A Dual Quantum Dot Fluorescent Probe for Time-Resolved Chemometric Detection of Chloramphenicolin Pharmaceuticals

Abstract

Dual-emission photoluminescence (PL) nanoprobes provide improved analytical performance to develop a reliable and sensitive sensing platform for quantifying chloramphenicol in pharmaceutical samples, thereby ensuring therapeutic efficacy and patient safety. In this work, a dual-emission PL sensing platform combining carbon dots (CDs) and AgInS₂ quantum dots (QDs) capped with mercaptopropionic acid (MPA) was developed for the quantitative determination of chloramphenicol, resorting to chemometric methods for data analysis. CDs, CdTe QDs, and AgInS₂ QDs were synthesized and individually evaluated considering their photostability, PL response and kinetics of their interaction with the antibiotic. After this, two dual-emission probes, CDs/MPA-CdTe and CDs/MPA-AgInS₂, were prepared and assessed based on the complementarity of their individual emission features. The obtained kinetic PL dataset was processed using unfolded partial least squares (U-PLS) in order to explore the multidimensional information of the dual-emission systems and to evaluate the performance of both sensing platforms. CDs/MPA-AgInS₂ probe was demonstrated to be the most efficient sensing platform due to its better compromise between sensitivity and photostability, as well as its cadmium-free composition, allowing the implementation of a more environmentally friendly analytical methodology. The optimization of the U-PLS models involved the assessment of the kinetic acquisition time and different spectral regions. The results showed that reliable, sensitive and efficient quantification could be achieved within the first 5 min of interaction and using the full emission spectrum of the sensing probe. Additionally, different interaction mechanisms were observed for each nanomaterial in the combined probe, being static for the CDs/chloramphenicol interaction and dynamic for MPA-AgInS₂/chloramphenicol interaction, which supports the synergetic behavior of the combined probe. The proposed methodology was effectively applied to commercial pharmaceutical formulations, yielding accurate results with good figures of merit. Therefore, this approach can be used as a relevant alternative to existing methodologies for a rapid, robust, and environmentally friendly method for chloramphenicol quantification.

1. Introduction

Antibiotics play a crucial role in human health, providing effective treatment against a wide range of bacterial infections; however, their overuse and misuse pose significant problems concerning the spread of antibiotic-resistant bacteria, toxicity and quality control [1,2]. For this reason, the accurate and reliable determination of antibiotic content in pharmaceutical formulations is imperative to ensure therapeutic efficacy and patient safety.

Chloramphenicol is a broad-spectrum antibiotic used to treat bacterial ophthalmic infections, such as conjunctivitis. Its prolonged and widespread use in ophthalmic preparations, together with perceptions of a favourable safety and efficacy profile, has contributed to its misuse, raising public health concerns about the potential risks of antimicrobial resistance [3,4]. Additionally, several adverse effects have been associated with chloramphenicol, including aplastic anaemia, acute leukaemia, anaphylaxis and contact dermatitis [3]. These safety concerns led to the imposition of stringent regulations regarding the formulation and dosage of this substance, thus highlighting the need to develop sensitive analytical methods for its accurate quantification [5,6]. Several analytical methods have been reported for the determination of chloramphenicol in pharmaceutical formulations, including flow-injection analysis with chemiluminescence [7] or biamperometric [8] detection, UV spectrophotometry [9], visible colorimetric spectrophotometric assays based on chromogenic reactions [10], high-performance thin-layer chromatography (HPTLC) [11], gas chromatography [12], capillary electrophoresis [13], and voltametric techniques [14].

Although these analytical techniques provide results with good accuracy and selectivity, they often depend on expensive or specialized instrumentation, namely HPTLC, capillary electrophoresis, or gas chromatography. Other analytical techniques require laborious sample preparation steps, derivatization procedures, or the use of chromogenic reagents. These demands need skilled and specialized operators and lengthy procedures, making these methods less suitable for routine and rapid quality control of pharmaceutical formulations.

On the other hand, photoluminescence (PL)-based methods are increasingly being explored as promising alternatives for pharmaceutical analysis, offering several advantages, including low cost, rapidity, operational simplicity, high sensitivity, and superior signal-to-noise ratio [15,16,17,18,19,20]. Although most active pharmaceutical ingredients don’t have intrinsic fluorescence, recent advances in nanomaterials have considerably increased the possibilities for applying photoluminescence-based sensing approaches. Carbon dots (CDs) have emerged as attractive PL nanomaterials due to their excellent optical and physical properties, including strong and adjustable emission, good aqueous dispersibility, photostability, low toxicity, and superior biocompatibility [21]. Additionally, ternary AgInS₂ quantum dots (QDs), as heavy-metal-free semiconductor nanoparticles, also offer remarkable optical properties, namely highly tunable PL, a broad emission range, and strong optical stability, while exhibiting lower toxicity than binary Cd-based QDs [22]. These nanomaterials, when combined in a single dispersion, can provide complementary optical properties, enabling the development of dual-emission PL systems with enhanced analytical performance [15,16,23].

Although fluorescent nanomaterials have recently been used for the determination of chloramphenicol [24,25,26,27,28,29,30], most of these reported approaches required surface functionalization with molecular recognition elements to obtain selective detection of antibiotics in food, environmental, and biological samples. These strategies typically require multi-step synthesis or chemical modification, making them more laborious and less suitable for routine analytical applications. In contrast, this work aims to achieve selective determination without incorporating recognition elements into the nanomaterials, relying instead on the kinetic evolution of photoluminescence and chemometric analysis to circumvent possible interferences.

Recent advances in analytical sensing increasingly combine PL-based detection with multivariate chemometric modelling to address challenges associated with complex sample matrices and overlapping signals [31]. Chemometric modelling has been highlighted as a key strategy for discriminating different antibiotic classes in distinct matrices, such as fluoroquinolones, illustrating the broader applicability of these approaches to antibiotic analysis. [32]. Beyond PL-based nanoprobes, carbon dots have emerged as versatile sensing tools for antibiotic and pesticide residues, yet most reported works rely on steady-state fluorescence and functionalization strategies that require additional conjugation or separation steps [21]. Other related developments, such as the combination of chemometrics with paper-based analytical devices or engineered nanomaterials for rapid antibiotic detection, further demonstrate the trend towards combined sensing and data analysis [33]. Despite this progress, the exploration of dual-emission quantum dot systems, kinetic PL data, and the further analysis using second-order chemometric models remains underexplored.

