2.1. Solvent-Free Synthetic Approach
The present lead-based MOF has been obtained through a less habitual synthetic procedure, coined by us and others as solvent-free, which has been largely used in the preparation of organic molecules [
46,
47] but scarcely employed in the case of metal-organic compounds [
48,
49]. Nonetheless, this alternative procedure to the main solvothermal method, which consists of the mechano-chemical mixture of the reagents followed by thermal treatment, was successfully employed from 2012 on in the preparation of metal-azolate MOFs starting from metal oxide or hydroxides [
50,
51]. As observed in those works, solvent-free procedure has shown to render pure and crystalline materials in a very high yield, in such a way that it may consider as a cheaper and environmentally-friendly approach compared to those methods (evaporation or solvothermal methods) that require an organic solvent. When we first employed this methodology, we observed that its success is mainly due exclusively to the reagents used, particularly with regard to their ease to form a melt during the acid-base reaction taking place. As an example, diazole/triazole-like ligands are characterized for low melting points that favour the formation of melts in which metal oxides and hydroxides are reacted. The latter metal sources count on the advantage of consisting of anions that cannot remain in the mixture as impurities [
52]. Accordingly, we focused on other reagents to reproduce the formation of the melt by selecting metal salts that are melted upon relatively low temperature to extend the use of the solvent-free procedure to ligands possessing high melting point such as pmdc [
45,
53]. For instance, the use of hygroscopic metal salts (often hydrated salts) in combination with organic carboxylic acids may lead, after being ground, sealed and heated, to the complete mixture of the reagents and the presence of enough water resulting from the acid-base reaction to allow for the crystallization of a MOF. Moreover, some anions such as nitrates and acetates acknowledge the capacity to be decomposed during the heating process to avoid their presence in the final product [
54].
On that basis, the synthesis followed in the present case has been carefully selected to meet all the previous keypoints. by mixing Lead(II) nitrate and pyrimidine-4,6-dicarboxylic acid are mixed in a mortar and, once smoothly hand-ground, the resulting mixture is placed into an airtight glass vessel that is oven-heated for 3 days at 150 °C to allow the crystallization of a MOF while the nitric acid formed during the acid-base reaction is decomposed and released when the vessel is open at room temperature.
2.2. Structural Description of {[Pb5(μ3-OH)(μ3-NO3)3(μ6-pmdc)3]·H2O}n (1)
Compound
1 crystallises in the
Rc space group and consists of a highly condensed 3D framework where the Pb(II) atoms are linked through μ
6-pmdc ligands in addition to the μ
3-hydroxide and μ
3-nitrate anions. The asymmetric unit contains three different lead atoms, each one showing a different coordination environment. The Pb1 atom shows a heptacoordinated NO
6Pb chromophore that best resembles a capped trigonal prism though it is also close to a capped octahedron [S(ctpr) = 2.16 / S(coc) = 2.87] according to the continuous shape measurements calculated with SHAPE program [
55]. The environment is formed by one chelating and three monodentate pmdc ligands, a nitrate anion, and the hydroxide anion. The Pb2 atom coordinates to two
N,
O-chelating pmdc ligands, two nitrate anions, and to two monodentate pmdc ligands with slightly larger bond distances (
Table 1), thus rendering an octacoordinated polyhedron resembling a slightly distorted square antiprism [S(sapr) = 1.77]. At last, the coordination sphere of the Pb3 atom, meanwhile, is fully occupied by twelve nitrate oxygen atoms leading to a distorted cuboctahedron [S(coc) = 1.83] (
Figure 1). Despite the astonishing coordination environment with that high coordination number, it must be remarked that such a geometry has been previously reported for other lead(II)-based complexes [
56,
57,
58] deposited in the CCDC [
59], although this is the first example in which such a fragment pertains to an extended network of a MOF.
On its part, the crystallographically unique pmdc ligand exhibits a previously non-reported κ
2N11,
O171:κ
2N13,
O181:κ
O171:κ
O172:κ
O182:κ
O182 coordination mode in which it is coordinated to six Pb centres. In detail, pmdc acts as a bridge between equivalent Pb1 and Pb2 atoms imposing Pb···Pb distances ranging from 4.04 to 7.22 Å depending on the ligand group involved (only the carboxylate group or the carboxylate and also the pyrimidine ring). As a consequence of such tangle of links, cubic building units are formed, whose faces are occupied by the pmdc ligands and metal ions decorate the vertexes and edges. The Pb3 atom is placed in the centre of the cube linking the Pb1 and Pb2 atoms through the nitrato bridging ligands. It is also worth mentioning that Pb1 atoms exhibit a large mutual connectivity being connected themselves by means of the hydroxide anion that dispose them in a triangular shape (
Figure 2).
