4Instituto de Investigacion en Materiales, Universidad Nacional Autonoma de Mexico, Coyoacan, Mexico D.F., 04510, Mexico.

CdS nanoparticles (NP), 22.5±1.5 Å, have been synthesized from cadmium 2-ethylhexanoate in DMSO as a uniformly sized dispersion. After ripening they exhibit a sharp excitonic emission peak at 402 nm, while in freshly prepared dispersions a broad trapped emission at 510 nm dominates. By using one- and two-dimensional NMR spectroscopy the conformation of the stabilizer adsorbed to the NP has been determined. The long hexyl chain of 2-ethylhexanoate ions spreads over the surface of NP, whereas the short ethyl end is primarily surrounded by DMSO. Surface modification of CdS with nucleophiles such as 4,4-bipyridine, thiophene, trimethylamine, and thiomolybdate anion results in a partial replacement of the stabilizer and reorientation of the hexyl chain away from the surface. The difference in the degree of replacement and/or conformational changes of 2-ethylhexanoate ion depends on the electrondonor activity of the modifier.

The investigation of nanoparticle-based materials has witnessed a period of intense growth over the last decade. This field has attracted the attention of numerous research groups of diverse research backgrounds^1-5. One of the most significant aspects of these studies is the surface modification of nanoparticles (NP) which can be utilized as a method of variation of their optical properties and as a tool for their spatial organization^6-9. Theoretical and experimental study of the interactions between the electronic systems of the NP core and that of the modifier is expected to be one of the most interesting areas of research in this field.

Surface structure of NP is being studied by a variety of spectroscopic techniques including nuclear magnetic resonance spectroscopy (NMR). The latter has been applied to the investigation of nanoclusters quite extensively with the emphasis on the determination of surface density of stabilizer molecules and their chemical bonding with the cluster^10-19. Although NMR signals allow for registration of even small changes in both electronic interactions of various components and in their structure, the major concern for the broader application of this technique to NP research is the necessity to have relatively large quantities of an NMR-active nuclei in the specimen. In this respect, high particle per volume ratio in powders favor the application of the solid state NMR for NP studies in order to achieve suitable signal intensity. The trade-off for that is peak broadening and limited selectivity characteristic for the solid state NMR. Conversely, various techniques of NMR in liquids afford determination of fine details of the structure of large molecular assemblies such as proteins and polymers²⁰. Similar level of structural description would be quite desirable for NP, too. However, low molar concentration of NP in solutions, their partial agglomeration, and line broadening due to long relaxation times create problems for the acquisition and analysis of the NMR data.

The purpose of this publications is twofold: 1) to suggest a synthesis of NP with surface properties convenient for the competitive modification with large variety of organic substances and 2) to investigate the potential of the liquid state NMR for elucidating the structure of NP surface. The produced NP are stabilized by 2-ethylhexanoate anion (ethex) and DMSO held in adsorbed state by relatively weak non-covalent interactions. This affords facile displacement of the stabilizers by a variety of organic molecules. Along with apparent limitations, the weakly stabilized particles can be also used as a convenient model system for the description of electronic processes on the NP surface discussed in the subsequent publication. Relatively high mobility of ethex ions electrostatically adsorbed to the CdS surface alleviates relaxation time line broadening and allows for the accurate assignment and analysis of the NMR spectra. It is also demonstrated that NMR can be used to determine atomic groups residing in the vicinity of the semiconductor surface and to establish the conformation of ethex moieties.

II. EXPERIMENTAL

Cadmium 2-ethylhexanoate, Cd(ethex)₂, (Strem Chemicals), sodium sulfide (Ultra puris Fluka), (NH₄)₂MoS₄ (Strem, Aldrich), dimethylsulfoxide (DMSO, Aldrich), DMSO-d₆ (Aldrich), polyethylene powder (spectral grade, Aldrich), polyvinyl spectroscopic grade (Aldrich), high purity fused quartz plates (Monsanto), DMF (Fisher), and MeCN (Aldrich), 99.999%, and argon gas were used in their commercial form. Ultra pure water (18 MW) was obtained from a Milli-Q (Millipore) deionization system.

