Investigating adsorption of 5-fluorouracil and 6-mercaptopurine on icosahedral silver nanoparticles through DFT and TDDFT

Abstract

The favourable physicochemical properties and biocompatibility of silver nanoparticles (AgNPs) make them increasingly promising platforms for drug delivery. This study employs density functional theory (DFT) and time-dependent DFT (TDDFT) to investigate the adsorption of two anticancer drugs, 5-fluorouracil (FU) and 6-mercaptopurine (MP), on an icosahedral silver nanoparticle (AgNP). DFT was used to evaluate adsorption stability, electronic interactions, and charge transfer between the drugs and the silver surface. TDDFT simulations helped us understand how the optical absorption changes when the molecules attach. Both drugs exhibited strong and stable binding to the nanoparticle surface, with MP demonstrating a more pronounced interaction in terms of charge transfer and electronic coupling. Charge density difference maps and density of states analyses revealed significant interaction between the drugs and silver cluster. Additionally, the adsorption of the drugs led to noticeable modifications in the nanoparticle’s plasmonic and interband optical features. These findings highlight the potential of AgNPs as effective and tuneable nanocarriers for anticancer drugs and offer valuable guidance for the design of advanced drug delivery systems in nanomedicine.

Introduction

The development of innovative drug delivery systems leveraging nanotechnology continues to transform therapeutic strategies by enhancing drug stability, targeting, and controlled release. Nanocarriers such as liposomes, dendrimers, polymeric nanoparticles, and lipid-based systems have shown significant promise in overcoming biological barriers and delivering therapeutic agents with improved precision and efficacy^1–7. These nanoscale platforms can reduce off-target effects and enable site-specific delivery, particularly for cancer and neurological disorders^2,4,8,9. Recent studies have also explored stimuli-responsive carriers and multifunctional systems for combination therapies^3,9,10. As nanomedicine continues to advance, several nano formulations have entered clinical trials or received regulatory approval, reflecting their growing impact on personalized and minimally invasive therapies^10,11.

Silver nanoparticles (AgNPs) have gained prominence as effective drug delivery agents due to their unique physicochemical properties and multifunctional potential^12,13. Their nanoscale dimensions allow them to penetrate biological barriers and deliver therapeutics with high spatial precision. AgNPs exhibit intrinsic antimicrobial activity^14,15, which not only enhances therapeutic efficacy but also reduces the risk of secondary infections associated with drug delivery. Their high surface area-to-volume ratio enables substantial drug loading and facilitates sustained and controlled release profiles¹⁶. Surface functionalization of AgNPs with various ligands improves biocompatibility and enables site-specific targeting, minimizing systemic toxicity and off-target effects¹⁷. Recent studies have demonstrated their successful application in delivering antibiotics, anticancer agents, and anti-inflammatory drugs, highlighting their versatility across therapeutic areas^18–21. Collectively, these attributes position AgNPs as promising nanocarriers for next-generation drug delivery platforms.

5-Fluorouracil (FU) and 6-mercaptopurine (MP) are widely used anticancer agents that interfere with nucleic acid metabolism by mimicking endogenous DNA and RNA bases^22–25. FU, a pyrimidine analogue, inhibits thymidylate synthase and disrupts DNA synthesis, making it a cornerstone treatment for gastrointestinal cancers, including colorectal and gastric cancers^22,26. In contrast, MP, a purine analogue, impairs nucleotide biosynthesis and DNA replication and is primarily employed in the treatment of acute lymphoblastic leukemia and other hematological malignancies^27,28. Beyond oncology, MP’s immunosuppressive activity also supports its use in managing autoimmune disorders and in post-transplant care^29,30. Despite their efficacy, both drugs present notable side effects that demand careful therapeutic monitoring. Hepatotoxicity and myelosuppression are major concerns for MP^31,32, while FU is associated with gastrointestinal toxicity and off-target effects³³.

