Highly Fluorescent Metal-Organic-Framework. Nanocomposites for Photonic Applications

Transcription

1 Supporting Information Highly Fluorescent Metal-Organic-Framework Nanocomposites for Photonic Applications A. Monguzzi 1 *, M. Ballabio 1, N Yanai 2,3, N. Kimizuka 2, D. Fazzi 4, M. Campione 5, and F. Meinardi 1 * 1 Dipartimento di Scienza dei Materiali, Università degli Studi Milano Bicocca via R. Cozzi 55, Milan, Italy 2 Department of Chemistry and Biochemistry, Graduate School of Engineering, Center for Molecular Systems (CMS), Kyushu University, Moto-oka 744, Nishi-ku, Fukuoka , Japan 3 PRESTO, JST, Honcho 4-1-8, Kawaguchi, Saitama , Japan 4 Max-Planck-Institut für Kohlenforschung (MPI-KOFO) Kaiser-Wilhelm-Platz 1, D Mülheim an der Ruhr (Germany) 5 Department of Earth and Environmental Sciences, Università degli Studi Milano-Bicocca, Piazza della Scienza 4, Milano, Italy. Corresponding Authors * angelo.monguzzi@unimib.it * franco.meinardi@mater.unimib.it 1

2 1. Experimental Materials. Poly(methylmetacrylate) (PMMA, CAS number ) and solvents were purchased from Sigma-Aldrich and used as received. MOF nanocrystals were synthesized via the microwave method following the procedure previously published. The benzene dispersion have been prepared by dispersing 0.2 wt% of the MOF powder. Polymer films were prepared by using 1 ml of an analogous nanocrystal dispersion in toluene mixed with 150 mg of PMMA. The composite were stirred more than 12 hours then it was drop casted on a glass substrate and dried in a nitrogenfilled glove box. MOF Synthesis. The ligand 4,4 -(anthracene-9,10-diyl)dibenzoic acid (adba) was synthesized following the literature procedures. [1] The MOF crystals for single crystal X-ray analysis were synthesized by solvothermal method. For liquid dispersion samples, we synthesized MOF particles by using microwave instead of solvothermal method since it gives better particle dispersibility in benzene probably due to the smaller crystal size. Microwave reactions were performed using the Biotage Initiator 2.5 under continuous stirring. The power was around W with temperature rise 3-4 ºC/sec. No pressure increase was observed during the synthesis. The synthesis of nano- MOF: a mixture of adba (20 mg, mmol), Zn(NO 3 ) 2 6H 2 O (14 mg, mmol), and 2.5 ml N,N-diethylformamide (DEF) in a sealed glass vial and heated at 120 C for 3 h in microwave, filtration through 200 nm filter to remove small crystals, washing several times with DEF, and subsequent drying under vacuum at room temperature. Yield = 11 mg (33%). Elemental analysis for [Zn(C 28 H 16 O 4 )(C 5 H 11 NO) 2 ]: calcd. (%): C 66.72, H 5.6, N 4.1; found (%): C 66.23, H 5.55, N The synthesis of MOF-bpy: a mixture of adba (20 mg, mmol), Zn(NO 3 ) 2 6H 2 O (14 mg, mmol), 4,4 -bipyridine (bpy) (3.75 mg, mmol), and 2.5 ml N,N -dimethylformamide (DMF) were placed in a sealed glass vial and heated at 120 C for 3 h in microwave, filtration through 200 nm filter to remove small crystals, washing several times with DMF, and subsequent drying under vacuum at room temperature. Yield = 20.5 mg (35%). Elemental analysis for [Zn 2 (C 28 H 16 O 4 ) 2 (C 10 H 8 N 2 )] DMF H 2 O: calcd. (%): C 68.44, H 4.08, N 3.47; found (%): C 67.94, H 4.08, N The synthesis of MOF-dabco [1] : reaction conditions are similar to the microwave synthesis of MOF-bpy by just changing the co-ligand from bpy to 1,4-diazabicyclo[2.2.2]octane (dabco). Yield = 18.2 mg (32.3%). Elemental analysis for [Zn 2 (C 28 H 16 O 4 ) 2 (C 6 H 12 N 2 )] DMF H 2 O: calcd. (%): C 66.90, H 4.58, N 3.6; found (%): C 66.53, H 4.47, N For single crystal X-ray structure analysis, crystals were synthesized by solvothermal method using teflon autoclave and 2

