pH-responsive delivery of hydrophobic anticancer drugs using boron and nitrogen Co-doped carbon dots as fluorescent nanocarriers Authors: Seyed Mostafa Jafari, Saeed Masoum, Elahe Seyed Hosseini Published: 2026 ABSTRACT In recent years, carbon dots have attracted considerable attention within the nanocarbon family due to their remarkable advantages and interesting properties such as high solubility, biocompatibility, tunable photoluminescence, and so on. Carbon dots can also be designed and synthesized according to the desired functionalities (such as sensors, biosensors, or drug delivery systems). In the present study, nitrogen- and boron-co-doped carbon dots (abbreviated as “NBCDs”) were successfully used as highly fluorescent and efficient nanocarriers to load two hydrophobic anticancer drugs, namely curcumin (CUR) and paclitaxel (PTX). The NBCDs were characterized before and after conjugation with CUR and PTX using spectroscopic techniques. After calculating the drug loading capacity (LC%) and adsorption efficiency (AE%) of the NBCDs nanocarriers, the drug release behavior was evaluated in two buffer media (pH 5.0 and pH 7.4) using the dialysis bag method for 72 h. Finally, the pH-dependent release behavior of the drugs was analyzed using several kinetic models.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-59204-4. Introduction Cancer is one of the most devastating diseases and a prevailing threat to humans owing to its high incidence, severe mortality, and low survival rate, while the development of nanotechnology offers promising methods to improve cancer diagnosis and therapy1. However, chemotherapeutic agents have been used to treat cancers. Among them, curcumin (CUR) and paclitaxel (PTX), as the most effective model anticancer drugs, have been used for a variety of cancers. CUR, as a polyphenolic and bioactive natural compound, has numerous properties such as antioxidant, anticancer, anti-inflammatory, antiviral, antifungal, immunomodulatory, antiangiogenic, neuroprotective, antimicrobial, and inhibition of the proliferation of cancer cells2,3 and it is also considered a potential therapeutic agent for treating Alzheimer’s4 and Parkinson’s5 diseases. It has also been used for wound healing6 and in drug delivery systems7,8. In addition, PTX has been used as a potent anticancer drug to treat a wide range of tumors, including colon9, primary epithelial ovarian10, non-small cell lung11, breast cancer12, and prostate cancer13. Most anticancer drugs are hydrophobic and have low water solubility and thus suffer from unfavorable pharmacokinetics, rapid metabolism, and poor bioavailability8,14. By using alternative formulation systems such as liposomes, micelles, and nanoparticles, numerous attempts have been made to increase the water solubility of these drugs. The conventional and commercial approaches for solubilizing hydrophobic anticancer drugs are generally based on the use of surfactants and organic solvents. For instance, the paclitaxel injection formulation (Taxol®) uses a considerable amount of Cremophor EL and dehydrated ethanol to improve its solubility. However, Cremophor EL is associated with adverse effects in humans, and the use of Taxol® in cancer treatment has been linked to several side effects, including peripheral neuropathy, hypersensitivity reactions, and hyperlipidemia15,16. In recent years, nanoparticles have been used as promising nanocarriers in nanosized drug delivery systems to improve the water solubility and bioavailability of CUR and PTX, as hydrophobic anticancer drug models. Among the nanoparticles synthesized so far, fluorescent carbon dots have attracted tremendous attention owing to their attractive properties, including excellent biocompatibility, low toxicity, multiple functional groups, excitation-dependent emission, high resistance to photobleaching, and excellent photostability17. These fascinating materials are nanoparticles with particle sizes of several nanometers18,19. According to previous studies, carbon dots have been used as a desirable platform for the design of novel drug-delivery systems and therapeutic agents20,21. In drug delivery systems, carbon dots are particularly attractive because of their several features including adaptable surface chemistry, ease of cellular ingestion, luminescence properties, and high biocompatibility22. So far, many methods have been proposed to enhance the targeting and bioavailability of CUR and PTX such as the use of nanocarriers8,23. Thereby, nano-based drug delivery systems can improve the overall anticancer activity of CUR24. Its anticancer effect is ascribed to its free-radical scavenging and antioxidant properties25. In the selection of drug delivery vehicles for cancer therapy, nanoparticles size, shape, charge and surface chemistry play a key role26. In the present work, we synthesized N/B co-doped CDs (NBCDs) with a high quantum yield using the hydrothermal method, employing citric acid as a carbon source and ethylenediamine/boric acid as nitrogen and boron sources, as reported in previous studies27,28. In the following, we developed a desirable nanocarrier system to improve the water solubility and bioavailability of PTX and CUR. In the final step, the release kinetics of CUR and PTX from the NBCDs–CUR and NBCDs–PTX conjugates were studied at two different pH values (7.4 as physiological and 5.0 as tumor-cell conditions) using release-kinetics models over a 72 h time course with the KinetDS 3.0 software. Based on the obtained release profiles, the drug-release behavior from the surface of the NBCDs is a pH-dependent process. Although we have recently reported the cellular effects of these nanoconjugates (NBCDs–CUR, NBCDs–PTX) on ovarian cancer cells66, the present study focuses on the design, synthesis, comprehensive characterization, pH-dependent release kinetics, and mathematical modeling of drug release from NBCDs. This study provides a fundamental understanding of the drug–carrier