Despite the advantages of using AgInS₂ CDs and QDs simultaneously as a combined PL probe for antibiotic analysis, this approach has been scarcely explored. Furthermore, most PL-based methodologies rely on steady-state measurements, thereby neglecting the rich temporal information contained in PL evolution profiles. Time-resolved spectral acquisition provides additional data that, when combined with the use of appropriate chemometric tools, enables overcoming problems associated with a lack of selectivity and thus improves the analytical robustness of the method, especially when interfering species are present [34,35].

Chemometric methods offer powerful strategies for analyzing these multi-signal data, extracting relevant analytical information from multidimensional fluorescence measurements [34,36]. This allows for improved sensitivity, reduced noise, and accurate and selective quantification, even in systems with overlapping emissions [15]. To the best of our knowledge, the combination of chemometric analysis of PL-based kinetic data with a dual-emission probe has not yet been reported for the determination of chloramphenicol.

In this work, a dual-emission probe composed of carbon dots and MPA-capped AgInS₂ quantum dots was developed for the fluorometric determination of chloramphenicol in pharmaceutical formulations. The PL response of the combined probe was monitored over time, generating temporal emission profiles that were subsequently analysed using unfolded partial least squares (U-PLS). The proposed approach offers a rapid, cost-effective, and sensitive method for determining chloramphenicol, demonstrating the potential of a dual quantum dot PL system combined with chemometric tools for advanced pharmaceutical analysis.

2. Materials and Methods

2.1. Reagents and Solutions

All reagents were of analytical grade and used as received. Ultrapure water obtained from a Milli-Q purification system (conductivity ≤ 0.1 μS cm⁻¹) was used for the preparation of all solutions and standards.

A chloramphenicol stock solution with a concentration of 207.0 mg L⁻¹ was prepared by dissolving 20.7 mg of chloramphenicol standard (C₁₁H₁₂Cl₂N₂O₅, Sigma-Aldrich^®, ≥98%, St. Louis, MO, USA) in 100 mL of ultrapure water. Working standard solutions were freshly prepared by appropriate dilution of this stock solution with ultrapure water to obtain the required concentrations.

Two commercial pharmaceutical formulations of chloramphenicol in ophthalmic solution were analyzed at concentrations of 5 mg mL⁻¹ and 8 mg mL⁻¹. For each formulation, three different batches were evaluated in order to validate the proposed method. For each batch, an intermediate solution was prepared by accurately transferring an aliquot of the original ophthalmic solution to a 25 mL volumetric flask and completing the volume with ultrapure water to obtain a final chloramphenicol concentration of approximately 207.0 mg L⁻¹ (similar to the intermediate standard solution). From each intermediate solution, two further dilutions were prepared to yield final concentrations within the calibration range and suitable for quantitative analysis.

For the preparation of CdTe quantum dots (QDs), tellurium powder (200 mesh, 99.8%), sodium borohydride (NaBH₄, 99%), cadmium chloride hemi(pentahydrate) (CdCl₂·2.5H₂O, 99%), cysteine (C₃H₇NO₂S, 97%), reduced L-glutathione (GSH, ≥98%), 3-mercaptopropionic acid (MPA, ≥99%), and sodium 2-mercaptoethanesulfonate (MES, 98.0%) were obtained from Sigma-Aldrich^® (St. Louis, MO, USA). Absolute ethanol (99.5%) was supplied by Panreac^® (Barcelona, Spain).

For the synthesis of ternary AgInS₂ (AIS) QDs, silver nitrate (AgNO₃, 99.9999%), ammonium hydroxide (28–30% NH₃ basis), thiomalic acid (TMA, ≥99.0%), and MPA (99%) were purchased from Sigma-Aldrich^®, while sodium sulfide nonahydrate (Na₂S·9H₂O, 98+%) and indium chloride (InCl₃, 99.995%) were acquired from Acros Organics™ (Geel, Belgium). Carbon dots (CDs) were synthesised using citric acid (≥99.5%) and ethylenediamine (≥99.5%), both obtained from Sigma-Aldrich^®.

Binary CdTe QDs capped with MPA or Cys were synthesised via a two-pot hydrothermal approach [37], adapted from a previously reported procedure [19], while those stabilised with MES were prepared via the aqueous synthetic route described by Paim et al. [38]. The molar ratios of Cd:Te:ligand were adjusted to 1:0.1:1.7 (MPA), 1:0.05:2.4 (Cys), and 1:0.03:2.3 (MES), with the reaction pH set to 11.5, 10.5, and 11.5, respectively. CdTe QDs capped with GSH were prepared via a one-step microwave-assisted synthesis, with a Cd²⁺:Te²⁻:GSH molar ratio of 1:0.2:0.77 and the pH adjusted to 9.8 [39].

After synthesis, CdTe QDs were purified by precipitation with absolute ethanol to remove unreacted species and excess ligands. The resulting solids were recovered by centrifugation, vacuum-dried, and stored in amber glass vials under dark conditions. Intermediate QD solutions were prepared by dissolving appropriate amounts of the dried materials in deionised water.

MPA- and TMA-capped AgInS₂ QDs were synthesised using a microwave-assisted aqueous route following protocols previously developed by our group [40,41]. The intermediary solutions of the ternary QDs were prepared daily by appropriate dilution directly from the crude solution. CDs were prepared via a hydrothermal method proposed by our group [16,20,23], using citric acid and ethylenediamine as carbon and nitrogen sources, respectively. The crude CD solution was purified by dialysis against ultrapure water for five days using a membrane (Spectra/Por^® 6, MWCO 1000, Spectra Lab, Los Angeles, CA, USA). Intermediate CD solutions were freshly prepared by appropriate dilution of the purified stock solution.

