Physicochemical aspects of colloid and interface

ELSEVI E R

Current Opinion in Colloid & Interface Science 5 (2000) 70-73 www.elsevier.nl/locate/cocis

Editorial overview

Physicochemical aspects of colloid and interface science

..

1. Physicochemical.

The physicochemical aspects theme of this issue is very broad. The term of art, physicochemical, subsumes most topics that combine physical properties with some chemical step or some type of chemistry. Such aspects are further restricted in the present context, as they apply to colloid and interface science. Most physical measurement techniques that have been applied to bulk phases are directly applicable to colloidal solutions and suspensions, and further have been refined and modified in applications to interfaces. Techniques developed for intefaces, such as second harmonic generation methods for noncentrosymmetric systems (ideally articulated by interfaces), have been modified and applied successfully to colloidal suspensions [1°,200]. Various aspects of electrochemistry and dielectric spectroscopy [3] have been instituted as raster-scanning microscopies, augmenting AFM-related techniques. The contributions to this issue span a wide range of technology and instrumentation. The interfaces examined in these contributions are finite (particles, micelles, vesicles) through infinite (liquid-gas interfaces, electrode interfaces).

2. Planar interfaces

2.1. Spectroscopy A tour de force for non-linear optical techniques has been the extension of vibrational and electronic spectroscopy to planar interfaces using second harmonic generation (SHG) techniques. A key result is that these SHG methods probe vibrational, vibronic, and electronic structure, as conventionally probed in the bulk by infrared, Raman, and UV/Vis spectroscopy, but rely on fluorescence emission processes that yield high interfacial sensitivity. Allen et al. (this issue) extend the seminal contributions of Richmond's group in the analysis of

surfactant aggregation at liquid-liquid interfaces [4"] by vibrational sum frequency spectroscopy (VSFS) to the more detailed analysis of water at aqueous interfaces. Water, a non-normal liquid, is key to many interfacial chemistries, and the results reviewed here show how orientation of important solutes and interfacial structure may be directly probed by VSFS. 2.2. Dynamical manipulation of intefacial properties Marangoni effects arising from gradients in surface tension have been reported many times but have not been put to practical use. Ink jet droplet formation, emanating from thermal disruption of surface tension, is an active development area [51, and may offer a commercially important exception to this generalization. Rosslee and Abbott (this issue) demonstrate how photoactive and redox active amphiphiles can be used to control wetting and interfacial tension on different length and time scales. These efforts may lead to new methods of microcompartmentalization. These studies using electroactive surfactants are yielding molecular design criteria for creating surfactants and surfactant precursors that can provide surface tension gradients of sufficient magnitude to effect control of interfacial phenomena. Similarly, photolabile amphiphiles will lead to optically addressable materials and control processes.

2.3. Sufactant assembly and packing The structure of surfactant assemblies on surfaces has undergone some revolutionary changes in the past few years. Twenty years and more ago the accepted structure of hemimicelles was homologous to the structure of micelles, except for the imposition of a boundary facet from the surface. Hemimicelles were considered to be globular, axisymmetric assemblies growing out of a surface. With increased indirect experimentation, dogma shifted to a view that presumed hemimicelle, as well as individual surfactant

1359-0294/00/$ - see front matter 0 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 9 - 0 2 9 4 ( 0 0 1 0 0 0 3 1- 5