The cubic building units are self assembled by sharing the edges in such a way that each one is surrounded by twelve surrounding units, giving rise to a highly condensed 3D framework that consists of a six-nodal network [(3-c)2(3-c)6(6-c)6(6-c)6(6-c)3(6-c) stoichiometry] possessing a (4.6
2)
6(4
3)
2(4
5.6
10)
3(4
5.6
8.8
2)
6(4
6.6
9)
6(6
12.8
3) point symbol, which is a new topology that has been registered as
jcr4 in the TOPOS database [
60]. The crystallisation water molecules are occluded within isolated small voids that are generated in the framework between the pmdc ligands of two adjacent cubes, being highly disordered without any remarkable interaction with the 3D crystal building (
Figure 3).
2.3. Thermal Behaviour
The thermogravimetric measurement was performed in polycrystalline sample of compound
1 as an additional characterization to verify the chemical formula (
Figure 4). Starting from room temperature, the compound does not exhibit any mass loss upon heating up to 100 °C despite the presence of a crystallisation water molecule. However, the TG curve experiments a very slight decrease in the 150–300 °C range whose mass loss corresponds to a water molecule. Therefore, it can be confirmed that compound
1 possesses a complete lattice water molecule per formula unit (calc. 1.0, exp. 1.2% of mass loss), which finds it difficult to be released from the framework since it is occluded within the isolated pores. In the thermodiffractometric study conducted over sample from the same batch, it is observed that the compound preserves its crystal structure because the diffractograms are practically unchanged up to 210 °C. From this temperature on, the PXRD patterns show a significant shift of most of their maxima to higher 2θ angles, which is consistent with the gradual loss of the lattice molecules and subsequent compression of the framework. The loss of water, which presents a weak endothermic character in view of the DTA curve, is followed by the structural collapse in which pmdc ligand and nitrate and hydroxide anions are decomposed. This step leads to PbO as final residue of the combustion (cald. 36.4, exp. 37.0% of mass remained in the experiment) above 400 °C after two successive exothermic processes.
To end up with the thermal characterization, just to mention that the dehydrated amorphous compound obtained at ca. 300 °C cannot revert back into the pristine material neither by gas diffusion nor by water soaking procedures. In the first one, the anhydrous powder was placed into a closed vessel containing an isolated open vessel of water, in such a way that water saturated atmosphere is generated. In the second one, the product is soaked in water for 6 h and filtered. Amorphous products were obtained in both experiments, confirming the irreversibility of the thermal dehydration, which is in agreement with the possible structural collapse occurring to the compound.
2.4. Photoluminescence Properties
We explored the photoluminescence properties of compound
1 in view of the good emissive properties previously shown by CPs based on pmdc ligand [
44,
45,
61]. To start with, the emission spectrum recorded at room temperature on polycrystalline sample (
Figure 5a) presents a wide band covering the whole visible spectrum that is composed of two main contributions with the maxima peaking at ca. 460 and 550 nm. Focusing on the wavelength of the emission maximum, the excitation spectrum reveals the presence of a very weak band in the 300-360 nm range followed by the main band (peaking at 450 nm) that cannot be completely observed due to its overlap with the emission. The first excitation band is concordant with the diffuse reflectance spectrum (
Figure S12) in which the occurrence of at least two bands are observed in the 220-400 nm range. Compared to the spectroscopic data of the free H
2pmdc ligand (in which λ
ex,max = 335 and λ
em,max = 435 nm), compound
1 shows a strong red shift of 115 nm (
Figure S5), meaning that the photoluminescence properties of this compound cannot be fully explained on the basis of the usual electronic transitions of the π-π* electronic of pmdc. The emission lifetime was also measured for both emission maxima (
Figure S6). The signal was very weak for the first shoulder (λ
em,max = 460 nm) and the lifetime very short so it had to be estimated by deconvolution with an appropriate instrument refinement file (τ = 2.8(1) ns). The maxima of λ
em,max = 550 nm presented a larger lifetime of 72(3) µs (averaged over two components, see ESI) estimated from the general multiexponential expression (I
t = A
0 + A
1exp(t/
τ1) + A
2exp(t/
τ2)). Moreover, the emission quantum yield (QY) measured for compound
1 almost doubles that of the H
2pmdc ligand recorded under equivalent experimental setup, being of 1.7(2)% vs 0.8(2)%, respectively.