Recent research efforts have sought to enhance the delivery of these drugs and mitigate toxicity through advanced formulations, including nanoparticle-based systems^34–41. Computational and experimental studies have demonstrated the potential of nanomaterials such as graphene, silica, and silver nanoparticles to act as efficient carriers for FU and MP, offering improved targeting, stability, and controlled release. Previous studies have demonstrated the utility of computational methods such as Density Functional Theory (DFT) and Molecular Dynamics (MD) in elucidating the interaction mechanisms between silver nanoparticles and various therapeutic agents^42–49. These approaches have provided valuable insights into the electronic structure, adsorption behavior, and thermodynamic stability of drug–AgNP complexes, thereby reinforcing the role of AgNPs as promising platforms in nanomedicine. DFT-based investigations have been successfully employed to model drug loading, surface binding configurations, and charge transfer interactions for a wide range of compounds, including antiviral, antimalarial, and anticancer drugs^44–49. In addition to conventional DFT and MD methods, the Density Functional Tight Binding (DFTB) approach has emerged as a valuable tool for modelling large nanoparticle–biomolecule systems due to its favourable balance between accuracy and computational cost. DFTB has been successfully applied to study silver and gold nanoparticle interactions with biological ligands, peptides, and enzymes, offering insights into adsorption, charge transfer, and catalytic behaviour^50–52. Although not employed in the present study, DFTB represents a promising framework for extending such simulations to even larger drug–nanoparticle or nanoparticle–protein complexes. Such studies highlight the versatility of silver nanocarriers and emphasize the predictive power of computational modelling in guiding the design of effective nanoparticle-based drug delivery systems.

Building on previous computational studies, the present work focuses on two clinically important anticancer drugs, FU and MP, whose interactions with silver nanoparticles remain underexplored at the electronic and optical levels. While AgNPs have been studied in various drug-delivery contexts, a detailed understanding of their behavior with these specific chemotherapeutics is lacking.

Here, we apply an integrated DFT and TDDFT approach to investigate the adsorption, charge transfer, and optical response of FU and MP on icosahedral Ag₅₅ nanoclusters. DFT reveals the binding energetics and electronic structure changes, while TDDFT captures the impact of drug loading on plasmonic absorption. Beyond photothermal therapy, such plasmonic modulation is also relevant for surface-enhanced Raman scattering (SERS), as supported by recent theoretical studies using DFT-based methods to examine plasmon–molecule coupling and charge-transfer effects^53–56. These findings underscore the importance of understanding how drug adsorption alters nanoparticle optical behavior.

This combined approach provides atomic-level insight into the physicochemical interactions governing drug–AgNP complexation. The results contribute to the rational design of silver-based nanocarriers with improved stability, tuneable optical properties, and targeted delivery potential, supporting their broader application in precision nanomedicine for cancer therapy.

Computational methods

Initially, the molecular structures of the drug compounds were optimized in the gas phase using the B3LYP/6–311 + + G(d, p) method. At the same theoretical level, their electrostatic potential maps were generated using Gaussian software⁵⁷. The frontier molecular orbitals, namely the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), are essential in defining the chemical reactivity and molecular interactions of these compounds. The energy values of HOMO and LUMO were determined for the optimized structures of FU and MP molecules, enabling the calculation of their quantum molecular descriptors.

The adsorption of the drug molecule onto the silver nanoparticle was simulated using the Quantum ESPRESSO Package⁵⁸. The exchange-correlation (xc) functional was treated using the Generalized Gradient Approximation (GGA) in the form of PBE⁵⁹, with the D3-Grimme dispersion correction⁶⁰ included to account for van der Waals interactions. Ultra-soft pseudopotentials were employed to describe electron–ion interactions, incorporating scalar relativistic and non-linear core corrections. The wavefunctions and charge density were expanded in plane waves, with energy cutoffs set to 70 Ry and 650 Ry, respectively, after verifying convergence. A Gaussian smearing method with a smearing width of 0.01 Ry was employed. Electronic structure calculations were conducted at the Γ-point of the Brillouin Zone (BZ). All calculations were performed in a large cubic simulation box (30 × 30 × 30 ų) to eliminate interactions with periodic images and effectively treat the system as isolated.

The initial structure of an icosahedral silver cluster containing 55 atoms was constructed based on the experimentally determined interatomic metal–metal distance. This silver nanoparticle structure was optimized separately at the same DFT computational level prior to the drug adsorption process. During geometry optimizations involving drug adsorption, the positions of the silver atoms were fixed to model a stable and rigid nanoparticle, reflecting realistic experimental conditions where the internal structure of metallic nanoparticles typically remains intact. This approach also significantly reduces computational cost while preserving the accuracy of drug–surface interaction modelling. The drug molecules were fully relaxed, and geometry convergence was achieved with thresholds of 10⁻⁶ eV for total energy and 10⁻³ eV/Å for interatomic forces.

