Porous and luminescent Dy-doped Co-BTC MOFs for label-free detection of tetracycline and vanadium traces in water

Abstract

Continuous intake of antibiotic residues and trace heavy metals in food is a growing health concern that weakens immunity. Sensitive, selective detection methods are therefore essential, motivating the development of new sensing materials. The present study reports the synthesis of dysprosium (Dy)-doped cobalt (Co)-trimesic acid (BTC)-based luminescent metal–organic frameworks (MOFs) via a single-step solvothermal method and their utilization in detecting traces of antibiotics (ABx) and metal ions. Extensive microscopic studies have shown that the prepared MOFs exhibit a highly porous, spherical morphology. Our results show that 5% Dy-doped Co-BTC MOF exhibits superior photoluminescence (PL) than those of the other MOFs prepared with 10% and 15% Dy doping. In addition, this 5% Dy-doped Co-BTC MOF is also able to achieve label-free identification/detection of tetracycline (TC) ABx as well as vanadium ions (V³⁺) with high selectivity and sensitivity from water samples through a perceptible “turn-off” PL behaviour. The limits of detection of TC and V³⁺ are estimated to be 0.2427 and 0.1587 ppm, respectively. Plausible mechanistic insights into the highly selective detection of TC and V³⁺ have also been proposed based on the experimental results.

Dy:Co‑BTC MOFs photoluminescence enabled label‑free turn‑off sensing of tetracycline and V³⁺ in aqueous medium.

1. Introduction

Contamination by antibiotic residues and metal traces poses a major threat to the health of living beings on a global scale.^1,2 Excessive use of such substances degrades food and contributes to water and soil pollution. However, screening of antibiotics (ABx) and metal traces is often challenging due to their low concentrations and their tendency to diffuse and accumulate in environmental and biological systems. Thus, recently, researchers have shown a considerable amount of interest in one of the ABx, i.e., tetracycline (TC), and trace metals like vanadium as they result in wastewater pollution and potential toxicological effects.^3,4 The presence of excess TC in food and aqueous media can promote the growth of many hazardous antibiotic-resistant bacteria (ARB).⁵ Such ARBs are a serious threat to public health, and they can affect people of any age. Nowadays, ARB inhibition has become a very challenging and concerning topic for health agencies. Further, TC intake beyond its permissible limit induces teratogenic effects, renal damage, liver dysfunction, and bone abnormalities.^5,6 The MRL limits stipulated by various regulatory agencies for TC intake are tabulated in SI Table S2. In a similar fashion, metal traces are also well known for their toxic effects. Vanadium is predominantly or extensively utilized in many industries, including the steel industry, battery manufacturing, catalytic and electronics industries, for the last two centuries.⁷ Regrettably, extreme use of vanadium devastatingly causes water contamination and health issues, including kidney disorder, oxidative stress, and lung damage.⁸ The World Health Organization sets preventive measures on excessive use of vanadium and allows a maximum permissible concentration, i.e. 100 µg L⁻¹, in drinking water.⁴ Initially, the Italian National Institute of Health (legislative decree no. 31/2001) proposed a maximum permissible limit of 50 µg L⁻¹ for vanadium concentration in drinking water, which was later revised to 140 µg L⁻¹.^9,10 Thus, highly sensitive and selective methods are required for the detection of TC and vanadium ions.

Several methods, including sophisticated instrumental technologies, microbiological testing and immunoassays, have been adopted to monitor ABx and metal traces.^11–15 Analytical techniques such as high-performance liquid chromatography, surface-enhanced Raman scattering, inductively coupled plasma-mass spectrometry, and gas chromatography-mass spectrometry are highly sensitive for the detection of trace substances.¹⁶ However, these approaches are costly, time-consuming, and require expensive equipment, specialized facilities and trained personnel. The microbiological testing strategies are labour-intensive, require intricate sample preparation, and take a prolonged time for sample testing.¹⁷ Immunoassay kits may also s suffer from antibody stability issues that lead to inaccurate results. Therefore, there is a need for more convincing, unconventional, cheap, and ultrasensitive sensors for addressing these challenges in identifying TC and vanadium ions.

Metal–organic frameworks (MOFs) are promising candidates to address these problems due to their tunable physical (surface area, pore size, and pore volume) and optical (absorption and luminescence) properties.¹⁸ MOFs have rigid metal nodes supported by flexible organic branches, which make them usable for various applications, including drug delivery, sensing, and catalysis.¹⁹ Additionally, luminescent MOFs (LMOFs) are potential sensing units for detecting analytes owing to their exceptional optical, chemical, and physical properties. The incorporated photoactive centres (lanthanides or organic chromophores) and unique porous skeleton of LMOFs provide exceptional light emission properties and permit label-free sensing, and real-time sensitivity against analytes.²⁰ Convincingly, LMOFs with tunable properties offer a multifaceted platform to develop analytical schemes and facilitate innovation in the realms of environmental monitoring, food safety, and healthcare. In addition, LMOFs have the availability of unsaturated points near their metal nodes, which induce binding sites for selective interaction with small molecules like ABx and metal ions. These specific interactions of LMOFs with small molecules accurately define their sensing mechanism, which primarily depends upon photoluminescence (PL) quenching/bleaching (turn-off effect). Some dominating mechanisms for LMOFs' PL bleaching are cation exchange,²¹ photoinduced electron transfer (PET),²² MOF structural collapse,²³ inner filter effect (IEF),²⁴ host–guest weak interactions,²⁵ and fluorescence resonance energy transfer (FRET).²⁶ Recently, various studies have been reported on LMOF-based luminescent sensors for the identification of various ABx and metal ions. However, only a handful of studies are available on a single LMOF material utilized as a double-functional luminescent sensor to monitor both ABx and metal ion traces.^27,28

Among different LMOFs, this study involves Dy-doped Co-BTC frameworks because the integration of Co, BTC, and Dy offers synergistic structural and photophysical characteristics.^26,27 As per former research, solvothermally synthesized Co-BTC MOFs have great acid–base durability, strong thermal stability, and superior coordination ability, and have been examined as promising candidates to fabricate porous MOFs for sensing applications.^26,28,29 Co MOFs also exhibit flexible metal–ligand interactions and the ability to form adaptable electronic environments, which allow for enhanced photophysical and reversible host–guest behavior. Compared to the more rigid MOF architectures, Co-BTC frameworks possess greater flexible coordination and electronic environments, making them better suited to incorporate lanthanides and tune their luminescence.^26,30

