Semiconductor Quantum Dots

Organometallic and Inorganic Synthesis

By Mark Green

The Royal Society of Chemistry

Copyright © 2014 Mark Green
All rights reserved.
ISBN: 978-1-84973-985-6

Contents

Chapter 1 The Preparation of II–VI Semiconductor Nanomaterials, 01,
Chapter 2 The Preparation of III–V Semiconductor Nanomaterials, 53,
Chapter 3 The Preparation of IV–VI Semiconductor Nanomaterials, 77,
Chapter 4 The Preparation of Other Chalcogenides and Pnictide Nanomaterials, 92,
Chapter 5 The Synthesis of Core/Shell Quantum Dots, 113,
Chapter 6 Ligand Chemistry, 166,
Chapter 7 The Use of Single-Source Precursors in Nanoparticle Synthesis, 224,
Subject Index, 252,

CHAPTER 1

The Preparation of II–VI Semiconductor Nanomaterials

1.1 Origins of Organometallic Precursors

Although nanoparticles (notably metals) can be traced to antiquity, the origins of modern semiconductor quantum dots (QDs) is a much later development. Specific advances in quantum-confined semiconductor particles can be traced to Grätzel, who reported the synthesis of colloidal CdS to examine photo-corrosion, and Brus who notably reported the band edge luminescence of the same material. At approximately the same time, Ekimov reported quantum confinement in CuCl prepared in a silica glass, while Henglein, an early pioneer of colloidal semiconductors, reported the synthesis of CdS on colloidal SiO2. Fendler then reported the use of a reverse micelle to prepare CdS nanoparticles, which was importantly improved on by Henglein, who used polyphosphates as a well-defined passivating agent, allowing nanoparticles to be processed and redispersed.

The work described above utilised mainly inorganic salts as precursors in aqueous-based reactions. Consequently, the low temperatures used and the presence of air and water often resulted in polydispersed materials with relatively poor optical and crystalline properties. The small number of available suitable precursors also reduced the number of systems that could be explored. The need to improve these reactions by removing air and water dictated the use of organic-based starting materials; however, inorganic reagents such as Na2Se could not easily be used in organic solvents so alternatives were required.

Independently, Steigerwald pioneered many of the early chemical routes to bulk solid-state materials, inspired by the early work on the deposition of semiconductors by metal organic chemical vapour deposition (MOCVD). In a typical MOCVD reaction, metal alkyls and arsine gas were used; precursors that are hypothetically suitable in the preparation of nanomaterials in an organic solvent, but realistically improbable. Although effective in vapour processes, precursors designed for gaseous-phase reactions are not ideal for use in solution chemistry because of their toxic nature, air sensitivity, and the associated difficulties in handling. In addition, metal alkyls such as dimethyl- or diethyltelluride are also too stable to be effective precursors at temperatures as low as 200 °C, the region of interest for solution synthesis.

Initial reports on alternative precursors described the suitability of trimethylphosphine (Me3P) as a tellurium transport agent and as a potential replacement for dialkyltellurides. The reaction between triethylphosphine telluride and mercury metal was investigated, giving mercury telluride (HgTe) in almost quantitative yields. This reaction suggested the assignment of the phosphine telluride as a single coordinate complex of Te(0). Since elemental metals are not normally utilised as precursors, the reaction was repeated using the metal alkyl diethylmercury (Et2Hg) and the less volatile diphenylmercury (Ph2Hg). Refluxing an organic solution of the metal alkyl and a phosphine telluride complex lead to the preparation of HgTe via the formation of an intermediate, Hg(TePh)2. It was also observed that Hg(TePh)2 eliminated mercury metal during the reaction, which immediately reacted with excess phosphine telluride giving HgTe. This reaction is noteworthy due to the insertion of Te atoms into a Hg–C bond and highlights the important role phosphines can play in the low-temperature solution synthesis of semiconductors.

