Paramagnetic NMR Analysis of Substituted Biscyclooctatetraene

May 23, 2016 - ... almost all 1H and 13C NMR signals could be detected and assigned. ... Angewandte Chemie International Edition 2017 56 (25), 7238-72...
0 downloads 0 Views 712KB Size
Article pubs.acs.org/Organometallics

Paramagnetic NMR Analysis of Substituted Biscyclooctatetraene Lanthanide Complexes Markus Hiller, Martin Maier, Hubert Wadepohl, and Markus Enders* Institute of Inorganic Chemistry, Heidelberg University, Im Neuenheimer Feld 270, 69120 Heidelberg, Germany S Supporting Information *

ABSTRACT: Cyclooctatetraene derivatives (COTR) with two carbocycles attached to the central ring have been used as dianionic ligands for the synthesis of double-decker complexes of the type [(COTR)2M]−. Two ligand derivatives were combined with diamagnetic Y3+ and with the five paramagnetic lanthanide ions from Tb3+ to Tm3+. The more complex substitution pattern in comparison to the parent ligand COT or the popular bistrimethylsilyl derivative allows a sufficient number of signals to be obtained for a comprehensive paramagnetic NMR analysis. The anionic double-decker complexes gave well-resolved NMR spectra where almost all 1H and 13C NMR signals could be detected and assigned. With these data, it was possible to separate the two main contributions to the paramagnetic shift (pseudocontact and Fermi contact shifts, respectively) and to determine the magnetic anisotropy of the lanthanide ions in their ligand fields. We can easily obtain the sign and the magnitude of the anisotropy of the magnetic susceptibility, which is itself strongly related to the energy barrier for spin reversal in single-molecule magnets. Our results confirm that Bleaney factors are inadequate descriptors for magnetic anisotropy in these lanthanide complexes.



INTRODUCTION NMR analysis of paramagnetic compounds has been applied since the very beginning of NMR spectroscopy, and a broad theoretical basis was developed many years ago.1 In lanthanide complexes, the dominating contribution to the paramagnetic NMR shift is the pseudoconatct shift (δpc) because large magnetic anisotropies of several Ln3+ ions lead to strong dipolar interactions between the unpaired electrons and the NMR nuclei, whereas the small overlap of the contracted f-shell electrons with ligand orbitals renders the Fermi contact term (δcon) small. However, for a detailed analysis of the magnetic properties of such compounds, both contributions have to be taken into account; therefore, it is necessary to separate them. Several techniques for this task have been used in the past. Separation attempts assuming different temperature dependences of the Fermi contact and pseudocontact shifts (T−1 and T−2, respectively) were found to be unreliable. The different contributions can also be calculated by DFT methods; in particular, Fermi contact shifts of d-block metal complexes have often been analyzed in this way.2 In the case of lanthanides, this method has rarely been used, but recently methodology for the theoretical treatment of f-block elements has been developed.3 The similar size and chemical behavior of lanthanide ions allows another approach to be used that does not require knowledge of the structure of the compounds. By neglecting effects of contraction along the lanthanide series, in a first approximation identical structures for compounds of adjacent lanthanide ions can be assumed. This results in identical spin densities for equivalent nuclei, giving Fermi contact shifts © XXXX American Chemical Society

directly proportional to the reduced spin expectation values ⟨SZ⟩.4 If enough data points are available, then the magnetic susceptibility anisotropies of the individual lanthanide ions can be obtained by evaluation of the remaining shifts. This method was used herein in combination with the assumption of purely axial anisotropy, i.e., absence of rhombicity. This assumption is justified first by the high symmetry of the coordinating C8 ring. Furthermore, it has been found that possible rotations often lead to axially anisotropic systems, even if the required symmetry is formally absent.5 In NMR studies of related substituted uranocene derivatives, barriers limiting such rotations have been observed only at very low temperatures.6 By using the structure-independent method, not only can a reliable separation of δcon and δpc be achieved but also the magnetic anisotropy of the studied molecule or ion in solution can be determined simultaneously. This parameter is very important for the phenomenon of single-molecule magnet (SMM) behavior as it is related to the energy barrier for the reversal of the magnetic moment of SMMs. In recent years, it has become evident that magnetic anisotropy is inversely proportional to the total spin value of a SMM;7 therefore, the focus of SMM research has now shifted from large polynuclear systems to small molecules with only one or a few magnetic centers.8 Such complexes usually remain intact in solution and are therefore ideally suited for paramagnetic solution NMR analysis. However, only recently has this technique been used Received: March 24, 2016

A

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

Scheme 1. Synthesis of Ligands hdcCOT and odbCOT and Their Complexes Li[M(hdcCOT)2] and Li[M(odbCOT)2] Together with the Isolated Yields [%]a

a

hdcCOT = hexahydrodicyclopentacyclooctatetraene; odbCOT = octahydrodibenzocyclooctatetraene. Only one of the isomers of the odbCOT proligand is displayed for clarity.

for studying SMMs.9 Double-decker complexes with two dianionic cyclooctatetraene (COT) ligands have been known for a long time,10 and SMM behavior has been recently demonstrated for some complexes of this ligand class.11 NMR active 13C nuclei are directly attached to the lanthanide ion, and 1 H nuclei are only two bonds apart so that the fast NMR relaxation behavior makes NMR detection difficult. In addition, the small number of inequivalent NMR nuclei gives only a small set of data, which prohibits the extraction of detailed structural and magnetic data. We would like to improve paramagnetic solution NMR spectroscopy for studying complexes with large magnetic anisotropies in general and SMMs in particular; therefore, we synthesized substituted cyclooctatetraene-lanthanide complexes and analyzed their NMR data.

