The Synthesis and Characterisation of a Molecular Sea‐Serpent: Studies of a {Cr24Cu7} Chain

Abstract A finite chain of thirty‐one paramagnetic centers is reported, synthesized by reaction of hydrated chromium fluoride, copper carbonate and pivalic acid in the presence of 1,4,7,10‐tetrazacyclododecane (cyclen). Magnetic studies show predominantly anti‐ferromagnetic exchange leading to a high density of low‐lying spin states and large saturation field.

The designed approach to the synthesis of polymetallic complexes often uses rigid ligands with donor atoms organised to bind to metal sites with specific geometries. [1][2][3][4][5][6] The design uses our knowledge of the preferred coordination geometries of metals and that al igand is bi-or tri-dentate, rigid or flexible.T emplates can also be used to direct complexes towards specific metal nuclearities. [7,8] Another route recognises that there are always chemical factors that are outside our complete control, and allowing such factors to play ap art in reaching new structures.I nt his paper we take aw ell-studied system, [9] and include two reactants that lead to an unusual chain compound.
We have been studying the formation of cyclic and acyclic complexes from hydrated chromium fluoride reacted with pivalic acid in the presence of an amine and adivalent metal ion that favours ar egular octahedral coordination environment, typically nickel(II). [9] Where the amine is as econdary amine with linear side-chains,for example,di-n-propylamine, we form in very high yield ac ompound [ n Pr 2 NH 2 ][Cr 7 NiF 8 -(O 2 C t Bu) 16 ]. [10] Theformation of this eight-metal ring can be easily rationalised. Fore xample:t he bond angles at the sixcoordinate metal sites lead to octahedral coordination geometries without strain;t he carboxylates can adopt their preferred bridging mode;the fluorides bridge two metals;the protonated ammonium cation H-bonds to the fluorides and acts as atemplate for the ring. If we make the side-chains on the amine more sterically demanding,oruse aprotonated Nheterocycle,n ine-or ten-metal versions of this structure result. [11][12][13] Theb uilding block is always the {M 2 F(O 2 C t Bu) 2 } fragment, which can oligomerise readily.The eight-metal ring is the smallest closed structure,a nd hence favoured by entropy.E ntropic control matches the reaction temperature of 1408 8 Ci nmolten pivalic acid.
Inclusion of copper(II) in place of nickel(II) leads to aseries of larger rings {Cr x Cu 2 }(where x = 10, 11 or 12}, which can be rationalised as due to the propensity of Cu II to show five-coordinate geometry rather than six-coordinate. [14] If we replace the simple amine by cyclen we find it coordinates to Ni II producing an S-shaped molecule,[{Ni(cyclen)} 2 Cr 12 NiF 20 -(O 2 C t Bu) 22 ]which we have described as amolecular seahorse, 1. [13] We can rationalise this as coordination of the tetradentate ligand to Ni II leads to it acting as at erminal group and preventing ring closure.A nother factor is the crystal field stabilisation energy for nickel(II);once bound to the cyclen it tends to stay bound and will adopt asix-coordinate geometry.