Steigerwald also prepared cadmium chalcogenides using silylated precursors and metal alkyls, providing a solution-based organometallic route to CdSe. The reaction between metal alkyls and hydrogen chalcogenide gases had previously been reported, but most silylated chalcogens exist as a liquid at room temperature and are therefore easier to handle than a vapour-phase precursor. The reaction proceeded in simple organic solvents providing an amorphous red/brown solid, which could be converted to bulk crystalline CdSe upon annealing at 400 °C. The work was extended to cover ZnSe and CdTe. The solvent was found to have a distinct effect on the reaction rate; dealkylsilylation in dichloromethane proceeded almost instantaneously, while reactions in toluene took days to complete and reactions in saturated hydrocarbons required weeks. Interestingly, there was no reaction between Me2Cd and the silylated precursor in the absence of solvent. Although the materials resulting from these reactions were not necessarily nanometric in size, this molecular precursor chemistry can be thought as the origin of the solution-based organometallic routes to nanomaterials.

1.2 Inverse Micelles

Steigerwald then used the developed chemistry to prepare discrete CdSe nanoparticles in inverse micelles, which can be thought of as the first pseudo-organometallic route to nanomaterials. A dioctyl sodium sulfosuccinate/water/heptane microemulsion was prepared and an aqueous solution of cadmium perchlorate added, followed by a heptane solution of Se(SiMe3)2. The room-temperature reaction yielded particles of CdSe, which had to be reduced to dryness to remove water and avoid flocculation. The powders were then redissolvable in hydrocarbons. Addition of Cd2+ stock solution followed by phenyl(trimethylsilyl) selenide resulted in a phenyl-passivated surface, following which, the particles could be isolated by centrifugation. Importantly, this is the first example of monomer passivation of a nanoparticle surface and the phenyl-passivated clusters were soluble in pyridine, but insoluble in petroleum ether. Absorption spectroscopy showed materials with a shift in bandgap with decreasing particle size and a slight excitonic shoulder. The paper also reported growth of the particles with further addition of precursor, demonstrating Ostwald ripening-type growth. Following isolation, annealing of these particles in 4-ethylpyridine improved the crystallinity giving crystalline zinc blende (cubic) type clusters. Annealing of small (20 Å) particles in a mixture of tributylphosphine (TBP) and tributylphosphine oxide (TBPO) resulted in further particle growth to ca. 40 Å, giving particles with a wurtzite (hexagonal) crystalline core. This highlighted the suitability of phosphine oxides as capping agents and was used as the basis for further work.

Clusters of CdSe prepared by the inverse micelle route were then used in core/shell studies, where a ZnS shell was deposited on the surface of the particles, inorganically passivating the surface. The use of an inorganic layer rather than a surfactant to protect the surface is an area of immense interest, as the inorganic layer is not restricted by factors such as the surfactant cone angle, is generally more stable and is a more complete protecting layer. By choosing the correct materials, specific heterojunctions with engineered band mismatches can be prepared, which will be discussed later. Once the layer of ZnS had been deposited on the CdSe core, the emission spectrum, previously broad with luminescence attributed to surface defect states, became more band edge, suggesting blocking of the defects. In this case, the sulfide shell was deposited using inorganic rather than organo-metallic reagents: however, once the surface trapping states had been removed, the band edge emission observed originated from a nanoparticle core prepared by organometallic-based precursors.

1.3 Organometallic Routes to CdE (E = S, Se, Te)

The seminal paper describing solution-based organometallic routes to nanomaterials was published in 1993, and reported the first totally organometallic/organic-based synthesis of CdE (E = S, Se, Te) nanoparticles which resembled a solution analogue of the MOCVD process, using an inert atmosphere, appropriate precursors with origins in vapour deposition and coordinating solvents suitable for high-temperature reactions. Although organometallic precursors had previously been investigated as discussed, this route relied entirely on organometallic reagents thermolysed directly in a hot coordinating solvent (long-chain phosphine oxides, such as tri-n-octylphosphine oxide, TOPO) rather than a room-temperature reaction in an inverse micelle. The precursors employed, Me2Cd and either a silylated chalcogenide or a chalcogen dissolved in trioctylphosphine (TOP), were chosen with reference to the work of Steigerwald, and the route elegantly incorporated both the precursor (phosphine chalcogenide) and capping agent (phosphine oxide) chemistries described above. It should be assumed that all reactions and preparations described hereafter are carried out under inert atmosphere conditions unless stated otherwise.

