http://pubs.acs.org/journal/acsodf Article Influence of Capping Ligands, Solvent, and Thermal Effects on CdSe Quantum Dot Optical Properties by DFT Calculations Didac A. Fenoll, Mariona Sodupe, and Xavier Solans-Monfort* Cite This: ACS Omega 2023, 8, 11467−11478 Read Online ACCESS Metrics & More Article Recommendations * sı Supporting Information ABSTRACT: Cadmium selenide nanomaterials are very impor- tant materials in photonics, catalysis, and biomedical applications due to their optical properties that can be tuned through size, shape, and surface passivation. In this report, static and ab initio molecular dynamics density functional theory (DFT) simulations are used to characterize the effect of ligand adsorption on the electronic properties of the (110) surface of zinc blende and wurtzite CdSe and a (CdSe)33 nanoparticle. Adsorption energies depend on ligand surface coverage and result from a balance between chemical affinity and ligand−surface and ligand−ligand dispersive interactions. In addition, while little structural reorganization occurs upon slab formation, Cd···Cd distances become shorter and the Se−Cd−Se angles become smaller in the bare nanoparticle model. This originates mid-gap states that strongly influence the absorption optical spectra of nonpassivated (CdSe)33. Ligand passivation on both zinc blende and wurtzite surfaces does not induce a surface reorganization, and thus, the band gap remains nonaffected with respect to bare surfaces. In contrast, structural reconstruction is more apparent for the nanoparticle, which significantly increases its highest occupied molecular orbital (HOMO)−lowest unoccupied molecular orbital (LUMO) gap upon passivation. Solvent effects decrease the band gap difference between the passivated and nonpassivated nanoparticles, the maximum of the absorption spectra being blue-shifted around 20 nm by the effect of the ligands. Overall, calculations show that flexible surface cadmium sites are responsible for the appearance of mid-gap states that are partially localized on the most reconstructed regions of the nanoparticle that can be controlled through appropriate ligand adsorption. ■ INTRODUCTION Semiconductor low-dimensional materials (e.g., quantum dots, and this influences their electronic structure, which can result in the presence of hole traps and electronic states in the band two-dimensional (2D) single- or few-layered sheets) exhibit gap that ultimately impact the optical properties, often in an different electro-optical properties with respect to their bulk undesired manner. In this context, surface passivation appears counterparts.1,2 This has already been exploited in electronic as a suitable strategy.29 This passivation can be achieved either and photonic devices and biological and catalytic applica- by the effect of capping ligands30−33 or by adding a shell of an tions.3−14 The enhancement of quantum mechanical effects inorganic material exhibiting a larger band gap, thus containing through spatial confinement and the increase in surface− the photoprocess in the core.34−38 volume ratio lie behind these novel properties.2,15 Nanoma- A wide range of ligands has been used for surface terial composition, structure, size, shape, and surface chemistry passivation. Within the large set of ligands already used, the determine its potential applications, which can be further following groups can be distinguished (Scheme 1): (a) tailored through various synthetic methods;16−20 in this regard, phosphines and phosphine oxide ligands, such as trioctylphos- low-dimensional CdSe structures have garnered special phine and trioctylphosphine oxide; (b) amines and amino- attention.21−24 Wurtzite and zinc blende CdSe are the most acids, as is the case of dodecylamine; (c) phosphonic and frequently studied polymorphs. Nanofabrication involving carboxylic acids, including octylphosphonic, oleic, stearic, or 3- either polymorph requires great control of their growth and surface chemistry to adequately tune their chemical and electro-optical properties.21−27 Received: January 16, 2023 Despite the control of the optical properties that can be Accepted: March 7, 2023 achieved through detailed control of the size, shape, and Published: March 17, 2023 structure of the nanoparticle, quantum dot surface chemistry has emerged as an additional way to tune material proper- ties.2,17,28,29 Most of the surface atoms are undercoordinated, © 2023 The Authors. Published by American Chemical Society https://doi.org/10.1021/acsomega.3c00324 11467 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article Scheme 1 wavelength when compared to ligand-free models. In contrast, controversy exists on calculations performed with larger models. That is, while there is a consensus that both anionic and neutral ligands modify the band gap of large non- stoichiometric nanoparticles,60,69 the role of neutral ligands on stoichiometric models is less pronounced.59,61,64,65 In this context, the inclusion of thermal effects on these larger models through ab initio molecular dynamics has been shown to be important for better representing the nanoparticle−ligand interface60,62,68 and particularly identifying the increase of the band gap upon capping ligand adsorption.62 Finally, the presence of ligands influences the localization of the frontier orbitals,29 and it can increase material luminescence through two effects: first, it decreases the rate constant of the nonradiative decay, and second, the presence of electronic states delocalized both on the material surface and the ligand mercaptopropionic acids; and (d) alcohols and thiols, for enables new channels for the radiative decay from the excited which decanethiol is the most relevant example. The nature of state.50 the capping ligand is important, as it can modify the emission In this contribution, we use zinc blende and wurtzite slabs as wavelength and/or luminescence quenching.5,15,17 Under- well as a (CdSe)33 nanoparticle model to analyze ligand standing the capping ligand interaction and how this adsorption of small and large models of the most common interaction influences the adsorption energy and the electronic capping ligands (phosphonic and carboxylic acids, amines, structure of the surface is of crucial importance for the alcohols, and thiols) and its effect on the electronic properties development of new materials with improved optical proper- of the material. For that, we analyze for both polymorphs the ties, and in this context, many experimental19,20,28,30,32,39−42 intrinsic ligand−surface interaction, together with ligand− and computational contributions29,41−74 have been reported in ligand stabilizing dispersive interactions. Moreover, we the literature. describe the importance of thermal effects through ab initio Density functional theory (DFT) calculations have been molecular dynamics (AIMD) simulations. To the best of our carried out to obtain atomistic insights into the ligand− knowledge, none of the previous contributions included at the quantum dot surface interaction and how this surface same time such a variety of ligands, both surface and passivation modifies the electronic structure of the material. nanoparticle models, and the analysis of thermal effects. The influence of surface passivation on the optical properties Results show that ligand adsorption is stronger on zinc blende of the nanomaterial has been examined mainly through three than that on wurtzite and, regardless of the CdSe polymorph, different approaches: (i) analysis of the density of states and/ phosphonic acids and amines interact more strongly with the or highest occupied molecular orbital (HOMO)−lowest surface than carboxylic acids, alcohols, and thiols. The impact unoccupied molecular orbital (LUMO) energy difference to of ligand adsorption on the electronic structure of the material determine the effect of the ligands in the band is determined to be low in rigid systems, such as in crystalline gap;29,44,45,61,64,73 (ii) time-dependent (TD)-DFT calculations, periodic surfaces, yet significant in small nanoparticles. This is which allow determining the maximum adsorption and related to the different surface reconstructions of the slab emission wavelength of the nanoparticle in the presence/ models and the nanoparticle in the presence/absence of absence of a particular ligand;41,48,49,54,55,59,61,64 and (iii) the capping ligands and the