Maria Cristina Cassani*a,
Francesca Gambassia,
Barbara Ballarina,
Daniele Nanni*a,
Ilaria Ragazzinia,
Davide Barrecab,
Chiara Maccatoc,
Antonietta Guagliardid,
Norberto Masciocchie,
Alessandro Kovtunf,
Katia Rubinig and
Elisa Boaninig
aDept. of Industrial Chemistry “Toso Montanari”, Bologna University, Viale Risorgimento 4, I-40136, Bologna, Italy. E-mail: maria.cassani@unibo.it; daniele.nanni@unibo.it; Tel: +39 051 2093700 Tel: +39 051 2093623
bCNR-ICMATE, INSTM, Dept. of Chemical Sciences, Padova University, Via Marzolo 1, I-35131, Padova, Italy
cDept. of Chemical Sciences, Padova University, INSTM, Via Marzolo 1, I-35131, Padova, Italy
dInstitute of Crystallography, To.Sca.Lab, National Research Council, via Valleggio 11, I-22100 Como, Italy
eDept. of Science and High Technology, To.Sca.Lab., University of Insubria, via Valleggio 11, I-22100 Como, Italy
fInstitute of Organic Synthesis and Photoreactivity, (CNR-ISOF), Via P. Gobetti 101, I-40129 Bologna, Italy
gDept. of Chemistry “Giacomo Ciamician”, Bologna University, Via Selmi 2, I-40126, Bologna, Italy
First published on 7th June 2021
A copper-based metal–organic framework (MOF) was prepared using a new linker, a 5-substituted isophthalic acid bearing a propargyl carbamate group, intended to provide a terminal alkyne function protruding from the material surface to generate supported gold species for potential catalytic applications. The novel material was fully characterized by spectroscopic analyses of different kinds: FTIR, Raman, EDX, and XPS, as well as by thermal and surface area measurements. Synchrotron X-ray diffraction data analysis, in particular, revealed that this MOF, labelled [Cu(1,3-YBDC)]·xH2O (x ∼ 2), where Y stands for the pendant alkYne and BDC for benzene dicarboxylate, contains a complex network of 5-substituted isophthalate anions bound to Cu(II) centers, arranged in pairs within paddlewheel (or “Chinese lantern”) fragments of Cu2(μ-COO)4(D)2 formulation (D being a neutral Lewis base), with a short Cu⋯Cu distance of 2.633(4) Å. Quite unexpectedly, the apical atom in the paddlewheel structure belongs to the carbamate carbonyl oxygen atom. Such extra coordination by the propargyl carbamate groups drastically reduces the MOF porosity, a feature that was also confirmed by BET measurements. However, the MOF functionality is retained at the external crystal surface where 2% of active terminal alkynes is located.
As anticipated, benzene-polycarboxylato ligands are frequent choices for the construction of stable and performing metal–organic networks. One of the most common organic linkers is benzene-1,3,5-tricarboxylic acid (BTC). The related 3D-[Cu3(BTC)2(H2O)3] compound (also called HKUST-1 or Cu–BTC),12 one of the most extensively studied 3-dimensional porous coordination polymers, is a highly porous material (with a specific surface area, SSA, reported to be ca. 690 m2 g−1 in the original paper12 and much higher than 1500 m2 g−1 in the marketed materials) in which {Cu2} units are coordinated to four carboxylate groups to give the well-known paddle-wheel unit.6,13–17 Simpler ligands, like benzene-dicarboxylates (BDC), thanks to their structural rigidity (and simultaneous charge balancing effects of M(II) centers), have also been widely employed, as they can bear an additional, pendant and chemically tailored, functionality on the benzene ring.18–20
In recent years, our research group has shown that a propargyl carbamate [–N(H)C(O)O–CH2–CCH] group anchored on different oxide supports (SiO2, Al2O3, TiO2, Fe3O4) is capable of straightforwardly reducing Au(III) to Au(0), yielding supported gold nanoparticles (AuNPs) without the addition of any external reducing and/or stabilizing agent.21–23 The reactivity of the triple bond within this molecule has been additionally exploited to fabricate a novel amino-sulfide branched silica support by radical click thiol-yne (TYC) chemistry.24 Based on these results and taking into account that the chemical, structural, and functional behaviour of the gold nanoparticles reported above depend on the physico-chemical environment dictated by the support, we envisaged that the versatility demonstrated by the propargyl carbamate residue could be further exploited by anchoring it to different solid supports than oxides, for example within alkynyl-derivatized MOFs. Indeed, the isolation of a metal–organic framework starting from a suitably functionalized organic linker would ensure an even, dense, and possibly highly symmetric distribution of the reactive alkyne moiety in the material. Additionally, the material porosity and the presence of the metal nodes could affect the reactivity of the alkyne and induce different effects on the ensuing gold nanoparticles, than when the alkyne is attached onto the oxidic surfaces cited above.