2.2. Equipment

Microwave-assisted syntheses of QDs were carried out in a CEM Discover SP^® microwave reactor (Matthews, NC, USA), operated through Synergy™ software and equipped with an ActiVent™ pressure control system, an infrared temperature probe, and a PowerMAX™ cooling module. Solution pH measurements were conducted using a sensION+ pH 31 GLP laboratory pH meter (Hach^®, Loveland, CO, USA).

A Jasco FP-6500 fluorescence spectrometer (Easton, MD, USA) was used to perform all spectrofluorometric measurements, including PL evaluation and optical characterization of the synthesized nanomaterials. A DeltaFlex™ TCSPC lifetime spectrofluorometer (Horiba Scientific, Kyoto, Japan) was used to conduct time-resolved PL measurements, equipped with a NanoLED pulsed light source, a PPD detector, and DeltaHub timing electronics. A LUDOX^® scattering solution was used to calibrate the instrument response function (IRF).

2.3. Fluorometric Procedure

Kinetic fluorescence assays were carried out using freshly prepared intermediate solutions of CDs and MPA-AIS QDs. The MPA-AIS QDs intermediate solution was prepared daily by diluting 400 µL of the crude suspension to a final volume of 2.0 mL, while the CDs intermediate solution was obtained by diluting 100 µL of the stock dispersion to 150 mL with deionized water.

For calibration, appropriate volumes of chloramphenicol intermediate solution (207 mg L⁻¹) and deionized water were sequentially mixed in a quartz cuvette. Afterwards, 50 µL of γ-MPA-AIS QDs and 50 µL of CDs were added, and the total volume was adjusted to 2000 µL. The mixture was stirred, and PL emission spectra were recorded every minute for 5 min, using an excitation wavelength of 360 nm and an emission range of 400 to 700 nm (excitation and emission slit widths set to 5.0 nm).

The same procedure was applied to the analysis of pharmaceutical samples, replacing the standard solutions with the corresponding intermediate sample solutions containing the drug.

2.4. Data Analysis: Chemometric Modelling

The kinetic emission fluorescence data reported in this manuscript have a three-dimensional structure (A × B × C, where A represents the sample number, B the emission wavelength, and C the acquisition time). In this context, the unfolded partial least squares (U-PLS) model was employed to develop a quantitative model relating the PL data to the concentration of the chloramphenicol molecule [42]. Initially, this chemometric model refolds the kinetic PL emission data into an AB × C matrix and estimates the optimal number of latent variables for quantifying chloramphenicol using the Haaland and Thomas criterion [43], based only on the calibration samples and employing leave-one-out cross-validation. In this criterion, the optimal number of latent variables is selected when the prediction error sum of squares (PRESS), which measures the model’s fitness for a specific latent variable, shows no significant difference between consecutive latent variables. This is indicated by the F-ratio probability, which should be below 0.75. After selecting the optimal number of latent variables, the calibration model is complete, and the independent validation set can be projected to assess its accuracy. The parameters used for the evaluation of the accuracy of the developed models were: root-mean-square error of calibration (RMSEC), root-mean-square error of cross-validation (RMSECV), root-mean-square error of prediction (RMSEP), determination coefficient of calibration (R²_C), determination coefficient of cross-validation (R²_CV), determination coefficient of prediction (R²_P), relative error of prediction (REP), limit-of-determination (LOD), and limit-of-quantification (LOQ). The following equations were used to calculate the RMSEC, RMSECV, RMSEP, and REP.

$$RMSE=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^N{\left(\widehat{Y_i}-Y_i\right)}^2}{N}}}$$

(1)

where $Y_i$ stands for the predicted value for sample $i$, $\widehat{Y_i}$ is the experimental value for sample $i$, for calibration (RMSEC), cross-validation (RMSECV) and prediction (RMSEP) and $N$ represents the sample number.

$$REP=100\times {\displaystyle \frac{RMSEP}{{\overline{Y}}_{cal}}}$$

(2)

where ${\overline{Y}}_{cal}$ stands for the average calibration concentration, and RMSEP represents the root mean square error of prediction.

The kinetic emission fluorescence spectra were obtained within 400 and 700 nm, using a 1 nm wavelength interval, each minute over different intervals of time, which, in the cases of 20, 10, 15 and 5 min of spectral acquisition, yielded a total of 6321 (301 × 21 = 6321), 3311 (301 × 11 = 3311) and (301 × 6 = 1806) data points, respectively. All the kinetic emission fluorescence data were mean-centered before the application of the U-PLS model. The U-PLS model and its calculations, including figures of merit, were performed in MATLAB 2023a (MathWorks, Natick, MA, USA) using the MVC2 graphical interface [44].

3. Results and Discussion

3.1. Emission Characteristics and Photophysical Properties of the Synthesized QDs

The PL properties of as-prepared QDs, namely, carbon dots, CdTe QDs capped with different thiol ligands, and AgInS₂ QDs stabilized with MPA or TMA, were investigated by steady-state and time-resolved photoluminescence spectroscopy.

As can be seen in Figure 1, the emission spectra of the synthesized nanomaterials showed different PL profiles depending on their surface functionalization and composition. In the case of carbon dots, a blue emission centered at approximately 440 nm was observed, whereas CdTe QDs showed narrower and more symmetric emission bands with maximum emission wavelengths ranging from 541 to 583 nm. On the contrary, AgInS₂ QDs revealed broader emission bands, which is a typical characteristic of ternary semiconductor nanocrystals. The narrower and more symmetric emission bands of CdTe QDs demonstrate a more homogeneous particle size distribution, while the broader emission observed for AgInS₂ QDs is ascribed to multiple relaxation pathways involving electron–hole recombination and radiative transitions through donor–acceptor mid-gap energy states within the band gap [22].

Table 1. Summary of the maximum emission wavelengths (λ_max) and average PL lifetime values of each synthesized nanomaterial.

