2009 peng review

An Essay on Synthetic Chemistry of Colloidal Nanocrystals Xiaogang Peng（ ） Department of Chemistry and Biochemistry, University of Arkansas, Fayetteville, AR 72701, USA Received: 26 March 2009 / Accepted: 6 April 2009 ©Tsinghua University Press and Springer-Verlag 2009. This article is published with open access at Springerlink.com 00425 Nano Res (2009) 2: 425 447 DOI 10.1007/s12274-009-9047-2 Review Article Address correspondence to xpeng@uark.edu ABSTRACT The central goal of synthetic chemistry of colloidal nanocrystals at present is to discover functional materials. Such functional materials should help mankind to meet the tough challenges brought by the rapid depletion of natural resources and the signifi cant increase of population with higher and higher living standards. With this thought in mind, this essay discusses the basic guidelines for developing this new branch of synthetic chemistry, including rational synthetic strategies, functional performance, and green chemistry principles. KEYWORDS Colloidal nancrystal, synthetic chemistry, function materiols, green chemistry, crystallization Introduction The nanotechnology-based industrial revolution, if it is ever realized, will differ from any other industrial revolutions occurring in the last two centuries from a materials viewpoint. From the steam-engine, to electricity, to information technology, every previous industrial revolution was mainly founded on the innovation of physical concepts. However, nanotechnology is so diverse and the materials base, mostly nanomaterials, is so vast and new to mankind. As a result, the main efforts in nanotechnology and nanoscience must be synthesis, manipulation, and processing of nanomaterials, at least in its initial stage. Nanomaterials refer to numerous types of advanced materials with their physical dimensions in the nanosize regime that often matches the feature sizes associated with the targeted properties. The importance of nanomaterials can also be highlighted by the accelerated consumption of natural resources. Human society in the foreseeable future will always be limited to the earth, a more or less isolated system in the universe except for the seemingly endless photo-radiation from the sun [1]. As the population and living standard increase, we are under increasing pressure to uncover new and innovative means for utilizing all types of raw natural substances as functional materials, preferably with minimum impact to the environment on earth. Although it is too naïve to state that “nano”-materials imply the use of a tiny amount of materials, chemistry developed around synthesis and processing of nanomaterials should certainly offer human society smart pathways to build a much needed harmony with the natural world surrounding us, instead of poisoning our home and that of future generations. Among all types of nanomaterials, colloidal nanocrystals are probably the largest class at present. Colloidal nanocrystals are nanometer-sized fragments of the corresponding bulk crystals which are typically synthesized and processed as solution species. The properties of colloidal nanocrystals are Nano Research 426 Nano Res (2009) 2: 425 447 often found to be size dependent for various reasons. The first well known reason is that their intrinsic physical sizes are comparable to the critical sizes of many important properties of a given class of functional materials [2], such as the wavelength of the electron wavefunction, the diameter of photo- generated excitons, the domain size of magnetic single domains, etc. The second reason is their large surface-to-volume atom ratio, which considerably alters the chemical potential of the structural units in comparison to that for the corresponding bulk crystals [3]. The strongly size-dependent solubility of nanocrystals is a direct result of this property. The third reason is the size dependence of the structure in the nanometer regime, which includes electron band configuration, surface structure and reconstruction, and crystal structure, etc. The unique catalytic properties of gold nanocrystals [4] can be considered as an example of the third type of size- dependent properties. This variety of size-dependent properties coupled with solution-based processability make colloidal nanocrystals a major class of attractive “man-made” materials. Synthesis of colloidal nanocrystals with rationally controlled size and size distribution is obviously the first step for utilization of their size-dependent properties. It should be pointed out that, although we often talk about size-dependent properties, the reality is that size variations of nanocrystals can be in all three dimensions. Consequently, shape-control of nanocrystals is becoming an important topic in the synthetic chemistry of colloidal nanocrystals. In addition to rational control over size, shape, size/shape