In this paper we report the synthesis of a new organic linker belonging to the substituted BDC class, i.e. 5-(2-{[(prop-2-yn-1-yloxy)carbonyl]amino}ethoxy)isophthalic acid (which we have labelled 1,3-H2YBDC, where Y stands for alkYne and BDC for benzene dicarboxylate), bearing a propargyl carbamate residue; we also discuss the optimized conditions for the synthesis of the novel Cu(II)–MOF, [Cu(1,3-YBDC)]·xH2O derived therefrom, and, finally, we present a detailed, comprehensive multi-technique characterization by means of complementary analytical, structural and imaging tools.
As shown in Scheme 2, the final MOF was obtained by reacting the dicarboxylic species 1,3-H2YBDC with Cu(NO3)2 · 2.5H2O in refluxing 2-propanol for 24 h employing a Cu:L molar ratio of 1.8:1 (see Experimental section for further details).
Scheme 2 Synthesis of [Cu(1,3-YBDC)]·xH2O and image of a pellet made by pressing at 100 bar for 2 min its polycrystals. |
Employment of 2-propanol was found crucial: indeed the use of ethanol brought about partial esterification of the linker, hence causing erratic MOF yields and purity.6 After filtration, a turquoise polycrystalline powder was obtained with yields above 90%. Thanks to the nearly quantitative formation of compound 6, the overall yield of the entire process approaches 90%, making the scale-up of this efficient synthesis highly viable.
X-ray evidences, shown in the ESI† file and commented in the Experimental section, highlight the formation of Cu(OH)2 and Cu2(OH)3NO3, in variable (minimal, but visible) amounts. Even by keeping them at a minimum, still a few percent (w/w%, from TGA data) of these hydroxy-salts are always present and contaminate the MOF up to a maximum value of ca. 10 w/w%. X-ray powder diffraction, the most suitable technique for assessing the correct phase composition of the mixture, however, did not help in quantifying it, as the extreme texture of the contaminants made visible only a few (00) diffraction peaks. Indeed, crystal phase quantification by the Rietveld method25 relies on the experimentally determined scale factors and, if preferential orientation (texture) is present, on the accurate estimation of the so-called orientation distribution function.26 In the present case, where (00) diffraction peaks only were observed, neither the ODF nor the scale factor of the fully oriented (nanosized) crystal plates are accessible, as they mutually 100% correlate.
Finally, the target product is stable in water at r.t and at neutral pH and such stability was evaluated using XRD. After keeping the product in water for 24 h, no changes in the XRD pattern were observed (see the ESI file†).