Ligands such as benzene-1,3,5-tricarboxylate (BTC), which contain an inherently π-conjugated aromatic system, can effectively capture an excitation pulse and transfer that energy to lanthanide centers via ligand-to-metal energy transfer.²⁹ This process of Dy³⁺ ion sensitization is important as the direct 4f–4f electronic transitions are inherently weak. The Dy-incorporated MOFs have displayed considerable flexibility in coordination, coupled with an even more valuable fluorescence response to a variety of analytes.^27,31 Additionally, Co²⁺ ions in Co-BTC frameworks create dynamic metal–ligand configurations and flexible coordination environments, which support the Dy³⁺ ions' incorporation and stabilization in the Co-BTC framework.^26,32 Dy³⁺ ions are characterized by a strong affinity for coordination to the BTC's carboxylate groups and facilitate their substitution/coordination into the Co-BTC framework during solvothermal crystallization.^33,34 Dy³⁺ incorporation within the Co-BTC framework can change the local electronic environment and suppress the Co-related nonradiative relaxation processes, improving the framework's luminescence.³⁵

Thus, we reported a controlled solvothermal synthesis of Co-BTC MOFs (designated as MOF1), Dy-BTC MOFs (designated as MOF2), and Dy-doped Co-BTC MOFs (designated as MOF3). Moreover, different doping amounts of 5%, 10% and 15% were studied to understand the PL characteristics of Dy-doped Co-BTC MOFs (MOF3), and analytical detection of ABx and various metal ions using fluorescence titration experiments was explored further. Various characterization techniques, such as FTIR, SEM, BET, and EDX, have been employed to understand the PL sensing mechanism for selective detection of TC and V³⁺ ions. In the case of ABx sensing, the charge transfer process between ABx and BTC was further analysed by density functional theory (DFT) calculations. Overall, these MOFs demonstrate a selection of attributes for the proposed sensing application, including straightforward solvothermal synthesis, reproducibility, structural stability, and sustained modularity as luminescent sensors for TC and vanadium ions.

2. Materials and methodology

Detailed information about the materials used and the adopted experimental procedures is explained in the SI Section A. In addition, details about the characterization techniques and approaches for studying the sensing ability of MOF3 and related studies are also presented in SI Section A.

3. Results and discussion

In the present study, we attempted to prepare MOF1, MOF2, and MOF3 using the solvothermal technique to create materials with structurally different luminescent properties. The Dy-doped Co-BTC MOF (MOF3) was then used as a sensing platform for the selective detection of the presence of certain ABx and trace metal ions in aqueous solutions. The salient results obtained in the present study are discussed in this section.

3.1. X-ray diffraction (XRD) analysis

The XRD method was employed to investigate the crystallinity, phase purity, and orientation of MOF1, MOF2, and MOF3. The diffraction angles (2θ) helped in identifying the crystallographic information and structural atomic arrangement of MOF1, MOF2, and MOF3. SI Fig. S2(A and B) presents the diffraction patterns recorded for MOF1 and MOF2, respectively. The crystallographic data for MOF1 and MOF2 were well matched with the XRD database available in the Cambridge Crystallographic Data Center (CCDC). The respective CCDC reference numbers for MOF1 and MOF2 are 1274034 and 2255424.^36,37 It can be ascertained that the polar protic nature of ethanol assists the precursor's solubility through hydrogen bonding. It stabilizes the strong coordination complexes between Co/Dy ions and BTC molecules by their systematic arrangements, and ensures enhanced crystallinity for MOF1 frameworks.^38,39

In order to achieve a luminescent and porous MOF structure for the sensing application, different amounts of Dy, i.e. 1, 5 and 10% w.r.t. Co of Co(NO₃)₂·6H₂O, were loaded in Co-BTC MOF (MOF1) and are designated as MOF3(1%), MOF3(5%) and MOF3(10%), respectively, while keeping the other reaction parameters (solvent = EtOH, T = 120 °C, time = 24 hours, pH = 7, M/L = 1.8 : 1) constant. The compared XRD patterns of MOF3 samples prepared using different amounts of Dy are presented in Fig. 1A. It is clearly observed that 5% doping of Dy (MOF3(5%)) gives the best crystalline material as compared to doping of 1% and 10%. In addition, the diffraction peaks of MOF3(5%) are well matched with those of MOF1 and MOF2, as shown in Fig. 1B. This indicates that MOF3(5%) retains the crystallographic structure of MOF1, suggesting successful incorporation of Dy into the Co-BTC MOF without significant structural deformation. Moreover, the loading of an appropriate amount of Dy may result in the production of highly pure phase and enhanced structural stabilization of MOF3(5%), as indicated by the pronounced or sharp diffraction peaks in the XRD pattern.^19,40 Thus, MOF3(5%) material will be employed for the detection of analytes, including ABx and metal ions.

Fig. 1. XRD patterns of MOF3 (A) with different doping percentages (1%, 5% and 10%) of Dy in Co-BTC MOF (MOF1) and (B) comparison of XRD patterns of MOF3(5%), MOF1, and MOF2.

3.2. Ultraviolet (UV)-visible studies

The optical properties of MOF1, MOF2, and MOF3(5%) were studied using the UV-visible spectroscopic technique. The UV-visible spectra of MOF1, MOF2, and MOF3(5%) are presented in Fig. 2. For comparison, the UV-visible spectrum of BTC was also obtained and is presented as a black line. The UV-visible spectrum of BTC exhibits two absorption peaks at 230 and 289 nm, corresponding to π → π* electronic transitions of conjugated aromatic rings in BTC and n → π* electronic transitions of carboxylic (–C O) groups in BTC, respectively.⁴¹ In the case of MOF1, the UV-visible spectrum showed three absorption peaks at 230, 289, and 553 nm. The peaks at 230 and 289 nm in the spectrum of MOF1 also arise from π → π* and n → π* electronic transitions of BTC, clearly indicating that MOF1 has Co well coordinated with BTC. In addition, the absorption at the peak position of 289 nm is greatly reduced. This may be because the incorporation of Co with BTC through –C O bond alters the electronic environment of C O in BTC, causing a decrease in the intensity of its n → π* electronic transitions.^42,43 The additional peak at 553 nm is attributed to d–d electronic transitions of Co in the Co-BTC complex.⁴⁴

Fig. 2. The comparative UV-visible spectra of MOF1 (red), MOF2 (blue), MOF3 (green) and BTC (black).

