Processing and optical properties of spin-coated polystyrene films containing CdS nanoparticles Andrk Chevreau, Brian Phillips, Brian G. Higgins and Subhash H. Risbud* Department of Chemical Engineering and Materials Science, University of California, Davis, CA 95616, USA The processing, microstructure and optical properties of CdS semiconductor nanoparticles sequestered in spin-coated polymer films has been investigated. A simple processing protocol has been developed to form thin film structures consisting of CdS nanoparticles dispersed in the interstices created by a close-packed stacking of polystyrene spheres. We have decoupled the synthesis of the CdS nanoparticles from the polymer matrix by synthesizing a colloidal suspension of CdS nanoparticles in water that is compatible with a polymer latex water suspension.In this way we are able to vary the CdS concentration in the polymer film over a wide range and create concentrations of quantum dot particles two orders of magnitude larger than those reported in previous work. In contrast to the conventional reaction synthesis of nanoparticles in a polymer matrix, the present method is versatile enough to create several useful thin film polymer-semiconductor structures using a straightforward, environment-friendly spin-coating technology. The key to achieving quantum confinement effects in semicon- ductor nanoparticles lies in the chemical synthesis of stable suspensions of particles smaller than the exciton (electron-hole pair) diameter. Once these quantum dots' are formed they must be placed, isolated from each other, in a matrix of glass, polymer or crystalline hosts.Various approaches have been reported in the literature to achieve this Most of the nanocomposite structures containing quantum dots consist of a somewhat random dispersion of nanoparticles determined by, for example, arrested nucleation in glasses, precipitation from sol-gel solutions or porous sites in zeolite cavities. This work deals with the processing challenges associated with spin- coating a dispersion of polystyrene and CdS nanoparticles to create organized arrays of quantum dots in a polymer matrix. Using spin-coating, multilayered colloidal coatings of pre-scribed thickness and with a chosen composition of CdS nanoparticles have been deposited.Packing density, long range order and defects in the colloidal film were controlled through wetting properties, drying conditions and particle-particle interactions. We present results that demonstrate the viability and potential of this technique and draw attention to the versatility of this process for tailoring optoelectronic quantum devices by size-selection of the nanoparticles within each monolayer of the polymer thin film. Because more is under- stood about the surface chemistry and processing protocols for systems of nanoparticles dispersed in thin films, the practi- cability of obtaining tailored film properties through prescribed processing is enhanced. We also discuss the possibility of preparing organized arrays of quantum dots by the deliberate placement of nanocrystallites in the interstitial sites of a two- dimensional close-packed monolayer template created from colloidal polystyrene particles.Experimental Synthesis of CdS nanocrystallites The method of homogeneous nucleation and growth of CdS nanocrystallites in N,N-dimethylformamide (DMF), as reported by Chemseddine and Weller,' was used. In a 500ml three-necked flask, 308 ml DMF, 3.859 g cadmium acetate, 0.642 g thiourea (TU) and 1.428 ml thioglycerol(3-sulfanylpro-pane- 1,2-diol; TG) were combined and stirred under nitrogen for at least 1 h. The solution temperature was gradually raised to the dissociation temperature of thiourea (120"C), at which point the solution became yellow. Then, 92ml of water was added, the temperature was lowered to 100 "C, and the mixture was allowed to reflux for 15 h.The dissociation of thiourea initiates the burst of nucleation necessary to grow particles. Thioglycerol is used as the capping agent for the particle. As the particles grow, the sulfanyl group of thioglycerol competes with dissociated sulfur from TU for bonding sites with cad- mium ions on the surface of the particle. This competition for surface sites, as well as the quantitative ratio, [Cd2+]/[TU], and the type of counter-ion present, can be used to control the sizes of the nanoparticles during growth.8 To narrow the particle-size distribution resulting from the homogeneous nucleation step, the colloidal suspension was size-fractionated using size-selective precipitation (SSP).'