To achieve a closed-shell, non-magnetic ground state and eliminate spin polarization, the Ag₅₅ cluster was modeled with a + 5 charge, yielding 50 valence electrons consistent with the superatom shell closure (1 S²1P⁶1D¹⁰2 S²1 F¹⁴2P⁶2D¹⁰). This configuration ensures full HOMO occupation, avoiding spin artifacts and fractional occupations in the electronic structure. While such a charged cluster may not occur freely under physiological conditions, it effectively mimics the stabilized electronic environment of ligand- or counterion-protected systems. Similar superatomic stability has been discussed in clusters like Au₁₃³⁺, Au₂₅⁻, and Au₁₀₂(SR)₄₄, which exhibit enhanced electronic and optical properties due to shell closure⁶¹.

The adsorption energy (E_ads) was calculated using the formulae:

1

In this equation, E_Ag55−drugs represents the total energy of the Ag₅₅ cluster with n adsorbed drug molecules, E_Ag55 is the total energy of the isolated silver cluster, and E_drug is the total energy of a single drug molecule optimized in the gas phase. The term n corresponds to the number of adsorbed drug molecules, which in this study is three. This equation yields the average adsorption energy per drug molecule, allowing for direct comparison between different adsorption systems. A more negative adsorption energy indicates stronger binding and enhanced thermodynamic stability of the drug–nanoparticle complex. All energy values were obtained from fully relaxed structures using the same computational level to ensure consistency.

The charge density difference () was visualized using the following equation:

2

where is the total charge density of the drug–nanoparticle complex, and and are the charge densities of the isolated Ag₅₅ cluster and drug molecules, respectively, each calculated using the same atomic positions as in the combined system. These plots provide a qualitative visualization of electron redistribution upon adsorption, highlighting regions of charge accumulation and depletion.

Charge transfer between the drug molecules and the Ag₅₅ nanoparticle was quantified using both Mulliken and Bader charge analysis. Mulliken charges were extracted directly from the self-consistent DFT calculations using Quantum ESPRESSO’s post-processing utilities, and the charge transfer was calculated as:

3

where and are the total charge of the drug molecules (or nanoparticle) after and before the adsorption process, respectively. A positive value for nanoparticle indicates the electron withdrawal from drug molecules after adsorption.

To provide a more physically meaningful estimate of charge transfer, Bader charge analysis was also performed. The total charge density was computed on a high-resolution real-space grid, and the Bader volumes were determined using the grid-based algorithm developed by the Henkelman group⁶². Unlike Mulliken analysis, which is basis-set dependent, Bader analysis partitions electron density based on zero-flux surfaces, offering a more rigorous and reliable estimate, particularly for systems with delocalized electron density such as metallic clusters. Both methods were applied to fully relaxed geometries of the drug–nanoparticle complexes to ensure consistency and reliability of the results.

To investigate the impact of drug molecules on the electronic structures and plasmonic absorption spectra of noble metal nanoparticles, the turboTDDFT code⁶³, available within the Quantum ESPRESSO (QE) package, was utilized. This code employs the Liouville-Lanczos method within time-dependent density-functional theory (TDDFT), providing a computationally efficient approach for calculating absorption spectra in the frequency domain. It is particularly well suited for large systems containing hundreds of atoms. For each polarization of the external electric field, 10,000 Lanczos iterations were conducted, and the results were averaged over the three spatial coordinates to generate the absorption spectra.

Results and discussions

Structural optimization of FU, and MP drug

The optimized molecular structures and electronic properties of FU and MP were analyzed using DFT calculations, presented in Fig. 1. The molecular electrostatic potential (ESP) maps, frontier molecular orbitals (HOMO-LUMO), and quantum molecular descriptors provide critical insights into their chemical reactivity, stability, and potential interactions with other molecules or surfaces.

Fig. 1.

The molecular electrostatic potential (MEP) map, HOMO and LUMO orbitals for the (a) FU, (b) MP molecules calculated using the DFT at the B3LYP/6–311 + + g(d, p) level of theory.