appearance/disappearance of mid-gap radiative/nonradiative exciton decay mechanisms and their states on the nanoparticle. impact on, for instance, photoluminescence quenching.50,68 Both slab29,43−46 and nanoparticle mod- els41,42,47−61,63−66,68,69,71−75 have already been used in the ■ COMPUTATIONAL DETAILS All calculations were carried out with the VASP code,77,78 literature. Indeed, periodic slab models have been used in the which performs three-dimensional (3D) periodic calculations characterization of ligand adsorption on selected facets, albeit using plane wave-based basis sets. Therefore, regardless of the at the cost of forgoing finer details present in the structure of dimensionality of the model (3D, 2D, or zero-dimensional nanoparticles.29,43−46 Regarding the modeling with clusters (0D)), calculations include periodicity in the three directions. and nanoparticles, much effort has been centered on For 2D and 0D systems, the lattice parameters used guarantee nanoparticles with up to 30 CdSe units, with Cd13Se13 being a distance between the most external atoms of the images of at one of the most employed,41,47−59 as it has been reported to be least 10 Å in the nonperiodic directions. For the slabs, the c a significantly stable cluster according to mass spectrometry value is 30 or 40 Å depending on the length of the alkyl chain experiments.76 Recently, larger nanoparticle models have been of the ligand. In addition, the (CdSe)33 nanoparticle and the employed by different groups,42,61−64,66,68−75 which allowed associated capped models are included in a 35 Å large cubic representing the actual size of quantum dots in a more realistic box (see below). manner as well as analyzing the role of different ligands. Models. The ligand adsorption on CdSe and its effect on The appearance of trap states on the edges of the band gap the band gap were modeled with both slab and finite region seems to be linked to surface defects with incomplete nanoparticle model approaches (Figure 1). Regarding the passivation.67,69,73 In this context, most of the calculations with slab models, two different surfaces were considered: stoichio- a small nanoparticle model (i.e., (CdSe)13) agree with the fact metric (110) zinc blende (ZB) and stoichiometric (110) that the adsorption of capping ligands modifies the band wurtzite (WZ) surfaces, i.e., the two nonpolar stoichiometric gap,48,49,54−56,65 thus leading to shifts in the emission surfaces presenting lower surface energies. Layer-wise surface 11468 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article Scheme 2 implemented in VASP 5.4. PBE-D2 was used recently in our research group and shown to give cell parameters closer to experiments than PBE-D3.81,82 Spin-polarized calculations were performed for bulk and pristine surfaces to successfully confirm the absence of magnetization. Therefore, all other Figure 1. Optimized structures for the unit cells of (110) wurtzite (WZ) and (110) zinc blende (ZB) and for the (CdSe)33 nanoparticle systems and the results reported hereafter are based on non- model. Cadmium atoms are in gray and selenium atoms are in orange. spin-polarized calculations. The atomic ionic cores were represented with projector augmented wave (PAW) pseudo- potentials83,84 and the valence electrons were represented with energy convergence was used to select slab thickness for both a plane wave basis set with a 500 eV cutoff. For each system, ZB (110) and WZ (110) surfaces (see Figure S1 in the different Monkhorst−Pack85 reciprocal space k-point grids Supporting Information). The role of the CdSe polymorph was were employed depending on cell dimensions to ensure analyzed with (3×3) and (2×2) supercells of the (110) ZB and efficient cost−accuracy simulations (Figures S2 and S3 in the (110) WZ surfaces to ensure the absence of lateral interactions Supporting Information). The nanoparticle model was between ligands. Moreover, the importance of ligand coverage computed at the Γ point due to its nonperiodic nature. For on the adsorption energy and the band gap structure was the zinc blende (110) slab, 2×2×1 and 8×8×1 meshes were analyzed on the smallest (110) ZB and (110) WZ (1×1) employed for ZB(3×3) and ZB(1×1) cells, respectively. For surfaces. Additionally, one nanoparticle model constructed the wurtzite (110) facet, 2×2×1 and 4×4×1 meshes were used from wurtzite and already used by several authors was also for WZ(2×2) and WZ(1×1), respectively, owing to its larger considered.59,61,62,64,66,72−74 The model is obtained by spheri- unit cell parameters. Bader charge analysis86,87 was performed cally cutting an extended wurtzite-like crystalline bulk as to analyze charge transfer between ligands and the CdSe detailed in previous works.61 This model includes 33 CdSe material. units, corresponding to one of the clusters identified by mass Since generalized gradient approximation (GGA) func- spectrometry.76 Moreover, the nanoparticle model has 27 tionals tend to underestimate semiconductor band gaps, two surface and 6 internal Cd centers, thus it presents 27 vacancies, main strategies have been proposed to overcome this each being initially saturated with one capping ligand in the limitation: (a) the use of hybrid functionals and (b) the calculations representing the full coverage of the nanoparticle. inclusion of Hubbard′s correction (U). For the case of CdSe, During structure optimizations, all atoms were allowed to relax the hybrid PBE0 density functional has been shown to be both on the slab and nanoparticle models; cell parameters were appropriate,47,58 and also, other authors applied Hubbard′s kept fixed in all cases. correction with success.88 In this context, we decided to The effect of the ligand structure was analyzed by analyze the optical properties of the different CdSe-based considering four groups of ligands as summarized in Scheme materials using three different levels of theory: PBE, PBE-U, 2: (a) carboxylic acids, (b) phosphine oxides and phosphonic and PBE0. For that, we first calibrated the Ueff value following acids, (c) amines, and (d) alcohols and thiols. In each group, the Dudarev et al. scheme89 by trying to reproduce the band small−large pairs of ligands were considered. In the small gap of CdSe zinc blende bulk. We found that a Ueff of 14 eV at models, the alkyl chain is CH3CH2− (C2), whereas the largest Cd d orbitals was necessary to obtain reasonably accurate band species has CH3−(CH2)9− as an alkyl chain (C10), with few gaps (see Figure S4 in the Supporting Information). We also exceptions. Additionally, in the case of the amines and the analyzed the dependence of the band gap on the Monkhorst− phosphonic acids, the more sizeable triethyl amine and Pack mesh density (Figures S2 and S3 in the Supporting triethylphosphine oxide ligands were included to analyze how Information) and found that 32×32 × 32 and 32× 32 × 1 the ligand′s steric effects may influence ligand coverage on the meshes for the bulk and surface models, respectively, surface. warranted band gap convergence with respect to Brillouin Level of Theory. Geometry optimizations were carried out zone representation. Unfortunately, such dense Monkhorst− at the density functional theory (DFT) level using the Pack meshes are impractical with PBE0, which requires sparser Perdew−Burke−Erzernhof (PBE) exchange−correlation func- grids. Results for ZB and WZ bulk and surfaces as well as the tional79 along with Grimme′s D2 dispersion80 correction as (CdSe)33 nanoparticle using PBE, PBE + U, and PBE0 at the 11469 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article PBE-D2 optimized geometries are reported in Table 1 (and the AIMD simulation at the PBE0 level of theory as Table S1 in the Supporting Information (SI)). implemented in the Gaussian 16 package.94 Atoms were As expected, PBE values are systematically lower than PBE0 represented with def2-SVP basis sets95 and the corresponding and PBE-U values. However, while the difference between PBE pseudopotentials.96 Solvent effects were included with the and PBE0 with the same number of k-points is constant CPCM continuum model97,98 and considering water as the (1.42−1.47 eV) regardless of the material considered, the solvent. The reported absorption spectra are obtained by difference between PBE and PBE-U decreases significantly considering three representative structures of the AIMD with the decrease of material