distribution, and other structural aspects of the targeted nanocrystals (“rational” in short), there are two basic rules in judging the significance of a new synthetic development for colloidal nanocrystals. The second rule is the functional performance of the nanocrystals (“functional” in short) because, after all, materials chemists are synthesizing nanomaterials in order to exploit their properties for mankind, instead of merely making an object of beauty. The third rule is green chemistry (“green” in short). Though this last rule has been more or less in everybody’s mind, it is still a good idea to spell it out in order to remind scientists in the field constantly. In a certain sense, our generation of scientists has been offered a unique opportunity to reinvent ways for dealing with natural resources, and very likely, only those ones with limited environmental impact will have a future in real life. Numerous excellent reviews with their focus on synthetic chemistry of colloidal nanocrystals have been published in recent years. Different from these review articles, this essay describes a personal perspective of the field, with its focus on the basic principles for developing the synthetic chemistry of colloidal nanocrystals. For this purpose, I will not limit myself to a given type of colloidal nanocrystals but, as the most promising and most pursued systems, colloidal semiconductor nanocrystals will be used as examples in most cases. Rather than attempting a comprehensive coverage of the relevant literature, a significant portion of the experimental results discussed will be from our own publications. 1. “Rational” synthesis Synthesis of colloidal nanocrystals is a combination of solution chemistry and crystal growth. At the moment, it is not clear which one of these two aspects is predominant, and quite possibly, it differs case by case. There is plenty of knowledge about solution chemistry that can foster the development of the synthetic chemistry of colloidal nanocrystals. However, crystallization is not well understood at present. As pointed out repeatedly in the literature, theories on crysta l l izat ion nucleat ion and growth can both differ by orders of magnitude from the experimental results [5, 6]. Without a solid foundation in the theory of crystallization, will it be possible for us to develop the necessary rational synthetic chemistry of colloidal nanocrystals? More importantly, such rational syntheses must also follow the other two rules mentioned above, namely “functional” and “green”. I will argue that, as the fi eld develops, the answer to this challenging question becomes more and more optimistic. There are several reasons that support this view. 1.1 The driving force for crystal growth The driving force for crystal growth, as pointed out 427Nano Res (2009) 2: 425 447 by Gibbs, is to minimize the total surface free energy of the system [5]. With the constant specifi c surface free energy approximation, the total surface free energy of a system is proportional to the total number of surface atoms in the entire system. Ultimately, the Gibbs Law (Eq. (1)) implies that, without kinetic barriers, a crystallization system should result in one single crystal in equilibrium with its saturated solution. In Eq. (1), the sum operation should include every facet area (Ai), with its associated specific surface energy σi, on each crystal in the solution. ∑Gsurface =∑σ i Ai ≈ σ∑Ai =minimum (1) Equation (1) illustrates that the key thermo- dynamic parameter controlling a crystallization system is the total surface free energy, which is approximately linearly related to the total surface area. The chemical origin of the surface free energy of a crystal comes from the dangling bonds of the surface atoms. In comparison to the interior atoms, the surface atoms on a crystal are missing at least one nearest neighbor in the lattice. Each of such missing coordination sites on the surface is regarded as a surface dangling bond. Although the surface dangling bonds can be partially compensated by the surface ligands and/or solvent molecules, the free energy difference caused by dangling bonds is quite large in a typical system because the synthesis often requires relatively weak ligands as will be discussed below. Simple mathematical estimation can show that the surface-to-atom ratio, or the approximate surface free energy contribution to the molar free energy of a crystal, decreases rapidly with the increase in the size of the crystals (Fig. 1) [3]. For a pure substance, the molar free energy is its chemical potential. To further visualize this, let’s take 0.2 nm as the average inter-atom distance in a simple cubic lattice for a crystal with a cubic shape. When the cube edge size is 2 nm, the surface-to-volume atom ratio is about 54.2%, which is very significant. As the cube edge size increases to 