3 heated at 120 C for 24 h, then cooled to room temperature at 1 C/min. The amount of reagent and solvent are identical to the microwave method. Figure S1. X-ray power diffraction (XRPD) patterns (black) of (a) nano-mof, (b) MOF-bpy and (c) MOF-dabco with corresponding simulated patterns made from their crystal structures (blue). The phase purity of the MOFs was confirmed by the good agreement between the experimental and simulated patterns. Methods. XRPD of nano-mofs was performed with a PANalytucal X Pert PRO with Cu K α radiation in Bragg-Brentano geometry. Transmission electron microscopy (TEM) was performed with a Jeol JEM 1220 microscope, with an electron beam potential set to 80 kv. Photoluminescence (PL) and photoluminescence excitation (PLE) spectra were acquired from a Variant Eclipse (bandwidth 1 nm) using quartz cuvettes with 0.1 cm of optical path. Time-resolved PL spectra have been recorded by monitoring the samples emission decay at 440 nm (2.82 ev). The nanocrystals dispersions have been excited with a pulsed LED at 340 nm (3.65 ev, EP-LED 340 Edinburgh Instruments, pulse width 80 ps). The MOF nanocomposites were excited with a pulsed laser at 405 nm (3.06 ev, EPL-405 Edinburgh Instruments, pulse width 90 ps). In this case the 405 nm laser was preferred to avoid the excitation of the PMMA wrapping the nanocrystals. Data were collected with an Edinburgh Instruments FLS-980 spectrophotometer, with a 5 nm bandwidth and a time resolution better of 0.2 ns. Computational Methods. Excited state properties of DPA clusters (i.e. dimers, trimers or aggregates) as extracted from MOFs were studied in the framework of molecular exciton theory. [2] The molecular structure of each DPA molecular aggregate (see Section 6 of the Supporting 3

4 Information) was re-optimized at the DFT level of theory (CAM-B3LYP/6-31+G**) constraining the inter-molecular distances and relative orientations (i.e. as extracted from the crystalline structure), but relaxing the bond lengths and the valence angles. This procedure was applied for each aggregate extracted from MOF. Singlet excited states were investigated adopting the linearresponse TDDFT (TD-CAM-B3LYP/6-31+G**) theory and the semi-empirical approach ZINDO/S. The singlet exciton coupling V was computed adopting the supermolecular method as reported and deeply discussed in Refs [2,3,4], which consists in: (a) calculating the excitation energies of molecular dimers/aggregates, (b) identifying the dimer s states associated to each excited state of the isolated molecules and (c) evaluating the exciton coupling V as: = 1 2 ( ) ( ) where ω +/- are the energy of the dimer excited states and ω a/b are the on-site excitation energies of the isolated molecules. [3] To set the proper correspondence between the excited states of the isolated molecules and those of the molecular dimers/aggregates (step b), we adopted a methods based on the projection of the transition density matrix (TDM) of the dimer onto those of the monomers. Full details are reported, discussed and tested in Refs 3 and 4. The structural reoptimization was necessary to avoid unphysical results in the diagonalization of the total Hamiltonian due to the mismatch of the computed average potential (V mean ) and the real one (V real ) that results in different computed atomic positions from the real ones. Although differences can be very small, within 10-4 Å, they can lead, for instance, to negative energy eigenvalues. To avoid this problem, the re-optimization procedure was performed constraining the reciprocal positions of the dimers (i.e. without altering the inter-molecular interactions) and re-optimizing the bond lengths and angles of each molecule. Differences between the re-optimised structures and the experimental ones are minimal (10-3 Å), tough necessary in order to compute positive eigenvalues, hence physical couplings and electronic properties. 4

5 2. Fluorescence Quantum Yield determination. Figure S2a shows the bi-dimensional structure of ADB. The electronic properties are fully determined by the rigid anthracene core. This molecule can thus absorb light in near UV region, and then emit blue photons with a characteristic energy distribution peaked at 2.85 ev (Fig. S2b). The radiative time (τ rad ) of this transition, calculated from absorption data with the Strickler-Berg formula, is set to 6.4 ns, similar to DPA. [5] The τ rad can be calculated using ( ) ( ) ε ν τ rad = n ν fl dν (Eq. S1) ν ν 3 fl 1 = ( ) ( ν) PL ν dν PL 3 ν dν (Eq. S2) where n is the solvent refraction index, the frequency energy in cm -1 and ( ) is the fluorescence spectrum. The Strickler-Berg equation is often considered a little precise method to determine radiative lifetime of emitters, but the high oscillator strength and the rigidity of these molecules satisfy the requirements that the formula need to be predictive. From the ratio between τ and τ rad we can evaluate the emission QY of ADB by Eq. S1, which correspond to = (Eq. S3) By freezing the solution at 77 K, the emission lifetime increase to 6.5 ns with a corresponding yield equal to unity (Fig. S2c). As like as for DPA, the ABD molecule is poorly quenched by the intramolecular vibrations, showing a very high QY even at room temperature. 5

