Aqueous Based Semiconductor Nanocrystals

Preparing to load PDF file. please wait...

0 of 0
100%
Aqueous Based Semiconductor Nanocrystals

Transcript Of Aqueous Based Semiconductor Nanocrystals

Review pubs.acs.org/CR

Aqueous Based Semiconductor Nanocrystals
Lihong Jing,†,∥ Stephen V. Kershaw,‡,∥ Yilin Li,§ Xiaodan Huang,†,⊥ Yingying Li,†,⊥ Andrey L. Rogach,*,‡ and Mingyuan Gao*,†,⊥
†Institute of Chemistry, Chinese Academy of Sciences, Bei Yi Jie 2, Zhong Guan Cun, Beijing 100190, China ‡Department of Physics and Materials Science & Centre for Functional Photonics, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon, Hong Kong SAR §Key Laboratory of Carcinogenesis and Translational Research, Department of GI Oncology, Peking University Cancer Hospital and Institute, Fucheng Road 52, Beijing 100142, China ⊥School of Chemistry and Chemical Engineering, University of Chinese Academy of Sciences, Beijing 100049, China
ABSTRACT: This review summarizes traditional and recent nonconventional, bioinspired, methods for the aqueous synthesis of colloidal semiconductor quantum dots (QDs). The basic chemistry concepts are critically emphasized at the very beginning as these are strongly correlated with the selection of ligands and the optimal formation of aqueous QDs and their more sophisticated structures. The synergies of biomimetic and biosynthetic methods that can combine biospecific reactivity with the robust and strong optical responses of QDs have also resulted in new approaches to the synthesis of the nanoparticles themselves. A related new avenue is the recent extension of QD synthesis to form nanoparticles endowed with chiral optical properties. The optical characteristics of QD materials and their advanced forms such as core/shell heterostructures, alloys, and doped QDs are discussed: from the design considerations of optical band gap tuning, the control and reduction of the impact of surface traps, the consideration of charge carrier processes that affect emission and energy and charge transfer, to the impact and influence of lattice strain. We also describe the considerable progress in some selected QD applications such as in bioimaging and theranostics. The review concludes with future strategies and identification of key challenges that still need to be resolved in reaching very attractive, scalable, yet versatile aqueous syntheses that may widen the scope of commercial applications for semiconductor nanocrystals.

CONTENTS
1. Introduction 2. Basic Chemistry of Aqueous Synthesis
2.1. Thermodynamic Aspects 2.1.1. Solubility Product 2.1.2. pH
2.2. Chemistry for Forming Semiconductors 2.3. Surface Chemistry of Semiconductor NCs 3. Aqueous Synthesis of Semiconductor NCs 3.1. Historical Overview of Aqueous Synthesis 3.2. Growth Mechanism of Aqueous NCs
3.2.1. Ligand Effects on Growth Kinetics 3.2.2. pH Effects on Growth Kinetics 3.2.3. Ionic Strength Effect on Growth Kinetics 3.3. Core/Shell Structured NCs 3.3.1. Thiol Ligand Decomposition Route 3.3.2. Surface Cation Exchange Route 3.3.3. Direct Growth Route with Auxiliary
Cationic/Anionic Precursors 3.4. Alloyed NCs
3.4.1. Direct Synthesis 3.4.2. Ion Exchange 3.4.3. Core/Shell Transformation and Ther-
mally Activated Indiffusion

10624 10626 10626 10626 10627 10627 10628 10629 10629 10630 10632 10635 10636 10636 10637 10638
10639 10641 10641 10646
10648

3.5. Doped NCs 3.5.1. Core Doping 3.5.2. Shell Doping
3.6. Nonconventional Aqueous Syntheses of Semiconductor NCs
3.6.1. Biomimetic Synthesis 3.6.2. Biosynthesis 4. Optical Properties of Quantum Dots 4.1. Fundamental Aspects 4.1.1. Quantum Confinement Effect 4.1.2. Radiative Recombination 4.1.3. Trap States 4.1.4. Type I Core/Shell 4.1.5. Type II Core/Shell 4.2. Surface Passivation 4.2.1. Growth Kinetics Control over Surface
Passiviation 4.2.2. Ligand-Mediated Surface Passivation 4.2.3. Removal of Trap States by Photoillumi-
nation

10650 10651 10653
10655 10655 10660 10661 10661 10661 10662 10663 10668 10669 10669
10669 10670
10671

Special Issue: Nanoparticle Chemistry
Received: January 19, 2016 Published: September 2, 2016

© 2016 American Chemical Society

10623

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

4.2.4. Removal of Trap States by Fabricating Core/Shell Structures
4.3. Band Gap Engineering by Fabricating Type II Core/Shell Structures
4.3.1. Influence on Band Gap and Energy Levels
4.3.2. Effect on Absorption Spectra 4.3.3. Effect on Emission Spectra 4.3.4. Effect on PL QY and Lifetimes 4.3.5. Influence on FRET 4.3.6. Charge Transfer 4.3.7. Relation to Auger Recombination 4.3.8. Other Variations on Type II Heterostruc-
tures 4.4. Emissive Doping
4.4.1. Mn Doping 4.4.2. Cu Doping 4.4.3. Lanthanide Doping 4.5. Optical Chirality 4.5.1. Case I: Chiral Cores 4.5.2. Case II: Surface Chirality by Distortion or
Chiral Surface Bonding 4.5.3. Case III: Achiral QDs with Chiroptical
Effects Induced by Electronic Interactions with Chiral Ligands 4.5.4. Chiral Memory 4.5.5. Application to Biolabeling 5. Theoretical Aspects of Optical/Electronic Properties of QDs and Heterostructures 5.1. Modeling Methods 5.2. Strain Effects 5.2.1. Strain and Various Modeling Methods 5.2.2. Strain and Lattice Parameters 5.2.3. Effects of Strain 5.2.4. Strain Theory and Limits to Coherent Shell Growth 5.2.5. Strain Measurement and Mapping 5.2.6. Strain Reduction 5.3. Radiative Lifetimes, Oscillator Strengths, Carrier Localization, and Other Transition Rates 6. Bioapplications of Semiconductor NCs 6.1. Conjugation of Biomolecules with QDs 6.2. Bioimaging and Theranostic Applications 6.3. Biolabeling and Biosensing 7. Concluding Remarks and Perspectives Author Information Corresponding Authors Author Contributions Notes Biographies Acknowledgments References

10671
10673
10674 10675 10676 10676 10677 10677 10677
10678 10679 10679 10681 10682 10683 10683
10684
10685 10685 10687
10687 10687 10688 10688 10690 10690
10691 10692 10694
10695 10697 10697 10700 10704 10706 10707 10707 10707 10707 10708 10708 10708