interactions and release mechanisms, which are essential for the rational design of pH-responsive nanocarriers. Experimental section Materials and characterization Quinine sulfate was purchased from Aladdin Company. Curcumin and paclitaxel were purchased from Sigma-Aldrich Company. Deionized water was used throughout the experiments. All other chemicals and reagents were of at least analytical grade and were purchased from Merck. UV-Vis absorption spectra were recorded on a double-beam ultraviolet-visible spectrophotometer (OPTIZEN 3220UV, South Korea). Fourier transform infrared (FT-IR) spectra were obtained using the KBr pellet method, with powder carbon dots embedded in the pellet, on an FT-IR spectrometer (Magna 550 Nicolt, USA). Photoluminescence spectra were recorded using a luminescence spectrophotometer (Perkin-Elmer LS-55). The proton nuclear magnetic resonance (¹H-NMR) spectra of the samples were recorded on a Bruker Avance 400 MHz spectrometer. Zeta potential and dynamic light scattering (DLS) of the samples were obtained using a nanoparticle analyzer (SZ-100, HORBIA). A dialysis bag (MWCO 3500 Da) was used for the purification of the synthesized NBCDs and for drug-release tests. In addition, KinetDS 3.0 software was employed to identify the mathematical models of drug release, and the best fit of the experimental data was chosen among these models. Synthesis of NBCDs In our study, NBCDs were synthesized using citric acid, boric acid and ethylenediamine as precursors by the hydrothermal method27. The brown solution as hydrothermal product was purified using a dialysis membrane (MWCO of 3500 Da) against DI water for 24 h. The purified solution was dried at room temperature and then the collected powder was homogenized and stored in a closed container for future use. Drug loading studies Both the conjugation and the maximum ratio between NBCDs and the drug were investigated by the titration method. Under these conditions, different concentrations of NBCDs were separately added to a fixed concentration of methanolic CUR or PTX (1:1 w/v), and the complex formation between NBCDs and the drugs was monitored by recording absorption spectra. After determining the optimal ratio between the drug and nanocarrier, the resulting mixture was subjected to ultrasound and stirred for 24 h at room temperature in the dark. Then, the solvent was evaporated, and the residue was dried at ~ 50 °C in an oven. It was then dispersed again in 10 mL of DI water, centrifuged (5000 rpm, 10 min), and the supernatant was collected. The pellet was dissolved in a known amount of methanol, and the free drug concentration was determined using a UV-Vis spectrophotometer and the calibration curve for each drug. Thus, the loading capacity or adsorption efficiency of the drug onto the carrier can be calculated. According to the definitions, adsorption efficiency (AE) is defined as the amount of drug adsorbed/entrapped onto the NBCDs relative to the total amount of drug added (Eq. 1), whereas the loading capacity (LC) of the drug can be expressed as the amount of drug loaded per unit weight of the NBCDs as nanocarriers (Eq. 2), as shown below29:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{AE}}\left( \% \right) = ~~\frac{{TDA - FDS}}{{TDA}} \times 100$$\end{document}2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{LC}}\left( \% \right) = \frac{{{\text{weight of drug loaded onto CDs}}}}{{{\text{weight of CDs}}}} \times 100$$\end{document} where AE, LC, TDA, and FDS correspond to adsorption efficiency, loading capacity, total drug added, and free drug in solution, respectively. It should be noted that, in general, the internal entrapment/encapsulation capacity of carbon dots is almost zero; therefore, in this study, we report drug loading based on the amount of surface adsorption. Briefly, for the preparation of NBCDs-CUR, 50 mL of methanolic CUR solution (1:1 w/v) was added dropwise to 25 mL of NBCDs solution (2 mg/mL in PBS, 0.1 M, pH 7.4), with the mixture being stirred at 500 rpm for 30 min at room temperature. After sonication of the mixture in a water bath for ~ 30 min, stirring was continued for another 24 h at room temperature in the dark. In the next step, the methanol solvent was evaporated in an electric oven. To purify the prepared nanoconjugate as a drug delivery system, it was transferred to a dialysis bag and immersed in 50 mL of deionized water with 5 mL of methanol as a co-solvent. The purification was performed using the dialysis method for 2 h to remove unreacted materials, such as free CUR. The orange-colored residue was dried in an oven at 50 °C, and then it was redissolved in 10 mL of deionized water with the help of a magnetic stirrer. In the final step, the aqueous mixture was centrifuged at 5000 rpm for 10 min, and the supernatant containing NBCDs-CUR was collected. Likewise, for loading PTX onto the carbon dot surface, a saturated solution of PTX in methanol (~ 20 mg/mL) was prepared. The PTX solution was then gradually added to the NBCDs solution (60 mg/mL in PBS, 0.1 M, pH 7.4), and the resulting solution was sonicated for ~ 5 min and stirred at 500 rpm for 24 h at room temperature in the dark. Similar to the above-mentioned section, the desired volume of the PTX methanolic solution was added to the NBCDs solution (2 mg/mL in PBS, 0.1 M, pH 7.4). PTX was added to NBCDs at a 3:1 w/w ratio. After mixing NBCDs with the PTX solution, the methanol was evaporated in an electric oven, and the dried mixture was redissolved in an appropriate amount of deionized water after stirring for ~ 30 min. The mixture was then centrifuged at 5000 rpm for 10 min. To purify the prepared mixture as a drug delivery system, it was transferred to a dialysis bag and immersed in 50 mL of deionized water in a flask containing 5 mL of methanol as a co-solvent. The purification was performed using the dialysis bag method for 2 h to remove unreacted materials, such as free PTX. Finally, the purified