QDs	λmax. (nm)	PL Lifetimes	1st Decay Component	2nd Decay Component	3rd Decay Component	τaverage (ns)
CDs	440	τi (ns)	12.85 ± 0.08	5.53 ± 0.08	1.52 ± 0.02	6.55 ± 0.08
		Bi (%)	22.32	57.06	20.62
CYS-CdTe	541	τi (ns)	56.5 ± 0.3	19.7 ± 0.3	2.47 ± 0.03	46.7 ± 0.3
		Bi (%)	45.13	44.64	10.23
GSH-CdTe	568	τi (ns)	60.4 ± 0.3	23.6 ± 0.7	4.90 ± 0.06	50.9 ± 0.5
		Bi (%)	48.50	38.57	12.93
MES-CdTe	559	τi (ns)	32.2 ± 0.3	9.0 ± 0.1	2.12 ± 0.03	23.2 ± 0.2
		Bi (%)	26.45	52.96	20.59
MPA-CdTe	583	τi (ns)	122.6 ± 1	36.7 ± 0.2	13.1 ± 0.4	78.9 ± 0.8
		Bi (%)	19.59	52.38	28.03
MPA-AIS	577	τi (ns)	401 ± 1	112 ± 3	13.7 ± 0.4	388 ± 2
		Bi (%)	83.74	14.77	1.49
TMA-AIS	583	τi (ns)	362 ± 1	83 ± 1	6.2 ± 0.2	348 ± 1
		Bi (%)	80.48	17.75	1.76

summarizes the maximum emission wavelengths and PL lifetime values of all as-prepared QDs. Time-resolved PL measurements revealed that the PL decay profiles of the QDs were adequately fitted using multi-exponential functions, which required three exponential components. This behavior indicates that PL emission occurs through multiple radiative recombination processes [22].

Additionally, the results summarized in Table 1 demonstrate that the PL lifetime strongly depends on both the nanomaterial composition and the nature of the surface capping ligand. For example, CDs displayed a short average lifetime (τ ≈ 6.6 ns), whereas CdTe QDs showed longer lifetimes, ranging from approximately 23 to 79 ns, depending on the capping ligand used. On the other hand, AgInS₂ QDs exhibited much longer lifetimes, on the order of several hundred nanoseconds, which is consistent with the more complex electronic structure of ternary QDs. When comparing to binary CdTe QDs, the presence of additional energy levels within the band gap of AgInS₂ nanocrystals increases the number of available radiative pathways, resulting in longer PL lifetimes [22].

As mentioned above, the surface passivation of the QDs provided by the different ligands significantly affects the obtained PL lifetime values. The passivation of binary CdTe QDs with MES exhibited shorter lifetimes than those observed when the capping was TMA-, MPA- or GSH, which demonstrate a lower efficiency of surface passivation. In the case of AgInS₂ QDs, TMA-stabilized nanocrystals exhibited longer lifetimes than those capped with MPA. Considering that these ligands act as a surface passivation layer, any variation in their binding affinity to the QDs’ surface and the degree of surface passivation could affect non-radiative recombination processes and, consequently, the observed PL lifetimes.

3.2. PL Response and Kinetic Behavior of the QDs Toward Chloramphenicol

Considering that the reactivity and kinetic profile of QDs toward a given analyte depend on the nanomaterial composition, size, and surface capping ligand, interaction studies between each synthesised nanomaterial and chloramphenicol were performed to evaluate the PL response and the time-dependent evolution of the PL emission spectra in the absence and presence of the drug over 30 min.

For the interaction studies, the final concentrations of all synthesised PL nanomaterials were adjusted to obtain similar initial fluorescence intensities. The final working concentrations of the CdTe QDs stored as powdered material and capped with MPA, Cys, GSH, and MES were adjusted to 4.50, 7.00, 5.25, and 67.50 µg mL⁻¹ (w/v), respectively. For the ternary QDs stored as a solution, dilution factors of 1:20 and 1:200 were used for the TMA- and MPA-capped AgInS₂ QDs, respectively. In the case of carbon dots, the crude solution was diluted 60,000-fold prior to analysis.

PL spectra of each nanomaterial were acquired in the absence and presence of increasing concentrations of chloramphenicol. For each concentration, as well as for the blank, spectral acquisition was performed at 1-min intervals, for a maximum of 30 min.

3.2.1. CDs–Chloramphenicol Interaction

In the study of the interaction between CDs and chloramphenicol, the photoluminescence of the nanomaterial remained essentially unchanged immediately after mixing, even in the presence of increasing concentrations of the antibiotic (Figure 2A). The Stern–Volmer plot in Figure 2B shows that the slope is practically zero, indicating the low reactivity of the CDs at the initial interaction time.

The analysis of the emission spectra evolution over 30 min for the CDs in the absence of the antibiotic (Figure 2C) revealed that these nanomaterials are stable, with a slight decrease in emission intensity observed (~10.3%). Figure 2D illustrates the temporal evolution of the maximum emission intensity of CDs in the absence and presence of increasing concentrations of chloramphenicol. As mentioned above, at the initial time point (t = 0 min), no significant changes in the PL intensity of the CDs were observed, regardless of antibiotic concentration. This demonstrates that the interaction between the PL probe and the analyte is negligible immediately after mixing the nanomaterial and chloramphenicol. However, increasing the interaction time reveals a progressive concentration-dependent quenching effect of chloramphenicol on the CDs’ PL emission.

These results demonstrate that the interaction between CDs and chloramphenicol is not immediate but time-dependent, resulting in a kinetic fluorescence response. PL quenching increases not only with chloramphenicol concentration but also with interaction time, indicating a progressively stronger interaction.

This kinetic behavior can be observed in Figure 2E through the analysis of the Stern–Volmer plots performed at different interaction times. In fact, the Stern–Volmer slope increases with increasing interaction time (Table S1—Supplementary Material). This systematic increase in the Stern–Volmer constant indicates a progressive enhancement of the CDs’ PL quenching efficiency by the antibiotic as the interaction time increases, demonstrating that the PL probe/antibiotic interaction becomes more significant as the system progresses. In this way, the emission spectra of CDs acquired at the end of 20 min in the presence of increasing chloramphenicol concentrations (Figure 2F) show a more progressive and accentuated inhibition in the PL intensity compared to that obtained at the initial interaction time (t = 0 min), thereby confirming a concentration-dependent and kinetically controlled quenching process. The increase in the PL response’s sensitivity over time highlights the importance of kinetic information for the analytical determination of the antibiotic.