dimensionality. Moreover, simulation: (i) the structure presenting the lowest band gap; experimental data for (CdSe)33 suggest that the band gap of (ii) the structure with the highest band gap; and (iii) the this cluster should be around 3.0 eV,76 something that is not structure with the closest band gap to the weighted average well reproduced with PBE-U. This so large Ueff value also value. In each case, we considered the first 10 excited singlet modifies the orbitals contributing to the valence and states, and the intensity is computed considering both the conducting bands, the contribution from cadmium 4d orbitals oscillator strength and the number of structures of each cluster. to the conduction band becoming not negligible. Therefore, To ensure that our approach is reasonable, Figure S5 compares our data suggest that PBE-U is not a good strategy for the absorption spectrum of the nonpassivated nanoparticle in analyzing the electronic structure of CdSe nanoparticles and vacuum considering the representative structure of all clusters should be avoided in these kinds of systems. So, our best arising from the dynamics with that considering the extrema estimated band gap values (BEV) are obtained by adding 1.45 exclusively, and the results are reasonably similar. eV (mean difference between PBE and PBE0 values with the same number of k-points) to PBE band gaps with the larger number of k-points. In addition, we analyzed the effect of solvent on the band ■ RESULTS AND DISCUSSION Results are organized into four subsections. We first character- gap by including it implicitly with VASPSOL.90 We consider ize the pristine structures of zinc blende and wurtzite CdSe water solvent as the limiting case, but test calculations with bulk, the associated main stoichiometric surfaces, and the toluene and acetonitrile (Table S2 in the Supporting (CdSe)33 nanoparticle model. In the second stage, we analyze Information) show that the effects observed with water (see the intrinsic ligand−material interaction and the corresponding below) can be tuned by modifying solvent polarity. adsorption energy on the slab at low coverages. We considered Thermal effects on the full coverage nanoparticle the ligands reported in Scheme 2 and the largest supercell of [(CdSe)33−L27](L = capping ligand) models were taken into each surface. We paid attention to the role of the functional account by performing NVT ab initio molecular dynamics. A group of the capping ligand, the chain length, and the structure Nosé−Hoover thermostat, set at 300 K, was used.91,92 An of the material. Afterward, we study the adsorption of the optimized structure for each capping ligand was taken as the ligands considering full coverage to determine the cooperative/ initial structure in a 0.5 ps equilibration simulation. Once competitive effects and the ligand−ligand interactions. Finally, thermalization ensued, 1000 steps (1 ps) of AIMD were used we discuss the band gap shifts induced by the presence of a full for clusterization. This clusterization was performed with the layer of ligands on the surface of the slabs and the nanoparticle algorithm provided in UCSF Chimera93 and considering the model. We first present the results from static calculations in Cd center distribution. The most representative structures vacuum, and then we consider the influence of solvation and resulting from clusterization were used to analyze the thermal thermal effects on the band gap. effects on full coverage nanoparticle optical properties. CdSe Bulk, Surfaces, and Nanoparticles. The structural Finally, to further analyze the optical properties of the properties of bulk zinc blende and wurtzite CdSe, the largest (CdSe)33 passivated and nonpassivated nanoparticles, we optimized pristine (110) zinc blende and (110) wurtzite performed TD-DFT calculations on structures obtained during surfaces (ZB(3×3) and WZ(2×2), respectively) and the (CdSe)33 nanoparticle model are examined first. Table 2 Table 1. Band Gap Energies (in eV) for Zinc Blende (ZB), summarizes the cell parameters and the Cd−Se and Cd···Cd Wurtzite (WZ) Bulks, ZB (110) and WZ (110) Pristine distances; Figure 1 shows the optimized structures. Surfaces, and (CdSe)33 Nanoparticle Model Computed with Different Levels of Theory. BEV Stands for Best Estimated Table 2. PBE-D2 Cell Parameters and Cd−Se Distances (in Value Å) for Zinc Blende and Wurtzite Structures and the (110) PBE PBE-U PBE0 BEVg Main Crystallographic Facet of Each Polymorpha a a d ZB 0.66 1.56 2.49 2.08 model cell parameters d(Cd−Se) d(Cd−Cd) WZ 0.78a 1.65a 3.09d 2.24 WZ a = 4.394 (4.299) 2.648 4.322 (110)ZB 1.21b 1.92b 2.99e 2.66 c = 7.171 (7.010) (110)WZ 1.34b 2.01b 3.28f 2.79 ZB a = 6.022 (6.017) 2.647 4.323 (CdSe)33 1.61c 1.76c 3.08c 3.06 b = 6.022 (6.017) a WZ (110) a = 14.095 2.590b/2.660c 3.818 Band gaps obtained with a 32×32×32 k-point mesh. bBand gaps obtained with a 32×32×1 k-point mesh. cBand gap at the Γ point. b = 14.936 d Band gaps obtained with an 8×8×8 k-point mesh. eBand gaps ZB (110) a = 8.623 2.612b/2.657c 3.712 obtained with an 8×8×1 k-point mesh. fBand gaps obtained with a b = 12.195 4×4×1 k-point mesh. gBest estimated value is obtained by adding the Cd33Se33 - 2.677b/2.667c 3.16/3.42 difference between PBE and PBE0 with the number of k-points used a in the optimization to the PBE value with the largest number of k- X-ray-determined values are added in parenthesis. bSurface atoms. c points. BEV stands for best estimated value. Internal atoms. 11470 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article Table 3. Surface Cd−Capping Ligand X Distances (in Å) and Adsorption Energies (in kJ mol−1) on (110) Zinc Blende and (110) Wurtzite Surfacesa low coverage full coverage ZB(3×3) WZ(2×2) ZB(1×1) WZ(1×1)b type ligand d(Cd−X) ΔE d(Cd−X) ΔE d(Cd−X) ΔE d(Cd−X) ΔE R−CO2H C9−CO2H 2.415 −83.2 2.446 −63.2 2.415 −113.5 2.534/2.472 −99.7 C1−CO2H 2.400 −81.1 2.417 −70.1 2.425 −79.2 2.432/2.427 −73.1 R−P(X)2O C8−P(OH)2O 2.287 −111.7 2.309 −97.7 2.301 −106.6 2.574/2.610 −144.4d C2P(OH)2O 2.292 −125.9 2.330 −94.4 2.501 −85.2 2.396/2.524 −129.3 (C2)3−PO 2.297 −130.2 2.305 −123.1 --c --c --c --c R−NX2 C10−NH2 2.353 −112.7 2.380 −99.6 2.427 −141.3 2.397/2.403 −134.7 C2−NH2 2.363 −117.3 2.364 −107.0 2.386 −111.0 2.395/2.389 −105.2 (C2)3−N 2.550 −49.6 4.177 −54.1 --c --c --c --c R−OH H−OH 2.447 −62.7 2.485 −61.3 2.499 −66.2 2.474/2.489 −58.6 C10−OH 2.397 −80.2 2.412 −65.3 2.631 −109.3 2.463/2.477 −114.0 C2−OH 2.405 −83.0 2.428 −74.4 2.486 −81.1 2.464/2.473 −75.7 R−SH C10−SH 2.730 −76.0 2.745 −65.0 2.815 −112.6 2.851/2.790 −100.5 C2−SH 2.722 −73.0 2.769 −66.1 2.813 −85.7 2.811/2-800 −75.3 a Low and full coverage of the ligands are considered. bOnly the most stable configuration is reported. cThe unit cell is too small to accommodate the ligands in the full coverage situation. dThe most stable adsorption mode for C8−P(OH)2O differs from all other situations. PBE-D2 optimizations lead to cell parameters that are very close to the experimentally determined values for ZB and WZ. Indeed, the largest deviation is 0.16 Å for the largest c value of wurtzite. Furthermore, the computed Cd−Se and Cd−Cd distances in ZB and WZ bulk are in very good agreement with experimental values (see Table 2). Overall, this indicates that the methodology used is accurate enough for structure characterization. The Cd−Se of the slab models is similar to that of bulk, with major differences observed only for the surface atoms. This leads to a shorter surface Cd···Cd distance of about 3.7−3.8 Å. In contrast, a large reorganization is observed on the nanoparticle model, particularly at the surface, where several Cd−Cd interactions with much smaller distances and smaller Se−Cd−Se angles are observed (Table 2). Bader charge analysis (Table S3 in the Supporting Information) provides very similar values both for ZB and WZ CdSe bulk. These charges are not homogeneous in the slab models. The outermost Se and Cd atoms present a lower charge separation (−0.66 in ZB (110) and WZ (110)) than the inner centers, whose values are closer to those of the bulk (−0.71 and −0.72 for ZB and WZ bulk structures, respectively). Consistently, Se atoms are Lewis bases and Cd atoms are Lewis acids, and as such ligands preferentially bind to surface Cd sites. Low Surface Coverage Ligand Adsorption. Adsorption of capping ligands occurs mainly by the interaction of the basic heteroatom of the ligand with Cd surface atoms (perpendicular adsorption). At low coverages, an alternative adsorption mode Figure 2. Low-coverage ligands adsorbed on WZ22 (left) and ZB33 in which the capping ligand alkyl chain interacts with the (right), including (a) acetic acid, (b) ethylphosphonic acid, (c) surface through dispersion interactions is also plausible trimethylphosphine oxide, (d) ethylamine, (e) ethanol, and (f) ethanethiol. (parallel mode). First, we evaluated the intrinsic interaction between the ligand and the surface; that is, we considered large zinc blende (3 ×3) and wurtzite (2×2) surface models to avoid The adsorption through the perpendicular coordination lateral ligand interactions. We considered the different mode takes place in a very similar manner regardless of the adsorption modes (perpendicular and parallel) for all ligands ligand type. The ligand adsorbs through the coordination of shown in Scheme 2. Table 3 summarizes the Cd−capping the X atom to Cd, the Cd···X distances varying between 2.287 ligand distance and the adsorption energies on ZB(3×3) and and 2.769 Å. According to Bader charge analysis (Table S4 in WZ(2×2) through the perpendicular mode. Figure 2 shows the Supporting Information), the interaction is electrostatic the optimized structure for the perpendicular adsorption of the with negligible charge transfer. However, the alkyl chain is in smaller ligand of each type. All other structures can be found in all cases slightly tilted due to dispersive interactions between the Supporting Information (Figures S6−S12). the alkyl chain and the surface. A comparison between the 11471 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article different types of ligands shows that for the two surfaces, the the smallest (110) WZ (1×1) and (110) ZB(1×1) cells, which shortest Cd···X distances correspond to the phosphonic acid enclose two and one cadmium surface cation, respectively. and phosphine oxide type species, whereas the largest values Table 3 reports the adsorption energies of the most favorable are obtained for thiols. Furthermore, R−SH and R−OH conformation for all ligands shown in Scheme 2. Figure 3 ligands can additionally form hydrogen bond contacts with the shows the optimized structures for all considered conforma- surface selenium atoms. However, the H···Se distance varies tions for the decanoic acid, decanol, and decanethiol. Note that from 2.297 to 2.992 Å, showing that hydrogen bonding for the case of the (110) WZ (1×1) surface, we explored between the ligands and the surface is weak. Moreover, we different ligand adsorption modes to determine how the explored the possibility of proton transfer between these interaction between two ligands that are not related by ligands and the surface, but most attempts evolved toward the translational symmetry may affect the adsorption on the molecular form and when we succeed to obtain the dissociated surface. form, this structure lies very high in energy. The parallel The ligand adsorption on the (110) ZB (1×1) surface takes adsorption mode presents systematically larger Cd···X place through the perpendicular mode similar to the distances (by around 0.02 Å). The alkyl chain and the surface distances are around 3−4 Å, indicative of an alkyl chain− surface dispersive interaction. Overall, the local Cd···X interaction is weakened to reinforce the alkyl chain−surface dispersion. The adsorption energies through the perpendicular mode range between −130.2 and −61.3 kJ mol−1, with the exception of (C2)3N that presents lower adsorption energies (in absolute value) as a consequence of the absence of a direct N···Cd interaction. Comparison between the computed values allows establishing the three following trends. First, the adsorption energy of the small and large analogous ligands differs by less than 13.0 kJ mol−1, with most cases being between 1.3 and 6.6 kJ mol−1. This suggests that lateral interactions are negligible and thus no dispersion interactions between alkyl chains exist in these large supercells. Second, the strongest adsorption energies are computed for phosphonic acid, phosphine oxide, and primary amine ligands, and the weakest are associated with carboxylic acids, alcohols, and thiols. The specific ordering for the lighter models is (C2)3PO > (C2)P(OH)2O ∼ (C2)NH2 > (C1)COOH ∼ (C2)OH ∼ (C2)SH, in agreement with the proton affinity trends of the different ligands that can also be classified in these two major groups (Table S5 in the Supporting Information). Third, the adsorption energies on the zinc blende surface are systematically higher than those computed on the wurtzite slab by about 30.2−5.8 kJ mol−1 (most values range from 10.9 to 15.1 kJ mol−1). These higher adsorption energies for ZB are attributed to larger surface− ligand dispersive interactions due to surface morphology. Indeed, while the adsorption energies without Grimme′s dispersion term are very similar for both surfaces, the dispersive contribution to the adsorption energy is 11.1−25.0 kJ mol−1 larger in ZB than that in WZ (see Table S6 in the SI). The adsorption energies through the parallel mode range between −142.6 and −87.7 kJ mol−1 (see Table S7 in the SI) and they are systematically higher than those of the perpendicular mode. Thus, despite the fact that this adsorption mode implies a decrease in the direct Cd···X interaction (enlargement of the Cd−X distance of about 0.1−0.4 Å), it is compensated by the alkyl chain−surface interaction through dispersion. In this context, results suggest that at low coverages, ligands will mainly adopt a conformation that maximizes the interaction of the alkyl chain with the surface, regardless of the functional group. Full Surface Coverage Ligand Adsorption. Once the main factors determining the intrinsic ligand−CdSe surface interaction are determined, we address the ligand full coverage defined as the adsorption of one ligand per surface cadmium Figure 3. Optimized structures of the different conformations cation. This is the model that best reproduces the experimental resulting from full coverage ligands adsorption on the WZ (1×1) conditions and represents ligand−ligand interactions. We used surface: (a) decanethiol, (b) decanol, and (c) decanoic acid. 11472 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article perpendicular mode of the isolated ligands. The only models and the (CdSe)33 nanoparticle at full coverage. Table 4 exceptions are (C2)3PO and (C2)3N ligands, which are too reports the band gap of the pristine (110) ZB and (110) WZ large to fit inside the (110) ZB (1×1) unit cell. This outlines slab models, and the pristine (CdSe)33 nanoparticle as well as that these two ligands and their associated families (phosphine those of the passivated systems with the smaller model of each oxides and tertiary amines) are not suitable for saturating all type of ligands considered. Figure 4 shows the projected cadmium vacant sites. The Cd···X distance in the full coverage density of states (pDOS) of the pristine materials (bulk, is akin to that of the low coverage, albeit slightly larger in surfaces, and nanoparticle) and the pDOS of (CdSe)33 capped general. The Cd···X distance trends are the same as those with 27 ethanethiol molecules (Cd33Se33-27EtSH), centered observed for the isolated systems ((C2)P(OH)2O < (C2)NH2 around the band gap region. < (C1)COOH < (C2)OH < (C2)SH). This suggests that the The estimated band gap energies of the pristine models vary ligand−surface interaction is similar to that in the low-coverage between 2.08 and 3.06 eV depending on the model (bulk, slab, simulation with minimum competitive effects. As a conse- or nanoparticle) and the polymorph considered. The band gap quence, the adsorption energies can again be divided into the for ZB bulk is 0.12 eV smaller than that of the WZ same two groups as in the low-coverage scenario: those of polymorph.27,99−101 This agrees with the lower band gap phosphonic acid and amine ligands, which are systematically experimentally determined for ZB, albeit both absolute band adsorbed more strongly for a fixed alkyl chain length, and those gaps are 0.3−0.4 eV higher than those determined of carboxylic acids, alcohols, and thiols. The adsorption experimentally. The band gap energies of the slab models energies at full coverage for the smallest C2 models on ZB and the (CdSe)33 nanoparticle are higher than those of the are in general close to those of the low coverage, the