20 nm, the surface- to-volume atom ratio drops sharply to 5.9%. If the edge of the cube further increases to 2 μm, the surface-to-volume atom ratio will decrease to 0.06%! This indicates that, in the typical micron size range where scientists traditionally study crystallization, the surface free energy is close to being negligible in a crystal in comparison to the total free energy of the system. Consequently, it becomes very difficult to identify the surface free energy contribution in a crystallization system. This is likely to be one of the main reasons why crystallization has been so diffi cult to understand in the past. The above discussions in this sub-section suggest that a fundamental understanding of crystallization may be established by studying crystallization in the nanometer regime. In other words, although we have not reached a satisfactory understanding of crystallization at present, it is very possible that such a success is realistic in the nanometer regime. This will not only provide a necessary foundation for designing rational synthesis for high quality colloidal nanocrystals, but also solve the longstanding challenge to offer a quantitative framework for crystallization in general [7]. Unrelated to the topic but being an interesting perspective, one may argue that understanding crystallization will further impact other fundamental scientific fields, such as phase transitions, biomineralization, surface chemistry in solution, etc. 1.2 The first set of unique experimental tools for studying formation of colloidal nanocrystals The fi rst set of unique experimental tools for studying formation of colloidal nanocrystals originates from the size-dependent properties of colloidal nanocrystals. In fact, the very reason why scientists in the fi eld of crystallization were mostly limited to the micron size range is because of a lack of reliable and convenient tools to probe a crystallization Figure 1 The surface-to-volume atom ratio and chemical potential of nanocrystals calculated using InP lattice parameters Nano Research 428 Nano Res (2009) 2: 425 447 system in its initial stage that involves nanometer- sized clusters/crystals, namely nanoclusters and nanocrystals [5]. Although powerful microscopy techniques with atomic resolution have been continuously developed in the past several decades, such techniques are limited to a small number of crystals in a given set of experiments and often need to be performed under quite restricted conditions. For instance, the formation of “nuclei” (seeds of crystals with sizes of a few nanometers or sub- nanometer in size) would be very diffi cult to observe by microscopy studies. In a certain sense, however, one could argue that a crystallization system should be well defined by the initial boundary conditions, which is more or less its nucleation stage. As mentioned above, there is a broad spectrum of size-dependent properties for colloidal nanocrystals known to the fi eld. Among them, the most convenient ones are the size-dependent optical properties of semiconductor nanocrystals. This is so for several reasons. Firstly, spectroscopic methods became well developed in the last century and widely available in modern laboratories. Secondly, spectroscopic tools are non-invasive in nature and can explore a macroscopic system with ease. Thirdly, the size- dependent optical properties of semiconductor nanocrystals can be readily correlated with the size, shape, and size/shape distribution of a sample with great accuracy [8 10]. The s ize-dependent optical propert ies of semiconductors are due to quantum confinement. Detailed discussions of quantum confinement of colloidal semiconductor nanocrystals can be found in a great many publications [2, 11, 12], and here we will only provide a brief discussion of this phenomenon. In a piece of a semiconductor, no matter whether it is a bulk crystal or a nanocrystal, the valence electrons are largely delocalized over the entire body, instead of forming localized bonds. This is very much like the large conjugated systems encountered in organic chemistry. When an electron is excited by a photon with the right energy, the electron becomes free to move throughout the entire lattice, except for the fact that the atom losing this electron becomes positively charged (formation of a “hole”) and will try to hold the electron through electrostatic interaction. This electrostatic attraction makes the hole follow the electron, which is done by extracting an electron from a neighbor atom in the lattice. Obviously, movement of the hole is more diffi cult, and thus we say the hole is heavy. Overall, the electron moves rapidly around a slowly moving hole and this photo-generated electron-hole pair is called an exciton. An exciton is similar to a hydrogen atom but the average size of an exciton in semiconductor crystals is much larger than