product was dried at 50 °C, and the homogenized powder was stored. In vitro drug release studies The study of drug release patterns from drug delivery systems is one of the main analyses used to evaluate their efficiency. In this study, the in vitro drug release profile was investigated at two different pH values (5.0 and 7.4), simulating the physiological environments of cancer and normal cells, respectively. The amount of drug released (% cumulative drug release) from NBCDs was estimated over time using the calibration curve of the drug in buffer. The percentage of drug release was calculated using Eq. (3), which is shown below30:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{Drug release}}\left( \% \right) = \frac{{{\text{Drug release concentration}}}}{{{\text{Drug loading concentration}}}} \times 100$$\end{document} Owing to the lactic acid metabolism pathway, cancer cells possess a weakly acidic environment. In this study, drug release was carried out using the dialysis method. Briefly, the desired amount of each dried drug-loaded NBCDs was weighed and dispersed in phosphate-buffered saline (PBS) solution containing 2% (v/v) Tween 80, and then infused into a dialysis bag before being immersed in an appropriate volume of PBS containing 0.5% (v/v) Tween 80 as the release medium31. Then, 2 mL of the release medium were withdrawn and replaced with 2 mL of fresh buffer, and the absorbance of the released drug was measured at the maximum absorption wavelength (λmax) of each drug using a UV-Vis spectrophotometer. The concentrations of released drug were calculated from the drug standard curve. In detail, 8.38 mg of dried NBCDs-CUR was dispersed in 1 mL PBS containing 2% (v/v) Tween 80 and infused into a dialysis bag (the weighed amount for NBCDs-PTX was 7.73 mg). Then, the dialysis bag was immersed in a flask containing PBS (30 mL) with 0.5% (v/v) Tween 80 as the release medium, and the solution was stirred at 37 °C at a constant rotation speed (200 rpm) using a hot plate equipped with a magnetic stirrer. The sampling of the drug release medium was performed as follows: at different time intervals (up to 72 h), 2 mL of the release medium was withdrawn, and the absorbance of the released drug was measured using the calibration curve of CUR or PTX in PBS containing 0.5% (v/v) Tween 80 as the release medium, at 425 nm (λmax of CUR) and 225 nm (λmax of PTX). After sampling, 2 mL of fresh medium was added back into the flask. The drug release mechanism from the NBCDs–drug nanoconjugate was then investigated by fitting the data to various release models32. Kinetic release study In general, mathematical models are useful tools to optimize the design of novel pharmaceutical formulations and to evaluate in vitro and in vivo drug release processes33,34. These models help to understand the mechanisms of release. Some of the main release-kinetics models have been used to predict the release process. Well-known mathematical models, including zero-order, first-order, Korsmeyer-Peppas, Hill, Weibull, Hixson-Crowell, and Higuchi, have been used in most of the literature35. Results and discussion Characterization of NBCDs and drug-loaded NBCDs To study the absorption properties of NBCDs, CUR, NBCDs-CUR, PTX, and NBCDs-PTX samples, their absorption spectra were recorded using a UV-Vis spectrophotometer (see Fig. 1). According to Fig. 1A, the absorption spectra of NBCDs showed two characteristic peaks: one at 215 nm, which is assigned to the C = C bonds or aromatic π orbitals of the carbon core (π-π* transitions), and another at 350 nm, which is attributed to n-π* transitions of C = O or C = N bonds involving functional groups with lone-pair electrons29,36. In addition to the absorption peak at 215 nm, another shoulder peak appears around 230 nm. Typically, these absorption peaks arise from the aromatic π system with extended conjugation, as in the carbon dot structure. Thus, the peak around 230 nm can be attributed to π-π* transitions37. As a result of trapping excited-state energy at the surface states, the latter peak can commonly be attributed to the surface states in highly fluorescent carbon dots28,38. With respect to the existence of different groups on carbon dot surfaces, it can be suggested that each of these functional groups creates an energy-level gap between the LUMO and HOMO orbitals. The water-soluble synthesized NBCDs were brownish in daylight (inset in Fig. 1A, left) and bright blue under UV light (inset in Fig. 1A, right). The UV-Vis spectrum of CUR (dissolved in methanol, Fig. 1B) shows a weak π-π* peak at ~ 240 nm and a strong n-π* peak at ~ 425 nm39. According to Fig. 1B, the UV-Vis spectrum of NBCDs-CUR depicts two peaks at 215 and 340 nm, with increased intensity compared to the corresponding peaks in NBCDs. Meanwhile, a blue shift of ~ 10 nm (from 350 to 340 nm) is observed in the UV-Vis spectrum of NBCDs-CUR compared to the bare NBCDs spectrum. After loading CUR onto the NBCDs, a new broad peak appears between 450 and 500 nm. Hydrogen bonding after drug attachment to the carbon dots makes the electronic structure of the NBCD surface more rigid, increasing the electron transition energy and causing a blue shift40. Based on these spectra, a new peak appeared at ~ 275 nm in the absorption spectra of NBCDs-CUR. Also, the conjugated NBCDs-CUR in Fig. 1B shows a peak at around 430 nm, indicating the presence of CUR in this compound. Thus, the changes in the absorption peaks of the NBCDs nanocarrier before and after loading with CUR can be observed. The UV–Vis spectrum of NBCDs-CUR exhibited absorption bands of both NBCDs and free CUR, indicating their successful combination. In Fig. 1C, PTX in methanol exhibits a strong peak at 225 nm, corresponding to π-π* transitions. Thus, the NBCDs-PTX conjugation exhibited two peaks at 215 nm and 340 nm. With respect to Fig. 1C, the absorption peak of the NBCDs-PTX conjugate shows a 10 nm blue shift (225 → 215 nm) and a decrease in intensity