chosen in this investigation. Y3+, with an ionic radius closest to that of Ho3+,13 served as a diamagnetic reference. After the extraction described above, the compounds were usually obtained as the monoTHF adducts, as judged by elemental analysis. In the case of Li[Ho(odbCOT)2], prolonged reaction with acetonitrile led to the isolation of the monoacetonitrile adduct. Addition of hexane to the toluene/acetonitrile solution precipitated bisacetonitrile adducts in the cases of Li[Tb(odbCOT)2] and Li[Tm(odbCOT)2]. A corresponding Yb3+ compound could not be obtained; the formation of a precipitate was observed, possibly indicating the reduction of Yb3+ followed by formation of polymeric [YbII(odbCOT)]n.14 The compounds are insoluble in hexane and very poorly soluble in toluene, dioxane, and 1,2-dimethoxyethane (DME). Solubility in acetonitrile is limited but sufficient for NMR investigations. Good solubility was observed only in THF. However, in THF-d8, significantly larger line widths were observed, an indication of the formation of contact ion pairs. 7 Li NMR in THF shows large pseudocontact shifts due to the close proximity of the Li+ ions to the paramagnetic Ln center, whereas in acetonitrile-d3, solvent-separated ion pairs are present. Therefore, this solvent was chosen for the NMR analysis. Slow diffusion of DME into a THF solution of Li[Tb(odbCOT)2] yielded crystals that were suitable for singlecrystal X-ray diffraction (Figure 1). The compound crystallized as [Li(DME)(THF)][Tb(odbCOT)2] with two independent molecules per unit cell. One of the four anellated ring substituents of each molecule has a boat-like conformation, and the other three have chair-like conformations, which are partially found in two different arrangements. The lithium counterion is coordinated by one DME and one THF molecule and shows a close contact to one of the COT rings. This kind of coordination has previously been reported for related compounds.15 The distances of the COT centroids to the Tb3+ ion are very similar, with values from 1.869 to 1.929 Å, and the average centroid-Tb-centroid angle is 177.3°. These numbers are comparable to values found in the literature for [Li(DME)3][Er(COT″)2]11b (1.89 Å, 178.2°). NMR Analysis. The paramagnetism of the investigated complexes induces fast NMR relaxation and therefore considerable line broadening. However, in comparison to many paramagnetic d-block complexes, the fast electron relaxation of the Ln3+ ions leads to relatively small NMR line widths for Ln3+ complexes, with the exception for Gd3+.16 Increased temperature further reduces the NMR line widths; therefore, all NMR investigations in this article have been conducted at 340 K. With a few exceptions, all expected 1H and



RESULTS AND DISCUSSION We selected 1,2,5,6-tetrasubstituted cyclooctatetraenes with two condensed five-membered (hdcCOT = hexahydrodicyclopentacyclooctatetraene) and six-membered (odbCOT = octahydrodibenzo-cyclooctatetraene) rings, respectively, as these substituents have reduced flexibility compared with that of linear alkyl chains, which is advantageous for the envisaged NMR analysis. The synthesis of these molecules has been reported by Wender et al.,12 who utilized the Ni0-catalyzed dimerization of terminal dialkynes. To the best of our knowledge, the resulting 1,2,5,6-tetrasubstituted cyclooctatetraenes have not been used as dianionic ligands so far. Both proligands were obtained in moderate to good yields (odbCOT: 50%; hdcCOT: 66%) from commercially available dialkynes. Whereas hdcCOT is obtained as a single isomer (double bonds as shown in Scheme 1), two double-bond isomers of odbCOT are obtained in the reaction. However, this mixture gives a uniform product upon reduction. Synthesis of the lanthanide compounds was achieved as sketched in Scheme 1 by reaction of the COT derivatives with lithium metal and subsequent addition of the corresponding anhydrous lanthanide trichloride. The target compounds could be extracted with a mixture of toluene and acetonitrile, which allowed separation from lithium chloride, which is slightly soluble in THF/toluene mixtures. The identity of all compounds was confirmed by elemental analysis and 1H and 13C NMR spectroscopies. We used the lanthanide series (Tb3+−Yb 3+), which corresponds to an f-shell that is more than half filled, with 8−12 f electrons. These ions show larger magnetic anisotropies compared to those of the f1−f7 series and have therefore been B

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

below and eq 4). The magnetic anisotropy is a property of the whole molecule and therefore is the same for all nuclei in a given complex. Consequently, the chemical shift of nucleus i is proportional to the so-called geometric factor Gi, which is large for atoms that are both close to the metal and close to the 90° or 0° angle abc (with a = COT center, b = Ln3+ ion, and c = nucleus i). Thus, C3 should exhibit higher shifts than C4, whereas inner groups H3i and H4i should be significantly more shifted than H3o and H4o. As the H3i and H3o positions are closer to the lanthanide ion, their line widths are significantly larger than those of the H4i and H4o groups. The line widths of these signals are dominated by dipolar interactions with the unpaired electrons (r−6 dependence1a,b). Due to the close proximity of the lanthanide ion to C1, C2, and H1 (in terms of distance and number of bonds), these signals are severely broadened and, in some cases, their detection was impossible. Dy3+ produces only small shifts, indicating small magnetic susceptibility anisotropy. As a result, the signals lie in a narrow range and one signal (probably H3o) is hidden by the residual solvent signal in the case of Li[Dy(hdcCOT)2]. By the above procedure, most signals can be assigned unambiguously, but a few assignments remain doubtful (signals C1 and C2). However, the three nuclei plots (see text below and Figure 3) allow the