Combining both Cu II as the divalent ion and cyclen as the template introduces af urther degree of uncertainty in our design and here it leads to aremarkable result. Ther eaction of chromium(III) fluoride hydrate,c yclen, basic copper carbonate,a nd pivalic acid at 140-160 8 8Cf or 5h produced [Cu(H 2 O)(cyclen)] 2 [Cr 24 Cu 5 {Cu(cyclen)} 2 F 40 (O 2 C t Bu) 50 ], 2.It was obtained as an isolated product with low but reproducible yield, and can be separated from the neutral byproduct [CrF(O 2 C t Bu) 2 ] 8 by extraction or crystallization. Crystals can be grown from am ixture of Et 2 O/MeCN solvent. The structure of 2 ( Figure 1) is related to the S-shaped {Cr 12 Ni 3 } complex [13] but is far longer. We are unaware of any finite 1Dchain containing more metal centres.An{Fe 18 }chain has been reported by Christou and co-workers. [14] Thea symmetric unit comprises one half of the oligomer, with the central copper (Cu4) residing on an inversion centre. There are seven Cu II sites within the 31-metal chain. Cu5, and its symmetry equivalent are bound to cyclen and awater and is not attached to the chain. Theother Cu sites are linked by av ariety of Cr-F fragments within the chain. Beginning at Cu1, there is a{Cr 5 }chain (Cr1…Cr5) that links to Cu2. There is then a{Cr 3 }(Cr6…Cr8) that links to Cu3, and finally a{Cr 4 } chain (Cr9…Cr12) that links to the central Cu4 site.W ehave only previously seen more than one Cr-chain in the same molecule in a{Cr 11 Cu 2 }r ing. [15] Each Cr···Cr contact in the three distinct chains is bridged by afluoride and two pivalates,the same motif we have seen in previous rings and chains. [9][10][11][12][13]15] Thelinking of the Cu sites to the Cr chains varies.C u1 that terminates the chain (see Figure 2) is five-coordinate,a dopting as quare-pyramidal geometry,b ound to 4N -atoms from cyclen in as quare with af luoride on the axial site bridging to Cr1. TheC u ÀFb ond length of 2.094 (7) is longest establishing this as the electronic z-axis for the Cu II site.B ond length ranges are given in Table 1.
Cu2 is bridged by two fluorides and apivalate to Cr5 and by as ingle fluoride and two pivalates to Cr6. Tw oo ft he fluorides are on the elongated Jahn-Teller (JT) axis.C u3 is bridged to Cr8 by asingle fluoride and single pivalate,and to Cr9 by afluoride and two carboxylates;this latter fluoride is on the JT axis.Cu4 is on the inversion centre and is bridged to Cr12 by two fluorides and as ingle pivalate.T he JT-axis for Cu4 is to two F-atoms.
Cu5 is part of a[ CuF(cyclen)] + cation, which forms N-H···F hydrogen bonds to the chain with N···F separations of 2.8-2.9 .
Thefive unique copper sites have four distinct geometries. Cu1 and Cu5 are square pyramidal, the former bound to 4N and 1F and the latter to O4 and 1O.C u3 is also square pyramidal but bound to two Fand three Od onors,a sn oted above the fluoride is on the apex of the square pyramid. Cu2 is six-coordinate with a mer-arrangement of three Oand three F donors,while Cu4 is six coordinate with a trans-arrangement of four Fand two Od onors.T his capability of Cu II to adopt five different coordination environments in one complex is am ajor reason for the formation of 2.A ll the Cr III sites are six-coordinate,w ith Cr2, Cr3, Cr3, Cr4, Cr6, Cr7, Cr9, Cr10 and Cr11 having a cis-arrangement of two Fa nd four O donors.Cr5, Cr8 and Cr12 have a mer-arrangement of three O and three Fdonors.Cr1 is unique in having four Fand two Odonors with a cis-geometry.
No significant intermolecular interactions were observed in the solid state,w hich is similar to related rings. [9] This is ac onsequence of the chain possessing fifty bridging pivalate groups,w hich effectively block any interactions between the neighbouring chains.
There is considerable resemblance between compounds 1 and 2.I n1 all nickel sites are six-coordinate and it is the coordinative flexibility of Cu II that allows formation of the longer chain. Examining the core of 2 we can see an almost planar metal S-chain {Cr 8 Cu 3 }r unning from Cu3 to its symmetry equivalent via two {Cr 4 }chains and Cu4 (the plane shown in peach in Figure 2). This is very similar to 1,b ut in 1 there are {Cr 6

Forschungsartikel
Magnetic studies of 2 show predominantly anti-ferromagnetic exchange is present. Susceptibility (c m )w as measured from 300 to 2Kin a5000 Oe field and the magnetisation (M) was measured from 0t o7Ta tv arious temperatures (Figure 3). Thep roduct c m T shows as teady decline with falling temperature,w hile M increases steadily with field at all temperatures in the range 2to4K, and has not saturated even at the lowest T and highest H. As expected, this is consistent with multiple low-lying paramagnetic states as well as sizable antiferromagnetic exchange for this long chain of unpaired spins.