The ESP maps display the distribution of electrostatic potential across the surfaces of FU and MP, highlighting regions of varying electron density. Blue areas indicate electron-deficient (electrophilic) regions, while red areas represent electron-rich (nucleophilic) regions. In the case of FU, the most negative (red) potential is observed around the carbonyl oxygen and fluorine atom, showing that these sites are highly electron-rich. This suggests they can attract electron-deficient species and are likely to be involved in nucleophilic attacks, consistent with the strong electron-withdrawing character of these functional groups. In contrast, MP exhibits its most negative ESP regions around the sulfur and nitrogen atoms in the purine ring, suggesting that these functional groups play a key role in intermolecular interactions. The distinct ESP patterns imply that FU may have stronger hydrogen bonding capabilities, while MP’s electron-rich sites could facilitate stronger charge transfer interactions with metal surfaces.

The highest occupied molecular orbital (HOMO) represents the electron-donating ability of a molecule, while the lowest unoccupied molecular orbital (LUMO) corresponds to its electron-accepting capability. The energy gap (E_gap) between these orbitals is a fundamental descriptor of chemical reactivity, where a smaller gap indicates higher reactivity and greater charge transfer potential. The calculated HOMO and LUMO energy values for FU are − 6.28 eV and − 2.62 eV, respectively resulting in an energy gap of 3.65 eV. Similarly, for MP, the HOMO and LUMO values are − 4.87 eV and − 2.83 eV, yielding a smaller energy gap of 2.04 eV. These values are summarized in Table 1. This significantly lower energy gap for MP suggests that it is more chemically reactive than FU, making it more likely to participate in electron transfer or adsorption interactions.

Table 1.

The HOMO and LUMO energy, energy gap (eV), chemical hardness (), chemical softness (s), electronegativity (), and electrophilicity (), (in eV) of FU and MP drugs in gas phase calculated at the B3LYP/6–311 + + g(d, p) level of theory.

Quantum chemical descriptors		FU	MP
	HOMO energy: Indicates the electron-donating ability of a molecule.	−6.28	−4.87
	LUMO energy: Reflects the electron-accepting ability of a molecule.	−2.62	−2.83
	Band gap energy: a smaller gap suggests higher chemical reactivity.	3.65	2.04
	Chemical hardness: Measures resistance to charge transfer	1.82	1.02
	Electronegativity: The tendency of a molecule to attract electrons	4.45	3.85
	Electrophilicity: Indicates the stabilization energy when the molecule gains additional electronic charge	5.42	7.26

Visual inspection of the frontier molecular orbitals reveals distinct spatial distributions that further elucidate the reactivity of these molecules. In FU, the HOMO is primarily localized around the carbonyl oxygen and fluorine-substituted ring positions, indicating that these sites serve as the main electron-donating regions. The LUMO is concentrated over the carbonyl oxygen and nitrogen atoms, suggesting that these are the preferred electron-accepting sites. This distribution highlights localized charge transfer pathways, favoring interactions at these specific functional groups. In MP, the HOMO is mainly concentrated on the sulfur and nitrogen atoms within the purine ring, indicating a greater capacity for electron donation. The LUMO is more delocalized across the purine framework, particularly around nitrogen atoms, enhancing MP’s ability to accept electrons from external donors. This broader LUMO distribution facilitates stronger and more flexible charge transfer interactions, making MP more chemically reactive than FU.

Additional quantum descriptors provide further insights into the electronic characteristics and reactivity of FU and MP. The electronegativity (χ), which represents the molecule’s tendency to attract electrons, is 4.45 eV for FU and 3.85 eV for MP, suggesting that FU is more electron-withdrawing and exhibits a stronger affinity for electron-dense environments. The chemical hardness (η), which defines the resistance to charge transfer, is 1.82 eV for FU and 1.02 eV for MP, confirming that FU is less reactive and more stable, while MP is chemically more flexible. The electrophilicity index (ω), which indicates the stabilization energy when a molecule gains additional charge, is significantly higher for MP (7.26 eV) compared to FU (5.42 eV), demonstrating that MP is more prone to electrophilic interactions, enhancing its adsorption and charge-transfer potential. These quantum reactivity descriptors are also summarized in Table 1.