Like MOF1, BTC-associated peaks in the MOF2 spectrum were observed nearly the same at 230 and 289 nm. A little new hump at 349 nm can be attributed to the LMCT concept, which is unique to the Dy-BTC framework.⁴⁵ Such transitions are documented in the literature as an important sign of the excitation pathways for PL in Dy MOFs. Notably, the delocalized electronic states in the MOF frameworks are likely to strengthen this low-energy absorption further. Similarly, the UV-visible spectrum of MOF3(5%) also exhibits peaks at 230 and 289 nm, corresponding to π → π* and n → π* electronic transitions of BTC, along with a cobalt absorption peak at 562 nm due to d–d electronic transitions. The synergistic integration of Co and Dy with BTC contributes to slight changes in the electronic environments around them, causing higher absorbance of MOF3(5%) in the λ range of 325–450 nm in comparison to that of MOF1. Available reports on UV-visible and PL excitation spectra of Dy-based materials evidence light absorption characteristics in this spectral range due to f–f electronic transitions.^45,46 Thus, this change signifies the successful incorporation of Dy units with Co-BTC complexes.

3.3. Photoluminescence spectroscopic studies

The luminescent nature of MOFs in aqueous solutions was carefully analyzed using a PL spectrofluorometer at room temperature. For comparison, the PL emission spectra of all the prepared MOF3 samples having 1, 5, and 10% of Dy were obtained, with MOF3(5%) exhibiting the most intense PL emission spectra, as depicted in Fig. S3. Thus, MOF3(5%) was further studied for its luminescent properties and used for sensing applications also. The comparative excitation and emission spectra of BTC, MOF1, MOF2, and MOF3(5%) are presented in Fig. 3A–D. Free BTC molecules exhibit the maximum excitation (λ_ex) and emission (λ_em) wavelength at 278 nm (UV region) and 435 nm (blue region), respectively, arising due to the transition between n and π* within the aromatic rings of BTC (Fig. 3A).⁴⁷ In the case of MOF1, the λ_ex band was observed at 280 nm; the λ_em of MOF1 was observed to remain at the same position (λ_em = 435 nm) w.r.t. free BTC molecules (Fig. 3B). It indicates negligible to no charge transfer from BTC energy levels to Co energy levels, suggesting minimal electronic interactions between the Co and BTC. Hence, only the BTC molecules are responsible for the PL characteristics in MOF1 through the ligand-to-ligand charge transfer (LLCT) mechanism.⁴⁸

Fig. 3. The comparative PL excitation and emission spectra of (A) free or isolated BTC, (B) MOF1, (C) MOF2, and (D) MOF3 having 5% doping of Dy.

The PL excitation and emission spectra of MOF2 are also presented in Fig. 3C. As expected from the literature, for Dy³⁺-based coordination materials, the MOF2 sample showed an emission profile with a peak at 573 nm, with shoulder peaks at 482 nm, when excited at 349 nm.⁴⁹ The emission peaks at 480 nm and 573 nm are due to the f–f transitions of Dy³⁺, namely, the ⁴F_9/2 → ⁶H_15/2 (blue emission) and ⁴F_9/2 → ⁶H_13/2 (yellow emission) transitions. These transitions are Laporte-forbidden, but asymmetric ligand fields predominant in coordination environments like MOFs enable them to occur partially. Fig. 3D presents the PL excitation peak of the prepared MOF3(5%) at λ_ex = 345 nm, which is in a comparable range (i.e. λ_ex: 320–350 nm) of reported literature for Dy-MOFs.⁴⁹ At λ_ex = 345 nm, the electronic transitions from ⁶H_15/2 (ground state) → ⁴F_7/2 (higher vibronic state) are common within Dy, because of its electronic configuration.^49,50 During relaxation, Dy electrons first de-excite to ⁴F_9/2 (non-radiative metastable state), which is further de-excited by radiative emissions. Consequently, MOF3 λ_ex at 345 nm gives two emission peaks at 480 nm (shoulder peak) and 573 nm (major characteristic peak) in the PL scan range of 450–600 nm. The shielding of 4f orbitals by 5s and p sub-shells produces definite energy gaps in electronic levels, permitting narrow line-like PLE at 480 and 573 nm for Dy. In the case of MOF2 and MOF3, the shoulder at λ_em = 480 nm was less intense and submerged within the BTC emission spectrum (Fig. 3C and D). The less intense blue emissions of λ_em = 480 might be due to magnetic dipole-based ⁴F_9/2 → ⁶H_15/2 (ΔJ = ±3) transitions in these MOFs.⁴⁵ In comparison, the high-intensity yellow-green emissions at λ_em = 573 nm were attributed to electric dipole-related ⁴F_9/2 → ⁶H_13/2 (ΔJ = ±2) transitions. Thus, all analytical studies based on PL features of MOF3(5%) were recorded on the λ_em band of 573 nm.⁷⁴ Overall, these characteristics suggest the successful incorporation of Dy into Co-BTC structures.

3.3.1. Mechanism of PL enhancement

The occurrence of PL of lanthanides in MOFs depends on the efficiency of the ligands to transfer energy to the metal as opposed to the intrinsic oscillator strength of the 4f–4f transitions. The BTC ligand in MOF2 is an antenna; however, because of the energy transfer from the ligand triplet state to Dy³⁺ being weak, the emission is weak and spectrally broadened. For Dy-BTC systems, reports state that the emission of Dy³⁺ is low due to inefficient sensitization.³⁵ In contrast to this, Dy-doped Co-BTC (MOF3) shows an enhancement of 3.3 times in intensity of the Dy³⁺ emission at 573 nm compared to Dy-BTC in MOF2 (see Fig. S4). The Dy³⁺ cooperative mixed metal framework enhancement is due to the modification and energy redistribution of Dy³⁺ in the local coordination lattice in the dissipative excitatory environment. Co²⁺ is known to be a luminescence quencher because of its d orbital, but the partial Dy³⁺ substitution acts to provide Co²⁺ with an efficient radiative relaxation pathway that competes with non-radiative Co²⁺ decay. Hence, Dy³⁺ captures the ligand-derived excitation energy that would be lost to Co²⁺ centers.