*9 A beaker containing the CdS suspension was stirred continuously in a sealed dessicator, in the presence of acetone vapour.Although acetone is miscible with DMF, it is a 'poor solvent' for the capped CdS particles, and as a result, larger CdS particles precipitate preferentially as acetone diffuses into the DMF solution. The extent of precipitation is determined by the mole fraction of acetone present. The amount of acetone placed in the dessicator with the CdS-DMF suspension was chosen based on trial-and-error visual observation of the amount of solids precipitated. To avoid over-precipitation in any single step, the amount of acetone used was approximately 10-20% of the volume of the CdS-DMF suspension.During this process, there also may be agglomeration of smaller particles to form larger ones that subsequently precipitate. After precipitation was observed, the CdS suspension was centrifuged and the supernatant was decanted off. The process was then repeated using the supernatant solution until the desired particle-size distribution was achieved. When the size distribution of CdS nanocrystals in DMF was narrowed satisfactorily, it was necessary to resuspend the CdS particles in water, since DMF is not compatible with the aqueous latex suspension. This was achieved by adding ethanol (a poor solvent for the particles) dropwise to the suspension to cause precipitation of CdS nanoparticles.The solution was centrifuged, decanted and the precipitate redissolved in water. There are several non-solvents that can be added dropwise to precipitate the CdS from this solution, including acetone and cyclohexanone. In this study we chose ethanol because these solids redissolved in water interacted most favourably with the polystyrene nanosphere suspensions in water. J. Muter. Chem., 1996,6( lo), 1643-1647 1643 Optical, TEM and NMR characterization of CdS nanocr ystallites A Cary 3 UV-VIS spectrophotometer from Varian was used to measure optical absorbance in liquid samples and deposited film samples The liquid samples were placed in 3 ml quartz cuvettes with a path length of 1cm Optical absorbance measurements were also taken for film samples, which were prepared on 3 81 cm diameter, 0 159 cm thick quartz wafers (from G M Associates, Inc) Optical excitation experiments were performed on both the cast and the spin-coated films containing CdS nanoparticles dispersed in a polystyrene matrix The photoluminescence (PL) spectra were obtained using a 100 fs self-mode-locked Ti sapphire laser Excitation pulses varying from 355 to 400 nm at a 82 MHz repetition rate were generated by fre- quency doubling through a KDP crystal Room-temperature PL spectra were recorded using an optical multichannel ana- lyser and appropriate bandpass filters Photoluminescence experiments were also conducted with tuned light from a xenon arc lamp striking the film surface at about 45" from perpendicular, after which it was detected at about 15" from perpendicular Transmission electron microscopy (TEM) of the CdS nano- particles was conducted using a Phillips 400 transmission electron microscope with a point-to-point spatial resolution of O34nm Samples suspended in water were placed on TEM grids and the solvent was allowed to evaporate prior to insertion in the microscope Both bright-field imaging and selected area electron diffraction were performed to determine the size and crystallinity of the CdS nanoparticles Cadmium- 1 13 single-pulse magic angle spinning (SP MAS) and 113Cd('H} cross-polarization MAS (CP MAS) spectra were obtained under ambient conditions at 66 5 MHz with a Chemagnetics CMX-300 spectrometer The sample spinning frequency was 4 5-5 kHz We report the chemical shifts relative