In view of the observed spectra, TDDFT calculations were performed on a suitable model of compound
1 consisting of a monomeric anionic fragment containing all ligands, which was previously optimized at DFT level (see Computational details and
Figure S7). According to the results obtained with Gaussian software (
Figure 5b), the main excitations occurring in the compound are related with two theoretical electronic transitions: i) LUMO
← HOMO-5 occurring at 395 nm and ii) LUMO
← HOMO-7 at 335 nm, which would give rise to the seventh (S
7) and tenth (S
10) excited states. Although both excitations possess high oscillator strengths, it must be highlighted that the former is much higher than the latter (see
Table S3). Moreover, both HOMO-5 and HOMO-7 orbitals are mainly located over the coordinated hydroxide and carboxylate groups, while HOMO-7 also possesses some participation of the non-bonding electrons of Pb(II). Considering that LUMO is assigned as a π* orbital of one of the pmdc ligands, the PL excitation of this compound may have both a mixed ligand-to-metal charge transfer (LMCT) and ligand-to-ligand charge transfer (LLCT) origin. These results mean that, under the employed experimental setup with the λ
ex = 325 nm laser excitation, and though the absorbed energy could be enough as to excite the S
10 state, the S
7 state could more probably be involved in the PL excitation process. In any case, given that these states are unlikely to be directly involved in the emission process according to Kasha’s rule [
62,
63], which anticipates the lowest-lying state to be the emission donor, we focused on the optimization of the first excited state (S
1) to elucidate the theoretical luminescence energy. According to this calculation, the S
1 lies 18868 cm
-1 (530 nm) above the ground state, meaning that the PL emission occurs from that state and that there must be an energy transfer from S
7 and S
10 during the structure relaxation.
To confirm the validity of the previous calculations, we also computed the excitation and emission spectra by means of the ESD module in ORCA using the same mononuclear model. The calculated excitation spectrum shows a symmetric band peaking at 350 nm, suggesting that there is a slight blue shift of the calculated band with respect to the experimentally measured one (
Figure 6), whose maximum seems to be over 400 nm although the band might be somewhat mixed with the emission. Nonetheless, the excitation band for the main emission (λ
em = 550 nm) may be well observed experimentally at 15 K, in which the excitation band is centred at 390 nm, thus confirming the slight blue shift for the calculated spectrum. On its part, the calculated emission spectrum reproduces very well the band maximum of the experimental measurement with a small blue shifted of ∆ = 15 nm. However, the shape is not that accurate although the calculated spectrum presents a distinguishable shoulder in the region of 460 nm.
The PL characterization of the material was also conducted at low temperature by placing the sample into a close cryostat linked to a helium pump. The emission spectrum recorded at the same conditions (λ
ex = 325 nm) resembles that of RT although the first shoulder (λ
em = 460 nm) becomes a well-defined and somewhat structured band (with higher relative intensity compared to RT,
Figure S8). Moreover, the main band, at λ
em,max = 525 nm at 15 K, is somewhat split in such a way that the resulting wide maxima is broadened and covers the region of 500-600 nm. The occurrence of the band at 600 nm as well as the increasing intensity of the band at 460 nm seems to be a consequence of the decreasing non-radiative vibrational quenching due to the low temperature, because the excitation spectra collected at all emission wavelengths show the same pattern (
Figure S9). This fact means that there is no significant change in the radiative pathways by the effect of the temperature. The experimental lifetimes were also estimated for the two main bands of the emission spectrum for comparative purposes. Surprisingly, the first band (λ
em = 460 nm) consists of a slightly shorter fluorescent process (τ = 0.8(2) ns), an opposite result to the usual trend by which lowering the temperature brings an enlargement of the lifetimes due to a decrease of the vibrational quenching (
Figure S10). However, this unexpected behaviour may be understood with the analysis of the second and dominant band of λ
em,max = 525 nm, for which the average lifetime is substantially lengthened (2337(56) µs at 15 K vs 72(3) µs at RT). Such a large lifetime, in the range of milliseconds, may be classified within the low part of long-lived phosphorescence [
19], an emission that leads to an afterglow that is traced by human eye once the excitation source is turned off. The large difference in the emission lifetime between the first and second bands is clearly observed in the time-resolved emission spectra (TRES) experiment conducted at low temperature (
Figure S11). Immediately after the absorption of the pulsed light, the first band fully disappears and the emission spectrum consists of a unique and symmetrical band centred at 525 nm, which is maintained along the whole experiment. In this regard, it is worth noticing that the second band (at low temperature) contains at least two bands of different lifetimes given that the high intensity around 550-600 nm is rapidly vanished after 1 ms in the TRES.