1. INTRODUCTION
Initiated in large part by the leading efforts of Henglein1−3 and Fendler4−7 in the 1980s, the aqueous synthesis of semiconductor nanocrystals (NCs), or colloidal quantum dots (QDs), has remained an attractive subject of research for over three decades owing to the ensuing efforts by Henglein’s line of successors including notably Weller,8−11 Rogach,12−22 Eychmüller,8,23−27 and Gao,28−34 and of course many more aqueous

enthusiasts from the QD community at large. Early success in the preparation of aqueous CdS NCs not only provided an excellent QD model for investigating the physics underlying the unique electronic features of QDs characterized by strong particle size-dependent optical properties, but also established the basis of a set of aqueous synthetic routes which then expanded to cover the synthesis of other cadmium, zinc, mercury, silver, and copper chalcogenide NCs and ternary NCs as well. Governed by the quantum confinement effect, semiconductor NCs exhibit unique optical properties such as narrow, relatively symmetric, and particle size-/compositiondependent fluorescence ranging from the UV-blue to the midIR.19−21,35−45 This highly attractive flexibility has triggered extensive and detailed exploration of their applications in various and diverse areas including the commercially and societally important biomedical11,39,40,42,46−74 and optoelectronic fields.72,75−92
Nonaqueous synthetic approaches, initially represented by the so-called TOP−TOPO (TOP for trioctylphosphine and TOPO for TOP oxide), or “hot-injection” method developed by Bawendi, Alivisatos, and Peng in the 1990s,35,37,93−96 have been demonstrated to be very powerful for synthesizing highly fluorescent QDs with excellent morphological uniformity and engineerable semiconductor/semiconductor core/shell heterostructures. The nonaqueous synthesis performed at high temperatures in organic solvents, for some semiconductors, yields better quality NCs and lower defect densities. The growth temperature and choice of ligands may also allow selectivity in the lattice morphology in some cases.96,97 A more recent development, mentioned later in this review, is the socalled “heat-up method” which, although using organic solvent and organically soluble precursors, does not rely on the rapid temperature changes upon injection of one of the precursors to transit through nucleation and into QD growth conditions. Instead a thermal ramp (typically a few tens of degrees Celsius per minute) is used and with suitably temperature sensitive precursors a nucleation and subsequent growth phase can be arranged. This approach is easier to use than hot injection when synthesizing larger volumes of material. While the heat-up method has not been applied to all organic solvent syntheses, it has met with reasonable and in several cases good success for a range of QD materials. The review article by van Embden et al.98 gives a comprehensive discussion of the thermodynamics and reaction kinetics of the method and a detailed survey and assessment of the range of binary, ternary, and quarternary QDs reported to have been synthesized using this technique. Nevertheless, the aqueous synthesis of colloidal semiconductor NCs remains attractive owing to the scope for the use of more environmentally friendly materials, potentially better scalability, and cost-effectiveness. The chemistry of the aqueous synthetic route is also very attractive owing to a much richer choice of functionalities for the particle surface capping agents that allow the QDs to be very specifically tailored to applications. For example, the water solubility makes the as-prepared semiconductor NCs readily functionalizable for biolabeling and bioimaging purposes. The excellent aqueous compatibility of biomolecules such as amino acids, nucleotides, nucleic acids, and proteins enables them to also serve as particle surface capping agents during synthesis in order to produce biofunctional nanomaterials directly. In fact, the aqueous synthesis approach can even be expanded as a biosynthesis of semiconductor NCs within microorganisms through mineralization processes.71,99,100 By whichever of these means, the

10624

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

processing of the resulting water-soluble NCs for the fabrication of solid film devices can be more environmentally benign than for the corresponding organic solvent soluble QDs, simply due to the greater ease and lower cost of solvent waste reclamation or disposal.
Apart from the basic biocompatibility benefits, more recent research progress on the aqueous synthesis of NCs has capitalized on their versatile surface chemistry, bringing further insight into their optical properties, and applications as well.101,102 For example, by using chiral (bio-)molecule ligands that are optically active as surface capping agents during synthesis, size-dependent chiral optical properties of semiconductor NCs can readily be realized.101,102 This imprints conventional circular dichroism with quantum confinement effects in the visible-light region by combining the optical activity of chiral molecules with the strong size-dependent absorption of semiconductor NCs. A further example of a topical area of research which benefits from the mild conditions in the aqueous phase is the nucleic acid directed one-pot synthesis of biocompatible NCs conducted in the presence of mononucleotides, DNA or RNA, metal ions, and a chalcogenide source.103−107 The use of nucleic acids as a programmable and versatile ligand community for NC synthesis provides a powerful tool for the rational design of colloidal QDs via nucleic acid sequence variation. NCs synthesized using this approach possess useful spectral characteristics as well as high specificity to DNA, protein, and cancer cell targets, rendering such bioconjugated QDs close to meeting the requirements for translational research in biological sciences.105,106
In the optoelectronic application areas concerning photovoltaic devices (solar cells81,84,88,90,108−111 photodetectors83,112−117) and photocatalysis,5,118−126 after the photogeneration of charges within QDs, the initially formed excitons must be dissociated and the charges extracted from the QDs and transported to other sites such as external electrodes. Extraction and transport in solid QD films are favored by using short ligands to reduce the dielectric barrier and to make interdot separation distances shorter. NCs grown in organic solvents do not benefit from ionic stabilization but must make use of the steric repulsions at short distance arising from the use of long chain ligands. Long alkyl chains have poor or no solubility in water, so QD growth in the aqueous phase relies upon short chain (water-soluble) and charged ligands to prevent aggregation of the colloid by electrostatic repulsion at short range. In order to either make thin dense films of QDs or closely couple QDs to other surfaces or nanoparticles such as TiO2 or other oxide NCs, short chain linker molecules such as mercapto acids or dithiols are commonly used. In order to make good use of organic solvent grown NCs in optoelectronic devices, the long chain growth ligand must be removed and replaced (i.e., exchanged) by a short chain ligand that both continues the job of surface passivation (filling or preventing the formation of surface traps) and may also act as a crosslinking molecule to hold the QD assembly or film together. QDs grown in water come ready-equipped with such ligands, so the challenge there is to produce water grown materials that have optical properties and stabilities that are as good, or better, for devices as for those grown in organic solvents.
The current status is that III−V QDs require high growth temperatures and are exclusively grown in organic solvents. High photoluminescence quantum yield (PL QY) lead chalcogenides are likewise better grown in organic solvents,