largest bulk structures, consistent with a decrease in the dimension- deviation being between −10.0 and +12.7 kJ mol−1. In ality of the system. The projected density of states (pDOS) contrast, the adsorption energy for the largest models is indicates that the valence band for all materials is mainly systematically larger than the value for the low coverage by formed by selenium 4p orbitals, while the conducting band has about 30 kJ mol−1 (range 28.6−36.6 kJ mol−1). Since the Cd··· contributions of both cadmium 5s and selenium 4p orbitals. X distance does not vary significantly, this arises from the Due to the different dimensionalities of the systems, the dispersive stabilizing ligand−ligand interactions, which are conducting and valence band edge states appear at different particularly important in the large models. Indeed, calculations energies for each system. However, when referenced to the for several thiols with increasing alkyl chains show that the Fermi level, the pDOS of bulk and slab models are relatively ligand−ligand stabilizing interaction increases by about 5.2 kJ similar and none of them presents mid-gap states (Figure 4a− mol−1 per CH2 (Figure S13). d). In contrast, the pDOS of the nanoparticle model (Figure In the case of the (110) WZ surface, the presence of two 4e) is significantly different and is characterized by the ligands per unit cell gives rise to a larger configurational presence of mid-gap states. The presence of these mid-gap variability. That is, for thiols, alcohols, and amines, we have states arises from the nanoparticle reconstruction that leads to identified two different modes of adsorption: one in which the short Cd···Cd distances and small Se−Cd−Se angles. two ligands interact with cadmium atoms, and another one Adsorption of capping ligands on both ZB and WZ surface where one ligand is directly adsorbed on the surface, while the models does not significantly alter the band gap, with almost all latter is hydrogen-bonded to the former (Figure 3a,b). differences being within 0.04 eV. This is linked to a lowly Meanwhile, for carboxylic and phosphonic acids, there are reorganization of the surface Cd···Cd distances. In contrast, five potential different coordination modes depending on adsorption on the nanoparticle induces an increase of the band whether the adsorption of each ligand takes place through the gap by at least 0.2 eV. In particular, the computed value for �O group or the OH and the number of ligands directly (CdSe)33−(C2−NH2)27 is 3.44 eV in reasonable agreement adsorbed (1 or 2). However, some of these possibilities were with the reported adsorption spectrum of this cluster with not stable and during the optimization evolved to one of the decylamine in toluene (λ = 415 nm).76 Indeed, the band gap other conformations (Figure 3c). In all cases, the most stable obtained from calculations correctly accounts for the differ- configuration is that where the two ligands interact with the ences between the materials studied here, but it is 0.4 eV surface, although the configuration where one ligand is higher than that of the experiments as already found for bulk desorbed, and hydrogen bonded with the other ligand has an materials. Remarkably, TD-DFT calculations (Table S9 in the adsorption energy per ligand that is only between 3.1 and 17.4 Supporting Information) at the PBE0/def2-SVP level of theory kJ mol−1 lower in absolute value (adsorption energies for all suggest that the experimental and computational discrepancy conformations are reported in Table S8 in the Supporting does not arise from the selected functional but from the fact Information). Moreover, for the acids, the preferred adsorption mode is through the �O group by at least 5.5 kJ mol−1 per Table 4. Best Estimated (See the Computational Details ligand. Finally, as for ZB, adsorption energies for the long- Section) Band Gap Energies in the Gas Phase and Water (in alkyl-chain ligands are systematically larger by 35−50 kJ mol−1 eV) for Pristine and Passivated Slab and Nanoparticle due to ligand−ligand interactions. This increase is larger than Models in ZB due to a shorter surface Cd···Cd distance in WZ, which favors lateral dispersive interactions. As a result, adsorption ZB (110) WZ (110) (CdSe)33 energies on ZB and WZ at full coverages become more similar. Gp water Gp water Gp water Optical Properties. As mentioned in the introduction, CdSe quantum dots show excellent optical properties for a pristine 2.66 2.67 2.79 2.84 3.06 3.43 large variety of applications such as optical devices, photo- C2−COOH 2.64 2.60 2.81 2.77 3.25 3.49 catalysis, and biomedical diagnosis.3−14 Ligand passivation can C2−NH2 2.65 2.58 2.82 2.76 3.44 3.51 tune the electronic structure of the material; thus, we analyzed C2−OH 2.56 2.55 2.83 2.83 3.30 3.43 how the adsorbed ligands modify the band gap for the slab C2−SH 2.70 2.65 2.83 2.79 3.42 3.49 11473 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article The inclusion of solvent effects at the gas-phase-optimized structures using an implicit continuum model has little influence on the band gap of pristine and passivated surfaces (±0.06 eV). Consequently, the computed band gaps with and without surface passivation are very similar and suggest that in the absence of surface reconstruction, neither solvation nor ligand passivation tunes the band gap. On the other hand, solvent effects have a large influence on the band gap of the pristine nanoparticle that becomes 3.43 eV in water solution (3.06 eV in the gas phase). The presence of ligands passivating (CdSe)33 diminishes the influence of the solvent regardless of the nature of the ligand, and thus, the band gap of the passivated nanoparticles only increases by about 0.1−0.2 eV. Overall, the band gaps of the nonpassivated and passivated nanoparticles become very similar in water. Similar trends are found when considering less polar solvents (acetonitrile or toluene, Table S2 in the Supporting Information), but the effect is less pronounced. Finally, since the results obtained from static calculations Figure 4. Projected density of states (Cd blue line, Se red line) of show that the band gap of the nanoparticle is very sensitive to wurtzite (a) and zinc blende (b) bulk, the main (110) wurtzite (c) solvation as well as its structure, subtle changes associated with and zinc blende surfaces (d), the (CdSe)33 nanoparticle model (e), thermal effects may influence the band gap and the role of and the (CdSe)33 nanoparticle with 27 C2−SH ligands adsorbed on capping ligands. To evaluate these thermal effects, ab initio its surface (f). molecular dynamics calculations have been performed for the five nanoparticle models: pristine CdSe 33 , CdSe 33 − that we are analyzing the band gap instead of the optical (C2COOH)27, CdSe33−(C2NH2)27, CdSe33−(C2OH)27, and transitions. The band gap increase is related to the displace- CdSe33−(C2SH)27. For each trajectory, clusterization based on ment of the mid-gap states to higher values with respect to the the cadmium centers has been performed and the band gap of Fermi Level due to the interaction between the nanoparticle the representative structure of each cluster has been computed and the capping ligand (Figure 4). with and without solvation. In addition, we computed the Figure 5. Weighted band gap and UV−vis spectra of the representative AIMD structures of pristine (CdSe)33 (with [s] and without solvation) and the full covered nanoparticle models with 27 acetic (C2−COOH), ethylamine (C2−NH2), ethanol (C2−OH), and ethanethiol (C2−SH) capping ligands. 11474 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article absorption spectra of these systems by considering 10 excited Results on the optical properties of bare materials (bulk, states of three representative structures (see the Computa- surfaces, and nanoparticles) show that decreasing model tional Details section). Results show that the band gap energy dimensionality on going from bulk to surfaces and nano- values as well as the absorption wavelengths are sensitive to the particles increases the energy band gap. However, the density structure of the nanoparticle (Figure 5), with variations being of states of the (CdSe)33 nanoparticle presents mid-gap states as large as 0.5 eV (120 nm) in agreement with the work of that decrease the overall HOMO−LUMO energy difference. Tretiak and co-workers. 