compared to the PTX spectrum. This is likely due to π-π stacking interactions between the π-domains of PTX and NBCDs. According to some studies, the electronegativity of the C atom is higher than that of the B atom. Thus, the doped B atom in B-CDs could lead to a positive charge distribution around B atoms, which facilitates biomolecule adsorption28. Additionally, a strong peak is observed relative to the NBCDs spectrum, accompanied by a 10 nm blue shift (350 → 340 nm). This shift may result from interactions between electron lone pairs in PTX (e.g., the N in the –NH group) and functional groups of NBCDs (e.g., the B in the –B(OH)₂ group), as well as their participation in bond formation. After PTX loading onto the NBCDs and the formation of NBCDs-PTX, the shoulder peak at 230 nm, present in the NBCDs spectrum, disappears. The absorption spectra confirm that NBCDs can be successfully conjugated to both drugs. Also, the photoluminescence (PL) spectra of NBCDs (Fig. 1D) were recorded at different excitation wavelengths with 10 nm increments. The PL spectra of NBCDs showed a maximum fluorescence peak at 440 nm under an excitation wavelength of 360 nm. According to previous studies, the emissive behavior of carbon dots is highly dependent on their surface functionalities41. The fluorescence quantum yield (QY) of NBCDs and NBCDs-drug was measured using the gradient method, as shown in Figures S1A and B. The QY of NBCDs, NBCDs-CUR, and NBCDs-PTX were approximately 60%, 21%, and 57%, respectively. The fluorescence quenching of NBCDs can be explained by the spectral overlap between the absorption spectrum of the drug (CUR or PTX) and the excitation or emission spectra of NBCDs (see Figures S2A and B). The fluorescence quenching observed in NBCDs-CUR can be explained by efficient energy transfer from NBCDs to CUR, in which NBCDs act as energy donors and CUR serves as the acceptor through a Förster resonance energy transfer (FRET) mechanism42. Systematic evaluation of the effects of pH and ionic strength on the fluorescence behavior of NBCDs demonstrated that the maximum fluorescence intensity is achieved at pH 5, while variations in ionic strength between 0 and 2.4 M have no significant effect on the fluorescence intensity27. These characteristics highlight the suitability of NBCDs for biomedical applications, particularly in tumor microenvironments, which are mildly acidic (pH 4–6) and exhibit an ionic strength of approximately 0.3 M in the human body43. Furthermore, the negligible change in fluorescence intensity at pH 5 and high salt concentration (2.4 M) indicates minimal aggregation of NBCDs under these conditions44. Figure S2C illustrates the fluorescence spectra of NBCDs, NBCDs-CUR, and NBCDs-PTX, showing that the fluorescence peak of NBCDs is red-shifted after conjugation with the drug. Fig. 1(A) UV-Vis absorption spectra of the NBCDs (inset: images of NBCDs solution under daylight (left) and ultraviolet light (right) using a 365 nm UV lamp); (B) UV-Vis absorption spectra of NBCDs, CUR, and their conjugate; (C) UV-Vis absorption spectra of NBCDs, PTX, and their conjugate; and (D) Fluorescence spectra of NBCDs (slit widths for excitation/emission: 15/3 nm) at different excitation wavelengths. (A) UV-Vis absorption spectra of the NBCDs (inset: images of NBCDs solution under daylight (left) and ultraviolet light (right) using a 365 nm UV lamp); (B) UV-Vis absorption spectra of NBCDs, CUR, and their conjugate; (C) UV-Vis absorption spectra of NBCDs, PTX, and their conjugate; and (D) Fluorescence spectra of NBCDs (slit widths for excitation/emission: 15/3 nm) at different excitation wavelengths. The FT-IR spectra of the drugs, NBCDs, and drug-loaded NBCDs are presented in Fig. 2A and B. The FT-IR results of NBCDs, CUR, PTX, and their conjugated products are summarized in Table S1 and Table S2. Several characteristic bands of NBCDs are also observed in both drug-loaded carbon dots. The peak at 1747 cm⁻¹ corresponds to the C = O stretching vibration. The IR bands at 1081 cm⁻¹ and 1342 cm⁻¹ are assigned to the C–B and B–O stretching vibration modes, respectively45. The absorption peak at around 1020 cm⁻¹ can be attributed to B–O–C or B–O–H deformation vibrations, or to C–O stretching vibrations45. The IR band at approximately 1237 cm⁻¹ is assigned to the C–N stretching vibration, confirming the successful incorporation of nitrogen atoms into the NBCDs46,47. The band near 1554 cm⁻¹ corresponds to the N–H bending vibration of amide groups. The peaks appearing around 3000 cm⁻¹ are associated with Ar–C–H and N–H stretching vibrations47. In detail, the presence of a broad peak at approximately 3185 cm⁻¹ together with a peak at 798 cm⁻¹ confirms the existence of primary amide groups on the surface of the carbon dots. In addition, the peaks observed at 3185 and 3386 cm⁻¹ are attributed to B–OH / –NH₂ and O–H groups, respectively. The excellent aqueous stability and hydrophilicity of the NBCDs in water can be ascribed to the abundance of polar functional groups on their surface48,49. The FT-IR spectrum of CUR (Fig. 2A) shows absorption bands at 963 cm⁻¹ (benzoate trans-CH vibration / C=C stretching of the benzene ring / in-plane bending of enolic O–H), 1028 cm⁻¹ (C–O stretching), 1274 cm⁻¹ (C–O–C stretching), 1433 and 1585 cm⁻¹ (C = C stretching vibrations), 2932 cm⁻¹ (C–H stretching of methyl groups on the aromatic ring), and 3506 cm⁻¹ (phenolic O–H stretching)50. After the addition of CUR to the NBCDs, the B atoms in the –B–OH functional groups on the surface of the carbon dots can interact with the enol/keto moieties of CUR through coordination to the empty orbital of boron. Consequently, the bending vibration peaks at 1183 cm⁻¹ (B–O–H in the carbon dots) and 1274 cm⁻¹ (enolic stretching in CUR) disappear upon the formation of a six-membered ring structure (Scheme S1). Furthermore, after the formation of NBCDs–CUR (Fig. 2A), the characteristic peaks at 1554 and 1081 cm⁻¹ (assigned to the aromatic