that of a hydrogen atom. This is so because the space between the photo-generated electron and hole is full of other atoms and electrons, instead of being a vacuum as in the case of a hydrogen atom. Typically, the size of an exciton is determined by the dielectric constant of the given semiconductor, the origins of the molecular orbital of the excited state and the ground state, the sizes of the atoms, etc. For example, the Bohr diameter of excitons in a bulk CdSe crystal is approximately 12 nm. When the physical size of a crystal becomes smaller than the intrinsic size of the corresponding bulk exciton, an exciton is effectively confi ned inside a box. As a result, just as for a typical particle-in-the- box solution, the energy levels of the exciton (particle) become discrete and the energy separation between the ground state and the fi rst excited state increases markedly as the physical size of the nanocrystal (box) decreases. This is the origin of quantum confi nement and quantum size effects in the case of semiconductor nanocrystals. Traditionally, when the size of a semiconductor nanocrystal is within the quantum confi nement size regime, we call it a quantum dot. For typical semiconductors, their bandgaps are in the optically active window. The first absorption peak and photoluminescence (PL) of the excitons lie slightly below the bandgap in a bulk semiconductor due to the contribution of the bonding energy of the excitons (or, the weak electrostatic interaction between the photo-generated electron and hole). Because of quantum confi nement, the absorption and PL spectra of quantum dots shift to the blue upon reduction of their sizes. As an example, Fig. 2 (a) shows a series of absorption and PL spectra of nearly monodisperse CdSe nanocrystals in the size range from about 2 nm to about 10 nm. The lowest excitonic 429Nano Res (2009) 2: 425 447 absorption peak in each spectrum is determined by the size of the nanocrystals (Fig. 2(b)). The sharp absorption features in each absorption spectrum and narrow peak width of the corresponding PL spectrum confi rm that the ensembles of nanocrystals used for recording the spectra were nearly monodisperse. For a nearly monodisperse sample, the particle concentration in the solution can be readily determined by the molar extinction coeffi cient of the nanocrystals [13] (see Fig. 2 (c) as an example). For studying crystallization, most nanocrystal samples cannot be truly monodisperse. As a result, it is necessary to extract size distribution information about the nanocrystals from the corresponding optical spectra if one wants to quantitatively define a crystallization system using the spectroscopic method outlined in the above paragraph. If the optical quality of the nanocrystals is decent, the PL of the nanocrystals should only have bandgap emission and it is thus a single peak (Fig. 2(a)). However, because of the uncertainty of the PL quantum yield of the nanocrystals with respect to the different sizes in the ensemble, history of the sample, and environment [14, 15], the single-peak feature of the PL spectra can only offer semi-quantitative information about the size distribution profi le for a given sample [16]. The absorption spectra of semiconductor nanocrystals, however, always have multiple and overlapping features (Fig. 2(a)). The size distribution information can only be extracted by computer deconvolution of the entire spectrum. However, i t i s impossible to f ind one set of standard spectra of truly monodisperse nanocrystals for any type of semiconductor nanocrystals in the literature. The recent deconvolution scheme reported by our group represents a solution to this problem [10]. Instead of using truly monodisperse samples as the reference samples, the UV vis spectra (see representative ones in Fig. 3(a)) of the best available quality samples of CdS nanocrystals were recorded. Transmission electron microscope (TEM) measurements indicated the particle Figure 2 (a) Absorption and photoluminescence (PL) spectra of different sizes of CdSe nanocrystals; (b) plot of size as determined by TEM vs. the first exciton absorption peak; (c) plot of molar extinction coefficient (per mole of particles) vs size of CdSe nanocrystals sizes of these samples had a standard deviation of about 5% 7%, which was used to build up a Gaussian distribution for each standard (Fig. 3(b)). By deconvoluting the UV vis spectra of a sample, a series of contribution factors for the standard spectra were obtained. The corresponding size distribution profi le of the test sample could thus be obtained by summing up the product of each contribution factor and the corresponding Gaussian distribution of particle size of the standard sample. As demonstrations, three samples with known size distribution and spectra were examined and comparisons of the expected and simulated results are sh