62 Remarkably, the band gap These mid-gap states arise from the nanoparticle reconstruc- fluctuation decreases significantly both by ligand adsorption tion that leads to shorter Cd···Cd distances and smaller Se− and the inclusion of solvent effects, the latter factor being Cd−Se angles. Ligand passivation on both zinc blende and predominant. Similarly, the bandwidth of the absorption wurtzite surfaces is not associated with an important surface spectra decreases when including solvent effects. Moreover, reconstruction and, consequently, the band gap is very similar the weighted band gaps are smaller than those of the static to that of the bare surfaces. In contrast, ligand adsorption on calculations due to the changes in the HOMO and LUMO (CdSe)33 induces a significant structural reconstruction, which energies associated with the general increase of the bond is related to a general increase of the Cd···Cd distances. distances. Interestingly, the inclusion of thermal effects does Therefore, mid-gap states move to higher energies so that the not change the trends observed in the static calculations as HOMO−LUMO gap increases, and the maximum of the evidenced by the following two observations. First, even absorption spectrum is blue-shifted upon passivation. though the difference between the pristine and passivated Overall, the combination of slab and nanoparticle models nanoparticles is smaller than the static calculations, there is an suggests that while rigid and crystalline unsaturated cadmium increase in the gas phase band gap upon adsorption of 27 atoms (represented with the slab models) are essentially not capping ligands. Second, addition of solvent effects increases affected neither by the adsorption of organic ligands nor the the weighted band gap of the pristine nanoparticle significantly solvent effects, centers located at flexible environments but almost does not modify that of the passivated models. (represented by (CdSe)33) are more sensitive to the presence Overall, the maximum wavelength of the absorption spectra of the solvated pristine nanoparticle is blue-shifted by 80 nm of both the capping ligands and the solvent, although the when compared to the values in vacuum, while the difference effects do not depend on the nature of the capping ligand. This between the maximum absorption wavelength of passivated makes the tunning of the optical properties through ligand and nonpassivated nanoparticles including solvation is 20 nm. passivation rather limited. Thus, passivation through core− shell materials and nanoparticle size appears to be more ■ CONCLUSIONS suitable strategies. Cadmium selenide nanoparticles are among the most widely used quantum dots in photonics, solar cells, catalysis, and bioimaging. Surface passivation through the adsorption of ■ * ASSOCIATED CONTENT sı Supporting Information capping ligands is key for tuning the quantum dot’s optical The Supporting Information is available free of charge at properties, and atomistic insights into how ligand alters their https://pubs.acs.org/doi/10.1021/acsomega.3c00324. properties are important for the design of new quantum dots. The present computational work provides a broad, compre- All calibration tests for defining the computational hensive insight into the ligand passivation on CdSe nanoma- methodology; adsorption energies for the parallel mode terials through a combination of static and AIMD DFT as well as for the different conformations of the WZ full simulations. The zinc blende (ZB) and wurtzite (WZ) (110) coverage scenario; TD-DFT calculations; and the crude surfaces, the (CdSe)33 nanoparticle, and five types of ligands data of the AIMD simulations (PDF) used in experiments (carboxylic acids, phosphonic acids, amines, alcohols, and thiols) were considered to analyze how their adsorption influences the optical properties of the material. ■ AUTHOR INFORMATION Corresponding Author Ligand adsorption is mainly electrostatic with almost no Xavier Solans-Monfort − Departament de Química, charge transfer from the ligand to the surface. At low Universitat Autònoma de Barcelona, 08195 Bellaterra, coverages, ligand adsorption energies on ZB are computed to Spain; orcid.org/0000-0002-2172-3895; be higher than those on WZ by about 10−15 kJ mol−1 due to Email: xavier.solans@uab.cat the larger surface−ligand dispersive interactions. Moreover, for both polymorphs, phosphine oxides, phosphonic acids, and Authors amines adsorb more strongly than carboxylic acids, alcohols, Didac A. Fenoll − Departament de Química, Universitat and thiols, and this correlates well with their proton affinities, Autònoma de Barcelona, 08195 Bellaterra, Spain; reinforcing the electrostatic nature of the interaction. orcid.org/0000-0003-1141-8114 Increasing the ligand coverage up to full coverage does not Mariona Sodupe − Departament de Química, Universitat change the trends found at low coverages. However, the steric Autònoma de Barcelona, 08195 Bellaterra, Spain; hindrance of phosphine oxides and tertiary amines hampers orcid.org/0000-0003-0276-0524 their adsorption on each undercoordinated surface Cd site. Moreover, the adsorption difference between ZB and WZ Complete contact information is available at: decreases due to larger ligand−ligand dispersive interactions in https://pubs.acs.org/10.1021/acsomega.3c00324 the latter. Indeed, dispersive ligand−ligand interactions are important for large ligands mainly, and their contribution is Notes estimated to be about 5.2 kJ mol−1 per CH2 unit. The authors declare no competing financial interest. 11475 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article ■ ACKNOWLEDGMENTS The authors gratefully acknowledge financial support from (20) Zhou, Q.; Cho, Y.; Yang, S.; Weiss, E. A.; Berkelbach, T. C.; Darancet, P. Large Band Edge Tunability in Colloidal Nanoplatelets. Nano Lett. 2019, 19, 7124−7129. MICINN [PID2020-112715GB-I00], the Generalitat de (21) Cho, W.; Kim, S.; Coropceanu, I.; Srivastava, V.; Diroll, B. T.; Catalunya [2017SGR1323]. D.A.F. acknowledges MICINN Hazarika, A.; Fedin, I.; Galli, G.; Schaller, R. D.; Talapin, D. V. Direct for his FPU PhD fellowship [FPU18/03612]. X.S.-M. is Synthesis of Six-Monolayer (1.9 Nm) Thick Zinc-Blende CdSe grateful for Professor Agregat Serra Húnter position. Nanoplatelets Emitting at 585 Nm. Chem. Mater. 2018, 30, 6957− 6960. ■ REFERENCES (1) Alivisatos, A. P. Semiconductor Clusters, Nanocrystals, and (22) Asokan, S.; Krueger, K. M.; Colvin, V. L.; Wong, M. S. Shape- Controlled Synthesis of CdSe Tetrapods Using Cationic Surfactant Ligands. Small 2007, 3, 1164−1169. Quantum Dots. Science 1996, 271, 933−937. (23) Meyns, M.; Iacono, F.; Palencia, C.; Geweke, J.; Coderch, M. (2) Boles, M. A.; Ling, D.; Hyeon, T.; Talapin, D. V. The Surface D.; Fittschen, U. E. A.; Gallego, J. M.; Otero, R.; Juárez, B. H.; Klinke, Science of Nanocrystals. Nat. Mater. 2016, 15, 141−153. C. Shape Evolution of CdSe Nanoparticles Controlled by Halogen (3) Polman, A.; Knight, M.; Garnett, E. C.; Ehrler, B.; Sinke, W. C. Compounds. Chem. Mater. 2014, 26, 1813−1821. Photovoltaic Materials: Present Efficiencies and Future Challenges. (24) Peng, X.; Manna, L.; Yang, W.; Wickham, J.; Scher, E.; Science 2016, 352, No. aad4424. Kadavanich, A.; Alivisatos, A. P. Shape Control of CdSe Nanocrystals. (4) Yang, Y.; Zheng, Y.; Cao, W.; Titov, A.; Hyvonen, J.; Manders, J. Nature 2000, 404, 59−61. R.; Xue, J.; Holloway, P. H.; Qian, L. High-Efficiency Light-Emitting (25) Liu, L.; Zhuang, Z.; Xie, T.; Wang, Y. G.; Li, J.; Peng, Q.; Li, Y. Devices Based on Quantum Dots with Tailored Nanostructures. Nat. Shape Control of CdSe Nanocrystals with Zinc Blende Structure. J. Photonics 2015, 9, 259−265. Am. Chem. Soc. 2009, 131, 16423−16429. (5) Carey, G. H.; Abdelhady, A. L.; Ning, Z.; Thon, S. M.; Bakr, O. (26) Shan, W.; Song, J. J.; Luo, H.; Furdyna, J. K. Determination of M.; Sargent, E. H. Colloidal Quantum Dot Solar Cells. Chem. Rev. the Fundamental and Split-off Band Gaps in Zinc-Blende CdSe by 2015, 115, 12732−12763. Photomodulation Spectroscopy. Phys. Rev. B 1994, 50, 8012−8015. (6) Kamat, P. V. Quantum Dot Solar Cells. The next Big Thing in (27) Lunz, U.; Kuhn, J.; Goschenhofer, F.; Schüssler, U.; Einfeldt, S.; Photovoltaics. J. Phys. Chem. Lett. 2013, 4, 908−918. Becker, C. R.; Landwehr, G. Temperature Dependence of the Energy (7) Mora-Seró, I.; Giménez, S.; Fabregat-Santiago, F.; Gómez, R.; Gap of Zinc-blende CdSe and Cd 1− x Zn x Se Epitaxial Layers. J. Shen, Q.; Toyoda, T.; Bisquert, J. Recombination in Quantum Dot Appl. Phys. 1996, 80, 6861−6863. Sensitized Solar Cells. Acc. Chem. Res. 2009, 42, 1848−1857. (28) Hartmann, L.; Kumar, A.; Welker, M.; Fiore, A.; Julien-Rabant, (8) Jia, J.; Seitz, L. C.; Benck, J. D.; Huo, Y.; Chen, Y.; Ng, J. W. D.; C.; Gromova, M.; Bardet, M.; Reiss, P.; Baxter, P. N. W.; Chandezon, Bilir, T.; Harris, J. S.; Jaramillo, T. F. Solar Water Splitting by F.; Pansu, R. B. Quenching Dynamics in CdSe Nanoparticles: Surface- Photovoltaic-Electrolysis with a Solar-to-Hydrogen Efficiency over Induced Defects upon Dilution. ACS Nano 2012, 6, 9033−9041. (29) Yang, S.; Prendergast, D.; Neaton, J. B. Tuning Semiconductor 30%. Nat. Commun. 