but can be formed in aqueous solutions, though with generally lower quality to date. For the II−VI QDs in most cases there is more flexibility, particularly for the lower band gap materials, leading to an overlap of the two synthetic approaches for many materials. Because the final device performance is dependent upon many QD properties, not just those that affect the solution fluorescence QY, the latter is not necessarily the definitive measure of which synthetic approach will produce the best solar cell, etc., in a given choice of QD material, though it is undoubtedly a major factor.
In terms of the detailed formation mechanisms, aqueous synthetic routes are generally more complicated than the nonaqueous counterparts owing to the involvement of H+, OH−, and H2O moieties that introduce a large number of thermodynamic and kinetic parameters to be taken into consideration. In general, the chemical reactions leading to semiconductor NCs in aqueous systems are dominated by double displacement reactions that produce nanometer sized precipitates with particle growth immediately arrested by stabilizing agents. Therefore, the hydrolysis of metal ions and the protonation/deprotonation of small water-soluble organic stabilizing agents strongly affect the nucleation, growth, and stabilization of the NCs formed in aqueous systems. In addition, water is a strong polar solvent that heavily influences the dynamics of precursors at the surface during reactions. This may lead to the introduction of surface trap (defect) states that may strongly affect the fluorescence properties. Additionally this may bring some problems when constructing core/shell semiconductor NCs to tailor their optical properties, as the TOP−TOPO method, for example, does for CdSe NCs.127 However, taking advantage of the nature of properly selected short water-soluble ligands, significant results have been achieved for the direct aqueous synthesis of heterostructures, such as core/shell or core/shell/shell particles, by exerting control over the shell growth kinetics. Being able to fabricate such heterostructures constitutes a large step forward, vastly improving the optical quality of the NCs and putting water synthesized materials on an equal footing with those obtained from organometallic synthetic routes.
The evolutions in the chemistry of NC synthesis and the better understanding and control of the growth mechanisms underpinning the improvements in the quality and sophistication of NC heterostructures have done much to bring NCbased devices closer to commercial performance specifications from both the materials and the devices perspectives. However, in many cases there remains much to be done to achieve a still deeper understanding of the underlying theoretical aspects behind the optical characteristics of QDs in order to be able to practice the rational design of high performance light-emitting nanomaterials and consequently fully meet the requirements of these applications.
This review will first focus on the basic chemistry of the aqueous synthesis of both simple and sophisticated structures (e.g., core/shell, alloyed, and doped QDs), together with an introduction to the recent valuable achievements in the field of biocompatible NCs via the biosynthesis approach. Then particularly special attention is given to the optical properties of QDs and the corresponding current understanding of the underlying theoretical aspects (e.g., electronic structure modeling, the impact of strain effects in heterostructures, Auger recombination processes) to show how this knowledge is starting to be used to rationally design advanced light-emitting nanomaterials. In addition, the optical chirality of aqueous QDs

10625

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews
will also be briefly discussed. Subsequently, we review the progress in a few selected biological and biomedical applications, while broader application areas for QDs, such as solar cells, light emitters and displays, photodetectors, and photocatalysis, are dealt with in detail by other authors in this special issue. The applications are treated with a chemist’s perspective; i.e., what are the materials used, how are they integrated together, how do they perform, and can they be improved by making better materials or more complex or better tailored nanostructures? We conclude with an outlook on the future perspectives with regard to NC aqueous synthesis, optical properties, and promising future applications.

Review

Table 1. Hard/Soft Classification of Different Species Used for the Aqueous Synthesis of Semiconductor NCs

hard
H+, Na+, In3+, Mn2+, Ln3+
Cl− H2O, OH−, O2− ROH, RCOO−, RO− NO3−, ClO4− NH3, RNH2, N2H4

borderline
Acids Cu2+, Zn2+, Pb2+
Bases

soft
Cu+, Ag+, Cd2+, Hg2+
H2S, HS−, S2− RSH, RS− Se2−, Te2−

2. BASIC CHEMISTRY OF AQUEOUS SYNTHESIS
2.1. Thermodynamic Aspects
The aqueous synthesis of a given semiconductor NC is strongly associated with at least the following four major parameters: the solubility product of semiconductor compounds in water; the binding affinity of particle surface capping ligands, generally for metal ions, forming the NCs; the binding affinity of water and hydroxyl ion for the metals; and the pH of the aqueous media. The role of the above thermodynamic parameters can qualitatively be understood through the hard and soft (Lewis) acids and bases (HSAB) theory,128,129 as discussed below.
The aqueous synthesis of semiconductor NCs is largely based on Lewis acid/base reactions often inevitably involving H+, OH−, and H2O. A Lewis base is a chemical species that has a pair of electrons to donate, and a Lewis acid is a chemical species that accepts the electron pair of bases. According to the HSAB concept, Lewis acids/bases can further be classified as hard or soft according to their polarizability. “Hard” acids or bases are species which are relatively small and possess high charge states, while being weakly polarizable, whereas “Soft” species on the contrary are large, have low charge, and are strongly polarizable. To a large degree most of the hard−soft distinction hinges upon the polarizability, the extent to which the charge distribution of a molecule or ion is perturbed by interaction with neighboring chemical species. Moreover, larger atoms or ions tend to be softer due to inner electron shielding of the nuclear charge, lessening the influence of the latter on the outer electrons that engage in bond formation. The electronic charge distributions in easily polarizable molecules can readily be distorted by attraction or repulsion by charges on other molecules. In so doing, the distortion in turn forms a slightly polar species that can then interact with the other molecules. Although such a distinction is made in a qualitative way, the essence of HSAB theory is that hard acids react faster and form stronger bonds with hard bases, while soft acids react faster and form stronger bonds with soft bases. The aqueous synthesis of semiconductor NCs involves not only double replacement reactions, often for forming metal chalcogenides, but also coordination reactions of the NC surface capping agents that are mostly small organic compounds. HSAB theory in this context is relevant for understanding the inorganic and organic reaction chemistry involved, as well as qualitatively explaining the stability of compounds, the different complexation behaviors, and reaction pathways. Table 1 shows the hard/soft classification of the species often used in the aqueous synthesis of semiconductor NCs according to HSAB concepts.
2.1.1. Solubility Product. From the standpoint of thermodynamics, the aqueous synthesis of colloidal semi-

conducting materials can be guided and rationalized by solubility product principles. In aqueous synthesis, water as a strong polar coordinating solvent inevitably interferes with the formation of the target semiconductor compounds, through either the hydration of both cations and anions or the formation of insoluble metal hydroxide precipitates. Therefore, the solubility of the target compound and its balance with that of the corresponding metal hydroxide becomes the first criterion in predicting the formation of a desired semiconductor compound formed in aqueous media under specified conditions.
Many thermodynamic factors influence solubility, including ionic size and charge via the hardness or softness of the ions (HSAB), the crystal structure of the solid, and the electronic structure of each solvated ion. The dissolution of an ionic solid includes the processes of splitting the crystal lattice and solvating the cations and anions. Thus, the solubility product constant strongly depends on the lattice and hydration energies. The former (endothermic, positive sign) relates to the energy required to break the solid apart into gaseous cations and anions, and the latter (exothermic, negative sign) relates to energy variation in consequence of the hydration of gaseous ions. One of the important applications of the HSAB principle is to understand the solubility of compounds of various types. According to HSAB theory, compounds formed by hard−hard interaction between two small ions or soft−soft interaction between two large ions are generally expected to have lower solubility than compounds containing one large ion and one small ion. This is particularly so when the two ions have the same charge magnitude. For inorganic compounds comprised of small ions, the exceptionally large lattice energy apparently cannot be compensated for by the relatively large hydration enthalpies, rendering these compounds less soluble. For salts of large ions, the lower solubility is rationalized by relatively small hydration enthalpies that cannot compensate for the lattice energy. Even though the lattice energy in the latter case is relatively low in magnitude, the smaller hydration enthalpies nonetheless mean that the lattice energy remains the dominant factor.
Chalcogen ions (S2−, Se2−, Te2−) are typical soft bases, while most transition metal ions, depending on the oxidation state, can be classified as soft acids, e.g., Cd2+, or borderline acids, e.g., Pb2+ and Zn2+, as shown in Table 1. According to the HSAB concepts, chalcogens readily form insoluble compounds with the majority of transition metal ions in aqueous media due to the low solubility products. In comparison with solids formed via hard−hard interactions, the covalent character of the soft− soft interactions is increased, which also leads to semiconducting properties for the transition metal chalcogenides. These principles form the fundamental basis for the aqueous