In addition, the IR spectrum (Fig. S13†) possesses two intense bands at 1585 and 1374 cm−1 which are particularly diagnostic for the asymmetric and symmetric stretching mode of the carboxylate group (RCO2−). The difference Δν (213 cm−1) between the νas(1585 cm−1) and νs(1372 cm−1) stretching vibrations indicates that each carboxylate group is bonded to two different copper atoms in a bridging bidentate mode, whereas the band at 731 cm−1 can be attributed to bonding between copper and oxygen of YBDC.15,28–30 Another band, attributed to the ν(CO) stretching of the carbamate residue, bound to the Chinese lantern in apical position (see Crystallochemical analysis), can be found at 1628 cm−1, i.e. at ca. 59 cm−1 lower frequencies than in the pristine ligand. This red shift is in line with previous observations reported for analogous fragments coordinating, through their carbonylic oxygen atoms, to Cu(II) ions of molecular (mononuclear) complexes.31
Complementary information was gained by Raman spectroscopy; Fig. S14 in the ESI† file compares the spectra measured for [Cu(1,3-YBDC)]·xH2O, 1,3-H2YBDC, and Cu(NO3)2 · 2.5H2O. The appearance of the Raman active band at 2135 cm−1, unobserved in the IR spectra, is assigned to –CC– stretching of the alkyne triple bond,28 and confirms that, during complexation, no changes in the structure of the organic linker have occurred nor is the triple bond involved in π–metal bond. The signals located at 1605 and 1004 cm−1 are associated with the benzene ring ν(CC) stretching modes, whereas the peaks at 807 and 745 cm−1 are ascribed to out-of-plane δ(C–H) and δ(CC) ring bending vibrations, respectively.32,33 In the low-frequencies region (600–200 cm−1 range), [Cu(1,3-YBDC)]·xH2O exhibits, in agreement with previous studies, two peaks at 495 and 276 cm−1, respectively assigned to the vibrational stretching modes of equatorial and axial Cu–O bonds.34,35
The elemental analysis of the synthesized sample is in good agreement with the theoretical chemical composition, showing a Cu/N molar ratio of ca. 1 (see Experimental part). Thermogravimetric analyses (TGA) were carried out in air to determine the thermal stability of [Cu(1,3-YBDC)]·xH2O. The corresponding TGA profiles of 1,3-H2YBDC and [Cu(1,3-YBDC)]·xH2O are reported in Fig. 1. The two contaminants found in small quantities in the synchrotron X-ray diffraction analysis (vide infra) decompose at ca. 132 °C (Tonset for Cu(OH)2)36 and at ca. 175 °C (Tonset for Cu2(OH)3NO3)37 and are not evidenced in the TG trace, confirming their low percentage.
Fig. 1 TGA curves (continuous red lines) and their first derivatives (dashed lines) for 1,3-H2YBDC (top) and [Cu(1,3-YBDC)]·xH2O (bottom). |
In the ideal absence of contaminants, TGA experiments can be safely used to estimate the amount of water content, but, as [Cu(1,3-YBDC)]·xH2O contains water molecules loosely bound to the framework (vide infra), very much as the prototypical channel hydrates found in zeolites and in molecular organic species of pharmaceutical interest, their amount can vary depending on the history of the sample and on the relative humidity conditions.38 Accordingly, the stoichiometry of this crystal phase is somewhat undefined, and may vary depending upon the external conditions.
The TGA plot for 1,3-H2YBDC (of C14H13NO7 formula, mw 307.25 g mol−1) shows three main weight losses between 200 and 600 °C, due to organic material decomposition, and zero residual weight. The first two steps are interpreted by progressive loss of the propargyl fragments (obs. 48.5%, calc. for C6H8NO3 46.3%). Similarly, the copper-containing product (of C14H11CuNO7 formula, mw 368.79 g mol−1) shows two decomposition steps in the 200–350 °C range (attributed, as above, to the loss of the C6H8NO3 residue, obs. 38.2%, calc. 38.5%), terminating at a temperature ca. 150 °C lower than in the pristine organic ligand. Such lower thermal stability of the organic skeleton within the MOF is tentatively attributed to assistance, during decomposition, of redox process(es) catalyzed by Cu(II) ions. Furthermore, a residue of ≈26 wt% is present at 400 °C, with no significant variation up to 800 °C, which is presumably due to residual CuO. As a residual 21.6% only is calculated if the starting material were pure, the excess residue at high T speaks for the presence of carbonaceous residuals and only marginally to Cu-rich contaminants, the nature of which is presented in the Experimental section.
Residue a | Residue b | |
---|---|---|
a Labelling of atoms is shown in Fig. 2A. | ||
Chain sof | 0.51(1) | 0.49(1) |
Torsion angles | ||
C2–C3–O5–C9, ° | 26(1) | 53(1) |
C3–O5–C9–C10, ° | 93(1) | 163(1) |
O5–C9–C10–N1, ° | 35(3) | −108(3) |
C9–C10–N1–C11, ° | −104(4) | 24(6) |
C10–N1–C11–O7, ° | 18(7) | −48(8) |
N1–C11–O7–C12, ° | 163(6) | −53(9) |
C11–O7–C12–C13, ° | 156(9) | −162(10) |
Bond distance | ||
Cu–O6, Å | 2.26(10) | 2.14(8) |
The extended fragments shown in panels C and D highlight the location of the carbonyl oxygen atoms (O6a and O6b) completing the Cu2(μ-carboxylate)4 coordination sphere in apical positions. The complete structures (E and F panels) for the idealized and periodic a and b conformers (not including the weakly bound water atoms) show the terminal alkyne moieties pointing coherently into the same (tetrad axis-generated) crystal cavity. The remaining channels, accounting for ≈20% void space in both a and b models, are partially filled by water molecules, possibly disordered and of non-stoichiometric character. As powder diffraction of such complex material can provide only limited information on these guests, their amount and geometrical properties will not be discussed any further.