Figure 1. Molecular structure of one of the two independent molecules in the lattice of [Li(DME)(THF)][Tb(odbCOT)2] (left) and two views of the anion with the lithium counterion omitted (right). Thermal ellipsoids are drawn at the 50% probability level. Hydrogen atoms have been omitted, and for disordered substituents, only one arrangement is shown for clarity. Please note the different conformations of the cyclohexene substituents visible on the right side. Selected bond lengths (Å; values in square brackets refer to the second independent molecule): Li51−O51: 1.938(6) [1.933(5)]; Li51−O52: 1976(6) [2.035(5)]; Li51−O53: 2.003(7) [2.004(5)]; Li51−C67: 2.383(7) [2.382(5)]; Li51−C74: 2.343(7) [2.362(5)]; Tb51−C68: 2.718(3) [2.682(3)]; Tb51−C67: 2.690(3) [2.717(2)]; Tb51−C74: 2.628(2) [2.661(3)]; Tb51−C73: 2.654(3) [2.628(3)]; Tb51−C72: 2.715(3) [2.682(3)]; Tb51−C55: 2.661(3) [2.633(3)]; Tb51−C56: 2.627(3) [2.671(2)]; Tb51−C57: 2.587(3) [2.632(2)]; Tb51−C58: 2.613(3) [2.605(3)]; C68−C67: 1.416(4) [1.419(3)]; C67−C74: 1.425(4) [1.425(3)]; C74−C73: 1.421(3) [1.420(3)]; C73−C72: 1.421(3) [1.420(3)]; C55−C56: 1.420(4) [1.421(3)]; C56−C57: 1.421(5) [1.411(3)]; C57−C58: 1.413(5) [1.412(4)]. More details can be found in the Supporting Information. 13

C NMR signals could be observed, including the COTR ring carbon atoms C1 and C2 for most complexes (not for the Er3+ and Tm3+ compounds) and their corresponding 1H resonances. For both ligands, a maximum of five 1H and four 13C NMR signals indicates a relatively high-symmetric molecule. Schematic representations and the numbering of the individual positions in the metal complexes of both ligands are given in Figure 2. The basis for a detailed NMR analysis is an unambiguous assignment of the NMR signals. This was successfully performed in a multistep process: First, we evaluated positions where the Fermi contact contribution must be small as the nuclei are separated by at least three bonds from the lanthanide ion. For these nuclei, the relative hyperfine shift is determined by the pseudocontact shift so that geometric considerations allow the estimation of a relative shift order (see discussion

Figure 3. Three-nuclei-plot of Li[Ln(hdcCOT)2] according to Reuben21 with linear fits for the individual positions. The Ho3+ ion was chosen as reference, and parameter Y was calculated accordingly as YLn,iHo = δHFLn,i(⟨SZ⟩Ho/⟨SZ⟩Ln) − δHFHo,i. The values plotted on the x axis belong to the H4o position, with all other values plotted on the y axis.

final assignment of the remaining signals as the plot can be done using only the assured signals in a first step and adding the doubtful signals afterward. Only when the correct assignment is chosen do both signal series give linear

Figure 2. Schematic representations of the [M(hdcCOT)2]− and [M(odbCOT)2]− anions. All nonlabeled positions are equivalent to the labeled ones. C

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

above). If these assumptions are correct, then a plot of δHF/ ⟨SZ⟩Ln versus CBLn/⟨SZ⟩Ln should give a straight line. This method, which was introduced by Reilley,19 can be used not only for the determination of isostructurality but also for a separation of Fermi contact and pseudocontact shift. However, this simplified approach relies on the Bleaney factors that have been derived for a specific crystal-field environment. There are a number of reports that show limitations of the applicability of Bleaneys theory, and the corresponding plots (see the Supporting Information) reveal that the compounds investigated herein are poorly described by Bleaney factors.4b,20 This is in line with recent ab initio calculations which have shown that the crystal field parameter B40 is considerably large in [Ln(COT)2]− complexes. The use of Bleaney factors can be avoided when more NMR data are available, leading to plots with the so-called “three-nuclei method”.21 Application of the latter gives excellent isostructurality plots for all positions (Figure 3). The reason for the very shallow slope of the H3o data points is the small angle of the vector Ln3+−H3o versus the magnetic axis, which is close to 54.74°, leading to a very small geometric factor Gi. The combination of eqs 2−4 leads to eq 5, which describes the hyperfine shifts of all nuclei.

correlations (i.e., one line for C1 signals and a second for C2 signals), whereas with the wrong assignments, they do not. On the basis of the signal assignment, the hyperfine shift δHFLn,i of nucleus i in a Ln3+ complex was calculated by subtraction of the observed chemical shifts of the Y3+ complex from the corresponding observed chemical shift of the Ln3+ complex according to eq 1 Ln, i Ln, i Y, i δ HF = δobs − δobs

(1)

The resulting hyperfine shifts δHF are interpreted as the sum of the Fermi contact shift δcon and the pseudocontact shift δpc (eq 2). Ln, i Ln, i δ HF = δcon + δpcLn, i

(2)