Cadmium selenide (CdSe) is generally considered the prototypical QD material as size quantisation effects result in tuneable emission across the entire visible range of the electromagnetic spectrum, making the material attractive for a wide range of optoelectronic applications. A notable report by Donegá et al. highlighted the importance of reaction temperature, reagents and the ratio of reagents. By carefully selecting optimum condition, CdSe particles with quantum yields as high as 85% were prepared. In contrast, CdS, one of the earliest materials to be studied because of its ease of preparation, demands less attention as it displays only small changes in the optical properties when prepared on the nano scale. This can be attributed to the difference in Bohr radius of the exciton (CdSe, aB = 32 Å; CdS aB = 19 Å) which means CdS particles have to be significantly smaller than CdSe to exhibit size quantisation effects and therefore display a smaller shift in the band edge. (In comparison, TiO2, a wide-bandgap semiconductor, has an excitonic radius of only 8 Å and would require crystals to be little more than clusters to display any optical effects from the confinement of charge carriers. TiO2 displays little, if any, shift in the optical band edge. In this case, size quantisation effects are manifest as variations in the oscillator strength.)

In a typical reaction, Me2Cd and a trioctylphosphine chalcogenide (such as trioctylphosphine selenide, TOPSe) were dissolved in TOP. The Lewis base phosphine, a liquid at room temperature, served as both a solvent for precursor delivery and a capping agent once the nanoparticles had formed. The precursor solution was then injected into TOPO (which had been rigorously dried and degassed), under an inert atmosphere at temperatures of between 100 and 350 °C, with lower temperatures producing smaller particles. The sudden introduction of reagents into a hot solvent and the subsequent immediate supersaturation resulted in the formation of nuclei. The drop in temperature after the injection of room-temperature reagents prevented further nucleation, and further heating resulted in growth of particles by Ostwald ripening. The sudden nucleation and slow growth steps, originally described by La Mer, resulted in a monodispersed product. The surfactants, TOPO and TOP coordinated to the surface of the nanoparticles, providing physical and electronic passivation. The labile nature of the surfactants is a key requirement, desorbing from the particle surface to allow growth, yet coordinating strongly enough to allow particle isolation and provide the required protection for the nanoparticle.

After injection and during the growth stage, the reaction could be monitored by removing aliquots and recording the emergence of a band edge via absorption spectroscopy, and the growth could also hence be tuned by altering the reaction temperature. After the particles had grown to the required size, the reagents were then left to cool to ca. 60 °C, followed by addition of a polar solvent (termed a non-solvent) such as methanol, which induced precipitation. The precipitate, collected as a waxy powder by centrifugation, was dispersed in non-polar solvents producing an optically clear solution. Addition of small amounts of non-solvent increased the average polarity of the solvent and resulted in the precipitation of the larger particles. This size-selective precipitation resulted in a solution of nanoparticles with an extremely narrow size distribution.

The materials prepared were monodispersed (<5% standard deviation), 1.2–11.5 nm in diameter and slightly prolate with aspect ratios up to 1.3. The particles were also crystalline, capped with a monolayer of surfactant molecules (TOPO and TOP) and displayed excellent optical properties. Figure 1.1 shows typical absorption spectra from CdE (E = S, Se, Te) nanoparticles prepared by this method, with the first excitonic transitions clearly visible. The emission from TOPO-capped particles was initially band edge without the need for a further inorganic shell, highlighting the high optical quality of materials prepared under an inert atmosphere.

In this case, the steric properties of TOPO/TOP play a key role; reducing the chain length to butyl groups or smaller resulted in uncontrolled growth. The actual coordination of the capping agent to the crystallite slowed growth kinetics, allowing for the controlled growth at elevated temperatures. The affinity of the phosphine oxide group for the surface metal (controlled by tuning the chain length of the alkyls chains) also effected the growth rate; the increased Lewis base character, i.e., the more electron donating, the stronger the binding and the slower the growth. The presence of the capping agent was also essential for the band edge emission and relatively high initial quantum yields of ca. 10% (for CdSe) were observed, which dropped by orders of magnitude upon removal of the surface ligand. Notably, it has since been discovered that the emission quantum yield of QDs can be increased up to a maximum of 75% via the surface treatment with sodium borohydride (NaBH4). The addition of small amounts of the reagent reduced the ligand, removing the surfactant while allowing the surface cadmium to oxidise, yielding a cadmium oxide layer that enhanced the emission. Prolonged exposure or an excess of reducing agents resulted in the precipitation of the nanoparticles, although addition of the optimum amount of NaBH4 was achievable and the treated nanoparticles were found to be stable for up to a year in ambient conditions.