The narrowing of MOF3's emission bands (567–581 nm) relative to MOF2's (563–583 nm) is consistent with this hypothesis of decreased vibrational coupling and increased rigidity of Dy³⁺ coordination. Enhanced emission via mixed-metal sensitization is observed in Zn_xDy_1−xBTC systems, where the addition of a secondary metal increases the ligand-to-Dy³⁺ energy transfer.³⁵ Although Co²⁺ is electronically different from Zn²⁺, the theory of mixed-metal modulation of energy transfer holds in this case.³⁵ More evidence consistent with this explanation is found with the Co-BTC multi-metal frameworks with lanthanide ions, where the secondary rare earth was found to induce favorable charge redistribution and to mitigate the detrimental decay pathways associated with Co-based MOFs.³⁰ Thus, the enhanced luminescence observed in MOF3 may be advantageous for PL-based sensing applications. In this regard, Xu et al. (2024) demonstrated the use of a recently developed Co-MOFs as fluorescent sensor for Fe³⁺ and TC detection.⁵¹ Collectively, these results articulate that Dy³⁺ does not function as a solitary emitter; rather, it acts in conjunction with the Co-BTC framework to alleviate the detrimental effects of quenching and to promote the luminescence emission of the lanthanide.

The role of Dy in adjusting the emission characteristics of the Co-BTC framework is captured using time-resolved PL measurements. Co-BTC (MOF1) exhibits a short lifetime of 8.56 ns (see Fig. S5), which is associated with Co²⁺ centers and non-radiative decay. This includes ligand to metal charge transfer and d–d related quenching. Incorporation of Dy³⁺ alone (Dy-BTC, MOF2) increases the lifetime by a mere 1.06 ns to 9.62 ns. Moreover, the emission is still low, which means that the ligand to Dy³⁺ transfer could not be efficient. This is analogous to other studies where the triplet energy of BTC is known to be compatible with Dy³⁺ and also that the sensitization process in Dy-BTC is weak, which results in a broad and weak emission at the Dy center.

In contrast, MOF3 shows a further increase in lifetime to 10.74 ns along with a notable increase in steady-state PL. The simultaneous increase in lifetime and intensity serves as a compelling indicator of Dy³⁺ incorporation, changing the dynamics of excited-state relaxation. The process, in MOF3, involves the harvesting of Dy³⁺ as the BTC ligand-generated excitation energy and transferring it to Dy³⁺ centers, which are radiative energy traps and relax through the characteristic 4f–4f transitions. This Dy³⁺ energy transfer process is more effective than Co²⁺ non-radiative quenching, thus elevating the excited-state lifetime. Similar effects have been documented in mixed metal BTC systems, Zn_xDy_1−x-BTC, where the addition of a secondary metal is known to reduce non-radiative quenching to enhance Dy³⁺ emission and increase ligand to lanthanide energy transfer to cause an enhancement.³⁵ Therefore, the observed (MOF1 < MOF2 < MOF3) changes in lifetime, along with increased intensity of Dy-centered emission, confirm that Dy³⁺ doping does not serve a role of a mere emitter. Instead, it reconfigures the energy transfer architecture of the Co-BTC MOF, the PL improvement being a direct manifestation of this change.

3.4. FTIR spectroscopic studies

The presence of various organic functional groups in the prepared MOF1 and MOF3(5%) was also analyzed using FTIR spectroscopic studies. The FTIR spectra of MOF1 and MOF3(5%) are presented in Fig. S6. For comparison, the FTIR spectrum of BTC was also obtained and is presented in Fig. S6. The FTIR analysis of BTC reveals that its spectrum exhibits peaks at 903, 748, 708, and 678 cm⁻¹, corresponding to the bending vibrations of aromatic C–H bonds, confirming the meta-substitution of the aromatic ring in BTC.⁵² Moreover, the BTC spectrum also exhibits peaks at 1265 and 1108 cm⁻¹, arising from C–O stretching vibrations resulting from deprotonation of carboxylic acid in BTC. The peak at 1697 cm⁻¹ is assigned to the stretching frequency of the carbonyl (C O) bond in the carboxylic acid (–COOH) group. Further, the peaks at 1465 and 1403 cm⁻¹ indicate the presence of –COOH moieties in BTC due to the symmetric (sym) and asymmetric (asym) stretching of –COO⁻ ions, respectively.

In the case of the FTIR studies of MOF1 and MOF3(5%), no significant differences in peak nature and peak positions of MOFs w.r.t. BTC were observed, indicating that MOF1 and MOF3(5%) both have BTC in their structures.^52,62 A broad peak of MOFs at 3375 cm⁻¹ can be assigned to the hydroxyl (–OH) stretch of coordinated H₂O molecules in the MOF structure. The splitting and shifting of peaks ranging from 1618 to 1520 cm⁻¹ for the MOF structures demonstrate the asymmetric stretching of –COO⁻ ions, suggesting successful coordination between metal centers (Co and Dy) and BTC molecules. Further, a minute shift in peak range of 1434–1373 cm⁻¹ for MOFs indicates symmetric stretching of –COO⁻ ions. Thus, the study clearly indicates the proper coordination of BTC with the metals, implying the formation of the MOF structure.