Contact shifts are expected to be rather small for lanthanide ions with the exception of atoms very close (in terms of bonds) to the lanthanide ion (C1, C2, and possibly C3 and H1). As the substituents are attached only by σ-bonds, proliferation of spin density along the substituent backbone should be weak. Hence, contact contributions are expected to decrease with increasing number of bonds from the metal ion. The contact contribution of lanthanide ions can be described by eq 3 μ 2πAi 6 Ln, i = ⟨Sz⟩Ln B 10 δcon 3kTγI h (3)

Ln, i = ⟨Sz⟩Ln δ HF

= Pi⟨SZ⟩Ln + Q Ln·Gi′

where Ai represents the hyperfine coupling constant to a specific nucleus i and γI is the gyromagnetic ratio of the investigated nucleus. ⟨SZ⟩Ln is the mean reduced value of the average spin polarization for the lanthanide ion. This property has been calculated for the individual Ln3+ ions at different temperatures, and the values used herein are directly taken from the literature (see also Table S6, Supporting Information).4a If the investigated compounds are isostructural, then the hyperfine coupling constants Ai for different Ln3+ ions should not vary, giving contact shift values that are directly proportional to the ⟨SZ⟩Ln value of the corresponding ion. The pseudocontact shift, which is expected to dominate the overall hyperfine shift, is linked to the axial and rhombic anisotropies of the magnetic susceptibility (Δχax and Δχrh, respectively). As the coordination environment of the Ln3+ ions in the studied compounds is axial in the first coordination sphere, no significant rhombic component is expected. In this case, the pseudocontact shift is described by eq 4. δpcLn, i =

⎛ 3 cos2 θ − 1 ⎞ 1 1 i ⎟= ΔχaxLn ·⎜ ΔχaxLn ·Gi 12π ri3 ⎝ ⎠ 12π

μB 2πAi 6 1 10 + Δχ Ln Gi 3kTγI h 12π ax (5)

In eq 5, we have replaced the first term by the variable Pi (for simplicity), which is proportional to the hyperfine coupling constant Ai. The variable QLn is related to the axial anisotropy of the lanthanide ions. The physically meaningful geometric factor Gi is replaced by the relative values G′i, thus not requiring the possibly false assumption of any structural parameter. The relative values G′i can be used for comparison with different structural models. This approach is particularly useful for systems where several conformations are possible, as in these cases, where the G′i values obtained correspond to the average structure in solution. As the last product in eq 5 has to correspond to the pseudocontact shift, using relative G′i values produces QLn values, which also are inequivalent to the actual susceptibility anisotropies. Their relative values, however, can be directly compared to the Bleaney factors. As we have determined 9 hyperfine shift values (5 1H and 4 13 C positions) for 5 different lanthanide ions, we have a maximum of 45 data points and 23 variables (5 for QLn, 9 for Pi, and 9 for G′i). As one of the G′i values is chosen as reference, the number of variables is reduced by 1. Furthermore, the contact shift contributions for the H4i and H4o positions were determined to be very small and thus fixed to zero, further reducing the number of variables by 2. As 4 signals could not be observed, our final fit used 41 data points to give 20 variables. A corresponding second set of data points and variables have been obtained for the second ligand derivative. All measured hyperfine shift values were fitted simultaneously by employing a custom function in the software package OriginPro 9.1. From the obtained values, the contact and pseudocontact contributions to the hyperfine shift were back-calculated. Detailed tables and plots can be found in the Supporting Information. The results for Li[Tb(odbCOT)2] are presented in Figure 4. In order to demonstrate the possibility of distinguishing between several structural models, we considered the different conformations found in the solid-state structure of [Li(DME)-

(4)

Herein, θi is the azimuth of investigated nucleus i with respect to the main magnetic axis, which is expected to lie along the connection of the centroids of the cyclooctatetraenyl systems. This assumption is supported by recent ab initio calculations of this system.17 The value ri is the distance from the metal ion to the investigated nucleus. The geometry-dependent parameter (expression within the brackets in eq 4) is called the geometric factor and abbreviated by the symbol Gi. According to Bleaneys theory of pseudocontact shift,18 the value ΔχaxLn depends on the f electron configuration of the lanthanide ion and the crystal field parameters. These values can be combined to give the so-called Bleaney factor CBLn, which in a given crystal field (an isostructural series) should be proportional to the pseudocontact shift δpcLn,i. In contrast to that, the contact shift δconLn,i should scale with ⟨SZ⟩Ln (see D

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

the Ln(COT)2− system17 along with the observation that Tm3+-based single-ion magnets are possible.22



CONCLUSIONS We synthesized a new group of 1,2,5,6-substituted [Ln(COTR)2]− compounds for detailed NMR analysis. Employment of a simultaneous fit of all observed paramagnetic shifts shows that the overall paramagnetic shift is dominated by pseudocontact shifts. Only for the COTR ring carbon atoms are sizable contact contributions found. Using the classic Reilley plot approach, it could be shown that Bleaney’s factors are not useful for the description of the magnetic properties of the investigated system. Instead, the fitting approach reveals that the Dy3+ ion shows only very small anisotropy, whereas the largest value is found for Tm3+. This last result, together with other recent findings, makes Tm3+ a promising candidate in SMM research.

Figure 4. Contributions of contact and pseudocontact shifts to the overall paramagnetic shifts for all positions in Li[Tb(odbCOT)2]. Values and their respective errors are taken from the results of the fitting procedure.