3.2.2. CdTe QDs–Chloramphenicol Interaction

The evaluation of the interaction between chloramphenicol and CdTe QDs stabilized with distinct thiol capping ligands seeks to assess their reactivity, as well as to monitor the PL spectra changes over time to appraise the intrinsic PL stability of each nanomaterial. Different responses and behaviors were expected, given the nature of the surface-capping ligand.

The results from the interaction between chloramphenicol and MPA-CdTe QDs, depicted in Figure 3A, clearly showed concentration-dependent PL quenching of the nanomaterial. The corresponding Stern–Volmer plot displayed a linear relationship (Figure 3B), with a Stern–Volmer constant (K_sv) of 0.0065 L mg⁻¹, demonstrating a significant interaction between the antibiotic and the nanomaterial. Figure 3C showed good PL stability for MPA-CdTe QDs, with a moderate decrease of about 18.6% in PL intensity after 30 min. This result suggests that this kind of nanocrystal retains sufficient photostability to allow implementing kinetic measurements.

Analysis of the evolution of the maximum PL intensity of MPA-CdTe QDs in the absence and presence of increasing concentrations of chloramphenicol, shown in Figure 3D, indicates that the interaction between MPA-CdTe QDs and chloramphenicol occurs immediately after mixing, reaching equilibrium almost instantaneously. Indeed, the quenching effect caused by the antibiotic occurs immediately at the initial interaction time (t = 0 min) and does not become more pronounced as the interaction time is prolonged.

The PL intensity remains relatively constant over 30 min, following the same trend observed for the blank. This result demonstrates that, in contrast to the behavior observed for CDs, the quenching effect of the chloramphenicol on the PL intensity of MPA-CdTe QDs is concentration-dependent rather than time-dependent, indicating the absence of a kinetically controlled fluorescence response.

In the interaction process with GSH-capped CdTe QDs (Figure S1—Supplementary Material), a gradual PL quenching effect was also observed by adding increasing concentrations of chloramphenicol. The Stern–Volmer plot with a K_sv value of 0.0077 L mg⁻¹ shows a slightly higher quenching efficiency for GSH-capped CdTe QDs than for MPA-CdTe QDs. However, despite this apparent higher sensitivity, GSH-CdTe QDs exhibited very low PL stability over time. In fact, a pronounced decrease in emission intensity of around 74% was observed over 30 min in the absence of the antibiotic. Therefore, these nanomaterials are not suitable for kinetic-based analytical applications due to their low photostability.

The interaction process with Cys-capped CdTe QDs also resulted in PL inhibition of the nanoprobe as the antibiotic concentration increased (Figure S2—Supplementary Material). Nevertheless, a lower quenching efficiency was observed in the Stern–Volmer analysis, with a K_sv value of 0.0039 L mg⁻¹, indicating a weaker reactivity than that observed with MPA- and GSH-capped CdTe QDs. In addition, these nanomaterials exhibited poor photostability, with a substantial decrease in PL intensity (~64%) over 30 min in the absence of chloramphenicol, which limits their applicability for time-resolved or kinetic fluorescence measurements.

On the contrary, regarding MES-capped CdTe QDs, no significant reactivity was noticed because their PL intensity remained practically unaltered by increasing chloramphenicol concentrations (Figure S3—Supplementary Material). This low reactivity was confirmed by the Stern–Volmer plot, which showed an almost negligible slope and a K_sv value close to zero. In terms of PL stability, the emission intensity of MES-CdTe QDs remained nearly constant over 30 min in the absence of chloramphenicol, indicating good photostability. Despite this favorable stability, the lack of reactivity impairs their use for the determination of chloramphenicol.

In summary, although MPA- and GSH-capped CdTe QDs exhibit noteworthy reactivity towards chloramphenicol, only MPA-capped CdTe QDs allow for achieving sufficient PL quenching efficiency, maintaining at the same time suitable photostability. Therefore, among the binary QDs analyzed, the MPA-CdTe QD is the only one that enables the development of a robust and sensitive sensing platform for chloramphenicol because it allows a better balance between reactivity and temporal photostability.

3.2.3. AgInS₂ QDs–Chloramphenicol Interaction

The photostability and reactivity of ternary AgInS₂ QDs stabilized with different capping ligands (MPA and TMA) towards chloramphenicol were also investigated. Both MPA- and TMA-AgInS₂ QDs showed concentration-dependent PL quenching in the presence of chloramphenicol (Figure 4A and Figure S4—Supplementary Material). Additionally, the corresponding Stern–Volmer plots for both nanomaterials showed linear behavior. However, for TMA-AgInS₂ QDs, a slightly higher K_sv value was observed (0.0057 L mg⁻¹) than that obtained for MPA-capped AgInS₂ QDs (K_sv = 0.0048 L mg⁻¹) (Figure 4B). Therefore, TMA-stabilized nanomaterials showed a slightly higher quenching efficiency and greater sensitivity toward the antibiotic. However, in terms of photostability, TMA-AIS QDs showed lower PL stability over time in the absence of chloramphenicol, with a 32.8% decrease in emission intensity over 30 min, thereby impairing their use for kinetic or time-resolved PL approaches. In contrast, Figure 4C reveals a moderate PL intensity inhibition of MPA-capped AgInS₂ QDs (~14.6%) over 30 min in the absence of the antibiotic, demonstrating a superior photostability. Consequently, these nanomaterials stabilized with MPA are more suitable for methodologies that involve time-dependent fluorescence monitoring.

The monitoring of the maximum PL emission intensity of MPA-AIS QDs in the absence and presence of chloramphenicol over 30 min, depicted in Figure 4D, showed that PL quenching does not change significantly with interaction time but rather depends on the analyte concentration. These results demonstrate that, similarly to what occurs with MPA-CdTe QDs and unlike what is observed with CDs, the interaction between chloramphenicol and MPA-AgInS₂ QDs occurs instantaneously, without exhibiting any kinetic behavior.