2016, 7, No. 13237. Band Edge Energies for Solar Photocatalysis via Surface Ligand (9) Kang, D.; Kim, T. W.; Kubota, S. R.; Cardiel, A. C.; Cha, H. G.; Passivation. Nano Lett. 2012, 12, 383−388. Choi, K. S. Electrochemical Synthesis of Photoelectrodes and (30) Elimelech, O.; Aviv, O.; Oded, M.; Banin, U. A Tale of Tails: Catalysts for Use in Solar Water Splitting. Chem. Rev. 2015, 115, Thermodynamics of CdSe Nanocrystal Surface Ligand Exchange. 12839−12887. Nano Lett. 2020, 20, 6396−6403. (10) Das, A.; Han, Z.; Haghighi, M. G.; Eisenberg, R. Photo- (31) Lin, Q.; Yun, H. J.; Liu, W.; Song, H. J.; Makarov, N. S.; generation of Hydrogen from Water Using CdSe Nanocrystals Isaienko, O.; Nakotte, T.; Chen, G.; Luo, H.; Klimov, V. I.; Pietryga, J. Demonstrating the Importance of Surface Exchange. Proc. Natl. M. Phase-Transfer Ligand Exchange of Lead Chalcogenide Quantum Acad. Sci. U.S.A. 2013, 110, 16716−16723. Dots for Direct Deposition of Thick, Highly Conductive Films. J. Am. (11) Medintz, I. L.; Uyeda, H. T.; Goldman, E. R.; Mattoussi, H. Chem. Soc. 2017, 139, 6644−6653. Quantum Dot Bioconjugates for Imaging, Labelling and Sensing. Nat. (32) Anderson, N. C.; Hendricks, M. P.; Choi, J. J.; Owen, J. S. Mater. 2005, 4, 435−446. Ligand Exchange and the Stoichiometry of Metal Chalcogenide (12) Wolfbeis, O. S. An Overview of Nanoparticles Commonly Used Nanocrystals: Spectroscopic Observation of Facile Metal-Carboxylate in Fluorescent Bioimaging. Chem. Soc. Rev. 2015, 44, 4743−4768. Displacement and Binding. J. Am. Chem. Soc. 2013, 135, 18536− (13) Wegner, K. D.; Hildebrandt, N. Quantum Dots: Bright and 18548. Versatile in Vitro and in Vivo Fluorescence Imaging Biosensors. (33) Ben-Shahar, Y.; Scotognella, F.; Waiskopf, N.; Kriegel, I.; Dal Chem. Soc. Rev. 2015, 44, 4792−4834. Conte, S.; Cerullo, G.; Banin, U. Effect of Surface Coating on the (14) Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, Photocatalytic Function of Hybrid CdS-Au Nanorods. Small 2015, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. 11, 462−471. Quantum Dots for Live Cells, in Vivo Imaging, and Diagnostics. (34) Č apek, R. K.; Lambert, K.; Dorfs, D.; Smet, P. F.; Poelman, D.; Science 2005, 307, 538−544. Eychmüller, A.; Hens, Z. Synthesis of Extremely Small CdSe and (15) Hartley, C. L.; Kessler, M. L.; Dempsey, J. L. Molecular-Level Bright Blue Luminescent CdSe/ZnS Nanoparticles by a Prefocused Insight into Semiconductor Nanocrystal Surfaces. J. Am. Chem. Soc. Hot-Injection Approach. Chem. Mater. 2009, 21, 1743−1749. 2021, 143, 1251−1266. (35) Cho, J.; Jung, Y. K.; Lee, J. K.; Jung, H. S. Highly Efficient Blue- (16) Jasieniak, J.; Califano, M.; Watkins, S. E. Size-Dependent Emitting CdSe-Derived Core/Shell Gradient Alloy Quantum Dots Valence and Conduction Band-Edge Energies of Semiconductor with Improved Photoluminescent Quantum Yield and Enhanced Nanocrystals. ACS Nano 2011, 5, 5888−5902. Photostability. Langmuir 2017, 33, 3711−3719. (17) Hines, D. A.; Kamat, P. V. Recent Advances in Quantum Dot (36) Moreels, I.; Rainò, G.; Gomes, R.; Hens, Z.; Stöferle, T.; Mahrt, Surface Chemistry. ACS Appl. Mater. Interfaces 2014, 6, 3041−3057. R. F. Band-Edge Exciton Fine Structure of Small, Nearly Spherical (18) Hens, Z.; de Roo, J. Atomically Precise Nanocrystals. J. Am. Colloidal CdSe/ZnS Quantum Dots. ACS Nano 2011, 5, 8033−8039. Chem. Soc. 2020, 142, 15627−15637. (37) Boldt, K.; Kirkwood, N.; Beane, G. A.; Mulvaney, P. Synthesis (19) Kroupa, D. M.; Vörös, M.; Brawand, N. P.; McNichols, B. W.; of Highly Luminescent and Photo-Stable, Graded Shell CdSe/Cd Miller, E. M.; Gu, J.; Nozik, A. J.; Sellinger, A.; Galli, G.; Beard, M. C. XZn1-XS Nanoparticles by in Situ Alloying. Chem. Mater. 2013, 25, Tuning Colloidal Quantum Dot Band Edge Positions through 4731−4738. Solution-Phase Surface Chemistry Modification. Nat. Commun. (38) Peng, X.; Schlamp, M. C.; Kadavanich, A. V.; Alivisatos, A. P. 2017, 8, No. 15257. Epitaxial Growth of Highly Luminescent CdSe/CdS Core/Shell 11476 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article Nanocrystals with Photostability and Electronic Accessibility. J. Am. and Stabilities of CdSe Nanoclusters. J. Mol. Struct. 2016, 1114, 123− Chem. Soc. 1997, 119, 7019−7029. 131. (39) Elimelech, O.; Aviv, O.; Oded, M.; Peng, X.; Harries, D.; Banin, (56) Voznyy, O.; Mokkath, J. H.; Jain, A.; Sargent, E. H.; U. Entropy of Branching Out: Linear versus Branched Alkylthiols Schwingenschlögl, U. Computational Study of Magic-Size Cdse Ligands on CdSe Nanocrystals. ACS Nano 2022, 16, 4308−4321. Clusters with Complementary Passivation by Carboxylic and Amine (40) Gomes, R.; Hassinen, A.; Szczygiel, A.; Zhao, Q.; Vantomme, Ligands. J. Phys. Chem. C 2016, 120, 10015−10019. A.; Martins, J. C.; Hens, Z. Binding of Phosphonic Acids to CdSe (57) Liu, H. Tuning the Binding Energy of Surfactant to Cdse Quantum Dots: A Solution NMR Study. J. Phys. Chem. Lett. 2011, 2, Nanocrystal: A Theoretical Study. J. Phys. Chem. C 2009, 113, 3116− 145−152. 3119. (41) Jose, R.; Zhanpeisov, N. U.; Fukumura, H.; Baba, Y.; Ishikawa, (58) Nguyen, K. A.; Pachter, R.; Day, P. N. Computational I. Structure-Property Correlation of CdSe Clusters Using Exper- Prediction of Structures and Optical Excitations for Nanoscale imental Results and First-Principles DFT Calculations. J. Am. Chem. Ultrasmall ZnS and CdSe Clusters. J. Chem. Theory Comput. 2013, Soc. 2006, 128, 629−636. 9, 3581−3596. (42) Singh, S.; Leemans, J.; Zaccaria, F.; Infante, I.; Hens, Z. Ligand (59) del Ben, M.; Havenith, R. W. A.; Broer, R.; Stener, M. Density Adsorption Energy and the Postpurification Surface Chemistry of Functional Study on the Morphology and Photoabsorption of CdSe Colloidal Metal Chalcogenide Nanocrystals. Chem. Mater. 2021, 33, Nanoclusters. J. Phys. Chem. C 2011, 115, 16782−16796. 2796−2803. (60) Azpiroz, J. M.; de Angelis, F. Ligand Induced Spectral Changes (43) Csik, I.; Russo, S. P.; Mulvaney, P. Density Functional Study of in CdSe Quantum Dots. ACS Appl. Mater. Interfaces 2015, 7, 19736− Surface Passivation of Nonpolar Wurtzite CdSe Surfaces. J. Phys. 19745. Chem. C 2008, 112, 20413−20417. (61) Kilina, S.; Ivanov, S.; Tretiak, S. Effect of Surface Ligands on (44) Virgili, T.; Calzolari, A.; Suárez López, I.; Ruini, A.; Catellani, Optical and Electronic Spectra of Semiconductor Nanoclusters. J. Am. A.; Vercelli, B.; Tassone, F. Hybridized Electronic States between Chem. Soc. 2009, 131, 7717−7726. CdSe Nanoparticles and Conjugated Organic Ligands: A Theoretical (62) Ghosh, D.; Ivanov, S. A.; Tretiak, S. Structural Dynamics and and Ultrafast Photo-Excited Carrier Dynamics Study. Nano Res. 2018, Electronic Properties of Semiconductor Quantum Dots: Computa- 11, 142−150. tional Insights. Chem. Mater. 2021, 33, 7848−7857. (45) Virgili, T.; Calzolari, A.; Suárez López, I.; Vercelli, B.; Zotti, G.; (63) Albert, V. V.; Ivanov, S. A.; Tretiak, S.; Kilina, S. V. Electronic Catellani, A.; Ruini, A.; Tassone, F. Charge Separation in the Hybrid Structure of Ligated Cdse Clusters: Dependence on DFT Method- Cdse Nanocrystal-Organic Interface: Role of the Ligands Studied by ology. J. Phys. Chem. C 2011, 115, 15793−15800. Ultrafast Spectroscopy and Density Functional Theory. J. Phys. Chem. (64) Fischer, S. A.; Crotty, A. M.; Kilina, S. V.; Ivanov, S. A.; Tretiak, S. Passivating Ligand and Solvent Contributions to the Electronic C 2013, 117, 5969−5974. (46) Manna, L.; Wang, L. W.; Cingolani, R.; Alivisatos, A. P. First- Properties of Semiconductor Nanocrystals. Nanoscale 2012, 4, 904− 914. Principles Modeling of Unpassivated and Surfactant-Passivated Bulk (65) Kilina, S. V.; Tamukong, P. K.; Kilin, D. S. Surface Chemistry of Facets of Wurtzite CdSe: A Model System for Studying the Semiconducting Quantum Dots: Theoretical Perspectives. Acc. Chem. Anisotropic Growth of CdSe Nanocrystals. J. Phys. Chem. B 2005, Res. 2016, 49, 2127−2135. 