(THF)][Tb(odbCOT)2]. This comparison shows that the chair-like conformation agrees much better with the fitted values (see the Supporting Information). Taking the averaged physical geometric factor for the C1 position from the X-ray structure (3.463 ± 0.340 × 1028 m−3), an estimate of the Δχax value for the different ions can be obtained. The values are summarized in Table 1. These values are interesting, as they Table 1. Δχax Values for the Different Ln3+ Ions in Li[Ln(odbCOT)2] Obtained by a Combination of the Fitting Procedure and the X-ray Structure Δχax (10−31 m3)

Ln3+

−8.42 0.40 −2.34 5.74 11.47

Tb3+ Dy3+ Ho3+ Er3+ Tm3+

± ± ± ± ±

EXPERIMENTAL SECTION

Proligands hdcCOT and odbCOT were synthesized by slight modifications of the literature procedure, using activated Zn dust (1% Cu) without addition of water. It was found to be necessary to degas the employed alkyne, especially after prolonged storage under air. All reactions were carried out in oven-dried glassware. Standard Schlenk and glovebox techniques were employed. Celite was dried by heating at 140 °C for 5 h at high vacuum. Used solvents (THF, hexane, toluene) were purified by a MBraun SPS-800 solvent purification system. Acetonitrile was degassed before use, and water content was checked by Karl Fischer titration. Deuterated acetonitrile was purchased from Deutero GmbH and distilled over CaH2 before use. Reported line widths were taken from deconvolution of the spectra using Topspin 3.1. Synthesis of Hexahydrodicyclopentacyclooctatetraene (hdcCOT). 1,6-Heptadiyne (4.75 g, 51.6 mmol) was degassed twice by the freeze−pump−thaw technique. NiBr2(DME) (3.17 g, 10.3 mmol, 0.25 equiv), activated Zn dust (1% Cu, 1.68 g, 25.8 mmol, 0.5 equiv), and anhydrous THF (100 mL) were added in a glovebox. The resulting deep purple solution was placed in a 60 °C oil bath and allowed to stir for 2 h. After cooling, the solution was filtered through a pad of Celite using dichloromethane as eluent. The resulting dark red solution was absorbed on silica, dried in vacuum, and purified by column chromatography (petroleum ether, Rf = 0.55) to give a yellow oil (3.31 g, 18.0 mmol, 69%) that solidified upon refrigeration (−40 °C) and remained solid at room temperature. The product decomposes slowly at room temperature when kept as the yellow oil after chromatography. Analytical data were identical to the reported values in the literature.

1.33 0.46 0.58 0.98 1.74

show striking inconsistencies with Bleaney’s factors. In particular, for the Dy3+ ion, which has the largest Bleaney factor, the smallest magnetic susceptibility anisotropy was found. The largest Δχax value is found for Tm3+, which is experimental confirmation for recent ab initio calculations of

Table 2. Observed NMR Data for Li[Ln(hdcCOT)2] in MeCN-d3 at 340 K Y3+

C1 C2 C3 C4 H1 H3i H3o H4i H4o

Tb3+

Dy3+

δ [ppm]

Irel

δ [ppm]

ν1/2 [Hz]

Irel

88.9 104.4 46.0 26.1 5.8 3.4 3.3 2.1 1.9

1.98 0.87b 2.13 1 2.08 2.17 2.19 1.15 1

−653.5 −560.0 195.9 130.7 122.1 241.3 −14.2 224.9 50.8

370 290 180 280 640 410 120 340 95

1.10 0.93 1.64 1 1.87 1.87 1.57 0.94 1

Ho3+

Er3+

Tm3+

δ [ppm]

ν1/2 [Hz]

Irel

δ [ppm]

ν1/2 [Hz]

Irel

δ [ppm]

ν1/2 [Hz]

Irel

δ [ppm]

ν1/2 [Hz]

Irel

−70.0 20.1 47.5 −3.3 17.1 −21.8

580 500 110 90 2000 820 − 550 170

0.48 0.41 1.50 1 1.47 1.99 − 1.34 1

−223.7 −142.6 63.0 37.2 27.6 47.8 5.0 39.0 11.1

1000 1000 130 75 2400 940 470 480 190

1.33 1.04 1.78 1 1.28 2.00 1.46 0.98 1

−a − −41.2 −76.7 −70.2 −178 −8.0 −147.3 −30.3

− − 240 120 4700 2000 960 1000 420

− − 2.17 1 1.04 1.34 1.64 0.76 1

− − −158e −159e −158.3 −341.1 −5.6 −281.8 −60.9

− − ∼400e ∼700e 10 000 4300 1900 2300 980

− − 1.6e 1e 0.5 1.4 0.5d 0.9 1

c

−20.6 −1.5

a Dashes indicate signals that were not observed. bQuaternary carbon in diamagnetic molecule leads to low integral value. cSignal probably obscured by the residual solvent signal. dSignal close to the solvent peak and heavily influenced by baseline distortions (see Figure S12 in the Supporting Information). eValues separated by deconvolution analysis.