6 Figure. S2 a) Bidimensional representation of ADB molecule. b) Absorption (indigo line) of ADB in diluted DMF solution ( M). c) PL decay of ADB at room temperature (black circles) and at 77 K (red triangles). Both the decays show a single exponential behavior, with a characteristic lifetime of 5.3 ns and 6.5 ns, which lead to a QY of 0.81 and 1.01 respectively. Relative and Absolute QY Measurements. The fluorescence QY of the ADB ligand derived from time-resolved PL data has been verified has been with relative and absolute methods. First, the QY was determined relative to 10-5 solution of DPA in ethanol (QY= 0.95), [6] which has used as secondary standard according to the following equation QY =QY, Eq. S4 where QY, As unk, I unk, P unk and η unk represent the quantum yield, absorptance at the excitation intensity, integrated photoluminescence spectral profile, excitation power density, and refractive index of the investigated material. The corresponding terms for the subscript std are for the reference quantum counter DPA at the identical excitation wavelength. To avoid any detrimental effect of the scattering of the excitation light that could avoid a proper determination of the absorptance of the samples we measured the absolute QY of nanocomposite films using an integrating sphere. Adapting the protocol by DeMello et al. [7], the amount of absorbed photons has been calculated by considering the ratio between the laser intensity in the sphere with and without the sample. The light generated by self-absorption and by re-absorption of the scattered laser light has been taken into account by measuring the signal light generated in the sphere while the laser beam did not hit directly the sample. 6

7 Table S1. Fluorescence QY data for the ABD ligand and MOF nanocrystals in benzene suspension and embedded in a PMMA matrix obtained with the Strickler-Berg analysis and measured with the relative method and with an integrating sphere. The asterisk * indicate a MOF-dabco nanocrystal with increased size (500 nm, Fig. S7) in respect to the first MOF-dabco sample analyzed (200 nm, Fig. 3 in the main text) Benzene suspension PMMA matrix Sample ADB MOF MOF-bpy MOF-dabco MOF-dabco* Method Stickler-Berg Relative Stickler-Berg Int. sphere 3. MOFs Co-ligand molecular structure Figure S3. a) Molecular structure of 4-4 bipiridine (bpy) and b) 1,4-diazabicyclo[2.2.2]octane, used as co-ligands in MOF. 4. MOFs Crystallographic Data Table S2. Lattice parameters for MOF nanocrystals. MOF MOF-bpy MOF-dabco Symmetry Monoclinic Monoclinic Orthorhombic a (Å) b (Å) c (Å) β Volume (Å 3 ) Molecules per unit cell

8 5. Penetration Length and Forster Hopping Rate Calculation The penetration depth of the incident light x was evaluated in the Lambert-Beer approximation. Using Eq. S5 it is possible to found x as a function of the molar extinction coefficient ε, the density M of ADB ligands within in the crystal in units of moles*dm -3, and the optical absorbance A at fixed photon wavelength. =10 = (Eq. S5) In order to estimate the one dimensional diffusion length of the singlet excitons in the crystal lattice, we need to know the probability that the energy jumps from a molecular site to another. This probability is the hopping rate, which we suppose equal to a Förster-like energy transfer rate between donor and acceptor, as shown in Eq. S3. = = (Eq. S6) depends on the spontaneous decay rate of the donor, on the intermolecular distance and on the Förster radius. The Förster radius depends on both geometric and spectroscopic quantities since the transfer is mediated by the interaction of two transition dipoles. According to Eq. S3, 4. The dipole orientation factor is obtained by XRD analysis and is set equal to 0 when dipoles are orthogonal, 1 when are parallel and 4 when collinear. is the refractive index of the medium, which has been approximated to [8] Finally, QY is the emission quantum yield of the MOF and ( ) is the overlap integral between absorption ( ) and emission spectra ( ). =0.211 ( ) (Eq. S7) ( )= ( ) ( ) (Eq. S8) 8

9 The results of the calculation are summarized in Tab. 2. The hopping rate was evaluated between every first neighbour in each crystallographic direction. We point the attention to the first neighbours along the z-axis of MOF-bpy and MOF-dabco. In this case, in the point-dipole approximation, the two adjacent molecules are orthogonal, therefore = 0, and consequently the Förster transfer is suppressed. In order to have a non-zero rate in those directions we took into account the second neighbour. We notice that our approach is conservative about the calculated diffusion lengths along the z- axis, because for closely packed molecules the point-dipole approximation is not fulfilled. Therefore, we do not consider the next neighbours contribution to the overall Förster-mediated diffusion of singlet excitons. Table S3. Constants and parameters used in the calculation of the Förster transfer-mediated hopping rate in nano- MOFs. The z superscript indicate if the calculated quantities are referred to the first neighbour (') or to the second neighbour (") along the z-axis. Sample Axis R (Å) Θ 2 R 0 (nm) k hop (THz) D(cm 2 s -1 ) L(nm) MOF MOF-bpy MOF-dabco x y z' x y z' z'' x y z' z'' Quantum-chemical calculations: ZINDO/S and TDDFT singlet exciton couplings. In this section we evaluation of the exciton coupling at the quantum chemical level exclusively done in order to validate the hypothesis of weak interacting systems. This result allows us considering any fluorescent ligand as an isolated molecule, enabling the estimation of the energy 9