CdS nanoparticles were synthesized according to the following procedure. Typically, Cd(ethex)₂, 0.004 g, was dissolved in 49.9 mL of DMSO previously purged with nitrogen for at least 30 min in order to remove trace amounts of oxygen and hydrogen sulfide. Following sonication for 15-30 minutes, 0.10 mL of a stock Na₂S solution was rapidly injected into the vigorously stirred Cd²⁺ solution. The sodium sulfide solution was prepared fresh every time from 0.6 g of Na₂S·9H₂O dissolved in 25 mL of deoxygenated water (Ar bubbling for 30 minutes). Synthesis and subsequent ripening required about 6 hours for completion. If stored in the dark, UV-VIS absorption characteristics of thus prepared CdS dispersions did not change for several months, whereas emission parameters did as described in the Results and Discussion section. Exposure to the ambient light, the peak at 365 nm became broader and gradually disappeared over a period of one week. A short exposure, for example, during the preparation of UV and NMR samples, did not cause any apparent change of its UV-VIS spectrum. High purity Na₂S and a low concentration of cadmium ions were important factors in producing CdS crystallites with a narrow size distribution. Cd²⁺ ions were taken in a 5% excess compared to S^2- ions to promote adsorption of the negatively charged 2-ethylhexanoate anion and therefore, stabilization of the NP.

UV-visible absorption spectra were taken by using a 8453 and 8452A diode array Hewlett-Packard spectrophotometers. Fluorescence spectra were taken on a modular Fluorolog 3 SPEX spectrofluorometer.

Nuclear magnetic resonance (NMR) spectra were taken by using a Varian Inova 400 and 600 MHz instruments operating at H¹ frequencies of 399.96 MHz and 599.98 MHz, respectively. NP samples were dissolved in 99.9% DMSO-d⁶ (Aldrich), with 1% v/v TMS as an internal standard. Presaturation of the residual solvent line was used to improve the dynamic range of the spectrum. The synthetic procedure described above was found to be essential for obtaining a well resolved NMR spectrum.

Powder X-ray diffraction (XRD) spectra were collected by using a Siemens D5000 instrument. Samples for X-ray study were prepared by spreading a dispersion of NP in DMSO over fused quartz plates and evaporating the solvent at 10^-5 mm of Hg in a Mag III Cook vacuum system.

High resolution transmission electron microphotographs (HR-TEM) were obtained by using a JEOL 4000 Ex instrument operating at 400 kV. A 200 mesh copper grid, was coated with a layer of carbon. A drop of DMSO dispersion was deposited onto Cu grids and then the solvent was evaporated under vacuum.

Instead of covalently bound organic molecules, two relatively labile stabilizers have been employed: 2-ethylhexanoate anion (ethex) and DMSO. They are expected to be in a constant dynamic exchange between adsorbed and dissolved state and, therefore can be easily replaced with other molecules. This property can be utilized for surface modification of NP with organic groups with low affinity to CdS surface.

At sufficiently low concentrations of CdS, these stabilizers allow for the crystallization of narrow-sized NP with sharp absorption and emission peaks. After ripening, ethex/DMSO-stabilized CdS reveals a maximum of absorption at 365±5 nm (3.4±0.06 eV) (Fig. 1) characteristic of the first excitonic state. A weak shoulder corresponding to the second excitonic transition, located in the valley of the absorption edge can be observed around 335 nm. According to the experimental correlations between the position of the first absorption feature and the nanoparticle diameter for CdS²¹, 365 nm absorption maximum corresponds to NP with a diameter of 22.5±1.5 Å.

As-synthesized NP display a broad luminescence peak at l_max= 510 nm, Dl_1/2 = 100 nm, which is commonly attributed to the recombination of charge carriers trapped in the surface states. Within a period of three weeks, a sharp peak at l_max= 402 nm, Dl_1/2= 40 nm, nm develops, while the intensity of the trapped emission significantly decreases (Fig. 1). After that period of time, the emission characteristics do not change if the solution is stored in the dark. Annealing of the starting dispersion at 70^o C for 10 hours, only broadens and shifts the absorption peak to 400 nm.

The kinetics of CdS growth is found to be quite slow, which, in general, simplifies the synthesis of uniformly sized nanoparticles^22,23. Typically, it takes about six hours until the UV-VIS absorption spectrum ceases changing. The sluggishness of the process can be attributed to the low concentration of the precursors as well as to the surface activity of 2-ethylhexanoate anion adsorbing on CdS. It also permits the calculation of the growth rate constant from the UV absorption data (Fig. 2a). A second order kinetic process provides adequate description for the evolution of UV absorption (Fig. 2b). A rate constant²⁴ averaged for three different wavelengths, 340 nm, 346 nm and 350 nm, is found to be 44±5 mol^-1 dm³ s^{-1 25}. The compliance with the second order kinetics indicates that interparticle collisions constitute the rate-determining step in a mechanism of ripening, which is probably not surprising when the NP are present at low concentration.