E

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Table 3. Observed NMR Data for Li[Ln(odbCOT)2] in MeCN-d3 at 340 K Y3+ δ [ppm] 1

C C2 C3 C4 H1 H3i H3o H4i H4o

94.7 100.7 39.2 26.2 5.7 3.0 2.8 2.0 1.7

Tb3+ Irel

δ [ppm]

0.97 0.39b 0.99 1 1.00 1.03 1.03 1.05 1

−756.2 −455.9 205.3 101.6 85.6 254.1 32.4 141.8 26.6

ν1/2 [Hz] 280 270 180 130 510 460 140 220 63

Dy3+ Irel

δ [ppm]

0.71 1.05 1.04 1 0.96 0.96 0.98 1.17 1

−89.7 24.3 44.2 26.0 19.7 −18.8 0.5 −10.3 2.1

ν1/2 [Hz] 670 650 100 58 1800 680 420 240 200

Ho3+ δ [ppm]

Irel

−259.2 −107.4 61.1 50.1 19.0 51.6 25.6 12.9 7.4

0.63 0.88 0.90 1 0.72 0.91 0.97 1 c

ν1/2 [Hz] 1000 570 120 52 1600 1000 270 500 170

Er3+

Tm3+

Irel

δ [ppm]

ν1/2 [Hz]

Irel

δ [ppm]

ν1/2 [Hz]

Irel

0.65 0.35 0.98 1 0.18 0.93 0.93 0.78 1

− − −55.3 −30.4 −51.6 −188.6 −34.6 −87.8 −10.6

− − 190 130 3700 2100 1200 610 410

− − 0.83 1 0.33 0.68 0.97 1.06 1

− − −177.5 −71.2 −116.3 −361.6 −65.6 −164.5 −19.9

− − 490 190 14 000 5100 3100 1400 1100

− − 0.7 1 0.7 0.3 0.7 0.9 1

a

a Dashes indicate signals that were not observed. bQuaternary carbon in the diamagnetic molecule leads to a low integral value. cIntegral of this signal could not be determined reliably due to partial overlap with the solvent residual signal.

Synthesis of Octahydrodibenzocyclooctatetraene (odbCOT). The same procedure as that described above using 1,7octadiyne (4.0 g, 37.7 mmol), NiBr2(DME) (2.3 g, 7.54 mmol, 0.2 equiv), activated Zn dust (1% Cu, 1.02 g, 15.6 mmol, 0.41 equiv), and anhydrous THF (100 mL) was employed. The resulting dark brown solution was absorbed on silica, dried in vacuum, and purified by column chromatography (petroleum ether, Rf = 0.5) to give a colorless oil (1.90 g, 8.95 mmol, 47%) that solidified upon refrigeration (−40 °C) and remained a solid at room temperature. The product decomposes slowly at room temperature when kept as oil after chromatography. Analytical data were identical to the reported values in the literature. Synthesis of Li[Ln(hdcCOT)2]·THF. A solution of hexahydrodicyclopentacyclooctatetraene (200 mg, 1.09 mmol) in anhydrous THF (5 mL) was added to lithium foil (excess, typically 20.0 mg, 2.86 mmol, 2.6 equiv). The mixture was allowed to stir overnight before excess lithium was removed and weighed to ensure the consumption of 2 equiv. Anhydrous lanthanide trichloride (0.54 mmol, 0.5 equiv) was added, and the solution was allowed to stir for at least 4 h at ambient temperature. n-Hexane (5 mL) was added, resulting in the formation of an off-white precipitate. All volatiles were removed under reduced pressure to give a powdery residue, which was washed with toluene (4 mL). The remaining solid was extracted with a mixture of toluene and acetonitrile (8 mL, 10:1 v/v), and the extract was filtered over a short plug of Celite. All volatiles were removed under reduced pressure to give the desired compounds as distinctly colored powders (vibrant yellow for Y3+, orange for Tb3+, brown for Tm3+, and yellow for Dy3+, Ho3+, and Er3+) in moderate to good yields (47−98%). The reported yields refer to the material after workup. NMR data for Li[Ln(hdcCOT)2] are given in Table 2. Li[Y(hdcCOT)2]·THF. Yield: 298 mg (0.535 mmol, 98%). Anal. Calcd for C32H40LiOY: C: 71.64, H: 7.51. Found: C: 71.94, H: 7.38. Li[Tb(hdcCOT)2]·2MeCN. Yield: 224 mg (0.363 mmol, 67%). Anal. Calcd for C32H38LiN2Tb: C: 62.34, H: 6.21, N: 4.54. Found: C: 61.31, H: 6.29, N: 4.82. Li[Dy(hdcCOT)2]·THF. Yield: 155 mg (0.254 mmol, 47%). Anal. Calcd for C32H40DyLiO: C: 63.00, H: 6.61. Found: C: 62.64, H: 6.66. Li[Ho(hdcCOT)2]·MeCN. Yield: 255 mg (0.439 mmol, 80%). Anal. Calcd for C30H35HoLiN: C: 61.97, H: 6.07, N: 2.41. Found: C: 61.61, H: 6.29, N: 2.42. Li[Er(hdcCOT)2]·THF. Yield: 283 mg (0.460 mmol, 84%). Anal. Calcd for C32H40ErLiO: C: 62.51, H: 6.56. Found: C: 62.73, H: 6.55. Li[Tm(hdcCOT)2]·2MeCN. Yield: 301 mg (0.480 mmol, 88%). Anal. Calcd for C32H40LiN2Tm: C: 61.34, H: 6.11, N: 4.47. Found: C: 61.05, H: 6.31, N: 4.84. Synthesis of Li[Ln(odbCOT)2]·THF. A solution of octahydrodibenzocyclooctatetraene (200 mg, 0.942 mmol) in anhydrous THF (5 mL) was added to lithium foil (excess, typically 20 mg, 2.86 mmol, 3.0 equiv). The mixture was allowed to stir overnight before excess lithium was removed and weighed to ensure the consumption of 2 equiv. Anhydrous lanthanide trichloride (0.471 mmol, 0.5 equiv) was