10 transfer probability from directly measured structural and spectroscopy data using the classic Förster model (Section 5). The couplings were evaluated at the quantum chemical level on molecular dimers, as extracted from crystals (see Computational Methods in section 1 of the Supporting Information). Dimers are classified in accordance to the notation reported in Table 3. For instance: Dimer along x for MOF is the molecular dimer extracted from the crystal structure MOF along the axis x. For the cases of MOF and MOF-dabco, in Figure S4 we report two dimers, along the z direction (Dimer along z and Dimer* along z ), which show slightly different geometries in terms of bond lengths and relative orientations. This is due to the DFT constrained optimization procedure, which found two local minima by starting from the crystal structure as a guess. For the sake of clearness, however, we report the calculation on both the structures. Dimer along z R (6.0 Å) Dimer along y R (9.0 Å) MOF Dimer along x - R (9.2 Å) Dimer* along z R (6.0 Å) MOF-bpy Dimer along z R (4.9 Å) 10

11 Dimer along z R (4.8 Å) Dimer along y R (13.1 Å) MOF-dabco Dimer* along z R (4.8 Å) Figure S4. Nearest neighbor DPA dimers as extracted from MOF nanocrystals. The asterisk * marks an alternative dimer structure along the z-axis given by the DFT constrained optimization procedure, which found two local minima. Table S4. Singlet exciton coupling V (ev) computed at the ZINDO/S and TDDFT (CAM-B3LYP/6-31+G**) level for each DPA dimer as extracted from nano-mofs crystalline structure and partially re-optimized at the DFT level (see computational methods in Section 1 of the Supporting Information). For the case of MOF-dabco, the exciton coupling computed with respect to the entire crystalline unit cell is also reported. The asterisk * marks an alternative dimer structure along the z-axis given by the DFT constrained optimization procedure, which found two local minima. ZINDO/S (ev) TDDFT (ev) MOF Dimer along z Dimer along y: Dimer along x: Dimer* along z Unit cell MOF-bpy Dimer along z MOF-dabco Dimer along z Dimer along y Dimer* along z

12 Figure S5: ZINDO/S singlet exciton couplings computed amongst nearest neighbor (e.g. V 1-2 ) and not (V 1-3 and V 1-4 ) of a stack of 6 DPAs as extracted from MOF-dabco ev 2.83 ev PL Intensity Time (ns) Figure S6. Time-resolved PL decay of nano-mofs:pmma at room temperature (triangles) recorded at the emission maximum at 2.83 ev and at 2.20 ev under pulsed laser excitation. 12

13 Figure S7. a) Comparison between L exc, d and the calculated exciton diffusion lengths L i for the MOF-dabco* nanocrystal. i. e. a system with the same crystalline structure of MOF-dabco but increased crystals size. b) PL, PLE and time-resolved luminescence spectra of MOF-dabco* nanocomposites in PMMA. The triangles show the PL decay for bare nanocrystals in benzene suspension. References [1] I. M. Hauptvogel, R. Biedermann, N. Klein, I. Senkovska, A. Cadiau, D. Wallacher, R. Feyerherm, S. Kaskel, Inorg Chem 2011, 50, [2] D. Beljonne, J. Cornil, R. J. Silbey, P. Millie, J. L. Brédas, J. Chem. Phys. 2000, 112, [3] C. Quarti, D. Fazzi, M. Del Zoppo, Physical Chemistry Chemical Physics 2011, 13, [4] C. Quarti, D. Fazzi, M. Tommasini, Chemical Physics Letters 2010, 496, 284. [5] S. J. Strickler, R. A. Berg, The Journal of Chemical Physics 1962, 37, 814. [6] J. V. Morris, M. A. Mahaney, J. R. Huber, The Journal of Physical Chemistry 1976, 80, 969. [7] J. C. de Mello, H. F. Wittmann, R. H. Friend, Advanced Materials 1997, 9, 230. [8] A. Muthuraja, S. Kalainathan, Materials Research Innovations 2016, 20,