3.5. XPS study

In order to find out the chemical composition and electronic structure of the MOFs, XPS measurements of both MOF1, prepared under optimized conditions, and MOF3(5%) were carried out. The observed full-scan XPS spectra of MOF1 and MOF3(5%) are presented in Fig. S7(A) and S7(B), respectively. The spectra clearly show the presence of C, O, and Co in MOF1 and C, O, Co and Dy in MOF3(5%), confirming the high purity of the MOF samples. Because of BTC formulations, C and O were present in each MOF. Fig. S7(C and D) displays the high-resolution spectra of O 1s in the range of 526–538 eV with a single peak corresponding to each of the binding energies of 532.13 eV and 532.18 eV of MOF1 and MOF3(5%), respectively, indicating the surface hydroxyl group (–OH) substituted with MOF's chemical moieties.^53,54 Some unfitted areas of O 1s in the range of 529–531 eV are believed to be due to the association with the available O²⁻ or O⁻ anionic vacancies near MOFs' surface lattice.⁵⁵ Fig. S7(E and F) presents the high-resolution spectra of C 1s of the MOFs, exhibiting three peaks with binding energies at 284.6 ± 0.15, 286.36 ± 0.02, and 288.28 eV. The binding energies of C 1s at 284.6 ± 0.15 and 286.36 ± 0.02 eV are considered to arise from sp² and sp³ hybridization for C–C or C–H bonds, respectively, whereas the binding energy at 288.28 eV could be assigned to C–O–C or C–O–H bonding,^53,54 confirming the existence of C(O)OH or HO–C–OH bonding in the MOF structure. Fig. S7(G and H) presents the high-resolution XPS spectra of Co 2p, showing a broad peak. The spectra were further deconvoluted into four peaks of binding energies at 781, 786.6, 796.6, and 802.2 eV. The peaks at 781 and 796.6 eV correspond to Co 2p_3/2 and Co 2p_1/2, respectively, whereas peaks at 786.6 and 802.2 eV represent satellite characteristics of Co²⁺ ionic species in MOF1 and MOF3(5%).⁶⁹ Similarly, Fig. S7(I and J) displays the XPS spectra of Dy in MOF3(5%). The peaks at 154.20 and 156.36 eV correspond to the Dy 4d_5/2 and Dy 4d_3/2 states in MOF2,³⁴ whereas the peaks at 1296.6 and 1335.25 eV correspond to two binding energies of Dy 3d_5/2 and Dy 3d_3/2 states present in MOF3(5%).⁵⁴

3.6. FESEM investigation

The detailed surface morphology of the prepared MOF3(5%) was extensively examined with FESEM (JEOL JSM-7600F). The representative FESEM images of MOF1 (prepared under optimized reaction conditions) and MOF3(5%) are presented in Fig. 4. The FESEM images in Fig. 4A and C clearly reveal that both MOF1 and MOF3(5%) have quasi-spherical morphologies with high surface roughness. The particle sizes are determined to be in the range of 500–1000 nm and 150–300 nm for MOF1 and MOF3(5%), respectively. In addition, upon careful observation under highly magnified conditions, it is clearly observed that each particle of MOF1 and MOF3(5%) has high surface roughness and porous features, as shown in Fig. 4B and D. This suggests that the integration or loading of Dy in the MOF1 structure does not create any adverse effect on the MOF1 frameworks. This coordinated and highly porous structure of MOF3(5%) makes it a highly promising candidate for hosting suitable guest molecules like ABx or metal traces. This is the first time of observation where a MOF has a highly porous and spherical morphology in comparison with the reported literature of MOF1 having different morphologies, such as nanorods, needles, or layered structures.⁵⁶ It is believed that the spherical and porous nature of MOF units will be more favorable for hosting guest molecules in comparison with the reported MOF structure of needle- or rod-like shapes.

Fig. 4. Low- and high-magnification FESEM images of (A and B) MOF1 and (C and D) MOF3(5%).

3.7. Transmission electron microscopy (TEM) analysis

To decipher the finer details of the prepared MOFs, careful TEM investigations were carried out for MOF1 and MOF3(5%). The representative TEM images at low and high magnifications and the diffraction pattern captured in TEM are presented in Fig. S8. From the images (Fig. S8A), it can be clearly observed that MOF1 particles appear to be spheroidal in shape with diameters of 200–500 nm. Moreover, the clusters or the particles of MOF1 are observed to resemble a water-bubble-like feature having different contrast in the cluster, indicating the highly porous nature of the material. Generally, these pores are formed by the diffusion-driven growth process, where the redistribution of reactant species during crystal growth leads to the development of internal voids.⁵⁷ From Fig. S8B, one can clearly see the cluster comprised of uniform Co nanocubes (20–55 nm) surrounded by BTC, which acted as a binder. The HRTEM images of MOF1, presented in Fig. S8C, also confirm that the nanocubes are of Co, exhibiting lattice fringes having inter-planar spacing of 0.274 and 0.280 nm, corresponding to diffraction planes of (5,1,−2) and (4,2,−2) of Co respectively.⁵⁸ The ring diffraction pattern obtained for the MOF1 cluster also clearly suggests that MOF1 is polycrystalline in nature (Fig. S8D).

Similarly, the bright-field TEM images of MOF3(5%) also appear to show quasi-spheroidal structures with more dense packing due to the loading effect of Dy atoms (Fig. S8E).⁵¹ The magnified image (Fig. S8F) showed that each cluster of MOF3(5%) also consists of nanocubic units of Co and monoclinic units of Dy. The HRTEM pattern reveals different planes of Co and Dy with different d-spacing values, illustrated in Fig. S8G. Multiple lattice fringes of different Co crystals have different d-spacing values, including 0.268, 0.274, 0.315, 0.340, 0.483, and 0.536 nm. The corresponding planes of these inter-planar spacings are (5,3,−1), (5,1,−2), (1,1,−2), (3,3,0), (1,1,1), and (1,1,−1), respectively. The Dy units in MOF3(5%) exhibit (5,1,3) and (5,0,1) planes having fringe widths of 0.239 and 0.290 nm, respectively. Further, the polycrystalline nature of MOF3(5%) is confirmed with the ring diffraction pattern captured in TEM, as shown in Fig. S8H.