added, and the solution was allowed to stir for at least 4 h at ambient temperature. n-Hexane (mL) was added, resulting in the formation of an off-white precipitate. All volatiles were removed under reduced pressure to give a powdery residue, which was washed with toluene (5 mL). The remaining solid was extracted with a mixture of toluene and acetonitrile (8 mL, 10:1 v/v), and the extract was filtered over a short plug of Celite. All volatiles were removed under reduced pressure to give the desired compounds as distinctly colored powders (vibrant yellow for Y3+, orange for Tb3+, brown for Tm3+, and yellow for Dy3+, Ho3+, and Er3+) in moderate to good yields (38−87%). NMR data for Li[Ln(odbCOT)2] are given in Table 3. Crystals suitable for X-ray diffraction experiments were grown from Li[Tb(odbCOT)2]·THF, obtained by the above procedure. A sample was dissolved in THF and slow diffusion of a mixture of hexane and DME (10:1 v/v) was allowed. After 1 week, orange crystals were obtained. Li[Y(odbCOT)2]·THF. Yield: 244 mg (0.412 mmol, 87%). Anal. Calcd for C36H48LiOY: C: 72.96, H: 8.16. Found: C: 73.18, H: 8.19. Li[Tb(odbCOT)2]·THF. Yield: 384 mg (0.580 mmol, 82%) starting from 300 mg (1.413 mmol) odbCOT. Anal. Calcd for C36H48LiOTb: C: 65.25, H: 7.30. Found: C: 64.39, H: 6.98. Li[Dy(odbCOT)2]·THF. Yield: 55 mg (0.083 mmol, 70%) starting from 50 mg (0.236 mmol) odbCOT. Anal. Calcd for C36H48DyLiO: C: 64.90, H: 7.26. Found: C: 65.38, H: 7.10. Li[Ho(odbCOT)2]·THF. Yield: 229 mg (0.342 mmol, 73%). Anal. Calcd for C36H48HoLiO: C: 64.67, H: 7.24. Found: C: 64.30, H: 6.87. Li[Er(odbCOT)2]·THF. Yield: 250 mg (0.373 mmol, 79%). Anal. Calcd for C36H48ErLiO: C: 64.44, H: 7.21. Found: C: 64.70, H: 7.19. Li[Tm(odbCOT)2]·THF. Yield: 120 mg (0.178 mmol, 38%). Anal. Calcd for C36H48LiOTm: C: 64.28, H: 7.19. Found: C: 64.74, H: 7.16.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.6b00241. NMR spectra; details of X-ray crystal structure determinations; structural considerations and the question of separation of ion pairs; experimental NMR chemicals shifts; isostructurality plots; description of the fitting procedure and fitted shifts; and paramagnetic shift composition charts (PDF) Crystallographic data (CIF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: +49-6221546247. Fax: +49-6221-541616247. F

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Notes

(14) Kurikawa, T.; Negishi, Y.; Hayakawa, F.; Nagao, S.; Miyajima, K.; Nakajima, A.; Kaya, K. J. Am. Chem. Soc. 1998, 120, 11766−11772. (15) Lorenz, V.; Edelmann, A.; Blaurock, S.; Freise, F.; Edelmann, F. T. Organometallics 2007, 26, 6681−6683. (16) (a) Ivano Bertini, C. L. Physical Methods for Chemists; Saunders College Publishing, 1992. (b) Enders, M. In Modeling of Molecular Properties; Comba, P., Ed.; Wiley-VCH Verlag GmbH & Co. KGaA, 2011; pp 49−63. (c) Funk, A. M.; Fries, P. H.; Harvey, P.; Kenwright, A. M.; Parker, D. J. Phys. Chem. A 2013, 117, 905−917. (17) Gendron, F.; Pritchard, B.; Bolvin, H.; Autschbach, J. Dalton Trans 2015, 44, 19886−19900. (18) Bleaney, B. J. Magn. Reson. 1972, 8, 91−100. (19) Reilley, C. N.; Good, B. W.; Allendoerfer, R. D. Anal. Chem. 1976, 48, 1446−1458. (20) (a) Blackburn, O. A.; Edkins, R. M.; Faulkner, S.; Kenwright, A. M.; Parker, D.; Rogers, N. J.; Shuvaev, S. Dalton Trans 2016, 45, 6782−6800. (b) Funk, A. M.; Finney, K.-L. N. A.; Harvey, P.; Kenwright, A. M.; Neil, E. R.; Rogers, N. J.; Kanthi Senanayake, P.; Parker, D. Chem. Sci. 2015, 6, 1655−1662. (c) Zhang, Y.; Krylov, D.; Rosenkranz, M.; Schiemenz, S.; Popov, A. A. Chem. Sci. 2015, 6, 2328−2341. (21) (a) Geraldes, C. F. G. C.; Zhang, S.; Sherry, A. D. Bioinorg. Chem. Appl. 2003, 1, 1−23. (b) Reuben, J. J. Magn. Reson. 1982, 50, 233−236. (22) Meng, Y. S.; Qiao, Y. S.; Zhang, Y. Q.; Jiang, S. D.; Meng, Z. S.; Wang, B. W.; Wang, Z. M.; Gao, S. Chem. - Eur. J. 2016, 22, 4704− 4708.

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank the Fonds der Chemischen Industrie for financial support and for a scholarship for M.H.



