. 2013,
42, 11047.
(6) (a) Wang, J.; He, C.; Wu, P.; Wang, J.; Duan, C. J. Am. Chem. Soc.
2011, 133, 12402. (b) El Aroussi, B.; Guenee, L.; Pal, P.; Hamacek, J.
Inorg. Chem. 2011, 50, 8588.
(7) (a) Chen, X.-Y.; Bretonnier
́
́
́
́
̀
́
e, Y.; Pecaut, J.; Imbert, D.; Bunzli, J.-
̈
C.; Mazzanti, M. Inorg. Chem. 2007, 46, 625. (b) El Aroussi, B.; Zebret,
S.; Besnard, C.; Perrottet, P.; Hamacek, J. J. Am. Chem. Soc. 2011, 133,
10764.
(8) (a) Gorden, A. E. V.; DeVore, M. A., II; Maynard, B. A. Inorg. Chem.
2013, 52, 3445. (b) Lewis, F. W.; Hudson, M. J.; Harwood, L. M. Synlett
2011, 18, 2609. (c) Paiva, A. P.; Malik, P. J. Radioanal. Nucl. Chem. 2004,
261, 485. (d) Panak, P. J.; Geist, A. Chem. Rev. 2013, 113, 1199.
(9) Johnson, A. M.; Young, M. C.; Hooley, R. J. Dalton Trans. 2013, 42,
8394.
(10) CRC Handbook of Chemistry and Physics, 87th ed.; Lide, D. R., Ed.;
CRC Press: Boca Raton, FL, 2006; Sect. 4, p 132.
(11) Piguet, C.; Bunzli, J.-C. G.; Bernardinelli, G.; Hopfgartner, G.;
̈
Williams, A. F. J. Am. Chem. Soc. 1993, 115, 8197.
(12) (a) Petoud, S.; Bunzli, J.-C. G.; Renaud, F.; Piguet, C.; Schenk, K.
̈
J.; Hopfgartner, G. Inorg. Chem. 1997, 36, 5750. (b) Le Borgne, T.;
́
Benech, J.-M.; Floquet, S.; Bernardinelli, G.; Aliprandini, C.; Bettens, P.;
Piguet, C. Dalton Trans. 2003, 3856.
(13) (a) Young, M. C.; Johnson, A. M.; Gamboa, A. S.; Hooley, R. J.
Chem. Commun. 2013, 49, 1627. (b) Ousaka, N.; Grunder, S.; Castilla, A.
M.; Whalley, A. C.; Stoddart, J. F.; Nitschke, J. R. J. Am. Chem. Soc. 2012,
134, 11528.
In conclusion, we have shown that self-assembled M2L3
complexes can discriminate among lanthanide ions with a kinetic
preference for smaller metals and a thermodynamic preference
for larger metals. Selectivity is obtained despite small differences
in Ln ion size and identical coordination environment of the
ligand, and a correlation is observed between distribution of
complex and difference in ionic radius. Cooperative effects are
observed in the bis-tridentate ligand, and the presence of two
coordination sites assists in equilibration to the larger complex.
Further studies of lanthanide- and actinide-based self-assembly
are underway in our laboratory.
(14) (a) Sorgho-Aboshyan, L.; Nozary, H.; Aebischer, A.; Bunzli, J.-C.
̈
G.; Morgantini, P.-Y.; Kittilstved, K. R. J. Am. Chem. Soc. 2012, 134,
12675. (b) Albrecht, M.; Osetska, O.; Bunzli, J.-C. G.; Gumy, F.;
̈
Frohlich, R. Chem.Eur. J. 2009, 15, 8791. (c) Zeckert, K.; Hamacek, J.;
̈
Rivera, J.-P.; Floquet, S.; Pinto, A.; Borkovec, M.; Piguet, C. J. Am. Chem.
Soc. 2004, 126, 11589. (d) Riis-Johannessen, T.; Bernardinelli, G.;
Filinchuk, Y.; Clifford, S.; Favera, N. D.; Piguet, C. Inorg. Chem. 2009,
48, 5512.
(15) Caravan, P.; Ellison, J. J.; McMurry, T. J.; Lauffer, R. B. Chem. Rev.
1999, 99, 2293.
ASSOCIATED CONTENT
* Supporting Information
Experimental details and characterization data. This material is
available free of charge via the Internet at http://pubs.acs.org.
■
S
17726
dx.doi.org/10.1021/ja409882k | J. Am. Chem. Soc. 2013, 135, 17723−17726