to 0 1 mol dm-3 Cd(C10,)2 (aq), but an external sample of cadmium acetate was used as a solid-state reference (6 51 6) For CP, the transverse 'H field was ramped approximately & 10 kHz about the Hartmann-Hahn match, which was meas- ured using a sample of cadmium acetate Thin film processing The final working suspensions were prepared by adding the CdS water suspension to the polystyrene (PS) nanosphere suspension Monodispersed PS latex suspensions containing 38 nm nanospheres were prepared by diluting with ultrapure water a stock suspension purchased from Interfacial Dynamics Corporation The suspensions as received were stabilized elec- trostatically through the dissociation of surface carboxylate terminal groups, with a surface charge density of 18 4 pC cm-2 The coefficient of variation for the particle diameter of the stock suspension was specified by the supplier to be 14 1% Solutions of 0 1mol dmW3 NaOH were used in preparations requiring adjustment of pH Cast films were prepared by depositing 225 ml of the working suspension with and without base onto 3 81 cm diameter quartz wafers, and then allowing the liquid to evapor- ate under vacuum The films were then annealed at 100°C The quality of the cast films was generally poor The films tended to crack and flake on drying To improve the film quality, spin-coating was used A desired film thickness was achieved by spin-coating multiple layers of thickness approxi- mately one monolayer (cu 38 nm) After each coating step, the film was heated to 100°C to prevent wash-off during spin- coating No cracks or defects were observed in the final film under an optical microscope at 50x magnification Thick-nesses of cast and spin-coated films were measured using a Dektak 3030 profilometer For each film, three thickness measurements were taken, and the average value was reported 1644 J Muter Chem , 1996, 6(lo), 1643-1647 Results and Discussion Characterizationof CdS nanocrystallites UV-VIS absorption spectra provide a semi-quantitative meas- ure of particle size, as well as the size distribution lo l2 The location of the excitonic peak identifies the average size of the nanocrystallites, and the shape of the curve (FWHM) indicates the width of the size distribution Fig 1 shows UV-VIS spectra for the suspension directly after reaction and for the super- natants of size-selective precipitations 1, 5 and 9 (Sl, S5 and S9, respectively) The spectrum for the suspension after reaction shows one main shoulder, attributed to the first excitonic transition (HOMO-LUMO transition), and a second shoulder attributable to higher order transitions The location of the absorption maximum for the spectra from the size fraction- ations (Sl-S9) was essentially constant Based on quantum mechanical calculations correlated with experimental evidence from transmission electron microscopy,1o the average size of the optically active core of the CdS nanocrystallites synthesized in this work was estimated to be 2 5 nm Fig 2 shows absorbance spectra for the reacted solution, the ninth size-selective precipitation and the nanocrystals redispersed in water The spectrum for CdS in water shows the re-emergence of the shoulder for higher excitonic transitions which may have been obscured owing to absorbance by DMF in the size-selected fractions Usually these higher transitions are not visible at room temperature or for polydisperse samples 'lo Our spectra suggest therefore that our samples are 14, 250 300 350 400 450 wave1ength/nm Fig.1 Evolution of CdS excitonic peak with size fractionation Suspension after reaction (R), after first size-selective precipitation (Sl), after fifth size-selective precipitation (SS) and after ninth size- selective precipitation (S9) 14 12 a10 gi06 04 02 0 250 300 350 400 450 wavelengthhm Fig.2 UV-VIS absorption of CdS nanocrystallites after reaction (-), after ninth size-selective precipitation (---) and after redispersion in water (--- ) monodisperse with a high degree of crystallinity for all CdS nanoparticles in the suspension. A further decrease in the FWHM from 40.1 nm for the final size-selected nanoparticles in DMF to 36.9nm for the