109, 6183−6192. (66) Kilina, S. V.; Kilin, D. S.; Prezhdo, O. V. Breaking the Phonon (47) Azpiroz, J. M.; Ugalde, J. M.; Infante, I. Benchmark Assessment Bottleneck in PbSe and CdSe Quantum Dots: Time-Domain Density of Density Functional Methods on Group II-VI MX (M = Zn, Cd; X Functional Theory of Charge Carrier Relaxation. ACS Nano 2009, 3, = S, Se, Te) Quantum Dots. J. Chem. Theory Comput. 2014, 10, 76− 93−99. 89. (67) Giansante, C.; Infante, I. Surface Traps in Colloidal Quantum (48) Azpiroz, J. M.; Matxain, J. M.; Infante, I.; Lopez, X.; Ugalde, J. Dots: A Combined Experimental and Theoretical Perspective. J. Phys. M. A DFT/TDDFT Study on the Optoelectronic Properties of the Chem. Lett. 2017, 8, 5209−5215. Amine-Capped Magic (CdSe)13 Nanocluster. Phys. Chem. Chem. (68) du Fossé, I.; Boehme, S. C.; Infante, I.; Houtepen, A. J. Phys. 2013, 15, 10996−11005. Dynamic Formation of Metal-Based Traps in Photoexcited Colloidal (49) Gao, Y.; Zhou, B.; Kang, S. G.; Xin, M.; Yang, P.; Dai, X.; Quantum Dots and Their Relevance for Photoluminescence. Chem. Wang, Z.; Zhou, R. Effect of Ligands on the Characteristics of Mater. 2021, 33, 3349−3358. (CdSe)13 Quantum Dots. RSC Adv. 2014, 4, 27146−27151. (69) Houtepen, A. J.; Hens, Z.; Owen, J. S.; Infante, I. On the Origin (50) Swenson, N. K.; Ratner, M. A.; Weiss, E. A. Computational of Surface Traps in Colloidal II-VI Semiconductor Nanocrystals. Study of the Influence of the Binding Geometries of Organic Ligands Chem. Mater. 2017, 29, 752−761. on the Photoluminescence Quantum Yield of CdSe Clusters. J. Phys. (70) Drijvers, E.; de Roo, J.; Martins, J. C.; Infante, I.; Hens, Z. Chem. C 2016, 120, 6859−6868. Ligand Displacement Exposes Binding Site Heterogeneity on CdSe (51) Tan, L.; Pickard, C. J.; Yu, K.; Sapelkin, A.; Misquitta, A. J.; Nanocrystal Surfaces. Chem. Mater. 2018, 30, 1178−1186. Dove, M. T. Structures of CdSe and CdS Nanoclusters from Ab Initio (71) du Fossé, I.; Lal, S.; Hossaini, A. N.; Infante, I.; Houtepen, A. J. Random Structure Searching. J. Phys. Chem. C 2019, 123, 29370− Effect of Ligands and Solvents on the Stability of Electron Charged 29378. CdSe Colloidal Quantum Dots. J. Phys. Chem. C 2021, 125, 23968− (52) Inerbaev, T. M.; Masunov, A. E.; Khondaker, S. I.; Dobrinescu, 23975. A.; Plamad, A. V.; Kawazoe, Y. Quantum Chemistry of Quantum (72) Kuznetsov, A. E.; Beratan, D. N. Structural and Electronic Dots: Effects of Ligands and Oxidation. J. Chem. Phys. 2009, 131, Properties of Bare and Capped Cd 33Se33 and Cd33Te33 Quantum 044106. Dots. J. Phys. Chem. C 2014, 118, 7094−7109. (53) Cui, Y.; Cui, X.; Zhang, L.; Xie, Y.; Yang, M. Theoretical (73) Puzder, A.; Williamson, A. J.; Gygi, F.; Galli, G. Self-Healing of Characterization on the Size-Dependent Electron and Hole Trapping CdSe Nanocrystals: First-Principles Calculations. Phys. Rev. Lett. Activity of Chloride-Passivated CdSe Nanoclusters. J. Chem. Phys. 2004, 92, 217401. 2018, 148, 134308. (74) Puzder, A.; Williamson, A. J.; Zaitseva, N.; Galli, G.; Manna, L.; (54) Cui, Y.; Lou, Z.; Wang, X.; Yu, S.; Yang, M. A Study of Optical Alivisatos, A. P. The Effect of Organic Ligand Binding on the Growth Absorption of Cysteine-Capped CdSe Nanoclusters Using First- of CdSe Nanoparticles Probed by Ab Initio Calculations. Nano Lett. Principles Calculations. Phys. Chem. Chem. Phys. 2015, 17, 9222− 2004, 4, 2361−2365. 9230. (75) Sarkar, P.; Springborg, M. Density-Functional Study of Size- (55) Sun, J.; Zheng, X.; He, H.; Chen, X.; Dong, B.; Fei, R. Dependent Properties of CdmSen Clusters. Phys. Rev. B 2003, 68, Theoretical Study of Ligand and Solvent Effects on Optical Properties No. 235409. 11477 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478 ACS Omega http://pubs.acs.org/journal/acsodf Article (76) Kasuya, A.; Sivamohan, R.; Barnakov, Y. A.; Dmitruk, I. M.; (96) Andrae, D.; Häußermann, U.; Dolg, M.; Stoll, H.; Preuß, H. Nirasawa, T.; Romanyuk, V. R.; Kumar, V.; Mamykin, S. V.; Tohji, K.; Energy-Adjusted ab Initio Pseudopotentials for the Second and Third Jeyadevan, B.; Shinoda, K.; Kudo, T.; Terasaki, O.; Liu, Z.; Row Transition Elements. Theor. Chim. Acta 1990, 77, 123−141. Belosludov, R. V.; Sundararajan, V.; Kawazoe, Y. Ultra-Stable (97) Barone, V.; Cossi, M. Quantum Calculation of Molecular Nanoparticles of CdSe Revealed from Mass Spectrometry. Nat. Energies and Energy Gradients in Solution by a Conductor Solvent Mater. 2004, 3, 99−102. Model. J. Phys. Chem. A 1998, 102, 1995−2001. (77) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab (98) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. Energies, Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Structures, and Electronic Properties of Molecules in Solution with Rev. B 1996, 54, 11169−11186. the C-PCM Solvation Model. J. Comput. Chem. 2003, 24, 669−681. (78) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for Liquid (99) Kaneta, A.; Adachi, S. Photoreflectance Study of Hexagonal Metals. Phys. Rev. B 1993, 47, 558−561. CdSe. J. Phys. D: Appl. Phys. 1999, 32, 2337−2341. (79) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient (100) Janowitz, C.; Günther, O.; Jungk, G.; Johnson, R. L.; Santos, Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. P. V.; Cardona, M.; Faschinger, W.; Sitter, H. Dielectric Function and (80) Grimme, S. Accurate Description of van Der Waals Complexes Critical Points of Cubic and Hexagonal CdSe. Phys. Rev. B 1994, 50, by Density Functional Theory Including Empirical Corrections. J. 2181−2187. Comput. Chem. 2004, 25, 1463−1473. (101) Ninomiya, S.; Adachi, S. Optical Properties of Cubic and (81) Heras-Domingo, J.; Sodupe, M.; Solans-Monfort, X. Interaction Hexagonal CdSe. J. Appl. Phys. 1995, 78, 4681−4689. between Ruthenium Oxide Surfaces and Water Molecules. Effect of Surface Morphology and Water Coverage. J. Phys. Chem. C 2019, 123, 7786. (82) González, D.; Heras-Domingo, J.; Pantaleone, S.; Rimola, A.; Rodríguez-Santiago, L.; Solans-Monfort, X.; Sodupe, M. Water Adsorption on MO 2 (M = Ti, Ru, and Ir) Surfaces. Importance of Octahedral Distortion and Cooperative Effects. ACS Omega 2019, 4, 2989−2999. (83) Blöchl, P. E.; Först, C. J.; Schimpl, J. Projector Augmented Wave Method: Ab Initio Molecular Dynamics with Full Wave Functions. Bull. Mater. Sci. 2003, 26, 33−41. (84) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method. Phys. Rev. B 1999, 59, 1758− 1775. (85) Monkhorst, H. J.; Pack, J. D. Special Points for Brillouin-Zone Integrations. Phys. Rev. B 1976, 13, 5188−5192. (86) Bader, R. F. W. A Quantum Theory of Molecular Structure and Its Applications. Chem. Rev. 1991, 91, 893−928. (87) Henkelman, G.; Arnaldsson, A.; Jónsson, H. A Fast and Robust Algorithm for Bader Decomposition of Charge Density. Comput. Mater. Sci. 2006, 36, 354−360. (88) Romanova, K. A.; Galyametdinov, Y. G.Quantum-Chemical Study of CdSe/CdS Core/Shell and CdSe/CdS/ZnS Core/Shell/ Shell Quantum Dots with Different Layers Ratio. In AIP Conference Proceedings; AIP Publishing, 2021; Vol. 2380, p 060001 DOI: 10.1063/5.0058295. (89) Dudarev, S. L.; Botton, G. A.; et al. Electron-Energy-Loss Spectra and the Structural Stability of Nickel Oxide: An LSDA+U Study. Phys. Rev. B 1998, 57, 1505−1509. (90) Mathew, K.; Sundararaman, R.; Letchworth-Weaver, K.; Arias, T. A.; Hennig, R. G. Implicit Solvation Model for Density-Functional Study of Nanocrystal Surfaces and Reaction Pathways. J. Chem. Phys. 2014, 140, 084106. (91) Nosé, S. A Unified Formulation of the Constant Temperature Molecular Dynamics Methods. J. Chem. Phys. 1984, 81, 511−519. (92) Hoover, W. G. Canonical Dynamics: Equilibrium Phase-Space Distributions. Phys. Rev. A 1985, 31, 1695−1697. (93) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF Chimera - A Visualization System for Exploratory Research and Analysis. J. Comput. Chem. 2004, 25, 1605−1612. (94) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.et al. Gaussian16 (Revision A.03), Gaussian Inc.: Wallingford CT, 2016. (95) Weigend, F.; Ahlrichs, R. Balanced Basis Sets of Split Valence, Triple Zeta Valence and Quadruple Zeta Valence Quality for H to Rn: Design and Assessment of Accuracy. Phys. Chem. Chem. Phys. 2005, 7, 3297. 11478 https://doi.org/10.1021/acsomega.3c00324 ACS Omega 2023, 8, 11467−11478