10626

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

synthesis of metal chalcogenide semiconductor materials, including group I−VI, II−VI, and I−III−VI semiconductors.
The solubility of inorganic solids may also be predicted, especially for those showing extremely low, near-undetectable levels of dissolved ions experimentally (e.g., metal tellurides). From empirical rules, low electronegativity differences between constituent atoms correspond to greater covalent bond character and lower aqueous solubilities, which are well exemplified by cadmium chalcogenides. Based on thermochemical data, the Ksp values of metal selenides and tellurides were also theoretically predicted, showing that the Ksp continuously decreases against the atomic weight of the chalcogens as shown in Figure 1.130 In principle and by analogy, the above knowledge helps to assess the stability of semiconductor compounds with little direct experimental information available. However, in real reaction systems, the solubility product value is not sufficient in isolation to predict the reaction pathway as the concentration of free cation may largely be altered upon the formation of nonsoluble metal hydroxides or complexes. With respect to the formation of NCs, the interaction between cations and the indispensable stabilizing agent will also shift the reaction equilibria involved since solubility product values in the literature are generally determined in the absence of any ligands.
2.1.2. pH. With respect to the aqueous synthesis of semiconductor NCs, the solution pH is apparently an important factor.19,28,30,134−137 The hydroxyl ion as a typical hard base can react with most transition metal ions forming metal hydroxides that occur as precipitates or soluble complexes. These reactions will compete against the reaction for forming the target semiconductor compounds. According to HSAB principles, “hard” OH− ion prefers to bind with “hard” cations. Therefore, binding affinity of OH− ions to cations, e.g., Zn2+ (log K1 = 4.4) > Cd2+ (log K1 = 4.17) > Ag+ (log K1 = 2.3), follows the sequence of the hardness of cations, i.e., Zn2+ > Cd2+ > Ag+, which is also reflected by the difference of the solubility product constants (Ksp) of the corresponding hydroxide compounds, i.e., Zn(OH)2 (Ksp = 3 × 10−17), Cd(OH)2 (Ksp = 7.2 × 10−15), and AgOH (Ksp = 2.0 × 10−8).
Figure 1. Comparison of pKsp values for metal sulfides, selenides, and tellurides calculated after Buketov et al.131 and from Licht132 and Moon et al.133 Data are compared against Licht’s sulfide data for which a S2− equilibrium is assumed. Reproduced with permission from ref 130. Copyright 2013 Wiley-VCH Verlag GmbH & Co. KGaA.

The difference in hydroxyl binding affinity to metal ions gives rise to different pH dependences with respect to the synthesis
of semiconductor compounds. For example, due to the relatively high affinity of hydroxide ions to Zn2+, careful pH
adjustment is needed to avoid the formation of Zn(OH)2 when preparing zinc chalcogenides.138,139 In comparison with Zn2+, Cd2+ is softer according to HSAB; therefore the solubility product constants of CdE (E2− = S2−, Se2−, Te2−) are much lower than that of Cd(OH)2 as E2− are soft bases. In consequence, less competing coordination of OH− ions to Cd2+ (Ksp = 7.2 × 10−15 for Cd(OH)2) permits the preparation of cadmium chalcogenides tolerating high concentrations of OH− ions. However, when the pH value goes beyond 12.5,
Cd(OH)2 precipitation and cadmium hydroxide complexes (e.g., Cd(OH)3− and Cd(OH)42−) will appear and interfere with the growth of CdTe NCs.24 Upon further lowering the softness of the cations, the pH effect on the preparation of
semiconductor NCs can generally be ignored. Silver ion belongs to such a category, being even softer than Cd2+
according to HSAB. The remarkably lowered Ksp constants of silver chalcogenides (i.e., Ag2S, Ag2Se, Ag2Te) enable the formation of silver chalcogenides relatively independent of pH.
All of the above examples suggest that, with respect to the
synthesis of semiconductor compounds in aqueous systems, the
competition against the formation of metal hydroxides can
generally be predicted by HSAB theory. Nonetheless, the pH effects on the formation of semiconductor NCs are far more
complicated since hydroxyl ion may induce heavy surface
coordinating dynamics apart from forming metal hydroxide
byproducts. This may also alter the optical properties and
colloidal properties of the resulting NCs. On the other hand, pH also shows a strong impact on the effect of stabilizing agents
and their coordinating behavior as well, which will be discussed
in the following sections.
2.2. Chemistry for Forming Semiconductors
The semiconductor NCs obtained so far through solution synthesis are mainly chosen from I−VI, I−III−VI, II−V, II−VI, III−V, and IV−VI semiconductors. The cations typically include Cu+, Ag+, Cd2+, Zn2+, Hg2+, Pb2+, Ga3+, In3+, etc., while the anions are mainly selected from S2−, Se2−, Te2−, N3−, P3−, As3−, etc. Chalcogens are soft bases, while most transition
metal ions are soft or borderline acids; therefore, the aqueous approaches are generally more suitable for synthesizing II−VI, I−III−VI, I−VI, and IV−VI semiconductor NCs, such as
Zn(S,Se,Te), Cd(S,Se,Te), Hg(S,Se,Te), Pb(S,Se), CuIn-
(S,Se)2, Ag2(S,Se,Te), and AgInS2 as well as some alloyed particles such as CdHgTe, CdSeTe, ZnSeS, ZnCdSe, ZnHgSe, and ZnSeTe, rather than III−V semiconductor NCs.
With respect to the synthesis of metal chalcogenides, the anion is often supplied in the form of HE− or E2− (E = S, Se,
Te). However, these anion forms do not translate for group V
elements such as P, due to the strong nonpolar character of the P−H bond in phosphine (PH3). In fact, phosphine dissolves more readily in nonpolar solvents than in water. It is technically
amphoteric in water, but acid and base activity is very poor. Proton exchange can proceed via a phosphonium (PH4+) ion in acidic solutions or via PH2− at high pH. Further deprotonation may occur but under extremely strong alkaline conditions for forming group III−V NCs such as GaP and InP. In addition, phosphine is readily oxidized as exemplified by the fact that the
element phosphorus largely exists in the form of oxides or phosphates. In comparison with group II cations such as Cd2+,