3.8. Energy-dispersive X-ray (EDX) spectroscopic analysis

In order to further confirm the elemental composition, EDX analyses of both MOF1 and MOF3(5%) were conducted with the EDX detector being combined with TEM. The representative EDX spectra and the corresponding area mapping of MOF1 and MOF3(5%) are presented in Fig. S9 and S10 respectively. The presence of only C, O and Co in the EDX spectrum of MOF1 (Fig. S9) indicates that the prepared MOF1 is highly pure. The corresponding elemental mapping of C, O and Co (Fig. S9(B–E)) also clearly shows that the elements are uniformly distributed all over the selected area of MOF1, confirming the formation of coordinated structures of Co with BTC. The observed result is also consistent with reported data.⁵⁹ The weight and atomic percentages of each element of MOF1 are also presented as an inset table in Fig. S9(A). For MOF1, the atomic percentages of C, O, and Co are given as 39.13, 39.08, and 22.79, respectively, which are in agreement with the expected Co-BTC framework composition. Similarly, MOF3(5%) has atomic percentages of 0.24, 68.87, 23.31, and 7.58 for Dy, C, O, and Co elements (displayed as an inset table in Fig. S10(A)). The resultant Dy/Co atomic ratio of 0.032 is in agreement with the incorporation of Dy-substituted Co in the Co-BTC framework. The lower Co content relative to MOF1 is further evidence of a substitution, as opposed to surface adsorption, of Dy. The uniform distribution of all the elements throughout the sample, as depicted by mapping analysis (Fig. S10(C–G)), clearly demonstrated the strong incorporation of all elements into the MOF3(5%) structure. Thus, the EDX quantification results confirm the degree of intended Dy doping into the framework.

3.9. BET analysis

Surface area and porosity play a major role in the study of adsorption or the holding capacity of any material. Thus, the surface area and the porosity of both MOF1 and MOF3(5%) were explored with BET analysis, and the observed isotherms are shown in Fig. S11. From the figure, it is clearly observed that both the MOFs exhibit a type-4 adsorption isotherm, indicating that both the MOF structures are mesoporous in nature. From the study, the surface area and pore size of MOF1 were estimated to be 48.396 m² g⁻¹ and 7.8 nm, whereas those of MOF3(5%) were estimated to be 45.391 m² g⁻¹ and 17.4 nm, respectively. One can clearly say that the surface areas are more or less similar, but there is a large difference in their pore sizes. The larger pore size of MOF3(5%) in comparison with MOF1 may lead to higher accommodation capacity of guest molecules, which is one of the required characteristics of the MOF for its sensing application.

3.10. Sensing applications of MOF3

3.10.1. Detection of tetracycline (TC) ABx

MOF3(5%) shows a “turn-off” PL response when ABx adsorb onto its active sites or into its pores.^60,61 We tested MOF3(5%) with kanamycin (KNM), neomycin (NMS), sulfadiazine (SDZ), cephalexin (CEX) and TC; results are shown in Fig. 5A (procedure and parameters are in SI, part d and Table S1A). All ABx reduced PL to varying degrees, with TC producing the largest quenching (86.3% PLE reduction; see Fig. S12A), likely due to non-covalent interactions (photoinduced electron transfer, π–π stacking, H-bonding) with the MOF.⁶² The other ABx showed weaker effects, indicating high selectivity of MOF3(5%) for TC.^63,64

Fig. 5. Comparative PL emission spectra of (A) MOF3(5%) treated with different antibiotics and (B) MOF3(5%) treated with different concentrations of TC, captured with excitation at λ_ex = 345 nm. (C) Quality assurance studies of MOF3(5%) detecting TC at various concentrations ranging from 0 to 1.4 ppm. (D) Interference studies for (i) OA: MOF3(5%) treated with antibiotics other than TC (red); and (ii) OAWT: MOF3(5%) treated with other antibiotics along with TC (blue).

Sensitivity tests used 200 ppm MOF3(5%) exposed to TC at 0.025–1.4 ppm (Fig. 5B). PL intensity decreased as TC concentration increased; 1.4 ppm TC produced 99.61% quenching versus the control (Fig. S12B). Quality assurance followed the Stern–Volmer analysis (eqn (1), SI). The I_o/I vs. TC concentration plot (Fig. 5C and Table S3) is linear up to 0.8 ppm and deviates above ∼0.9 ppm, indicating reliable trace detection in the lower range. The Stern–Volmer constant K_sv = 2 × 10⁶ M⁻¹ (slope from 0.0025 to 0.8 ppm; see Fig. 5C). Limits calculated from (LOD = 3σ/m) and (LOQ = 10σ/m) (σ = SD of ten measurements) gave LOD = 5.046 × 10⁻⁴ M (0.2427 ppm) and LOQ = 1.682 × 10⁻³ M (0.8085 ppm).

Interference tests compared (i) OA: mixture of other ABx (no TC); and (ii) OAWT: same mixture plus TC. Both were treated with 200 ppm MOF3(5%) (Fig. 5D). OA caused 49.17% PL reduction; OAWT caused 69.84%, i.e., TC adds ∼20.7% extra quenching, showing TC can overcome some interference (Fig. S12C).

Mechanistic studies support adsorption-driven quenching. BET analysis showed the surface area of TC-treated MOF3(5%) fell to 3.08 m² g⁻¹, consistent with TC filling pores (Fig. S12D). FESEM revealed dense aggregates after TC exposure versus the original porous spheres (Fig. 6A). UV-visible spectra show TC absorption (275, 357 nm) overlaps both Dy and BTC bands of MOF3(5%), creating competitive energy transfer that favours TC detection (Fig. 6B).

Fig. 6. (A) FESEM image of TC-treated MOF3(5%). (B) The UV-visible absorptive features of BTC, MOF3(5%), and all ABx. (C) Representation of the DFT calculations performed for BTC and all ABx. (D) FTIR spectra for MOF3(5%), TC-treated MOF3(5%), and TC ABx.

DFT calculations (B3LYP/6-31G(d)) indicate favourable photoinduced electron transfer: BTC and TC lowest unoccupied molecular orbital (LUMO) energies (−2.08 eV and −1.97 eV) and a small TC bandgap (3.99 eV) promote electron flow from BTC to TC;⁵¹ highest occupied molecular orbital (HOMO) separation (BTC: −7.69 eV; TC: −6.07 eV) supports charge separation and quenching (Fig. 6C). DFT trends match experimental selectivity: NMS and others show different bandgaps and overlaps, explaining⁵¹ their weaker quenching.