absorption of CdS in H,O is evident in Fig. 2. Analysis of transmission electron micrographs revealed an average CdS nanocrystallite size of about 4 nm. This obser- vation is in accord with size estimates of the CdS core of the nanocrystallite based on optical absorption measurements plus the added size from the attached thioglycerate ligands at the surface of the nanoparticles.Analysis of electron diffraction patterns gives d spacings corresponding to a cubic crystal structure of CdS. This result is consistent with data in the literature showing that solution-precipitated CdS nanoparticles have a cubic structure,' while similar particles formed in reheated glasses have a hexagonal crystal structure;' nanopart- icles synthesized in situ in polymer matrices show both types of crystalline structures.* Solid-state NMR spectroscopic characterization Both the SP and CP MAS spectra of nanocrystalline CdS show a relatively broad central peak near 6 + 700 with spinning sidebands as shown in Fig.3. This chemical shift is very similar to that reported previously for bulk crystalline CdSi3,l4 and suggests that crystallite size does not greatly affect the average chemical shift. However, the peak widths (62 f5 ppm FWHM, for both SP and CP MAS) are considerably greater than for bulk CdS.'3,'4 The large MAS peak widths could arise from a distribution of chemical shifts, due to a range of structural environments or dispersion of Knight shifts (k,frequency shift arising from conduction-band electrons). Additional work is needed to confirm the source of the line-broadening, although spatial inhomogeneities in the conduction-band electron density might be expected from the small cluster sizes.Although we expect the CP MAS experiment to give a signal preferentially from the Cd atoms nearest the thioglycerol capping agents, we found only subtle differences between the CP MAS and SP (bulk) NMR spectra. The principal differences are the slightly more shielded peak position (cu. 10ppm) and broader spinning sideband envelope observed by CP MAS. Only a small chemical shift anisotropy has been reported for bulk CdS (54 ppm),I3 and the small spinning sideband envelope observed in the SP spectrum is consistent with this value. The I\ 1200 1000 800 600 400 200 6 Fig. 3 Solid-state '13Cd CP MAS (a) and single-pulse (b) spectra of CdS nanocrystals. (a) 2 ms contact time, 2 s recycle delay, 40000 transients; (b)5.5 ps pulses (90"),100 s recycle delay, 1600 acquisitions.larger envelope observed in the CP MAS spectrum implies that the Cd atoms nearest the thioglycerol capping moieties experience greater chemical shift or Knight shift anisotropy, which might be expected for Cd near the particle surface. Modifying suspension chemistry to reduce CdS-PS particle-particle interactions As the loading of CdS was increased relative to PS, the solution became noticeably turbid when the [PS]/[CdS] mass ratio was <30, with a solid precipitate eventually settling out after several hours. For mass ratios in the range lo< [PS]/[CdS] <30 the absorption peak was still present, but was reduced considerably, and the background absorption due to scattering was increased considerably (see Fig.4). It seems plausible that the thioglycerate groups used to cap the CdS nanoparticles destabilize the working suspension, and thereby cause flocculation. This flocculation may occur if the OH group on the thioglycerate is sufficiently acidic to pro- tonate a carboxylate group on the polystyrene. In addition, hydrogen bonding between surface thioglycerate groups on CdS and carboxylate groups on PS may cause enough CdS particles to adsorb to the PS particle surface to shield the surface charge and cause flocculation. If these scenarios are correct, then it is necessary to increase the number of dis- sociated carboxylate groups at the PS particle surface to ensure that the working suspension is stabilized at higher CdS concen- trations.This can be accomplished easily with the addition of base, such as NaOH. The amount of OH-required to ionize a certain percentage of surface carboxylate groups can be estimated from the surface charge density of the