A simplified overall picture of the crystal structure is shown in Fig. 3 by removing the H atoms and the propargyl carbamate moiety from the crystal framework, but leaving the carbonylic oxygen atom bound to Cu. These sketches highlight the presence of separated (weakly corrugated) 2D layers, very similar to those found in [Cu(1,3-BDC) (C5H5N)], where D = pyridine N atom.41 However, in [Cu(1,3-YBDC)]·xH2O, the layers are not disjointed, but are interlinked by the long propargyl carbamate residues (in both a and b conformations) completing the Cu2(μ-COO)4 coordination sphere with the apical O6a and O6b atoms.
Additional important information was gained by XPS analysis, which was used to characterize material surface chemical composition and element chemical states. The analysis revealed the presence of C, N, O, and Cu, in line with the effective MOF composition (Fig. 5A). Quantitative analyses yielded the following data: C, 58.7 at%; N, 5.2 at%; O, 31.3 at%; Cu, 4.8 at% (Cu/C atomic percentage ratio = 0.08). The C 1s signal could be deconvoluted by means of four contributing components (see Fig. 5B) related to the structure of the MOF framework: (I) BE = 284.8 eV (typical value ≈40% of the total C signal), assigned both to adventitious carbon contaminations and to carbon atoms with homonuclear (all-carbon) contacts only; (II) BE = 286.2 eV (≈30% of the total C area), related mainly to contributions from carbon atoms bound to N and O atoms in the ligand skeleton; (III) BE = 287.9 eV (≈20% of the overall carbon content), attributable to Ar–COOH groups, and (IV) BE = 289.6 eV, attributable to –NCOO– moieties in the branching residue.29,42,43 Three bands contributed to the O 1s peak (Fig. 5C): (V), BE = 530.2 eV (typical value ≈8% of the total O signal), resulting from the concurrent contribution of O6x atoms (x = a,b) bound to copper;29,43–45 (VI), BE = 531.8 eV (≈40% of the total), due to O atoms bound to C and hydroxyl groups42,46,47 and (VII), BE = 533.3 eV, related to carboxylic oxygen atoms and/or a concurrent contribution from adsorbed water.42,43 The N 1s peaks (Fig. 5D) showed a single contribution centered at BE = 400.3 eV.48
Fig. 5 XPS analysis for the target [Cu(1,3-YBDC)]·xH2O specimen: (A) wide-scan spectrum; (B) C 1s, (C) O 1s, (D) N 1s, (E) Cu 2p photoelectron peaks. |
The analysis of copper chemical state (Fig. 5E) required particular attention. The energy positions of Cu 2p3/2 and Cu 2p1/2 spin–orbit components (BE = 934.8 eV and 954.7 eV, respectively) could be ascribed to Cu(II) centers in the copper-containing metal–organic framework and in Cu(II) hydroxides impurities.15,29,42,43,47 This attribution was in agreement with the presence of intense shake-up satellites centered at BE values ≈9.0 eV higher than the main spin–orbit components, that are considered as a finger-print for the predominant presence of d9 copper(II) centers and are not detected in the case of Cu(I) (d10, closed-shell).49,50
The BET model fits the isotherm slightly better than Langmuir in the low pressure region (Langmuir R2 = 0.99757, BET R2 = 0.99986); the Langmuir surface area was found to be 18.4 ± 0.9 m2 g−1, while the BET surface area was 14.5 ± 0.8 m2 g−1. The estimated pore volume was 46 mm3 g−1, while that derived from mercury52 was about two times larger (105 mm3 g−1).