moiety of CUR) and at 931 cm⁻¹ (associated with NBCDs) are shifted to 1589, 1087, and 937 cm⁻¹, respectively. The disappearance of the CUR-related bands at 1507, 1433, and 1274 cm⁻¹ following conjugation, together with the presence of the band at 1399 cm⁻¹ (assigned to C = C stretching) in both CUR and NBCDs, further confirms the successful conjugation between the two components. Indeed, after loading CUR onto the NBCDs, the characteristic band at 1183 cm⁻¹ (assigned to the B–O–H vibration in the carbon dots) disappears, indicating the formation of the NBCDs–CUR conjugate and confirming the removal of the B–O–H absorption band upon complex formation. In addition, the broad band at 3386 cm⁻¹ in the NBCDs spectrum is shifted to 3419 cm⁻¹ in the NBCDs–CUR system, further supporting the presence of hydrogen-bonding interactions between CUR and the functional groups on the NBCDs surface. In Fig. 2B, several characteristic bands of PTX are observed at 3390 cm⁻¹ (O–H stretching), 2486 cm⁻¹ (N–H stretching), 2927 cm⁻¹ (C–H stretching), 1736 and 1703 cm⁻¹ (C = O stretching of ketone carbonyls), 1639 cm⁻¹ (amide C = O stretching), 1432 cm⁻¹ (CH₂ scissoring mode), 1081 cm⁻¹ (C–O stretching or aromatic moiety), and 614 cm⁻¹ (out-of-plane bending of aromatic C–H bonds)51,52. After loading PTX onto the NBCDs, the FT-IR spectrum of the NBCDs–PTX conjugate shows characteristic bands at 3442 cm⁻¹, corresponding to the overlapping O–H and N–H stretching vibrations, and a band at 2537 cm⁻¹ attributed to the N–H stretching of the PTX amine group. In addition, several bands are observed at 1641, 1441, 1408, 1102, and 609 cm⁻¹, which are assigned to C = O stretching vibrations originating from both NBCDs and PTX, the CH₂ scissoring mode of PTX, C = C stretching, and an overlapping C–N stretching vibration of the NBCDs with the aromatic vibrational modes of PTX, respectively. These spectral features collectively confirm the successful loading of PTX onto the surface of the NBCDs. XRD analysis was also performed for the NBCDs powder. As shown in Fig. 2C, a broad amorphous diffraction peak appears in the range of 2θ = 15°–27°, which is characteristic of partially graphitic carbon structures and is consistent with previously reported carbon dots. In addition, the presence of a sharp and high-intensity reflection at 2θ ≈ 28°, together with several weak-intensity reflections, may be attributed to nitrogen incorporation and the formation of cubic B₂O₃, as confirmed by comparison with the JCPDS card No. 00-006-0297 in the peak list of Fig. 2C. The inset in Fig. 2C illustrates the formation of B₂O₃ from boric acid under the high-temperature and high-pressure conditions inside the autoclave53. Based on previous reports, carbon dots, due to their distinct optical and chemical properties, (1) exhibit optical behavior compatible with living cells, (2) can be modified with suitable exogenous chemicals, and (3) are biocompatible and non-toxic54. Likewise, the high and stable fluorescence of carbon dots can be a significant advantage in the development of drug delivery systems. To study the photostability of the NBCDs, the fluorescence intensity of the carbon dots was recorded at λmax = 360 nm (as excitation wavelength) at different time intervals. When NBCDs are used as fluorescent nanocarriers, the stability index is an important factor for such applications. According to Figure S3A, it is clearly evident that the fluorescence intensity, or optical stability, of the carbon dots and the drug-loaded carbon dots remains almost constant over different time intervals, showing only a slight decrease after several weeks and a more noticeable decrease after several months. The percentage decreases in fluorescence intensity for NBCDs-CUR were 2.1%, 7.2%, Fig. 2(A) FT-IR spectrum of dried and powdered NBCDs, CUR and NBCDs-CUR (B) FT-IR spectrum of dried and powdered NBCDs, PTX and NBCDs-PTX (both A and B spectrums were prepared using KBr disc method), and (C) XRD pattern of NBCDs powder. (A) FT-IR spectrum of dried and powdered NBCDs, CUR and NBCDs-CUR (B) FT-IR spectrum of dried and powdered NBCDs, PTX and NBCDs-PTX (both A and B spectrums were prepared using KBr disc method), and (C) XRD pattern of NBCDs powder. 17.1%, and 25.1% after one week, one month, three months, and six months, respectively. Similarly, these values for NBCDs-PTX were 2.0%, 3.8%, 11.0%, and 16.9%, respectively. Furthermore, the fluorescence of NBCDs remained stable for 50 min under continuous UV irradiation at 360 nm using a xenon lamp in a luminescence spectrometer, with the emission intensity monitored at 440 nm. This observation indicates that the NBCDs exhibit negligible photobleaching or blinking. Overall, the NBCDs demonstrate excellent fluorescence stability, as evidenced in Figure S3B. The elemental analysis of the synthesized NBCDs was performed by X-ray photoelectron spectroscopy (XPS) technique. According to XPS results (counts per second as signal intensity against binding energy), four predominant survey peaks were associated with elements O 1s (533 eV), N 1s (400 eV), C 1s (285 eV), and B 1s (192 eV) with relative contents of 33.0, 5.1, 47.4%, and 14.5%, respectively (according to Fig. 3A). In the high resolution XPS spectra (Fig. 3B), the O 1s spectrum reveals three peaks at 531, 532.8, and 534.6 eV, which are attributed to the O = C, O-B/O-C, O-C = O, respectively. The N 1s spectrum contains two peaks at 400.5 and 401.7 eV, representing N-H (primary and secondary Amino-N) and N-C/graphitic-N (N-C3, N bonded to three C nearest neighbors), respectively (see Fig. 3C). XPS analysis of the C 1s band can be deconvoluted into three components at 284.7, 286.6, and 288.8 eV, which correspond to the C = C/C-C, C-O/C-B, and C = O/C-N, respectively (Fig. 3D). According to Fig. 3E, the B 1s band can be deconvoluted into three peaks at 193.0, 193.5, and 193.7 eV exhibiting B-C, B-N, and B-O, respectively. These findings indicate that C and O are the