HR-TEM images give a visual representation of ethex/DMSO-stabilized CdS (Fig. 3). As expected, CdS NP are monocrystals with an average diameter close to that determined from UV-data. Notably, the size distribution on TEM images is broader than in the solution due to the destabilizing effect of DMSO removal. Interatomic distances match lattice spacings for zinc blende structure with d₂₀₀ = 2.35 Å, d₂₂₀ = 2.43 Å, d₁₁₁ = 3.30 Å, as determined from TEM images and electron diffraction. X-ray diffraction peaks (Fig. 4) are observed at 2q = 26.8^o (d = 3.32 Å), 28.7^o (d = 3.11 Å), 31.0^o (d = 2.88 Å), and 44.3^o (d = 2.04 Å). The signals at 2q = 26.8^o, 31.0^o, and 44.3^o correspond to the [111], [200], and [220] planes of cubic CdS, while the satellite peak at 2q = 28.7^o can be attributed to distorted [101] plane in hexagonally packed CdS of larger particles formed during the vacuum concentration of CdS dispersions for XRD measurements.

III. 2. ¹H-NMR.

For ethex/DMSO-d⁶-stabilized CdS, ¹H-NMR peaks observed are those arising from aliphatic protons in the 2-ethylhexanoate moiety. They are organized in four groups of signals located in the region between 0.4 and 1.8 ppm (Fig. 5). From the TOCSY spectrum of a sample of CdS NP bonded with ethex (Fig. 6), it is possible to obtain the corresponding resonance assignments. The most downfield peak at 1.74 ppm corresponds to the a-CH group (hydrogen 1), which is attached to the electron withdrawing carboxylate group. Particularly informative are the two well-resolved triplets at 0.81 and 0.75 ppm. Of these two triplets, the one resonating at 0.81 ppm is assigned to methyl-3, located on the ethyl chain. This assignment is based on the fact that this methyl group shows a scalar correlation (TOCSY cross peaks) only to methylene-2, which in turn shows a correlation to the a-CH group. On the other hand, methyl-7, located on the hexyl chain, resonates at 0.75 ppm and displays short and long range scalar correlations, e. g. to methylene-4, at 1.38 ppm, to methylene-6 and 5, and to the a-CH group at 1.74 ppm, thus clearly identifying this resonance as originating from a methyl group associated with the hexyl chain. Assignments of other signals are indicated in Fig. 5.

By comparison, in sodium 2-ethylhexanoate solution, the peaks from both -CH₃ groups of sodium 2-ethylhexanoate dissolved in DMSO-d⁶ are magnetically equivalent (superimposable chemical shifts) and produce a triplet centered at 0.81 ppm. In sharp contrast, the signal from the -CH₃ group in the hexyl chain shifts upfield to 0.75 ppm, when ethex is attached to nanoparticle, and back to 0.81 ppm, when ethex moiety is replaced by the modifier (Fig. 7). Consequently, it is concluded that the adsorption of ethex to NP changes the environment around the methyl-7, which indicates that this methyl group appears to be in the close proximity to the semiconductor core. Therefore, the chemical shift of methyl-7 demonstrates that ethex assumes a conformation presented in Fig. 5b, where the long hexyl chain spreads over the CdS surface, interacting with it presumably via van der Waals interactions. On the other hand, the ethyl side chain resides mainly in the solution. Note, in previous NMR studies covalently bound stabilizers in hydrophobic solvents are considered to be oriented perpendicularly to the CdS surface^14,16-19.

The NMR spectra of NP are likely to undergo a transformation upon surface modification with nucleophiles such as thiols and heterocycles. Indeed, the progressive modification of CdS with thiomolybdate anion, MoS₄^2-, causes the CH₃ triplets at 0.75 and 0.81 ppm to collapse gradually into a multiplet at 0.80 ppm (Fig. 7). Upon the modification of the NP surface, both CH₃ groups become magnetically equivalent, hence eliminating the shift in NMR lines corresponding to the methyl-7 group in NP-bonded ethex which is in the close proximity to CdS surface. Such behavior describes the partial replacement of ethex moieties from CdS surface by potent nucleophilic molecules such as thiomolybdate, and a concomitant change in the conformation of the CdS-adsorbed stabilizer. Within the degree of precision characteristic for NMR, one can conclude that the conformation change of ethex anion upon attachment of MoS₄^2- on the surface of NP is virtually complete. This includes both completely displaced molecules as well as the ones remaining in adsorbed state with the perpendicular orientation of the hexyl chains in respect to the CdS surface. It is necessary to point out that the modification with thiomolybdate strongly influences the electronic density of the NP surface. This factor must also affect the NMR line position. This effect is somewhat camouflaged by the straightening of the hydrocarbon chain and it should be more pronounced for carbon and hydrogen atoms in the close vicinity to CdS. However, their NMR signals overlap with those from other atoms (Figs. 5, 6) and are difficult to analyze.