The combined results from ESP, HOMO-LUMO analysis, and quantum descriptors confirm that MP exhibits greater chemical reactivity than FU. The smaller energy gap, lower hardness, and higher electrophilicity make MP more susceptible to charge transfer interactions with metal surfaces or biological environments. On the other hand, FU’s higher electronegativity and localized HOMO-LUMO distribution suggest that it may participate in more selective and stable adsorption interactions, making it potentially useful in controlled drug-delivery or surface-functionalization applications. These findings provide valuable insights into the electronic and chemical behavior of FU and MP, guiding their potential applications in adsorption studies, drug design, and nanomaterial interactions.

Adsorption of FU and MP on Ag₅₅ NP

Optimized adsorption geometries

Three molecules of each drug (FU and MP) were adsorbed onto an icosahedral Ag₅₅ cluster. Multiple initial orientations were tested by positioning the drug molecules near the nanoparticle with their key reactive sites—such as the carbonyl and fluorine groups in FU, and sulfur and nitrogen atoms in MP—facing the surface. Each configuration was subsequently optimized, and the most stable geometries, based on total energy minimization, were selected for analysis. This approach ensured a representative sampling of relevant interaction modes.

During geometry optimization, the positions of the Ag₅₅ atoms were fixed to model a rigid and stable metallic core, reflecting realistic experimental conditions where the internal structure of silver nanoparticles typically remains unchanged upon molecular adsorption. This approach also significantly reduces computational cost while preserving the accuracy of drug–surface interaction analysis. The drug molecules were fully relaxed, allowing them to reach the force convergence threshold on all atoms and adopt their most favourable binding geometries. The optimized adsorption structures and corresponding Ag–O, Ag–F, Ag–N, and Ag–S bond lengths are shown in Fig. 2.

Fig. 2.

Optimized adsorption geometries of (a) FU and (b) MP molecules on the Ag₅₅ nanoparticle. Key Ag–O, Ag–N, Ag–S, and Ag–F bond lengths are indicated, highlighting the primary adsorption sites and interaction distances.

All three FU molecules retained their planar geometry without noticeable bending upon adsorption, and each adopted a perpendicular orientation relative to the Ag₅₅ nanoparticle surface. The adsorption occurred via one of the two carbonyl oxygen atoms in each molecule, forming Ag–O bonds with lengths of 2.30 Å (O4), 2.28 Å (O1), and 2.25 Å (O5). Interestingly, two of the FU molecules adjusted their positions such that the fluorine atoms (F1 and F3) became oriented closer to the Ag surface, resulting in weak Ag–F contacts with distances of 3.02 and 3.07 Å, respectively. Although these are longer than typical covalent bonds, they fall within the range of van der Waals interactions. This suggests that dispersion forces, captured via the D3-Grimme correction, may contribute to stabilizing the adsorption configuration by favoring proximity between the electronegative fluorine atoms and the metallic surface.

Similarly, all three MP molecules maintained their rigid, planar conformation upon adsorption and adopted nearly perpendicular orientations with respect to the Ag₅₅ cluster. The first MP molecule interacted via a single nitrogen atom (N3), forming a direct Ag–N bond of 2.35 Å. The second MP molecule exhibited dual-site coordination, interacting through both sulfur (S3) and nitrogen (N12) atoms, with Ag–S and Ag–N bond lengths of 2.70 Å, 3.05 Å, and 2.34 Å, respectively. The third MP molecule also engaged in dual-site coordination through sulfur (S2) and nitrogen (N8), with corresponding Ag–S and Ag–N distances of 2.57 and 2.37 Å. These configurations highlight MP’s capability to establish multidentate interactions through its purine ring, utilizing multiple heteroatoms for surface anchoring. The presence of both Ag–N and Ag–S coordination supports the observed stronger binding affinity of MP compared to FU, as corroborated by its higher adsorption energy and charge transfer discussed in later sections.

Adsorption energies and binding strength

The computed adsorption energies reveal strong binding between both drug molecules and the Ag₅₅ cluster, with values of − 2.04 eV per molecule for FU and − 3.30 eV per molecule for MP (Table 2). These results indicate that both drugs form stable complexes with the nanoparticle surface, reinforcing their potential utility in nanoparticle-based drug delivery systems.

Table 2.

Interaction energy (E_ads), estimated residence time , mulliken charge transfer to nanoparticle (Δq), and bader charge transfer from FU and MP adsorbed on the Ag₅₅ surface.