dominant constituents, with minor contributions from N, and B. Also, the presence of graphitic-N nitrogen species confirms the incorporation of nitrogen into the carbon core of the NBCDs, evidencing successful N doping36. As a conclusion, the XPS results confirm the successful formation of –COOH, –OH, –NH₂, and –B(OH)₂ functional groups on the surface of semi-spherical NBCDs27, as well as B–O, N–B, and N–C bonding within the NBCDs cores28,38. Zeta potential is an important parameter for evaluating the stability (including photostability) of the produced dispersions. In general, a lower absolute value of the zeta potential indicates a higher tendency for particle agglomeration55. In this work, zeta potential analyses of the NBCDs and their drug conjugates (in water and PBS) were carried out, as shown in Figure S4 (A–F) and Table S3. A detailed analysis of the zeta potential results, including the effect of salt bridges in buffer, is provided in the Supplementary Information. In general, following the adsorption of drugs onto the surface of NBCDs, CUR and PTX molecules can partially mask the surface carboxylate groups and neutralize the overall surface charge, resulting in a positive shift of the zeta potential29. Dynamic light scattering (DLS) analysis was also performed. As illustrated in Figure S5, the particle size distributions of 14.5, 65.4, and 37 nm correspond to NBCDs, NBCDs-CUR, and NBCDs-PTX, respectively. These changes in particle size under identical conditions provide evidence for the successful loading of drugs onto the surface of the carbon dots. Additional confirmation of CUR and PTX presence in the drug-loaded NBCDs was obtained via proton nuclear magnetic resonance (1H-NMR) spectroscopy, along with other techniques. The 1H-NMR spectra of NBCDs (Figure S6A), CUR (Figure S6B), NBCDs-CUR (Figure S6C), NBCDs-PTX (Figure S6D), and PTX (Figure S6E) were recorded in dimethyl sulfoxide-d₆ (DMSO-d₆) using tetramethylsilane (TMS) as the internal standard. The detailed analysis of the 1H-NMR results is provided in the Supplementary Information, and a summary is presented in Table S4. Drug loading onto NBCDs nanocarrier The adsorption of the selected drugs onto NBCDs occurs through interactions such as van der Waals forces, electrostatic attraction, π–π stacking, and hydrophobic interactions56. After deprotonation of the carboxylate groups on the NBCDs at pH ~ 7.4, hydrogen bonding can be established between CUR (as a hydrogen bond donor) and the carbon dots (as hydrogen bond acceptors)29. The conjugation of PTX with NBCDs is achieved through the formation of amide bonds via the reaction between the amino groups of NBCDs and the carboxyl groups of PTX8, which are sensitive to acidic media. Fig. 3(A) XPS survey spectrum of NBCDs; high resolution XPS spectra. (B) O 1s, (C) N 1s, (D) C 1s, and (E) B 1s of the NBCDs. (A) XPS survey spectrum of NBCDs; high resolution XPS spectra. (B) O 1s, (C) N 1s, (D) C 1s, and (E) B 1s of the NBCDs. To study the conjugation of NBCDs with CUR, the maximum ratio between the two components was first determined. Different concentrations of NBCDs (0.002–0.03 mg/mL) were gradually added to a fixed concentration of CUR dissolved in methanol (0.02 mg/mL), and the process of complex formation between CUR and the carbon dots (by the acid-sensitive boronate ester “–O-B-O–” in this complex, and acid-sensitive imine bonding) was monitored by recording the changes in the absorption spectra after each addition. As shown in Figure S7A, the gradual addition of NBCDs induced a progressive decrease in the absorption intensity of the NBCDs–CUR complex at ~ 400–440 nm and a simultaneous increase at ~ 210–350 nm. Based on these results, the maximum [NBCDs]: [CUR] ratio is achieved at 1:1 (w/w). To study the conjugation between NBCDs and PTX (as a hydrophobic anticancer drug model), different concentrations of NBCDs (0.001–0.018 mg/mL) were gradually added to a fixed concentration of PTX dissolved in methanol (0.024 mg/mL), and the complex formation process was monitored using UV-Vis spectroscopy (Figure S7B). The gradual addition of NBCDs induced a progressive decrease in the absorption of the PTX–NBCDs complex at ~ 220–230 nm and a simultaneous increase at ~ 250–450 nm. Based on these results, the maximum [NBCDs]: [PTX] ratio was determined to be 1:3 (w/w). These findings also indicate strong interactions between the graphitic carbon cores or surface functional groups of NBCDs and the aromatic moieties of the drugs (CUR or PTX), leading to changes in their absorption spectra. Likewise, the residual pellet was dissolved in a known amount of methanol, and the free CUR concentration was determined using a standard curve (Figure S8A). Consequently, the percentage of drug loading capacity (LC%) and adsorption efficiency (AE%) of CUR onto NBCDs could be calculated, with a value of 67.7% obtained for both parameters. Since equal masses of the nanocarrier and curcumin were used, and the contribution of encapsulation/entrapment was considered negligible, the measured values of surface adsorption efficiency (AE) and loading capacity (LC) were effectively identical. According to the definition, adsorption efficiency (AE) is the ratio of the amount of drug adsorbed onto NBCDs to the total amount of drug added (Eq. 1), whereas the loading capacity (LC) of the drug is expressed as the amount of drug loaded per unit weight of NBCDs (Eq. 2). For CUR, the drug loading capacity on NBCDs was calculated using a CUR standard curve (Figure S8A). The same procedure was applied to determine the loading capacity of PTX on NBCDs using its calibration curve (Figure S9A) and UV-Vis spectrophotometry at 225 nm. Finally, the LC% and AE% for PTX on the carbon dots were determined to be 28.8% and 86.5%, respectively. pH-dependent release kinetic models Using release kinetic models, important physical parameters, such as the drug diffusion coefficient, can be determined. Understanding how to apply these