10627

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

group III cations are stronger acids and more readily react with strong bases such hydroxyl ions. All these features in the chemical nature of group III and V elements basically limit the development of aqueous synthetic routes for III−V NCs.
Apart from the chemical aspects for forming semiconductor solids in aqueous systems, the semiconducting properties characterized by band gap energy also strongly depend on the chemistry of the constituent cation and anion. The size and electronegativity of the atoms are two key factors determining the size of the band gap if it exists, and their chemical hardness is generally proportional to the band gap. Taking well-studied II−VI semiconductors (e.g., CdE, E = S, Se, Te) as an example, the bulk band gap of metal chalcogenide semiconducting materials decreases as the chalcogen electronegativity decreases (size increases). This is because chalcogens with low electronegativity do not hold their valence electrons tightly, there is a consequent tendency toward a more covalent bonding character which gives rise to a decreased band gap energy. The dependence of the band gap on the chemical properties of anions and cations involved conversely helps to predict the variation of band gap and consequently the optical properties and is particularly useful when alloying the semiconductors with different cations or anions. It should be mentioned that such comparisons and extrapolations of behavior are only possible based on the fact that all the systems possess the same crystalline structure. With respect to semiconductor NCs, the particle growth conditions and surface capping agents may also show a strong influence on the crystalline structures obtained.
2.3. Surface Chemistry of Semiconductor NCs
Apart from the fundamental aspects on the formation of semiconductor solids in aqueous media, the synthesis of semiconductor NCs further involves the surface chemistry of both particle core and surface stabilizing agents. Due to extremely high surface-to-volume ratios, the surface of NCs, comprised of dangling bonds relating to both anionic and cationic sites, strongly affects the physical properties of NCs. In solution synthesis, one of the most important strategies is to saturate the anion-associated dangling bonds with an excess of cations to suppress the formers’ impact, while the surface cations are further terminated with stabilizing agents. In this context, the chemistry of the metal−stabilizing agent interaction is an important subject of research.
Except for a few examples, stabilizing agents are indispensable in the aqueous synthesis of semiconductor NCs. Stabilizing agents are in a broad sense chemical species that can effectively bind via different types of interactions with the surface atoms of NCs, in most cases cations, to offer shortrange repulsive forces for the resulting NCs to be colloidally stable. Water-soluble small molecules, amphiphilic molecules, macromolecules, and polymers with suitable structures can be used as stabilizing agents, while those forming a quasimonolayer on the surface of NCs through covalent, dative, or ionic bonds are referred to as surface capping agents.37 They serve to mediate the growth of NCs, electrostatically or sterically stabilize the NCs in solution, and passivate surface electronic states in semiconductor NCs. The surface capping is analogous to the binding of ligands in more traditional coordination chemistry where ligands donate an electron pair to form chemical bonds with the metal atom. Therefore, surface capping agents are also called surface ligands in many cases in the literature. In fact, surface ligands affect not only the growth and colloidal stability of the resulting NCs, but also the optical

properties of the underlying NCs. Most importantly, they also play a crucial role in the applications that are discussed later in this review with respect to surface functionalization.
In nonaqueous synthesis, normally long alkyl chain molecules bearing one anchoring group at one side are used as surface ligands whereas, in contrast, those used in aqueous synthesis typically possess a strongly polar group linked through a short hydrocarbon chain with the capping group. The polar group in most cases offers the underlying NCs water dispersibility/solubility, but in some cases may interact with the cation on the surface of the NCs. Properly balancing the binding strength of the ligand molecule with the metal cation largely determines the validity of the aqueous synthesis approach. Overstrong binding affinity will prevent the metal cations from effectively reacting with the anions, while weak binding affinity will lead to insufficient colloidal stability and uncontrollable growth of the underlying NCs. An appropriate binding strength of the surface ligand is therefore necessary for tailoring the growth of semiconductor precipitates to form nanometer-sized crystals.
HSAB theory provides a qualitative way to assess the interactions between metal centers (Lewis acid) and coordinating ligands (Lewis base) through their hardness or softness. As mentioned above, most transition metal ions belong to the categories of borderline (e.g., Zn2+, Cu2+) or soft acids (e.g., Cu+, Ag+, Pb2+, Cd2+, Hg2+), and thus preferentially bind with soft bases rather than hard ones. By such reasoning, watersoluble thiol molecules as soft bases effectively meet the requirements for efficiently capping the NCs formed by the above-mentioned transition metal ions, and are much more effective than molecules bearing single carboxyl or amine groups as the latter belong to the set of hard bases that can be used in the nonaqueous synthesis of inorganic NCs. To this day, water-soluble thiol ligands are widely used in established aqueous synthetic routes for high quality metal chalcogenide semiconductor NCs. Nonetheless, multiple metal-chelating stabilizing agents offer increased binding strength generally following the sequence of monodentate < bidentate < tridentate < multidentate ligand. Macromolecules and polymers bearing multiple carboxyl groups or phosphate groups (polyphosphates) may also offer stabilization effects to semiconductor NCs containing the above cations.140,141
The binding affinity of capping groups can also be modulated by pH. Taking thiol molecules as an example, high pH pushes the thiol−thiolate equilibrium to form the RS− species which has stronger binding affinity to metal ions than the corresponding thiol.137 As mentioned above, the equilibrium for forming metal hydroxides also needs to be taken into consideration in designing the experimental conditions. In addition, small thiol capping agents can also form different types of molecular complexes with metal ions. This is strongly dependent on pH and determines the supply rate of free cations.
The other polar groups of the surface ligands serve to render the NC colloids stable via electrostatic repulsion, e.g., mercapto acids; strong solvation, e.g., mercapto alcohols; and steric hindrance, e.g., thiol-PEG (polyethylene glycol) or denatured bovine serum albumin (BSA). These stabilizing effects may sometimes coexist depending on the chemical structure of the surface capping agents. Since the polar groups are largely chosen from −COOH and −NH2 functionalities, apart from the pH effect on the capping group, the protonation− deprotonation of the polar group is also strongly pH-

10628

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

dependent. For example, 2-mercaptoethylamine requires slightly acidic pH to colloidally stabilize chalcogenide NCs, while mercapto acids need alkaline pH to obtain colloidally stable NCs.134
Apart from offering a stabilizing effect, the polar group may interact with the NC surface cation under appropriate pH conditions, which is exemplified by the formation of a secondary coordination between the carbonyl group of thioglycolic acid (TGA) or 3-mercaptopropionic acid (MPA) and the primary thiol-coordinating cadmium sites in TGA- and MPA-stabilized CdTe QDs.28,30,32,143 Such coordination can dramatically boost the fluorescence QY.28,30 Similarly, glutathione (GSH) tripeptide also shows pH-dependent interactions with metal ions such as Zn2+ due to the presence of −SH, −NH2, and −CONH− groups, as shown in Figure 2.142,144
In brief, most established aqueous synthetic routes are based on water-soluble thiols used as surface ligands. The interaction of the thiol capping group with metal ions of NCs under the influence of pH and the pH-dependent properties of the polar groups make the aqueous synthesis of semiconductor NCs more complicated, but the latter brings the attraction of more versatile surface functionalization/structures that offer a much richer choice than simple hydrocarbon chains with respect to many applications.
3. AQUEOUS SYNTHESIS OF SEMICONDUCTOR NCS
3.1. Historical Overview of Aqueous Synthesis Historically, the chemical synthesis of well-defined semiconductor NCs was first developed in aqueous media upon direct chemical reactions between cation and anion precursors in the presence of stabilizing agents. Owing to the early efforts of Henglein, Fendler, Graẗ zel, Brus, and Nozik, CdS NCs became the first well-established system for in-depth study on the physical and physiochemical properties of three-dimensional confinement materials that are significantly different from one-dimensional confinement materials previously developed in the epitaxial semiconductor industry. Benefiting from the knowledge derived from CdS NCs, a series of theoretical studies were pioneered by Efros145 and Brus,146−148 to disclose the electronic structures of the quantum-sized NCs for predicting and understanding their size-dependent optical properties.
From a chemistry point of view, the aqueous synthesis of semiconductor NCs is as simple as carrying out a precipitation reaction except that a stabilizing agent is essentially required to control the precipitation process during the early stage so as to limit growth to nanometer-sized particles. In early studies, micelles and vesicles formed by self-organized amphiphilic
Figure 2. Evolution of the complex structure of Zn2+-GSH against the increase of pH: (a) pH 6.5−8.3; (b) 8.3−10.3; (c) 10.3−11.5. Reproduced from ref 142. Copyright 2010 American Chemical Society.