PS particles. The effects of NaOH addition on the absorption spectra of the working suspension are summarized in Fig. 5. Without the addition of base the absorption peak of the first excitonic transition for [PS]/[CdS] = 10 is barely visible above the background. With the addition of base, the turbidity was reduced considerably, and the absorption peak became visible. This method of colloidal film formation allows for the selection of specific film properties. Various sizes of optically active semiconductor material and polymer colloid may be chosen and stabilized in a wide range of concentrations.CdS quantum-dot concentrations increased by orders of magnitude over previous work have been achieved. Table 1 summarizes some ways to look at the CdS concentration in the present films compared to values for comparable systems found in the literature. The film from this work used in the comparison has [PS]/[CdS] = 10 on a mass/mass basis, and the precur- sor suspension was stabilized by 4 x mol dm-3 OH-. It should be noted that this concentration of CdS is not an upper limiting value, but was chosen here as the most concentrated in CdS. 1.8 1.6 1.4 0.4 0 1 250 300 350 400 450 wavelengthhm Fig. 4 Absorption spectra of suspensions with varying [PS]/[CdS] mass ratios.--, [PS]/[CdS] = 10; ---, [PS]/[CdS] =50. J. Muter. Chern., 1996, 6(lo), 1643-1647 1645 1.6 1.4 1.2 81 t ([I 0.8 sD ([I 0.6 0.4 0.2 4 0 300 350 400 450 wavelength/nm Fig.5 Effect of the addition of base to suspensions containing both thioglycerate-capped CdS nanocrystals and carboxylate-capped PS nanospheres. --, [PS]/[CdS] = 10,4 x mol dm-3 OH-.7 ---, [PS]/[CdS] = 10, water added. Table 1 CdS concentrations in the spin-coated films compared with similar systems reported previously spin-coated films previous results ref. mass% 9.67 1-7 15-17 vol% 9.61 0.092-1 4, 6, 8, 18 q-dots cm-3 4.28 x 1OI8 3 x 10l6 2 q-dots cm-3 2.04 x 1015 (5.1-5.9) x lOI3 19 Optical activity of CdS nanocrystallites in polystyrene films UV-VIS absorbance measurements on the dried cast films showed no characteristic excitonic absorbance.When these films were heated to 100°C, just above the glass-transition temperature for the PS nanospheres, the excitonic absorption peaks became visible. The spectra are shown in Fig. 6. In the absence of base, the absorption spectrum exhibits a broad plateau that begins near the absorption band. When base is added the absorption peak is revealed. UV-VIS absorption spectra of the spin-coated films were measured after deposition of each layer. As layers are built up, a characteristic excitonic peak becomes more pronounced. Note that there is no visible absorption peak after the first layer is coated, owing to the low quantity of optically active CdS nanocrystallites in a single coating.As additional layers 1.4 1, \ a,05 0.8 \ \ I42 \ 8 0.6a (d 0.4 0.2 I 250 300 350 400 450 wavelengthhm Fig. 6 UV-VIS absorption in cast films with and without charge stabilization after heating. --, [PS]/[CdS] = 10, 4 x mol dmP3 OH-(approx. 4.77 pm); ---, [PS]/[CdS] = 10, 0 mol dm-3 OH-(approx. 13.33 pm). 1646 J. Muter. Chern., 1996,6(lo), 1643-1647 0.05 0.04 a, 0.03 g2 0.02 (d 0.01 ol I 250 300 350 400 450 wavelengthhm Fig. 7 UV-VIS absorption in step-wise spin-coated PS/CdS composite film ([ PS]/[CdS] =20, 1000 rpm, approx. 142 nm). (a) after one coat- ing; (b)after five coatings. 0.006 -0.005 cnc.-t7 0.004 -:0.003 .-22 v)s 0.002 c..-c 0.00 1 0 400 500 600 700 800 wavelength/nm Fig. 8 Representative photoluminescence spectra for spin-coated PS/CdS composite films ([PS]/[CdS] =20) showing peak lumi-nescence at ca. 540 nm: (a) 200 rpm, (b)500 rpm, (c) 1000 rpm are deposited the absorption peak becomes evident, as shown in Fig. 7. Luminescence spectra for the spin-coated films are shown in Fig. 8. The spectra show a maximum intensity around 540 nm. Spectra from charge-stabilized suspensions show the narrowest emission linewidths. There was an order of magnitude difference in measured luminescence intensity for the spin-coated films, depending on the deposition protocol and on the amount of material laid down.The nature of the luminescence from spin-coated films made from charge-stabilized cosuspensions closely resembles 0.0007 , 0.0006 n.