The surface ligand could be removed by refluxing the nanoparticles in pyridine, which resulted in ligand exchange, removing the TOPO and leaving the more labile pyridine coordinated to the particle surface. This was initially used as in intermediate for other capping agents and will be discussed later. Murray also reported that although Lewis base ligands are generally utilised for capping agents, Lewis acids, such as alkylboranes and alkylaluminium species could also be used to passivate the surface.

The particles were crystalline as determined both by powder X-ray diffraction (XRD) and by transmission electron microscopy (TEM). The XRD patterns displayed reflections consistent with wurtzite (hexagonal) structured nanoparticles, with the distinction between hexagonal and cubic structure being lost in particles below ca. 2 nm in diameter (Figure 1.2). Modelling studies of the diffraction patterns suggested each nanoparticle contained one stacking fault.

This route quickly became the standard method to prepare high-quality II–VI nanoparticles, with CdSe becoming the most studied material. A notable amendment to the preparation of CdSe was reported by Bowen Katari, where the selenium was dissolved in TBP rather than the longer TOP in an attempt to achieve a higher surface coverage, although comparisons between QDs prepared by both methods revealed little difference. In an attempt to prepare larger samples with absorption features beyond ca. 580 nm, the synthesis temperature was increased to 350 °C, which was followed by several minutes' continuous heating at ca. 320 °C after precursor injection. It is worth noting, however, that TOPO slowly decomposes above 330 °C. The nanoparticles prepared were larger, but possessed a larger size distribution. This was remedied by adding the precursor stock solution in 0.1 mL amounts. Particles prepared by this route were more spherical24 and had fewer stacking faults, although small variations in temperature resulted in large differences in size distribution.

The actual size range of particles from a single reaction is a key parameter and the literature offers conflicting reports as to which route provides the lowest size distribution. Alivisatos addressed the issue, based on the Gibbs–Thompson equation and reported a detailed study on the kinetics and growth model of CdSe particles, suggesting that diffusion-limited growth could reduce the size distribution if the diffusion area and particle size were considered. Alivisatos highlighted that the kinetics of crystal growth were also dependent on the variation of surface energy with size, and showed prolonged growth led to growth focusing, refocusing and defocusing (Ostwald ripening), with an intimate link to monomer concentration. During the focusing stage, the high concentration of the monomers exceeds the solubility of the nanoparticles: all particles grow, but the smaller particles grow faster than the larger ones and the size distribution can be focused to an almost monodispersed sample. Below a specific monomer concentration, the larger particles grow at the expense of the smaller ones, defocusing the size distribution. This highlighted that the size distribution could be controlled experimentally by maintaining a high monomer concentration. Further nucleation studies suggest the growth of CdSe nanoparticles is reaction limited, which is supported by the earlier detailed study by Dushkin et al. To complement experimental work, detailed theoretical studies of particle growth based on Monte Carlo simulations have also been carried out which make suggestions as to the favoured growth regime. Various models have been utilised in examining the kinetic and thermodynamics of particle growth; the different stages of nucleation and growth have been discussed in the context of classical nucleation theory, and a barrier diffusion model has also been developed that describes the growth kinetics in different solvents, highlighting Arrhenius behaviour dictated by solvent activation energies, which increase with molecular weight. The kinetics of nanocrystal growth is complicated and covers many variables, yet if controlled it can be tuned to allow the growth of different-sized particles. For example, controlling the growth of CdSe particles is possible by choosing ligands which form complexes of differing solubility, that act either as nucleating agents (resulting in higher particle yields and smaller particles) or as growth agents (resulting in early time ripening), allowing extremely small particles to be prepared and stabilised. Other work has suggested that the presence of water and oxygen in the reaction mixture can lead to the etching of the small particles that aggregate into larger particles during the Ostwald ripening process.

(Continues...)Excerpted from Semiconductor Quantum Dots by Mark Green. Copyright © 2014 Mark Green. Excerpted by permission of The Royal Society of Chemistry.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.