molecules were also used to confine the growth of semi-
conductor NCs. In principle, any chemical reaction leading to
nonsoluble substances can be adopted to obtain nanoparticles
upon a suitable choice of stabilizing agent. This strategy was
valid for synthesizing not only CdS NCs, but also for ZnS,149,150 PbS,151 Cd3P2,152 Zn3P2,152 Cd3As2,153,154 and CdTe155 QDs. With respect to CdS NCs, the direct reaction between Cd2+ and S2− was largely used in early studies. Later on, different types of anionic precursors that slowly release S2−
ions were also used for synthesizing CdS NCs rather than using
H2S as a gaseous sulfur precursor. Regarding the stabilizing agents, maleic anhydride/styrene copolymer, phosphate, and
polyphosphates were adopted in early studies. Later stabilizing
technique advances included the use of chelating peptides, thiol
molecules, and even biomolecues such as denatured BSA,
DNA, and RNA more recently. Among these stabilizing agents, thiol ligands are advantageous since they are not only effective
for forming monodisperse CdS NCs with very small size but
also equally applicable for a wider variety of semiconductor
NCs containing cadmium, zinc, lead, silver, copper, and mercury ions. With respect to cadmium sulfide, ultrasmall even molecule-like clusters with clearly defined structures and optical characteristics,10,156,157 e.g., [Cd17S4(SCH2CH2OH)26]10 and [Cd32S14(SCH2CH(OH)CH3)36](H2O)4,158 can be obtained, for which the degree of control highlights the advantage of thiol molecules as capping
agents. Benefiting from the early studies on CdS NCs, the aqueous
synthetic techniques were expanded to CdSe and CdTe NCs.
The resulting subsequent studies quickly made CdTe NCs a
key material of interest because the band gap energy of bulk
CdTe is much narrower, i.e., 1.43 eV for CdTe vs 2.45 eV for
CdS. Thus, it is possible to tune the optical properties of CdTe
NCs over a larger spectral window, especially from the visible to the near-infrared region, through the quantum confinement effect alone, which is beneficial for realizing fluorescence of different colors for many applications.
The earliest successful example of the aqueous synthesis of CdTe NCs employed the direct reaction of Cd2+ with Na2Te in the presence of a polyphosphate ((NaPO3)6) stabilizing agent.155 This followed a previous well-established procedure for the synthesis of fluorescent CdS NCs in aqueous solution differing only in that Na2Te was used rather than Na2S or H2S.155 But the CdTe nanoparticles obtained were not fluorescent under optical excitation. In another study, by
using polyphosphate and thioglycerol (TG) as costabilizing agents, fluorescent CdTe nanoparticles were synthesized in aqueous solution.159 Although the optical characterizations of fluorescence and PL QY in these early studies were not entirely
reliable, these early attempts led to a surge of interest in the
aqueous synthesis of NCs. A significant advance in stable and fluorescent CdTe NC
synthesis was achieved by Rogach et al. upon the use of TG or mercaptoethanol (ME) as the ligand.12,13 Since thiol molecules can firmly bind to the surface of CdTe NCs through a Cd−SR bond, they can effectively regulate growth of the nuclei, leading to effective size control. Under optimized conditions, exciton emission was observed, but the fluorescence efficiency of TGand ME-capped CdTe NCs remained lower than 3%.13
The following important milestone toward highly fluorescent
CdTe NCs was achieved by Gao and co-workers with mercapto acids as surface capping agents.28 Under optimized conditions, the fluorescence efficiency of CdTe NCs capped by thioglycolic

10629

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

acid (TGA) reached 18% at room temperature.28 In a similar way, the as-prepared CdTe NCs obtained with mercaptopropionic acid (MPA) as the capping agent was around 38%.30 In addition, the TGA- and MPA-stabilized CdTe NCs presented excellent colloidal stability in solutions over periods of years. These initial studies paved the way for more detailed and substantial explorations of the aqueous synthesis of CdTe NCs due to their potential for applications in bioassays and optoelectronic fields. The surface carboxyl moiety offers charge to the NCsgood for stabilization but also for forming solid films through ionic layer-by-layer self-assembly techniques. Moreover, the carboxyl group is also a useful reactive site for attaching bioligands. The following studies revealed that the PL QY of the as-prepared TGA- and MPA-capped CdTe NCs can further be increased to 40−60% by properly adjusting parameters such as pH, Te/Cd/thiol ratios, and precursor concentrations, etc.19,24,160−162 In addition, TGA and its derivatives were used not only for manipulating the size but also for shaping the CdTe NCs to obtain nanorods,163 nanowires,164 nanotubes,32 twisted ribbons,165 and nanosheets.166,167 MPA was effective for increasing the size of CdTe NCs up to 6 nm to further extend the PL emission from the visible to the near-infrared (NIR) region (700−800 nm), where the PL QY remains as high as 70−80%.19
To facilitate the growth of aqueous CdTe NCs, hydrothermal160,168 and microwave-assisted methods169−171 were also adopted based on the conventional synthetic recipe. The obvious benefit of the latter approach is that the microwave radiation as a heating source can rapidly and homogeneously heat the entire reaction system. We will discuss the influence of postsynthetic heating in later sections, but the reproducibility of preparation, particle uniformity, and corresponding optical properties were greatly improved.169,170,172 In addition, the reaction time was shortened producing CdTe NCs showing longer wavelength emission more rapidly.
The mercapto acid based aqueous routes for CdTe NCs have been widely extended to the majority of metal chalcogenide NCs, such as thiol-capped Zn(S,Se,Te),138,142,173−176 Cd(S,Se,Te),14,19,24,28,134,156,177 Pb(S,Se),178−180 Hg(S,Se,Te),14,22,181−183 Ag(S,Se,Te),184−190 CuInS2,191−194 and AgInS2,195−197 and NCs with alloy and core/shell structures. These advances not only enrich the aqueous synthesis of versatile semiconductor NCs, but also make the mercapto acid based aqueous synthesis a mainstream approach for achieving aqueous semiconductor NCs with rich and advanced optical properties, which have been reviewed previously by Weller,134 Rogach,19,21,79 Gao,33 and Gaponik and Eychmüller.25,26
3.2. Growth Mechanism of Aqueous NCs
The energy level density of states of semiconductor QDs becomes discrete rather than continuous, and for levels near the band edge, the energies are blue-shifted when their radius is smaller than or comparable to their bulk exciton Bohr radius. This regime is known as quantum (or size) confinement. In consequence, the optical properties of semiconductor NCs start to strongly depend on the crystal size, which is directly reflected in band edge spectral features in NC absorption spectra. Tunable absorption is very favorable for light harvesting in solar cell applications,198 Förster resonance energy transfer (FRET) in optical down-conversion devices,199,200 etc. Upon suitable surface modification, size-dependent exciton emission is achieved and characterized by narrow, symmetric features, apart from excellent robustness against photobleaching. The