FTIR analysis of MOF3(5%), TC, and TC-treated MOF3(5%) (Fig. 6D) confirms adsorption without irreversible MOF damage. Retained MOF peaks (1555, 1371, 1108, 769 cm⁻¹) coexist with new/shifted bands (new 1726 cm⁻¹ C O stretch; 1618 → 1610 cm⁻¹ shift) consistent with electrostatic interactions and π–π stacking. These interactions alter the local electronic environment and enable non-radiative pathways (e.g., PET), producing the observed PL quenching.

3.10.2. Detection of vanadium ion (V³⁺) traces

MOF3(5%) was tested against 20 metal ions (Ba²⁺, Bi³⁺, Li⁺, Cs⁺, Mn²⁺, Sn²⁺, V³⁺, Hg²⁺, Fe³⁺, Ni²⁺, Co²⁺, Zn²⁺, Cd²⁺, Cr³⁺, Ag²⁺, Pb²⁺, Na⁺, K⁺, Cu⁺, Pt²⁺). The representative and comparative PL spectra of metal ion-treated MOF3(5%) are presented in Fig. 7A. All ions reduced PL intensity, with V³⁺ causing the largest quenching (90.4%; see Fig. S13(A)), likely via non-radiative energy transfer and electrostatic interactions.^48,65,66

Fig. 7. Comparative PL emission spectra of (A) MOF3(5%) treated with different metal ions and (B) MOF3(5%) treated with different concentrations of V³⁺ ions. (C) Quality assurance studies of MOF3(5%) for detecting V³⁺ ions of various concentrations ranging from 0 to 1.4 ppm. (D) Competition tests for (i) OMT: MOF3(5%) treated with metal ions other than V³⁺ ions (OMT); and (ii) OMTV: MOF3(5%) treated with other metal ions along with V³⁺ ions.

Sensitivity tests, as presented in Fig. 7B, used V³⁺ at 0–1.4 ppm with 200 ppm MOF3(5%). PL decreased with concentration, but quenching at 1.4 ppm was slightly lower than at 1.2 ppm (see Fig. S13(B)), consistent with surface/pore saturation.⁶⁶ The Stern–Volmer plot was linear from 0.025 to 0.6 ppm and deviated above ∼0.8 ppm (see Fig. 7C). The quenching constant K_sv = 1 × 10⁶ M⁻¹. Limits were LOD = 1.009 × 10⁻³ M (0.1587 ppm) and LOQ = 3.364 × 10⁻³ M (0.5291 ppm) (see Table S4).

Competition experiments (Fig. 7D) compared the OMT (all metals except V³⁺) and OMTV (all metals including V³⁺) samples. OMT reduced PL by 47.13%, OMTV by 77.76% (see Fig. S13C), so V³⁺ adds ∼30.6% extra quenching, indicating selective interaction despite some interference.⁶⁶

Mechanistic evidence supports adsorption-driven quenching.⁶⁶ FESEM and elemental mapping results in Fig. 8(i–vii) show V³⁺ on the surface and in pores. BET surface area fell from 45.391 to 9.081 m² g⁻¹ after V³⁺ treatment, confirming uptake (see Fig. S13D). In Fig. 8(viii), FTIR shifts (765 → 753 cm⁻¹) and new peaks (1700, 974 cm⁻¹) indicate electrostatic binding to C O/OH groups and altered vibrational modes.⁶⁷ Close proximity of V³⁺ to Dy³⁺ in the doped MOF may enable short-range 3d–4f coupling,^68,69 promoting photoinduced electron transfer and strong non-radiative decay, which explains the pronounced PL quenching by V³⁺.

Fig. 8. Images illustrate FESEM and elemental mapping of V³⁺-treated MOF3(5%). Micrograph (i) shows the FESEM image of V³⁺-treated MOF3(5%) structures. Graphs (ii–vii) evidence the soaking of V³⁺ ions into/over MOF3(5%) pores and surfaces, including other compositional elements. (viii) The FTIR spectra of MOF3(5%) and V³⁺-treated MOF3(5%).

3.10.3. Reusability and structural integrity of MOF3(5%) for sensing application

The combined results from PL and PXRD demonstrate the structural integrity and recyclability of MOF3(5%) after sensing TC and V³⁺ ions (see Fig. 9). The sensing measurements show that with every increase in the number of cycles, there was a decrease in PL intensity. Pristine MOF3(5%) retained over 80% of its PL intensity after 3 cycles, demonstrating its good intrinsic PL stability. In contrast, the samples with TC had a steeper decline in PL intensity, which was due to the constant surface interactions (a weakly bound surface) that cause and/or trap the electrons via PET (Fig. 9a). The MOF3(5%) sample with V³⁺ ions had a similar decline in PL intensity, and that was due to a weaker and partially reversible drop in the coordination-driven quenching (Fig. 9b). The PXRD results in both cases show that after every successive cycle, the sample exhibits characteristic diffraction peaks and that the 2θ values did not shift, that are associated with the long-range crystalline structure (Fig. 9c and d). The minor decrease in PXRD peak intensity is associated with surface disorder and operational fatigue, but not structural collapse. The separate phenomena for PL and in PXRD illustrate that the repeated sensing changes are predominantly local electronic, and the MOF3(5%) crystal structure remains unchanged. That makes MOF3(5%) a durable and high-quality luminescent sensor.

Fig. 9. Recyclability of MOF3(5%) towards the sensing of TC and vanadium ions. (a and b) PL responses of the pristine and TC-/V³⁺ ion-treated MOF3(5%) over four consecutive sensing-regeneration cycles. Responses show consistent and reproducible quenching with decreasing intensity. (c and d) PXRD patterns of MOF3(5%) with TC/V³⁺ ions illustrating retention of characteristic diffraction peaks and absence of peak shifts, confirming the recycling process preserves the crystalline structure.