Overall, despite MPA-capped AIS QDs exhibiting slightly lower sensitivity to chloramphenicol than TMA-capped nanocrystals, they provide a better compromise between sensitivity and photostability, making them more suitable for implementing kinetic-based sensing strategies for reliable PL-based determination of chloramphenicol.

3.3. Development of Dual-Emission Fluorescent Probes for Chloramphenicol Monitoring

Taking into account the results obtained in the individual interaction and kinetic studies described in the previous sections, two dual-emission PL probes were developed by combining CDs with MPA-AIS QDs and MPA-CdTe QDs, which exhibited complementary PL behavior towards chloramphenicol. CDs were selected because of their time-dependent PL response, while MPA-capped CdTe QDs and MPA-stabilized AgInS₂ QDs were chosen due to their instantaneous and concentration-dependent quenching behavior, combined with suitable photostability. These nanomaterials were combined into a single dispersion to obtain dual-emission systems that provide more detailed fluorescence information, yielding kinetic and steady-state responses. In fact, the use of these dual-emission systems in combination with PL kinetic data enables the acquisition of multidimensional PL data, enhancing analytical robustness and providing additional selectivity through chemometric analysis.

The photostability of the two distinct combined probes, namely, CDs/MPA-CdTe and CDs/MPA-AgInS₂ dual-emission probes, and their PL response to chloramphenicol were carefully evaluated. Subsequently, the time-resolved dual-emission fluorescence data were analyzed using U-PLS chemometric models to compare the analytical performance of both dual-emission probes, thereby identifying the most effective sensing platform for chloramphenicol monitoring.

3.3.1. Development and Photostability of Dual-Emission Probes

The combination of CDs with MPA-CdTe QDs and MPA-AgInS₂ QDs resulted in two distinct combined PL nanoprobes, whose emission properties are different from those observed in the individual nanomaterials. Figure 5 shows the comparison between the emission spectra of each QD individually and in the mixture, aiming to assess the effect of probe combination on PL behavior, as well as the temporal stability of the resulting dual-emission probes.

In the case of the CDs/MPA-CdTe system (Figure 5A), the individual nanomaterials exhibited lower emission intensities than those observed after mixing. In fact, the resulting dual-emission probe showed a general enhancement of the PL signal, which can be explained by a more favorable photophysical environment that promotes changes in surface passivation or reduces non-radiative recombination pathways. However, this combined probe showed lower PL stability than the individual nanomaterials (Figure 5B) with a more pronounced decrease in emission intensity over 20 min. This behavior indicates that the PL enhancement observed immediately after the combination of nanomaterials is transient and progressively diminishes with time. This occurs probably due to gradual modifications in the surface electronic states of the nanomaterials within the mixed system.

Regarding the CDs/MPA-AgInS₂ system (Figure 5C), the dual-emission probe showed a decrease in PL intensity compared to that observed in the individual nanomaterials. This behavior can be related to the promotion of non-radiative deactivation pathways via energy or charge transfer processes, alterations in surface electronic states, or the inner filter effect. Compared to the CDs/MPA-CdTe system, CDs/MPA-AgInS₂ dual emission probe showed a higher photostability over time with negligible emission intensity variations (Figure 5D). The absorption and emission spectra of the CDs/MPA-AgInS₂ dual-emission probe exhibited minimal overlap, confirming the absence of significant mutual optical interference upon probe mixing (Figure S5—Supplementary Material).

The results indicate that combining distinct PL nanomaterials into a single probe can affect both emission intensity and stability. Despite the CDs/MPA-CdTe probe exhibiting an enhanced initial PL intensity, the CDs/MPA-AgInS₂ system offers superior photostability, which is important to implement kinetic PL-based analytical methodology.

3.3.2. PL Response of the Dual-Emission Probes Towards Chloramphenicol

The PL response of the CDs/MPA-CdTe and CDs/MPA-AgInS₂ dual-emission probes in the absence and presence of increasing chloramphenicol concentrations was evaluated, and the evolution of the emission spectra for each analyte concentration tested was monitored over 20 min. Figure 6 shows the PL spectra acquired immediately after mixing (t = 0 min) and after 20 min of interaction, allowing comparison of the initial and final stages of the PL response.

For the CDs/MPA-CdTe dual-emission probe, the PL emission spectra recorded at t = 0 min (Figure 6A) show clear concentration-dependent PL quenching, which is more pronounced in the emission band related to the inorganic MPA-CdTe QDs. This result is in accordance with the previously observed instantaneous quenching behaviour of this nanomaterial in the individual interaction studies. After 20 min of interaction (Figure 6B), no evident enhancement of the inhibition effect is observed when compared to t = 0 min, even in the emission band corresponding to CDs, which exhibited a clear time-dependent PL response when evaluated as an individual nanoprobe.

A similar PL behaviour was observed for the CDs/MPA-AgInS₂ probe. In fact, the PL spectra acquired at t = 0 min shown in Figure 6C reveal a progressive decrease in PL intensity with increasing chloramphenicol concentration, which affects more clearly the emission band corresponding to the inorganic MPA-AgInS₂ QDs. This PL response is in agreement with the instantaneous concentration-dependent PL quenching of MPA-AgInS₂ QDs when evaluated individually. Additionally, no substantial additional PL quenching effect is visually evident after 20 min (Figure 6D), at which point the overall spectral profiles remain similar to those observed at t = 0 min.

In both probes combined, the observable kinetic behavior in the case of the individual CDs/chloramphenicol interaction is not readily distinguishable by direct visualization or univariate analysis of the obtained data. In fact, the more sensitive and immediate PL response of the inorganic QDs prevails in the overall PL signal, hiding the slower kinetic contribution of the CDs. Consequently, the influence of the interaction time on the PL response of the dual-emission probes cannot be reliably assessed through univariate or qualitative spectral analyses.

For this reason, the kinetic dual-emission PL data were analyzed using chemometric models to explore their multidimensional information. In the following sections, the U-PLS model was applied to analyze the kinetic PL datasets obtained for both dual-emission probes, aiming to evaluate and compare their analytical performance and to determine if there is a kinetic contribution that improves the quantitative determination of chloramphenicol.