The conformational character of the observed NMR line shifts can also be seen from the similarity of the spectral changes for structurally different surface modifiers. Besides thiomolybdate, several other compounds have been tested, which are expected to strongly bind to the CdS surface: 4,4'-bipyridine, 2-carboxythiophene, and trimethylamine. Notably, the nature of the observed changes in NMR signals for the CH₃ triplets at 0.75 and 0.81 ppm was identical for all these compounds (Fig. 7). The difference between these modifiers was only in the concentrations of a particular additive causing comparable NMR line shifts (Fig. 8). As such, in order to reach 0.04 ppm separation between the peaks, the modifier per nanocluster ratio, N, for 4,4'-bipyridine should be about 100 times greater than for thiomolybdate anion. This should be attributed to the difference in strength of the modifiers as nucleophiles: large negative charge of sulfur atoms in MoS₄^2- results in much greater affinity to the electrophilic cadmium sites on the CdS surface than that of 4,4'-bipyridine. Interestingly, the data points corresponding to surface modification with trimethylamine and 4,4'-bipyridine fall on approximately the same line in Fig. 8, whereas one may expect to see a stronger effect from the heterocycle. While being different as nucleophiles, other factors such as steric interactions with the protective shell of ethex molecules should also be considered, which may result in partial compensation of their nucleophilic strength.

CdS nanoparticles with narrow size distribution stabilized by 2-ethylhexanoate and DMSO were synthesized and characterized. The emphasis here was made on the ¹H NMR description of the conformation of the stabilizer molecule in the liquid state. The assignment of the observed signals was accomplished by using TOCSY 2D NMR spectra. One of the important features of the prepared dispersions is the observation of two separate triplets corresponding to the terminal methyl groups located on the ethyl and hexyl chains. This is indicative of a "wrap-around" conformation of the ethex molecules, which is opposite to the traditional representation of the stabilizer as being perpendicularly oriented to the surface of NP. Packing of the hydrocarbon chains can be perturbed by the surface modification with various nucleophiles. This results in the detachment of hexyl chains from the surface of CdS. This study demonstrates the potential of ¹H NMR for elucidation of fine details of the surface structure of protective organic shell around semiconductor NP.

This work was supported by NSF-CAREER, NSF-EPSCoR, Energy Research Center, and OSU Sensor Center. Funds for the 400 and 600 MHz NMR spectrometers for the Oklahoma Statewide Shared NMR facility were provided by the National Science Foundation (grant BIR-9512269), the Oklahoma State Regents for Higher Education, the W. W. Keck Foundation, and Conoco Inc. This work was partially supported by CONACyT (Projects 400313-5-1016P to DD and 25059-E to JR). One of us, DD, expresses his gratitude to the Fullbrigth-Garcia-Robles Foundation and to DGAPA-UNAM for the financial support. We also, thank Luis Redon from IFUNAM for the HR-TEM images. Two of us, DD and JR are grateful to Guillermo Mendoza for useful comments.

Fig. 1. UV-VIS absorption (1) and excitonic emission (2) spectra of CdS nanoparticles after storing in a light protected environment for three weeks.

Fig. 2. Temporal profile (a) and second-order kinetics (b) of UV absorbance, A, during the synthesis of CdS nanoparticles stabilized by DMSO-ethex at 340 nm (1), 346 nm(2), and 350 nm(3).

Fig. 3. Transmission electron microscopy image of a selected CdS nanoparticle. During the TEM specimen preparation, the extensive agglomeration occurs due to removal of DMSO serving as one of the stabilizers. The image was selected from one of those of separately lying single nanoparticles, on which the effect of solvent removal is believed to be minimal.

Fig. 4. X-ray diffraction of nanoparticles obtained after vacuum evaporation of the solvent. The numbers by the peaks indicate the position of their maxima (in degrees) and atomic plane assignments.

Fig. 5. (a) Proton nuclear magnetic resonance spectra of CdS nanoparticles stabilized by d₆- DMSO/ethex. (b) Conformation of ethex molecules on CdS nanoparticles. The numbers by the peaks indicate their assignments to corresponding protons of ethex group.

Fig. 6. TOCSY nuclear magnetic resonance spectrum of ethex groups adsorbed on CdS. The standard, one-dimensional NMR spectrum is stretched along the top-left to bottom-right diagonal of the TOCSY spectrum. The 2D peaks located on the both sides of the diagonal are the mirror images of each other. Their presence reflects the fact that the protons corresponding to the signals located on the point of intersection of the diagonal with horizontal and vertical lines drawn from the 2D peak, are cross-polarized, which indicates their spatial closeness.

Fig. 7. Dependence of the center-to-center peak separation between two terminal methyl groups on the concentration of the surface modifier: thiomolybdate (l), 4,4'-bipyridine (D), and trimethylamine (n).