Complexes	Eads/mol (eV)		/molecule
Ag55 − 3 FU	−2.04	8.76 × 1021	0.34 e	0.50 e
Ag55 − 3 MP	−3.30	1.98 × 1043	0.74 e	1.41 e

The more negative adsorption energy of MP reflects a stronger binding affinity compared to FU. This is consistent with the optimized geometries discussed in Sect. “Optimized adsorption geometries”, where MP interacts with the Ag₅₅ cluster via multiple donor atoms—nitrogen and sulfur—allowing for bidentate coordination in two of the three adsorbed molecules. In contrast, FU interacts primarily through a single carbonyl oxygen, with additional weak Ag–F interactions in two cases. These energy differences suggest that MP may exhibit greater surface retention and slower desorption, making it more suitable for applications requiring sustained drug release. Meanwhile, FU’s moderate binding energy implies a more reversible interaction, which could be advantageous in scenarios demanding controlled or rapid release profiles.

To further assess the implications of adsorption strength, transition state theory (TST) was applied to estimate the desorption rate constant and the corresponding residence time. In this model, desorption is treated as a thermally activated process in which the adsorbed molecule must overcome an energy barrier—approximated here by the adsorption energy —to escape from the surface⁶⁴. The desorption rate constant is given by the Arrhenius-type expression:

4

where is the attempt frequency, is the Boltzmann constant, and is the absolute temperature. The attempt frequency was estimated from the lowest real vibrational mode using the relation , where is the vibrational frequency in cm⁻¹ and c is the speed of light. The residence time , representing the average time a molecule remains adsorbed before desorbing, is the inverse of the desorption rate constant:

5

While classical TST requires knowledge of the transition state energy barrier, in surface desorption studies a simplified approach is commonly adopted in which the adsorption energy serves as an effective barrier. This approximation assumes that desorption is primarily governed by the molecule–surface interaction strength, with entropy and surface dynamics treated as secondary factors. Although this model does not capture the full potential energy surface, it provides a widely accepted and practical framework for estimating surface retention in adsorbate–substrate systems.

In this study, the attempt frequencies () for FU and MP were calculated from the lowest real vibrational frequencies obtained from Gaussian frequency calculations. For FU, the lowest vibrational frequency was 120.49 cm⁻¹, corresponding to an attempt frequency of approximately 3.61 × 10¹² s^− 1, while for MP, the lowest frequency was 108.83 cm⁻¹, yielding an attempt frequency of about 3.26 × 10¹² s^− 1. Using these values and the adsorption energies in Eq. 4, the desorption rate constants were calculated as 1.14 × 10^− 22 s^− 1 for FU and 5.06 × 10^− 22 s^− 1 for MP at room temperature (298 K). These correspond to residence times of 8.76 × 10²¹ s for FU and 1.98 × 10⁴³ s for MP.

These findings confirm that MP exhibits significantly greater surface retention than FU, consistent with its more negative adsorption energy and multidentate coordination behavior. In physical terms, this means MP is more likely to remain immobilized on the nanoparticle under physiological conditions, while FU, with its shorter residence time, may be more suitable for controlled or targeted release applications.

Charge transfer analysis

The charge density difference (CDD) plots shown in Fig. 3 qualitatively illustrate the redistribution of electron density upon adsorption. In both FU and MP complexes, the plots clearly indicate a net electron transfer from the drug molecules to the Ag₅₅ nanoparticle, as evidenced by charge depletion near the drug adsorption sites and corresponding accumulation on the nanoparticle surface.

Fig. 3.

The charge density difference of adsorbed drug molecules on Ag₅₅ nanoparticles. (a) Ag-FU complex, (b) Ag-MP complex The isovalue for the charge transfer plot is fixed at 0.0005 e/a.u³. Yellow and blue colors indicate negative and positive levels correspond to the gain and loss of electron density.

To quantify this effect, Mulliken and Bader charge analyses for Ag nanoparticle were performed. While Mulliken analysis is computationally convenient, it is known to be highly basis-set dependent and therefore used here only for qualitative comparison. In contrast, Bader analysis partitions electron density using zero-flux surfaces and provides a more rigorous, basis-independent evaluation—particularly suitable for systems with delocalized electrons such as metallic clusters.