Chemometric Assessment of the Analytical Performance of CDs/MPA-CdTe and CDs/MPA-AgInS₂ Dual-Emission Probes

Considering the results obtained from the interaction and kinetic PL studies for CDs/MPA-CdTe and CDs/MPA-AgInS₂ dual-emission probes, U-PLS calibration models were developed using the kinetic dual-emission PL datasets acquired over 20 min for both probes, within a chloramphenicol concentration range of 10–196 mg L⁻¹, aiming to compare their analytical performance quantitatively. The resulting calibration curves for both combined probes are presented in Figure 7.

As can be observed, both dual-emission probes exhibited similar analytical sensitivities (0.83 and 0.84 for CDs/MPA-CdTe and CDs/MPA-AgInS₂, respectively) and comparable coefficients of determination. One calibration standard from the CDs/MPA-CdTe dataset was identified as an outlier and was therefore excluded from the calibration.

Taking into account the similar analytical performance of the two dual-emission systems, the CDs/MPA-AgInS₂ probe was selected as the sensing platform for chloramphenicol determination, due to its cadmium-free composition and, consequently, its lower environmental impact.

PL Quenching Mechanisms of the CDs/MPA-AgInS₂ Dual-Emission Probe

Considering that the CDs/MPA-AgInS₂ dual-emission probe was selected for antibiotic determination, the interaction mechanism between each nanomaterial of the sensing platform and chloramphenicol was carried out to clarify what PL quenching process occurs. Therefore, time-resolved PL measurements for CDs and MPA-AIS QDs, individually, in the absence and presence of increasing chloramphenicol concentrations, were performed (Figure 8).

The results obtained for CDs revealed that the PL decay profiles (Figure 8A) remained practically unchanged as the chloramphenicol concentration increased, whereas a progressive inhibition of the steady-state PL intensity was observed (Figure 8B). This behavior indicates that PL quenching occurs via a static mechanism, where the formation of non-fluorescent ground-state complexes between CDs and chloramphenicol occurs [45]. The resulting non-fluorescent complex does not affect the excited-state lifetime of the remaining emissive species [45].

On the other hand, PL decay profiles of MPA-AgInS₂ QDs became progressively faster with increasing chloramphenicol concentration (Figure 8C), leading to a decrease in the average PL lifetime.

Moreover, the Stern–Volmer constants obtained from steady-state fluorescence and lifetime-based analyses (Figure 8D) were very similar (k_SV (steady-state) = 0.0043 L μmol⁻¹ and k_SV (lifetime) = 0.0037 L μmol⁻¹), indicating that PL quenching occurs through a dynamic mechanism. This means that collisional interactions between excited AgInS₂ QDs and chloramphenicol molecules occur, which results in non-radiative deactivation of the excited states via charge-transfer processes involving exciplex formation [45,46].

Optimization of the Kinetic Fluorescence Acquisition Time

After selecting the CDs/MPA-AgInS₂ dual-emission nanoprobe as the most suitable sensing platform, the kinetic PL acquisition time was optimized in order to obtain the best compromise between analytical performance and rapid quantification of chloramphenicol. Therefore, U-PLS calibration models were constructed using different kinetic time windows (20, 15, 10, 5, and 2 min) within a concentration range of 10–175 mg L⁻¹. The three-dimensional and two-dimensional kinetic PL emission profiles obtained upon interaction of the CDs/MPA-AgInS₂ nanoprobe with a 71.4 mg L⁻¹ chloramphenicol standard are depicted in Figure 9. The analysis of these kinetic PL emission profiles reveals that the most pronounced PL quenching effect occurs within the first 5–10 min of interaction.

Nevertheless, aiming to evaluate the impact of acquisition time on analytical performance, calibration models were developed at different interval times (20, 15, 10, 5, and 2 min), and the corresponding calibration curves are presented in Figure 10. The analysis of the results in terms of sensitivity and coefficients of determination shows that similar analytical performance is achieved for time windows between 5 and 20 min. In contrast, the shortest acquisition time (2 min) resulted in a perceptible decrease in both sensitivity and linearity.

Bearing in mind the development of a rapid and reliable method for chloramphenicol quantification, a kinetic acquisition time of 5 min was selected to achieve an optimal balance between analysis rapidity and analytical performance. The next step was focused on evaluating whether the analytical performance could be improved by restricting the chemometric analysis to specific spectral regions, namely the emission band corresponding to carbon dots (400–525 nm), to MPA-AgInS₂ QDs (526–700 nm), or by using the full PL emission spectrum.

3.3.3. Optimization of the Spectral Region for Kinetic PL Analysis

In this context, U-PLS calibration models were developed using different spectral regions of the CDs/MPA-AgInS₂ dual-emission nanoprobe to evaluate the influence of the selected emission region on analytical performance. Three spectral ranges were analyzed, namely the full PL emission spectrum (400–700 nm), the emission band associated with CDs (400–525 nm), and the emission band corresponding to MPA-AgInS₂ QDs (526–700 nm). For this purpose, calibration curves for each spectral region were constructed within a chloramphenicol concentration range of 20–180 mg L⁻¹ using kinetic PL emission data acquired over 5 min (Figure 11).

The results (Figure 11 and Table S2) show that the emission region corresponding to MPA-AIS QDs (526–700 nm) provides analytical performance comparable to that obtained using the full spectrum, confirming that ternary QDs play a major role in the detection process. In contrast, the spectral region associated exclusively with CDs (400–525 nm) leads to slightly higher prediction errors, indicating a lower individual analytical contribution. Nevertheless, when the full PL emission spectrum (400–700 nm) is used, slightly lower RMSEC and RMSECV values are obtained, demonstrating that the combination of both emission bands improves the overall robustness of the chemometric model. The dual-emission nanoprobe does not rely on equal individual contributions from each nanomaterial, but rather on the complementary analytical information in the multiway dataset. This combined spectral information enhances model stability against potential uncalibrated interferents, which are unlikely to affect both emissive components equally.

This result reinforces the idea that, by combining the relevant analytical information from each emission band, the robustness of the chemometric model for chloramphenicol quantification is enhanced due to complementary PL responses.