The reported BET for Cu–BTC MOFs are usually above 1000 m2 g−1,53 however the presence of ligands that offer and extra coordination to Cu can significantly drop the surface area to 100 m2 g−1 or below.54,55 Using the computational approach proposed by Düren et al.56 we have computed, for an idealized ordered and anhydrous crystal phase (such as those provided in the separate a and b models cited above), a rather small specific surface area of the internal voids (of only 100 m2 g−1), accounting for 17.1% of the crystal volume. Given that this approach usually provides an estimate of an upper limit (experimental values being often less than one half of what predicted), our measurements are roughly in line with what expected, and possibly further lowered by the occluding effect of coprecipitated nanosized Cu hydroxide and hydroxonitrates lowering the channel accessibility. As a reference values for isostructural compound, we have taken the (pyridine free) copper isophthalate easily derived from the published coordinates of the MOF presented in literature; in this case, SSA = 1509 m2 g−1 and void percentage = 49.9%.57
Using these values and normalizing to the number of Cu2 dumbbells in the structure of [Cu(1,3-YBDC)], the size of the MOF cavities, arranged in poly-hourglass shaped channels running along a and b, is only 141 Å3. However, taking into account the approximate volume of the tetrachloroaurate anion [the Au(0) precursor, ≈142 Å3] and the tabulated size of a single Au atom [≈43(2) Å3],58 we hardly envisage that the MOF could be used for capturing gold species inside the material, particularly, due to charge-balancing effects by additional counterions and by the expected presence of a solvation sphere. Thus, Au species may bind primarily on the MOF external surface, which is covered by a dense and even array of protruding propargyl-carbamate residues. Considering the size of crystalline domain estimated by the XRD model (of the order of 100 nm for isotropic particles), nearly 2% of the ligands lie on the crystal surface, possibly acting as catalytic sites.
Field emission-scanning electron microscopy (FE-SEM) analysis was carried out by means of a Zeiss SUPRA 40VP instrument equipped with an INCAx-act PentaFET Precision spectrometer (Oxford Instruments) for energy dispersive X-ray spectroscopy (EDXS) characterization. The used primary beam acceleration voltages were comprised between 1 kV (for imaging) and 20 kV (for EDXS analyses).
X-ray photoelectron (XPS) characterization was performed using a Perkin-Elmer Φ 5600-ci instrument, at an operating pressure <10−8 mbar, using a standard AlKα excitation source (hν = 1486.6 eV) and an analysis area with a diameter of 800 μm. Survey scans were acquired in the 0–1300 eV range (187.8 eV pass energy, 0.8 eV per step, 0.02 s per step). Higher resolution scans for the single photopeaks were recorded using the following settings: 58.7 eV pass energy, 0.1 eV per step, 0.05 s per step. Binding energy values (BEs; uncertainty = ±0.2 eV) were corrected for charging by assigning to the adventitious C 1s peak associated with adventitious hydrocarbons a value of 284.8 eV.60 After a Shirley-type background subtraction,61 curve fitting was carried out by the XPS peak software.62 Atomic percentages (at%) were evaluated from integrated peak areas using sensitivity factors supplied by Perkin-Elmer.42
Raman spectra were recorded using a HORIBA Jobin Yvon T64000 spectrometer equipped with three monochromators in double subtractive configuration. The spectrometer was coupled to an Olympus BX40 confocal microscope equipped with 100×, 50×, 20×, and 10× objectives, for a lateral resolution lower than 1 μm with the 100× objective. An Ar+ laser emitting at 514.5 nm was used in which its output power was limited in order to avoid sample damaging (70–100 mW) and with long times and accumulations (about 50–60 min for spectrum).
The adsorption isotherm was measured by using a static volumetric apparatus (ASAP 2020, Micromeritics, USA). The absolute total surface area (SA) of the solid powder (>1.5 m2) was above the lowest measurable value (0.5 m2). The errors on SA were calculated by standard deviation on 5 consecutive measurements. The sample was degassed at 1 × 10−3 mbar at 323 K for 2 h prior the measurement and after this operation ca. 5% mass loss was observed, in according with TGA. The BET model is usually considered the reference for MOF,1 however, in order to obtain the highest agreement between the measured area and the effective geometrical area, the correct interval of pressure must be taken in account;63 in our case the optimal pressure range was between 0 and p/p0 < 0.075.
Footnote |
† Electronic supplementary information (ESI) available: For additional spectroscopic and analytical data for the materials characterization (ESI-MS, NMR, ATR-IR, AAS, X-ray crystallographic data) and crystallographic data in CIF format (CSD code 2054304). See DOI: 10.1039/d1ra02686k |
This journal is © The Royal Society of Chemistry 2021 |