equations is crucial for evaluating the various factors that influence drug dissolution behavior. According to the literature, drug release from a delivery system can be governed by several mechanisms, including diffusion, dissolution, osmosis, partitioning, swelling, and erosion. For example, the diffusion of an active agent strongly depends on structural characteristics, such as polymer morphology57. As shown in Fig. 4, samples of the release solution were collected at 2, 10, 20, 30, and 45 min, as well as at 1, 2, 3, 5, 10, 17, 25, 38, 45, 57, 68, and 72 h. Experimental conditions, including the volume of the release medium, continuous stirring, and pH adjustment to mimic physiological conditions, were carefully controlled to maintain sink conditions, ensuring that the obtained release profiles accurately reflect the behavior of the nanocarrier. All drug release experiments were conducted at 37 °C. The release profiles of CUR, PTX, NBCDs-CUR, and NBCDs-PTX were investigated in buffers at pH 5.0 and 7.4, simulating the environments of cancer and normal cells, respectively, using the dialysis bag method. Based on the obtained results for CUR release (Fig. 4A), the enhanced drug release under acidic conditions (pH 5.0) may be attributed to the protonation of the boron atom in the formed six-membered ring. Specifically, the six-membered ring between the keto–enol groups of CUR and the boron hydroxyl (–B(OH)₂) groups on the surface of carbon dots (Scheme S1) is opened under acidic conditions58. The increased release of curcumin under acidic conditions is likely due to protonation at lower pH, as curcumin readily interacts with the excess protons in the solution59. The enhanced CUR release at acidic pH is a desirable property for eliciting its therapeutic effect in the tumor environment. The lower release rate of CUR at pH 7.4 compared to pH 5.0 can be attributed to the stronger association between CUR and NBCDs under neutral conditions. The initial CUR release from NBCDs-CUR during the first 10 h was approximately 14% and 10% at pH 5.0 and pH 7.4, respectively. This feature is of practical importance for clinical therapy, as both extracellular (tumor tissues) and intracellular (endosome/lysosome) environments have low pH (≤ 5)60. After 72 h, the total cumulative CUR release from NBCDs-CUR was approximately 54% at pH 5.0 and 40% at pH 7.4. Similarly, PTX was selected as a hydrophobic drug model for conjugation to NBCDs via π–π stacking (hydrophobic interactions), dative bonding between the –NH group of PTX (electron lone pair) and the boron atom of –B(OH)₂ groups on the carbon dot surface (empty orbital), and hydrogen or amide bonding between the –NH group of NBCDs and the –COOH group of PTX, and vice versa, as illustrated in Scheme S1B. Figure 4B shows the cumulative release of PTX from NBCDs at different pH values. The release of PTX, similar to CUR, exhibits pH-dependent kinetics. A nonlinear release profile was observed for both drug-loaded NBCDs under both pH conditions. An initial burst release of PTX from NBCDs-PTX occurred during the first 5 h, with approximately 23% and 25% of PTX released at pH 5.0 and pH 7.4, respectively. After 72 h, the total cumulative PTX release from NBCDs-PTX was approximately 80% at pH 5.0 and 55% at pH 7.4. Drug release sampling was performed at different time intervals up to 72 h, where 2 mL of the release medium was withdrawn and replaced with 2 mL of fresh medium. The absorbance of the collected samples was measured at 225 nm, and PTX concentrations were determined using the calibration curve in PBS + 0.5% (v/v) Tween 80 (Figure S9B). The higher release rate of PTX at acidic pH may be explained by the protonation of amino groups (in both PTX and on the NBCDs surface) as well as boron hydroxyl and carboxylate groups (on the NBCDs), which leads to the disruption of electrostatic interactions between PTX and the carbon dots at pH 5.061,62. In general, the experimental release data were compared with well-known default models provided in the software. The studies of in vitro drug release can be helped us to predict drug release in vivo systems. Paired t-test analysis revealed a statistically significant difference in drug release between pH 5.0 and pH 7.4, indicating a pH-dependent release behavior (p = 0.001 and p = 0.047 for CUR and PTX, respectively, as listed in Tables S5 and S6). According to Table 1, the correlation coefficient (R²) of each model was evaluated to assess the accuracy of the proposed statistical models using the software. The drug release from a carrier depends on different factors such as drug molecular movement (mass transfer), diffusion, dissolution, swelling nature of carrier, erosion, interaction with the guest molecules and degradation (structural decomposition). On the other hand, the drug release kinetic is the combined results of these factors. The Weibull model (as an empirical and generalized form of the exponential function), describes many different types of drug release and has been previously shown to be the most applicable model for drug release from nanoparticles. The Hill model indicates that the drug release kinetics fits better for a combined model of diffusion, surface and bulk erosion. Also, the R2 values for Hixson-Crowell model are insignificant; thus, the bulk erosion is not occurred. The CUR and PTX release of NBCDs at pH 5.0, well fitted with the Higuchi model. The simplified Higuchi model describes release of drug as a square-root time-dependent diffusion process (based on Fick᾽s law). In the present study, for any drug and under any pH condition, the higher R² values (> 0.96) of the Hill, Weibull, Higuchi, and Korsmeyer–Peppas models (with the highest R² values) suggest that the drug release kinetics from the nanocarrier follow the corresponding kinetic model63. For example, Figure S10 shows the proposed release models of CUR from NBCDs at pH 5.0 over time (h), as generated by the software. The experimental