3.3.4. Quantification of Chloramphenicol in Commercial Pharmaceutical Samples Using the Optimized Dual-Emission Probe

The analytical performance of the CDs/MPA-AgInS₂ dual-emission nanoprobe was evaluated to quantify chloramphenicol in commercially available pharmaceutical samples under the optimized conditions. For this purpose, a definitive U-PLS calibration model within a chloramphenicol concentration range of 31.05–165.6 mg L⁻¹ was developed using the full PL emission range (400–700 nm) acquired during the first 5 min of interaction, and employing one latent variable. The corresponding calibration plot, which compares real versus cross-validated concentrations, is shown in Figure 12.

The excellent performance of the calibration method is confirmed by the achieved figures of merit, namely RMSEC and RMSECV values of 4.690 and 5.931 mg L⁻¹, respectively, and high coefficients of determination for calibration (R²_C = 0.989) and cross-validation (R²_CV = 0.991). Additionally, as shown in Figure 12, all calibration standards fall within the 95% confidence interval, thereby demonstrating the reliability and robustness of the U-PLS model.

After calibration, the developed method was applied to the quantification of chloramphenicol in commercial pharmaceutical formulations. Two commercial samples, each originating from three different batches and prepared at two dilution levels, were analyzed to evaluate the accuracy of the chemometric model. The predicted concentrations obtained using the developed calibration model are summarized in

Table 2. Predicted and real chloramphenicol concentrations in commercial pharmaceutical samples and corresponding validation figures of merit obtained using the optimized U-PLS model.

Commercial Sample	Batch	Dilution	Chloramphenicol Quantification (mg L−1)
			Real	Predicted
A	A1	1	88.0	90.5
A	A1	2	119.0	127.9
A	A2	1	88.0	89.0
A	A2	2	119.0	137.0
A	A3	1	88.0	85.1
A	A3	2	119.0	121.6
B	B1	1	88.0	89.6
B	B1	2	119.0	121.9
B	B2	1	88.0	84.5
B	B2	2	119.0	130.1
B	B3	1	88.0	86.2
B	B3	2	119.0	130.9
RMSEP (mg L−1)			7.702
R2P			0.980
REP (%)			7.4
LOD (mg L−1)			9.6
LOQ (mg L−1)			28.8

.

Through the analysis of the results for the commercial samples, good agreement between the real and predicted chloramphenicol concentrations was observed, which attests to the accuracy and applicability of the proposed methodology. The suitability of the method was validated by the good figures of merit obtained, including an RMSEP of 7.702 mg L⁻¹, a prediction coefficient of determination (R²_P) of 0.980, and a relative error of prediction (REP) of 7.4%. Additionally, the method displayed good limits of detection (LOD = 9.6 mg L⁻¹) and quantification (LOQ = 28.8 mg L⁻¹) for chloramphenicol quantification, as they are within the concentration levels usually found in pharmaceutical formulations.

To conclude, the combination of a dual-emission CDs/AgInS₂ QDs nanoprobe with kinetic response and multivariate chemometric analysis proved to be a reliable, accurate, and environmentally friendly approach for the determination of chloramphenicol. In fact, the application of the U-PLS model to handle kinetic PL data allowed for improved analytical performance, thus demonstrating its potential for practical pharmaceutical analysis and quality control applications.

4. Conclusions

In this work, an efficient, selective, and environmentally friendly dual-emission nanoprobe comprising CDs and MPA-AgInS₂ QDs was successfully developed for the quantification of chloramphenicol in pharmaceutical formulations.

PL response, stability, and kinetic behavior towards the antibiotic of different nanomaterials, including CDs, binary CdTe QDs, and ternary AgInS₂ QDs, were assessed. In fact, through the analysis of individual interaction studies, different quenching behaviors were observed depending on the nanomaterial composition and surface chemistry. CDs showed a clear time-dependent PL response, with PL intensity more markedly inhibited as the interaction time increased, demonstrating a kinetic response to the antibiotic. On the other hand, binary and ternary inorganic QDs showed no significant kinetic contribution, as both nanomaterials exhibited instantaneous, concentration-dependent PL quenching that did not become more pronounced with increasing interaction time. However, they can be used for time-dependent analysis due to their high photostability.

These complementary responses of the inorganic QDs and CDs towards chloramphenicol were explored to develop two dual-emission probes by combining CDs with MPA-CdTe QDs and MPA-AgInS₂ QDs. The analytical performance and kinetic contributions of both combined probes to the antibiotic were evaluated by analysing the kinetic dual-emission datasets with U-PLS chemometric models, which proved to be essential for extracting multidimensional information from the PL data and to compare which sensing platforms were the most efficient.

Among the dual-emission probes evaluated, the CDs/MPA-AgInS₂ QDs system was revealed to be the most suitable sensing platform because a better compromise between reactivity and photostability was achieved. Additionally, this dual-emission sensing probe is more environmentally friendly due to its cadmium-free composition.

Regarding the optimization process, a reliable and efficient analytical performance was obtained using only the first 5 min of interaction, which enables a rapid chloramphenicol quantification without compromising sensitivity or linearity. Additionally, the best performance was achieved using the full emission spectrum, which clearly indicates the advantage of combining complementary PL data from both emission bands within a single chemometric model.

In terms of the quenching mechanism process, the two individual nanomaterials of the selected dual-emission probe exhibited distinct behaviors. For CDs, PL quenching caused by the analyte occurred via a static mechanism, while for MPA-AgInS₂ QDs, the inhibition of their optical properties occurred due to a dynamic process involving charge-transfer phenomena. These distinct quenching mechanisms also contribute to a synergetic performance of the combined probe and its suitability for chemometric exploitation.

The proposed methodology was then successfully applied to the determination of chloramphenicol in commercial pharmaceutical formulations, yielding accurate and reliable results with good figures of merit, including low prediction errors and suitable limits of detection and quantification. To conclude, the combination of dual-emission nanoprobes with kinetic fluorescence monitoring and chemometric analysis proves to be a powerful and versatile approach for antibiotic determination with improved selectivity, robustness, and environmental compatibility. In addition, this methodology has strong potential for routine pharmaceutical analysis, which can be extended to other analytes of analytical and biomedical relevance.