As reported in Table 2, Mulliken analysis revealed an average charge transfer of 0.34 electrons per FU and 0.74 electrons per MP molecule the nanoparticle, while Bader analysis showed a net electron gain of 0.50 e and 1.41 e by the Ag₅₅ cluster in the FU and MP systems, respectively.

These values confirm a consistent trend of charge donation from the adsorbates to the nanoparticle. When considered alongside the computed adsorption energies (–2.04 eV for FU and − 3.30 eV for MP) and the observed bond lengths (typically 2.2–2.7 Å for Ag–O, Ag–N, and Ag–S), the nature of the interaction can be characterized as weak chemisorption. This indicates that the bonding is primarily driven by localized donor–acceptor interactions with partial covalent character, rather than purely electrostatic physisorption or strong covalent bonding. Such interactions are typical for metal–molecule systems involving soft metals like silver and functional groups with lone pair donors, as seen in the FU and MP.

Electronic structure analysis (DOS and PDOS)

Figure 4 presents the total density of states (DOS/PDOS) for the isolated systems and the Ag₅₅–drug complexes relative to their fermi energies. In the isolated state (Fig. 4a), the Ag₅₅ cluster—modelled with a + 5 charge to achieve superatomic shell closure—exhibits a pseudo-metallic character. While still metallic in nature, the + 5 charge reduces the density of states at the Fermi level due to the filled superatomic orbitals, resulting in a more discrete, shell-like electronic structure. In contrast, the DOS of the FU and MP molecules reveals large HOMO–LUMO gaps of approximately 3.5 eV and 2.0 eV, respectively, in excellent agreement with the molecular orbital analysis discussed in Sect. “Optimized adsorption geometries”.

Fig. 4.

UV-Vis absorption spectra of Ag₅₅before and after adsorption of FU and MP, computed using TDDFT (turboTDDFT code in Quantum ESPRESSO).

Upon adsorption of FU onto the Ag₅₅ nanoparticle (Fig. 4b), the projected density of states (PDOS) shows that the electronic states near the Fermi level remain largely dominated by the silver cluster. FU contributes molecular states primarily located away from E_F, with no significant features appearing near the Fermi level. This indicates that the adsorption of FU does not markedly alter the electronic structure of the Ag₅₅ cluster. The interaction is thus best described as weak, with limited orbital overlap and predominantly governed by charge redistribution. This is supported by Mulliken and Bader charge analyses, which show moderate electron transfer from the drug molecules to the nanoparticle surface.

A similar trend is observed for MP adsorption (Fig. 4c), where the PDOS again indicates that Ag-derived states dominate the electronic structure around the Fermi level. The contribution from MP is minimal in this region, with its orbitals primarily influencing states well below and above E_F. Although slight overlap between MP and Ag states is evident, it is insufficient to suggest strong hybridization. The Mulliken analysis reveals a larger charge transfer of approximately 0.74 electrons per MP molecule, indicating more substantial electrostatic interaction compared to FU. However, the absence of new features at the Fermi level and the minimal mixing of states confirm that the interaction mechanism remains dominated by charge transfer rather than covalent bonding.

Together, the PDOS results for both complexes support the conclusion that the drug–nanoparticle interactions occur through weak chemisorption, characterized by electron donation from the drug to the metal surface with limited electronic coupling.

Optical properties

The optical absorption spectra of the icosahedral Ag₅₅ nanoparticle, both pristine and after adsorption of FU and MP, were calculated using linear-response time-dependent density functional theory (TDDFT) as implemented in the turboTDDFT module of the Quantum ESPRESSO package (Fig. 5). The spectra were obtained using the Liouville-Lanczos approach, and the results were averaged over three orthogonal polarizations of the external electric field to capture the isotropic optical response.

The bare Ag₅₅ cluster displays two prominent absorption features: a peak centered around 2.5 eV (~ 496 nm) and another near 4.5 eV (~ 276 nm). The low-energy peak at 2.5 eV is attributed to a plasmon-like excitation, corresponding to the localized surface plasmon resonance (LSPR) of the cluster. However, due to the use of the PBE functional, which tends to place the Ag 4 d states too close to the Fermi level, the screening by d-electrons is overestimated. As a result, the LSPR energy is redshifted relative to experimental values. Previous theoretical and experimental studies reported LSPR peaks for Ag₅₅ near 3.6–3.8 eV⁶⁵ whereas our calculated value of ~ 2.5 eV reflects this known underestimation by GGA-level functionals.