emission color tunability and high PL QY are of particular interest for biolabeling/bioimaging applications,201,202 light emitting diodes,29 and lasing.203 In this context, understanding the growth mechanism of semiconductor NCs in aqueous media is important for achieving uniform NCs with attractive optical properties.
In the classical LaMer model,204 there are three recognized stages for forming colloidal particles, i.e., monomer accumulation, nucleation, and growth. According to this model, quick monomer accumulation is favored to allow a burst of nucleation that will depress the particle size dispersity. This in practice is often achieved through quick injection of the precursors into the reaction system. To effectively narrow the nucleation period, properly balancing the coordination of the surface ligand and hydroxyl ion to cations forming the target NCs is very important, which is related not only to the concentration and ratio among ligand, cationic, and anionic precursors, but also to the pH of the reaction system. The latter is largely dependent on the chemical nature of all the species involved. In the subsequent growth stage, a slow growth rate is also in favor of a narrow size distribution (or preserving a narrow distribution from the nucleation stage). The actual growth process of colloidal nanoparticles is generally believed to be strongly associated with the diffusion of monomers toward the surface of the NCs and, following that, surface deposition onto the growing NCs. Therefore, two different growth models associated with the above processes are proposed on the basis of classic kinetics theory and widely used to interpret the growth behavior of inorganic colloidal particles in solution, that is, a diffusion-controlled growth model95,96 and a surface reaction-controlled growth model.205−208 The former postulates that the particle growth rate depends on the flux of the monomers supplied to the particles, which leads to size focusing effects, whereas the latter postulates that the particle growth rate depends on the precipitation and dissolution of monomers which, depending on the chemistry for converting the monomer to NCs, often leads to size broadening. However, the detailed chemistry regarding particle growth still remains an open question.
In aqueous systems, the chemical reactions involving ligand, cationic, and anionic precursors often result in a rich variety of molecular complexes in the form of small molecular clusters. The ensuing growth stage first converts the small molecular complexes into crystal nuclei, which is characterized by a fast growth process followed by a slow growth process resulting in size distribution coarsening that is typically reflected by the broadening of any emission and, if sufficiently distinct, broadening of the band edge excitonic absorption peak.
An important point of the dynamics of the ensemble evolution is the intrinsic polydispersity of colloidal particles, which can result in Ostwald ripening. The major advance for describing this process was achieved by Lifshitz and Slyozov209 and Wagner,210 who developed a model for the evolution of the size distribution of an ensemble of particles during Ostwald ripening, known as LSW theory. This classical LSW theory predicts the ripening kinetics and the particle size distribution function in dilute systems and can be used to analyze the competition of growing particles for the monomer in the stationary regime. However, it fails to describe the earlier transient stages of the particle growth that are extremely important for achieving effective control over the particle size and optical properties as well, and the classic theory only

10630

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

treated the Gibbs−Thomson equation for the size dependent solubility by two expansion terms:

C(r)

=

Cf0lat

⎡ 2γVm ⎤ exp⎢ ⎥



0⎛ Cflat⎜1

+

2γVm ⎞ ⎟

⎣ rRT ⎦

⎝ rRT ⎠

(1)

Here, C(r) represents the solubility of a particle with radius r and Cf0lat is that of the bulk material. Vm is the molar volume and γ is the surface tension of the solid. The magnitude of the factor 2γVm is usually around 1 nm, and this term is referred to as the
RT
“capillary length”. For colloidal particles with radii greater than ∼20 nm, eq 1 is adequate. For radii in the range r = 1−5 nm, the capillary length and the particle radius are of similar values, and in practice the particle solubility becomes strongly nonlinear in terms of the inverse particle radius, r−1. Moreover, the chemical potential of nanoparticles and the rate of the surface reactions also depend nonlinearly on r−1. Both effects were not taken into account in the classical description of reaction-controlled Ostwald ripening.
Toward these problems, Talapin et al. developed a theoretical model by taking into account size-dependent activation energies of the growth and dissolution processes as well as the mass transport of monomers toward the particle surface to describe the complete particle ensemble presented in the colloidal solution as a function of time.206 In their model, the particle size dependent activation energies for growth and dissolution of the NP are given byΔ‡μg and Δ‡μd:

Δ‡μ (r) = Δμ∞ + α 2γVm

g

g

r

(2)

Δ‡μ (r) = Δμ∞ + β 2γVm

d

d

r

(3)

where α and β are the transfer coefficients (α + β = 1); Δμ∞ is an activation energy in the case of flat surface.
In the framework of their theory, the rate constant for particle growth and dissolution is expressed through the height of the activation barrier:

k = B exp(−Δμ/RT)

(4)

where B is a constant with the same dimension as k. Then the rate of change of particle radius r in the solution of
monomer, assuming a constant concentration [M] for the latter, is given as

dr

⎧⎪ [M]0bulk − exp⎡⎣ 2rγRVTm ⎤⎦ ⎫⎪



= VmDCf0lat⎨ Cflat ⎪r + D

⎬ exp⎡α 2γVm ⎤ ⎪

or



kgflat ⎣ rRT ⎦ ⎭

dr* = S − exp[1/r*]

dτ r* + K exp[α/r*]

(5)

in which

r* = RT r

2γVm

(6)

R2T 2DCf0lat τ= 2 t

4γ Vm

(7)

K = RT D

2γVm kgflat

(8)