To fit the magnetic data of such al arge molecule is not trivial, however because there is precedent for many of the super-exchange pathways then fitting is possible.W ef irst made the assumption that Cu5 and its symmetry equivalent act as independent s = 1/2 centers and are not coupled to the ring. Thenumber of super-exchange paths was kept to five,as listed in Table 2, giving the Hamiltonian (1) for the 31 coupled magnetic centres,that is,without Cu5 and Cu5' TheS ea-Serpent belongs to the largest spin systems of finite size;t he Hilbert space dimension is 36,028,797,018,963,968 (without Cu5 and Cu5'). Exact diagonalization works up to 10 5 ,finite-temperature Lanczos up to 10 10 .F ortunately,t he structure is one-dimensional and not frustrated. Thefirst property allows us to determine low-lying energy eigenvalues by means of density-matrix renormalization group (DMRG) methods,a nd the second enables quantum Monte Carlo (QMC) in order to evaluate thermal properties such as the magnetization as afunction of temperature and applied field. Forb oth types of calculations we employed the free ALPS library. [17,18] An average gv alue of 2.0 was assumed for the calculations.
There are many assumptions here,perhaps chiefly that the exchange of Cu1-Cr1 is equivalent to that of Cu2-Cr6 or Cu3-Cu9;t his assumption is based on the exchange being dominated by the F-bridge on the z-axis on the Cu site, regardless of the number of carboxylate bridges.I nT able 2 three of the five interactions were fixed, based on aprevious study of a{ Cr 12 Cu 2 }r ing, where the exchange interactions were determined by three techniques:magnetometry,inelastic neutron scattering and tunnel diode oscillator measurements. [16] Thefinal two parameters, J 3 and J 5 were allowed to vary with g fixed at 2.0. Theb est fit was then achieved with J 3 = 3K and J 5 = 10 K ( Figure 3). Given the limited data available introduction of still further parameters (e.g.making the Cu1-Cr1 exchange aseparate variable J 6 )isnot justified.
These parameters produce as pin state structure with multiple low-lying paramagnetic states.D MRG calculations [17,18] give ag round state with total spin S = 3/2. Of the many levels within 3Kof the ground state we targeted five for each magnetic quantum number (symbols in Figure 3c). They are grouped into spin multiplets when energetically degenerate and marked according to assigned total spin. Thet rue spectrum contains many more levels above those targeted by (c) Low-lying DMRG energy eigenvalues: for each magnetic quantum number (> 0) the lowest five levels were evaluated. These belong to multiplets marked by colored lines: S = 1/2 (red), S = 3/2 (green), S = 5/2 (blue), S = 7/2 (magenta) and S = 9/2 (light blue). Fits used parameters given in Table 2a nd text. DMRG;itissimply increasingly complicated to obtain them with reliable accuracy.
TheE PR spectra of 2 are broad with two interesting features ( Figure S1). Firstly,t he spectrum in frozen solution (1:1 CH 2 Cl 2 :toluene) and powder at 5Kare very similar. This suggests the structure of the chain and its supramolecular interaction with the cation containing Cu5 is maintained in solution. Secondly,there is copper hyperfine structure on the low field feature in the spectra. We believe this is due to the resonances from the isolated Cu5 complex. This lower field feature is centered at g = 2.13 while the higher field feature is at g = 1.960, 1.990, 1.990. Thes pectrum was simulated [19] as the sum of a{ Cr 3 }c hain with g = 1.960 and J = 16 Ka nd two Cu II sites with g values of 2.075, 2.055, 2.145 and 2.065, 2.045, 2.135.
Thesynthesis and structure of 2 demonstrates the richness and complexity of the chemistry that is found when chromium fluoride,pivalic acid and acopper(II) salt are present. [15] The yield of 2 is low,b ut reproducible.T here must be other oligomers formed that we have not crystallized to this point. It also suggests that inclusion of other N-donor macrocycles may lead to still further unusual structures.