REFERENCES

(1) (a) La Mar, G. N.; Horrocks, W. D., Jr; Holm, R. H. NMR of Paramagnetic Molecules: Principles and Applications; Academic Press, INC.: New York, 1973. (b) Bertini, I.; Luchinat, C.; Parigi, G. Solution NMR of Paramagnetic Molecules: Applications to Metallobiomolecules and Models; Elsevier, 2001; Vol. 2. (c) von Ammon, R.; Kanellakopulos, B.; Fischer, R. D.; Formacek, V. Z. Naturforsch., B: J. Chem. Sci. 1973, 28, 200−206. (2) (a) Mao, J.; Zhang, Y.; Oldfield, E. J. Am. Chem. Soc. 2002, 124, 13911−13920. (b) Fernández, P.; Pritzkow, H.; Carbó, J. J.; Hofmann, P.; Enders, M. Organometallics 2007, 26, 4402−4412. (c) Rastrelli, F.; Bagno, A. Chem. - Eur. J. 2009, 15, 7990−8004. (d) Hrobarik, P.; Reviakine, R.; Arbuznikov, A. V.; Malkina, O. L.; Malkin, V. G.; Kohler, F. H.; Kaupp, M. J. Chem. Phys. 2007, 126, 024107. (3) Gendron, F.; Sharkas, K.; Autschbach, J. J. Phys. Chem. Lett. 2015, 6, 2183−2188. (4) (a) Pinkerton, A. A.; Rossier, M.; Spiliadis, S. J. Magn. Reson. 1985, 64, 420−425. (b) Castro, G.; Regueiro-Figueroa, M.; EstebanGomez, D.; Perez-Lourido, P.; Platas-Iglesias, C.; Valencia, L. Inorg. Chem. 2016, 55, 3490−3497. (5) Horrocks, W. D. J. Am. Chem. Soc. 1974, 96, 3022−3024. (6) Luke, W. D.; Streitwieser, A. J. Am. Chem. Soc. 1981, 103, 3241− 3243. (7) Neese, F.; Pantazis, D. A. Faraday Discuss. 2011, 148, 229−238. (8) (a) Ishikawa, N.; Sugita, M.; Ishikawa, T.; Koshihara, S.-y.; Kaizu, Y. J. Am. Chem. Soc. 2003, 125, 8694−8695. (b) Freedman, D. E.; Harman, W. H.; Harris, T. D.; Long, G. J.; Chang, C. J.; Long, J. R. J. Am. Chem. Soc. 2010, 132, 1224−1225. (c) Jeletic, M.; Lin, P.-H.; Le Roy, J. J.; Korobkov, I.; Gorelsky, S. I.; Murugesu, M. J. Am. Chem. Soc. 2011, 133, 19286−19289. (d) Zadrozny, J. M.; Xiao, D. J.; Atanasov, M.; Long, G. J.; Grandjean, F.; Neese, F.; Long, J. R. Nat. Chem. 2013, 5, 577−581. (e) Gomez-Coca, S.; Cremades, E.; Aliaga-Alcalde, N.; Ruiz, E. J. Am. Chem. Soc. 2013, 135, 7010−7018. (f) Lin, C.-Y.; Fettinger, J. C.; Grandjean, F.; Long, G. J.; Power, P. P. Inorg. Chem. 2014, 53, 9400. (g) Samuel, P. P.; Mondal, K. C.; Amin, S. N.; Roesky, H. W.; Carl, E.; Neufeld, R.; Stalke, D.; Demeshko, S.; Meyer, F.; Ungur, L.; Chibotaru, L. F.; Christian, J.; Ramachandran, V.; van Tol, J.; Dalal, N. S. J. Am. Chem. Soc. 2014, 136, 11964−11971. (9) (a) Damjanovic, M.; Katoh, K.; Yamashita, M.; Enders, M. J. Am. Chem. Soc. 2013, 135, 14349−14358. (b) Gimenez-Agullo, N.; de Pipaon, C. S.; Adriaenssens, L.; Filibian, M.; Martinez-Belmonte, M.; Escudero-Adan, E. C.; Carretta, P.; Ballester, P.; Galan-Mascaros, J. R. Chem. - Eur. J. 2014, 20, 12817−12825. (c) Damjanovic, M.; Morita, T.; Katoh, K.; Yamashita, M.; Enders, M. Chem. - Eur. J. 2015, 21, 14421−14432. (d) Ishikawa, N.; Iino, T.; Kaizu, Y. J. Am. Chem. Soc. 2002, 124, 11440−11447. (10) (a) Luke, W. D.; Berryhill, S. R.; Streitwieser, A. Inorg. Chem. 1981, 20, 3086−3089. (b) Zalkin, A.; Templeton, D. H.; Luke, W. D.; Streitwieser, A. Organometallics 1982, 1, 618−622. (c) Streitwieser, A.; Kluttz, R. Q.; Smith, K. A.; Luke, W. D. Organometallics 1983, 2, 1873−1877. (d) Boussie, T. R.; Eisenberg, D. C.; Rigsbee, J.; Streitwieser, A.; Zalkin, A. Organometallics 1991, 10, 1922−1928. (11) (a) Meihaus, K. R.; Long, J. R. J. Am. Chem. Soc. 2013, 135, 17952−17957. (b) Le Roy, J. J.; Korobkov, I.; Murugesu, M. Chem. Commun. 2014, 50, 1602−1604. (c) Harriman, K. L. M.; Murugesu, M. Chem. Res. 2016, 10.1021/acs.accounts.6b00100. (12) Wender, P. A.; Christy, J. P. J. Am. Chem. Soc. 2007, 129, 13402−13403. (13) Shannon, R. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767. G

DOI: 10.1021/acs.organomet.6b00241 Organometallics XXXX, XXX, XXX−XXX