S = [M]bulk /Cf0lat

(9)

where r* is a dimensionless radius and τ is a dimensionless time constant. The dimensionless ratio K can effectively be said to
describe the type of the process involved, since it is set by the ratio between the rates of a purely diffusion-controlled process determined by D (the diffusion coefficient of the monomer) and a purely reaction controlled one, the latter given by kgflat (the first-order reaction rate constant for addition of a monomer at a flat surface). S is a dimensionless parameter for describing the oversaturation of the monomer in solution, α is the transfer coefficient of the activated complex (0 < α < 1).
According to the value of the parameter K, three types of case can arise:211 K < 0.01 corresponding to a virtually pure diffusion-controlled process; at the other extreme, K > 100
denoting a wholly reaction-controlled process; and in the
midrange, 0.01 < K < 100 corresponding to a regime of mixed control with comparable influences from both types of process. With this more flexible applicability, this model shows significant improvement in comparison with LSW theory for
predicting the size distribution and evolution of particle
ensemble against the particle growth if compared with
experimental results. The above theory is also suitable for the
description of growth rates dynamics where the ensemble of
particles is less than 10 nm in diameter, that is, in a typical size region for semiconductor NCs to exhibit quantum size effects.
Moreover, the theory allows the prediction of conditions leading to either “focusing” or “defocusing” of the particle size distributions and reveals that the diffusion-controlled growth
can result in narrower particle size distributions contrasting to
reaction-controlled growth. Under diffusion-controlled conditions, e.g., K = 0.01, a strong
narrowing effect of the size distribution is observed, and this becomes significant with the increase of the initial oversaturation of monomer. Such a narrowing effect was previously
observed in nonaqueous synthesized CdSe and InAs NCs but not in aqueous synthesized CdTe NCs. This difference can be
explained as follows. In the aqueous synthesis of CdTe NCs,
soluble cadmium salt and NaHTe are widely used as reactants. Under typical alkaline conditions, Te2− reacts with Cd2+ so
quickly that no inert atmosphere protection is required right after the injection of the Te2− source solution. This means most Te2− as one of the monomers is quickly depleted forming molecular complexes with Cd2+, which during the subsequent
heating up process quickly coagulate to form QD nuclei. In the typical synthesis of CdTe QDs, fluorescence starts to appear right after the reaction mixture is refluxed, indicating that most
monomers have been consumed by the nucleation process before the reaction solution is refluxed and remaining
monomers supply the particle growth that is about to step into the Ostwald regime. This hypothesis finds evidence from a
more recent study on ZnSe NCs prepared in aqueous phase in
the presence of a thiol capping agent. The concentration of Zn
in both QDs and solution remains constant after 30 min of reflux.175 This study suggests that the cation and anion as major
monomers can react so quickly in the aqueous phase that the
particle growth will quickly transition into the Ostwald ripening
regime at elevated temperature. It should be mentioned that the
time for the emission of CdTe QDs to shift typically from
green to red takes from tens to hundreds of hours dependent on the synthetic parameters. As a result, the fluorescence of
QDs prepared through aqueous synthetic routes often becomes

10631

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730

Chemical Reviews

Review

Figure 3. Growth curves of ZnSe QDs capped by the four ligands MPA, TGA, MTG, and TLA (a) using standard reaction conditions (24 mM
ligand, 9.6 mM Zn, and 5 mM Se concentrations at pH 11.9 for a reaction time of 1 h) and (b) using half of the standard concentrations of ligand, Zn, and Se for a reaction time of 2 h. The lines are fits to the data using the surface-reaction-controlled Ostwald ripening model. (c) QD growth rates for each ligand obtained from the slopes of the best fit lines shown in (a) and (b). Reproduced from ref 175. Copyright 2012 American Chemical
Society.

progressively broader for most aqueous synthesized QDs as the particle growth is dominantly governed by Ostwald ripening.
In contrast, for inorganic NCs prepared by nonaqueous synthetic methods such as organometallic routes for QDs and thermal decomposition method for iron oxide NCs (e.g., Fe2O3 and Fe3O4), the slow decomposition of metal−organic precursors limits the particle growth kinetics, which places the particle growth mainly in the diffusion-controlled regime. In consequence, the particle size distribution of NCs formed in a nonaqueous phase, irrespective of whether the exact method is “hot-injection” or “heat-up”, is typically narrower than that of NCs prepared in aqueous systems. If comparing CdSe QDs synthesized in organic solvents with CdTe QDs prepared in an aqueous phase, it generally also takes a much shorter time for the former NCs to extend the fluorescence into the red, because the particle growth is upon the consumption of monomer released by the precursor, while the latter growth is upon the dissolution and reinclusion of smaller particles.
Another important contribution of the theoretical study of Talapin et al. is that the correlation between growth rate and particle fluorescence QY can be established.211 Theoretically, in a polydisperse particle ensemble, the large particles have positive growth rates and the small ones have negative growth rates as a consequence of Ostwald ripening. Between these two cases, there are particles at the equilibrium established with the monomers in the solution showing nearly zero growth rate (ZGR). The critical radius (rZGR), which is identified as that at which there is net zero growth rate (dr*/dτ = 0), was derived from the above dr*/dτ expression (eq 5):

r = 2γVm

ZGR RT ln S

(10)

Monte Carlo modeling is thus used to simulate the temporal evolution of the size distribution of an ensemble of NCs during diffusion-controlled Ostwald ripening. The intrinsic growth rate vs particle size distribution of the ensemble can be calculated with a given monomer oversaturation and suggests that the particle size corresponding to ZGR lies to the lower size side of the entire ensemble during all growth stages.
It has been experimentally observed that there is strong nonmonotonic variation in the PL quantum yield of sizeselected fractions separated from the parent reaction solution size distribution.211 Moreover, the QD fraction showing a maximum PL quantum yield is always located to higher

emission energy than the PL peak of the corresponding parent solutions. In accordance with this experimental observation, the numerical simulations reveal that the ZGR size corresponds well to the particle fractions with the highest fluorescence efficiency within the ensemble. NCs are inclined to have relatively complete internal lattices, making the variation of the surface defects the main factor influencing the luminescent properties. Thus, the equilibrium conditions promote an optimal surface reconstruction for the NCs, leading to a surface free of defects. In this context, the above theoretical study provides an in-depth understanding of the PL quantum efficiency distribution in QD ensembles.
Coming back to synthetic chemistry, it is thus important to optimize the capping agents as their coordination with the surface of NCs largely affects the Ostwald ripening process and consequently the PL QY as well. In addition, water as a coordinating solvent makes the synthetic chemistry far more complicated than the nonaqueous synthesis, because the pH of the solution inevitably becomes an additional parameter for tuning particle growth kinetics and PL QY by mediating the binding strength of the capping ligands, which exactly highlights the beauty of aqueous synthetic routes for semiconductor NCs.
3.2.1. Ligand Effects on Growth Kinetics. 3.2.1.1. Ligand Affinity. The binding strength of a surface capping ligand substantially affects its reactivity with metal precursors and in consequence influences both nucleation and growth kinetics of NCs formed in solution.19,24,97,161,163,175 Taking the aqueous synthesis of CdTe NCs as an example, the well-established synthetic process generally consists of three major steps: (1) preparation of a stock solution containing stabilizing agent and metal ion typically from a soluble metal salt under a proper pH; (2) introduction of the anion typically in the form of Na2E, NaHE, or H2E as chalcogens to initiate the reaction; (3) the growth of the NCs by aging the reaction mixture at a desired temperature, normally through reflux. All these steps are generally applicable for preparing most aqueous NCs, but depending on the material system, in some cases NCs are quickly formed after step 2. Nevertheless, throughout the whole preparative procedure the capping ligand will interact with metal ions either from molecular complexes before the precipitation reaction or on the surface of NCs after the formation of the NCs. To achieve effective size control, properly balancing the ligand−cation−anion interaction is important. The former part is characterized by a formation

10632

DOI: 10.1021/acs.chemrev.6b00041 Chem. Rev. 2016, 116, 10623−10730
NcsSynthesisSemiconductor NcsPropertiesAgents