$ 0.0005 3$ 0.0004 v 0.0003 C Q,c1 .c 0.0002 0.0001 0 I-400 500 600 700 800 wavelength/nm Fig. 9 Detail of luminescence spectra for 500 and 1000 rpm films [(b) and (c) respectively from Fig. 81 showing the same emission linewidths that for cast films from charge-stabilized suspensions, indicat- ing retention of the narrow emission linewidth characteristics of a system which has uniformity of electro-optical properties in the quantum-confined CdS. Summary and Conclusions The viability of this processing scheme is exemplified by the ease with which film properties may be tailored, including CdS nanoparticle concentration and film height.The spin-coating of multiple layers presents a quick and reliable way to make durable, high-quality films in which the unagglomerated elec- tro-optical nature of the CdS nanoparticles is retained. Small chemical shift anisotropies observed in the CP MAS NMR spectra are consistent with the presence of hydroxy groups near surface groups on the CdS nanoparticles. These groups linked to the thioglycerol capping agents present a plausible explanation for the agglomerate-free nanoparticles synthesized in this work. These films have optical measurements which suggest that the CdS particles have not aggregated. It will be of interest to determine how the CdS nanoparticles are actually dispersed within the colloidal coating.The film thickness measurements for the 1000rpm film suggest that each layer is a single monolayer of particles. Previous work on PS particles by Steiner2' has shown that it is possible to sequester nanoparticles in the interstices of the two-dimensional close-packed monolayer template formed from colloidal PS. Electro-optical properties may be maximized by appropriate organization of semiconductor nanocrystallites into regular arrays. Our method also allows us to deposit multilayers of different PS particles with different CdS nanopar- ticles, and thereby to prepare stacked monolayers in which each layer has a different interstitial CdS nanoparticle size, allowing for the tuning of the optical properties of each layer.This work is based on Andre Chevreau's M. S. thesis and was supported by the MRSEC Program of the National Science Foundation under Award No. DMR-9400354. The NMR work was performed in laboratory facilities supported by the W. M. Keck Foundation. We are also grateful to Valerie Leppert (TEM work) and Christine Smith (luminescence work) for their assistance. References 1 A. Chemseddine and H. Weller, Ber. Bunsen-Ges. Phys. Chem., 1993,97,636. 2 K. Misawa, H. Yao, T. Hayashi and T. Kobayashi, J. Chem. Phys., 1991,94,4131. 3 V. Colvin, A. Goldstein and A. Alivisatos, J. Am. Chem. Soc., 1992, 114,5221. 4 Y. Wang, A. Suna, J. McHugh, E. Hilinski, P.Lucas and R. Johnson, J. Chem. Phys., 1990,92,6927. Y. Wang and W. Mahler, Opt. Commun., 1987,61,233. E. Hilinski, P. Lucas and Y. Wang, J. Chem. Phys., 1988,89,3435. L. C. Liu and S. Risbud, J. Appl. Phys., 1994,76,4576. Y. Wang, N. Herron, W. Mahler and A. Suna, J. Opt. SOC.Am. B, 1989, 6, 808. 9 C. B. Murray, D. J. Norris and M. G. Bawendi, J. Am. Chem. SOC., 1993,115,8706. 10 H. Weller, Angew. Chem., Int. Ed. Engl., 1993,32,41. 11 P. Lippens and M. Lannoo, Phys. Rev. B, 1989,39,10935. 12 H. Weller, H. Schmidt, U. Koch, A. Fojtik, S. Baral, A. Henglein, W. Kunath, K. Weiss and E. Dieman, Chem. Phys. Lett., 1986, 124, 557. 13 A. Nolle, Z. Naturforsch., Teil A, 1978,33,666. 14 P. DuBois Murphy and B. C. Gerstein, J. Am. Chem. Soc., 1981, 103,3282. 15 K. Choi and K. Shea, Chem. Muter., 1993,5,1067. 16 S. Yanagida, T. Enokida, A. Shindo, T. Shiragami, T. Ogata, T. Fukumi, T. Sakaguchi, H. Mori and T. Sakata, Chem. Lett., 1990,1773. 17 S. Yamazaki and Y. Kurokawa, Polym. Commun., 1991,32,524. 1s 0.Salata, P. Dobson, P. Hull and J. Hutchison, Thin Solid Films, 1994,251, 1. 19 R. Vogel, K. Pohl and H. Weller, Chem. Phys. Lett., 1990,174,241. 20 M. L. Steiner, PhD Thesis, University of California, Davis, 1996. Paper 6/03393H; Received 15th May, 1996 J. Mater. Chem., 1996, 6(lo), 1643-1647 1647