3.10.4. Leaching stability of Dy³⁺ in MOF3(5%)

An ICP MS leaching study monitored ¹⁶³Dy in Co-BTC to check for Dy³⁺ release during sensing (Fig. 10). Measurements used No Gas and He modes with external calibration over 10–500 ppb, yielding excellent linearity (R = 0.9998 No Gas; R = 1.0000 He), detection limits of 0.147 ppb (No Gas) and 0.183 ppb (He), and background equivalent concentrations >0.36 ppb, confirming method sensitivity and reliability. Pristine MOF3(5%) showed Dy in the leachate of 207.9 ppb (No Gas) and 198.4 ppb (He); after V³⁺ exposure, these rose to 280.9 ppb and 266.0 ppb, and after TC exposure to 286.1 ppb and 272.6 ppb, respectively. The close agreement between No Gas and He values indicates minimal matrix effects. The observed increases are small and are best explained by minor surface/topographical changes from host–guest interactions rather than substantial Dy loss. This implies that Dy is incorporated within the Co-BTC coordination network rather than merely physiosorbed on the surface. Therefore, PL quenching by TC or V³⁺ is more likely driven by supramolecular adsorption and photoinduced electron transfer than by framework breakdown or major Dy leaching, supporting the chemical resilience and reusability of the Dy-doped Co-BTC sensor.

Fig. 10. Optimized concentrations of Dy salts in (a) No Gas and (b) helium modes to check the leaching stability of Dy³⁺ in MOF3(5%). The inset table reports the leachate concentrations for MOF3(5%) in aqueous medium (at no. 10), after treatment with V³⁺ (at no. 11), and after treatment with TC (at no. 12).

Lastly, Table 1 presents a performance comparison of Dy-doped Co-BTC MOFs (MOF3(5%)) and selected MOF-based sensors for TC and vanadium ions to demonstrate the relative performance of MOF3. Recently created luminescent lanthanide MOFs (especially those composed of Eu and Tb) are often able to achieve low, absolute, and nanomolar detection limits for TC (due to the rapid and efficient antenna-assisted f–f emission pathways). However, most of these systems are solely focused on antibiotic sensing using turn-on mechanisms. In contrast, MOF3(5%) exhibits remarkable Stern–Volmer constants (∼10⁶ M⁻¹) for both TC and V³⁺, which demonstrates very efficient PL quenching due to electron transfer and weak host–guest interaction. Surprisingly, a recently described Zn-based luminescent MOF (PUC7) exhibited selective detection of V⁵⁺ with a limit of detection of 0.22 µM and a quenching constant of ∼10⁵ M⁻¹. In contrast, MOF3 demonstrates comparable or superior quenching behavior toward V³⁺ with a larger K_sv value. Moreover, colorimetric or optical detection of vanadium species at trace levels using ligand-anchored optical composite materials has been reported, although such systems are not crystalline MOF frameworks and therefore, typically do not possess the structural tunability, porosity, and recyclability of MOF-based sensors. The disparity in detectable oxidation states (V²⁺/V⁵⁺vs. V³⁺) further demonstrates the complementary sensing ability of MOF3. While electrochemical sensors based on MOFs can achieve lower detection limits, the Dy-doped Co-BTC system affords the most favourable combination of dual detection of analytes, a simple optical readout system, the greatest structural quenching efficiency, and structural quenching (robustness). This combination makes it the most suitable system for luminescence-based monitoring of antibiotic and metal ion contamination in surface waters.

Table 1. Benchmark comparison of MOF-based sensors for TC and V³⁺ ions.

	Material (MOF/sensor)	Analyte	LOD	Quenching constant (Ksv)	Reference
1	Dy-doped Co-BTC (MOF3(5%))	Tetracycline (TC)	0.2427 ppm (≈0.2427 mg L−1)	2.0 × 106 M−1	This work
2	Dy-doped Co-BTC (MOF3(5%))	V3+	0.1587 ppm (≈0.1587 mg L−1)	1.0 × 106 M−1	This work
3	Eu-MOF (water-stable, H4btec ligand)	TC, OTC, DOX	OTC: 78 nM; TC: 225 nM; DOX: 201 nM	—	70
4	MIL-53(Fe)-based MOF (electrochemical sensor/MOF-enabled electrode)	TC	0.0260 µM (≈0.026 µM)	—	71
5	Zn-MOF ([Zn3(BMP)2L2(H2O)4]·2H2O)—luminescent	TC	Reported as a sensitive luminescent sensor (typical LOD in sub-µM range)	—	72
6	In-SBDC/In-based SBDC MOF (fluorescent MOF reported in reviews)	Tetracyclines (various)	∼0.28–0.30 µM (reported ranges for In-SBDC)	—	73
7	Ligand-anchored optical composite (mesoporous silica/ligand anchored) — New J. Chem. work	V2+	∼0.15 µg L−1 (∼0.00015 ppm) (reported nanomolar sensitivity)	—	74
8	Zn-PUC7 luminescent MOF	V5+	0.22 µM; PL quenching	1.1 × 105 M−1	75
9	Al-MIL-53 derivative (Schiff-base functionalized)—spectrophotometric sensor	Various metal ions (Hg2+, Co2+, Al3+)	LODs in ppm → ppb range (example: Hg2+ 1.52 ppm reported)	—	76

4. Conclusions

We synthesized three porous MOFs—Co-BTC (MOF1), Dy-BTC (MOF2) and Dy-doped Co-BTC (MOF3)—by a simple, scalable solvothermal route (120 °C, 24 h, neutral pH, ethanol) and fully characterized their structure, morphology and optics (XRD, FESEM, HRTEM, EDX, XPS, BET, UV-visible, FTIR). The 5% Dy-doped sample, MOF3(5%), showed the strongest PL and served as a dual-mode sensor that selectively detects TC and V³⁺ among a range of antibiotics and metal ions, with LODs of 0.2427 ppm and 0.1587 ppm, respectively. Combined FESEM, BET, FTIR, DFT and spectroscopic evidence indicates that selectivity arises from strong adsorption, spectral overlap, π–π stacking and photoinduced electron transfer for TC, and electrostatic/dipolar 3d–4f interactions for V³⁺. The Dy ions remain structurally integrated, and the micro-sized, recyclable MOF3(5%) offers a robust, efficient platform for sensitive, selective environmental sensing.

Author contributions

Karanveer Singh: conceptualization and experimental work; manuscript drafting and editing. M. Manolata Devi: mentorship; experimental design guidance; manuscript review and editing. Suvankar Chakraverty: senior supervision and critical review of the manuscript. Bhanu Prakash: overall project supervision; manuscript review and final editing.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

Data supporting this article have been included as a part of the supplementary information (SI). Further clarifications can be sought from the authors concerning the data for this manuscript. Supplementary information is available. See DOI: https://doi.org/10.1039/d6ra04036e.