data and fitted curves are displayed in these plots. Besides the output results of the software listed in Table 1, the Korsmeyer–Peppas model was plotted for each drug and under each condition (see Fig. 5). This model can not only be used to predict the drug release mechanism from a polymeric system, but also accounts for diffusion and swelling as independent mechanisms of release64. The complicated release behavior can be described by this model. This model is applied for providing additional information (with more details) about release mechanism based on the ‘’n’’ value. This model is obtained by plotting the logarithm of cumulative drug release (%) against the logarithm of time (h). In this model, Mt/M∞, KKP, and n represent the fractional drug release, the Korsmeyer–Peppas rate constant, and the release exponent (the slope of the line in the graphs in Fig. 5), respectively. Notably, to determine the value of n, only data where Mt/M∞ < 0.6 should be used. Fig. 4CUR and PTX release profile from the NBCDs-CUR (A) and NBCDs-PTX (B) under pH 5.0 (citrate buffer) and pH 7.4 (PBS) with three replications (n = 3). CUR and PTX release profile from the NBCDs-CUR (A) and NBCDs-PTX (B) under pH 5.0 (citrate buffer) and pH 7.4 (PBS) with three replications (n = 3). The value of “n” characterizes the drug release mechanism; drug release is controlled by pure diffusion (Fickian model, Case I) if n = 0.5, by pure swelling (non-Fickian model, Case II) if n = 1, and corresponds to zero-order release kinetics when n = 1. When 0.5 < n < 1, the model represents anomalous (non-Fickian) transport. Finally, if n > 1, the release follows the Super Case II model, representing an extreme form of drug transport in which tension and breaking of the polymer occur. If n < 0.5, the release follow the quasi-Fickian diffusion60. The preferred Korsmeyer–Peppas model is the most complex among the selected empirical models, as it integrates different release mechanisms65. The fitting to the Korsmeyer–Peppas model (Fig. 5), demonstrates that drug release follows non-Fickian (anomalous) transport, indicating that both diffusion and structural interactions govern the release process. This confirms that the system is not governed by simple diffusion, but by a complex interplay of interactions, which is a key mechanistic insight. Thus, we concluded that the both “release models” and “release mechanisms” were different in diverse conditions, due to the different coupling ways between carrier and drug. This subject is suggested that drug release from nanocarrier is performed in a “complex manner”. To better understand the interactions of the drugs with carbon nanocarriers, the structures of the drugs (A: CUR and B: PTX) are shown in Figure S11. Based on Figure S11 and Scheme S1, the presence of different functional groups allows the formation of various bonds between the drugs and carbon dots, such as hydrogen bonding, π–π stacking, dative bonding, and imine bonds between the C = O groups of CUR and the –NH2 groups of NBCDs, among others. The drug loading capacity (LC), adsorption efficiency (AE), and encapsulation/entrapment efficiency (EE) of CUR and PTX on the NBCDs nanocarrier were compared with recent reports, as summarized in Tables S7 and S8. The biological effects of the synthesized drug carriers (NBCDs-CUR and NBCDs-PTX) on ovarian cancer cells were also evaluated, and the results were recently published by our group66. Table 1Kinetics models and R2-values for analysis of cumulative drug (CUR or PTX) release from NBCDs nanocarreir (as output data of KinetDS3.0 software).Release Model R 2 CUR (pH 5.0)CUR (pH 7.4)PTX (pH 5.0)PTX (pH 7.4)Hill0.98950.98930.93400.9836Weibull0.98570.98640.93550.9805Higuchi0.96170.95230.97290.3502Korsmeyer-Peppas0.97940.98240.97080.9738Hikson- Crowell0.69900.70430.72440.6692 Table 1Kinetics models and R2-values for analysis of cumulative drug (CUR or PTX) release from NBCDs nanocarreir (as output data of KinetDS3.0 software).Release Model R 2 CUR (pH 5.0)CUR (pH 7.4)PTX (pH 5.0)PTX (pH 7.4)Hill0.98950.98930.93400.9836Weibull0.98570.98640.93550.9805Higuchi0.96170.95230.97290.3502Korsmeyer-Peppas0.97940.98240.97080.9738Hikson- Crowell0.69900.70430.72440.6692 Kinetics models and R2-values for analysis of cumulative drug (CUR or PTX) release from NBCDs nanocarreir (as output data of KinetDS3.0 software). Fig. 5Drug releasing data at two different pHs based on the Korsmeyer-Peppas mathematical model. Drug releasing data at two different pHs based on the Korsmeyer-Peppas mathematical model. Conclusions In summary, highly fluorescent boron- and nitrogen-co-doped carbon dots (NBCDs) were synthesized by a hydrothermal method and used as suitable nanocarriers for curcumin and paclitaxel as hydrophobic drug models. Chemotherapeutic drugs can be conjugated to the carbon dots through covalent bonding via functional groups or other non-covalent interactions, such as π-π stacking, electrostatic interactions, and hydrogen bonding. After attaching the drugs to the carbon dots, the drug loading capacity and adsorption efficiency (67.7% for CUR, 28.8% and 86.5% for PTX) and the drug release profiles (CUR: 54% at pH 5.0 and 40% at pH 7.4; PTX: 80% at pH 5.0 and 55% at pH 7.4) were determined. The higher release efficiency at pH 5.0 indicates that the prepared NBCDs-drug nanoconjugates possess selective drug release under acidic conditions, which is a unique feature that can enhance their anticancer therapeutic potential. Consequently, the prepared nanoconjugates can enter cancer cells and release the drugs in the acidic tumor microenvironment, acting as effective therapeutic agents. Statistical analysis confirmed that drug release was significantly affected by pH, indicating a clear pH-dependent release behavior. These findings open a new channel for real-time studying of NBCDs-drug system in complex biological environments, in future investigations. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1