After adsorption of FU and MP, this first peak at ~ 2.5 eV vanishes completely, indicating strong plasmon damping. The disappearance of the LSPR feature suggests that the adsorption of drug molecules disrupts the coherence of collective oscillations of conduction electrons, likely due to significant charge transfer and electronic coupling at the interface. The effect is more pronounced for MP, consistent with its higher adsorption energy and stronger electron.

The second peak, initially located near 4.5 eV, undergoes a redshift and broadening upon adsorption of both drugs, shifting to around 4.0 eV. This shift indicates modification of interband transitions, likely due to the influence of molecular orbitals and charge redistribution following adsorption. The broader and more redshifted response observed for MP (Red curve) reflects its stronger interaction with the nanoparticle, as supported by its larger Mulliken and Bader charge transfers. These spectral changes point to enhanced plasmon damping and the introduction of additional excited-state pathways, although it remains difficult to unambiguously distinguish collective excitations from individual electron transitions within the LR-TDDFT framework⁶⁵.

A third absorption feature appears above 6 eV (~ 207 nm) and is visible in both the bare and drug-adsorbed systems. This high-energy peak originates from deep interband transitions involving electrons excited from lower valence states to higher unoccupied states. Unlike the plasmonic features, this region is less sensitive to adsorption, although some broadening is observed—particularly for the MP–Ag₅₅ system—suggesting subtle perturbations to the deeper electronic structure.

In summary, the optical response confirms that MP induces stronger modifications in the electronic and plasmonic properties of Ag₅₅ than FU. This is evident from the more significant redshift, damping, and spectral broadening across the low- and mid-energy regions. The observed trends correlate well with the adsorption energies and charge transfer analyses, reinforcing MP’s stronger interaction with the nanoparticle surface.

These spectral modifications may have important implications for biomedical applications, particularly plasmonic photothermal therapy (PPTT). The observed redshift and broadening of the absorption peak upon drug adsorption suggest a perturbation of the nanoparticle’s plasmonic response due to strong electronic coupling and charge transfer. While the plasmon features in our simulations lie in the UV region, similar red shifting effects in larger or ligand-stabilized AgNPs could shift the absorption toward the near-infrared (NIR) region, enhancing tissue penetration. Moreover, the peak broadening increases the spectral bandwidth for light absorption, while the damping of the 2.5 eV feature reflects enhanced non-radiative decay pathways—both of which are central to efficient photothermal heat generation. These findings suggest that drug-induced modulation of plasmonic properties could be exploited in designing photo-responsive nanocarriers for combined therapeutic and diagnostic (theranostic) applications.

Fig. 5.

UV-Vis absorption spectra of Ag₅₅before and after adsorption of FU and MP, computed using TDDFT (turboTDDFT code in Quantum ESPRESSO).

Summary

This study employed a combined DFT and TDDFT approach to investigate the interaction of 5-fluorouracil (FU) and 6-mercaptopurine (MP) with icosahedral Ag₅₅ nanoparticles, focusing on their structural, electronic, and optical behavior. Both drug molecules exhibited stable adsorption on the silver surface, with MP demonstrating stronger binding and more pronounced charge transfer, consistent with its multidentate coordination.

Electronic structure analysis revealed limited orbital hybridization, indicating that the adsorption is primarily governed by charge redistribution rather than strong covalent coupling. Optical spectra showed significant plasmon damping upon drug adsorption, with redshifted and broadened features particularly evident in the MP–Ag₅₅ complex, confirming its stronger perturbative effect on the nanoparticle’s optical response.

These findings underscore the potential of Ag₅₅ clusters as effective nanocarriers for anticancer drugs, particularly MP, which exhibits characteristics favourable for sustained or targeted delivery. The insights gained here provide a useful framework for designing plasmonically active and chemically stable nanoparticle–drug conjugates for biomedical applications.

Author contributions

R. M performed the simulations, analyzed the results, and wrote the manuscript.

Funding

This research received no external funding.

Data availability

The initial and optimized structure files (xyz format) used and/or analyzed during this study are available in the public repository at [https://github.com/RaziehMrd/Drug\_Delivery\_DFT](https:/github.com/RaziehMrd/Drug_Delivery_DFT).

Declarations

Competing interests

The authors declare no competing interests.