Editorial overview

adsorption, was nucleated and controlled by surfactant-surface coulombic interactions, and that surfactants preferred to adsorb head-down with a coulombic bond between opposing charges on the surfaces and the headgroup. This dogma was promulgated by many, mainly because of its ease of acceptance and because quantitative techniques such as neutron reflection, that lacked in-plane spatial resolution, could be more easily modeled assuming planar and bilayer types of adsorption modes. Indeed, the assumption of end-on monolayers as hemimicelles or of bilayer-type patches of admicelles is easier to conceptualize and to model, because of the greater order involved. This promulgation continued in spite of data readily available that showed the adsorption of charged surfactants onto surfaces of the same charge. Such data unequivocally support the real world picture of heterogeneous surfaces, and adsorption mechanisms involving dispersion forces rather than coulombic forces. Moreover, there are extant data (J-aggregation of dyes on variously charged surfaces) showing that cooperative dispersion force interactions can predominate in forming surface assemblies. In any case, the end of the last century brought to light data refuting the reigning dogma. Raman evidence [6"] showed unequivocally for the case of cationic pyridinium surfactants adsorbing onto negatively charged silver that in addition to a strong surface-headgroup interaction there was a strong surface-tail interaction. These surfactants did not adsorb as sticks normal to the surface, but adsorbed, if not flat on the surface, with at least two points of contact. Also, advances in double-layer force AFM, starting with the doctoral and postdoctoral work of Manne [7"] showed the existence of many different kinds of surfactant assemblies on well-ordered surfaces. In addition to some cases of 'end-on' and bilayer assembly (more the exception than the rule), globular, hemicylindrical, and cylindrical modes of surfactant surface assembly were observed. While now out of its infancy, this important approach to studying surfactants on surfaces is continuing, and Warr (this issue) gives us a very thorough review and update. He summarizes the extant data and provides several insightful correlations, such as the sensitivity of adsorbed layer morphology on surface energetics, effects of critical packing parameters, and competitive ionexchange effects. 2.4. Electrochemistry

Electrochemistry provides a venue for bringing together catalysis, interfacial science, and chemistry (forming and breaking covalent bonds). Exciting advances are steadily being made as microemulsions and layer-by-layer assembly techniques are put into practical applications for the design of novel electrocat-

71

alytic systems [8]. Lu and Wiekowski (this issue) give us an excellent overview of the breadth of such activities currently being pursued in fuel cells, water electrolysis, and hydrogenation. 3. Finite interfaces

3.1. Modeling of surfactant assemblies The modeling of surfactant packing has come a long way since the days when surfactants were constrained to form spherical micelles. Detailed all-atom simulations are still time consuming and restricted to very short time periods, but these constraints are steadily being eroded by advances in computing power and decreases in the associated costs. Model amphiphiles have been successfully simulated over sufficient time to demonstrate aggregate formation, but simulations are still too resource intensive to credibly model supramolecular aggregation and nanostructural transitions among competing equilibria such as reverse micelle and microemulsion cluster formation and transitions to irregular bicontinuous microstructure [9]. Shelley and Shelley (this issue) give us a very nice update on how issues on different length scales are currently being attacked. We're given a summary of atomistic simulations and key topical areas where such approaches have recently been applied, such as the hemimicelle formation discussed earlier by Warr. Also included are very useful discussions of coarse grain studies of longer lifetime phenomena and of dissipative particle dynamics applied to mesoscale processes of even longer lifetimes. 3.2. Enzymatic catalysis in surfactant systems The role of surfactant-protein interactions was shown to be pivotal in understanding the viscosity of aqueous gelatin as a function of added surfactant [lo]. Many of these interactions were well explained by the necklace model of Cabane [ll],where micelles are nucleated along the peptide backbone. Such micelles are also sites of cross-linking for two or more peptide strands, and the resulting networks provide interesting dynamics and dissipation. Savelli et al. (this issue) show how varied surfactant-enzyme interactions can be, and how these interactions affect all aspects of enzymatic activity. The details of these surfactantprotein interactions provide motivation for further extensive study of these systems. 3.3. Spectroscopy in surfactant systems

Most spectroscopies applicable to condensed solutions are applicable to surfactant solutions and

72

Editorial overview

mesophases, although some limitations are imposed by scattering effects. Adventitious dynamic light scattering [12] and acoustic scattering [13] provide basis for particle sizing. Fast processes probed by optical fluorescence in isotropic micellar and microemulsion solutions and in vesicles are reviewed by Levinger (this issue). It appears the introduction of mesoscales in length imposed by surfactant aggregation introduce longer relaxation times related to aggregate structural evolution.

4. Particle formation and characterization

4.1. Vesicles The impact of vesicles on technology development has been intense [14]. Drug delivery has been a major motivation [15]. Smart liposomes for chemical detoxification [ 161 represent another innovative application area. The discovery of thermodynamically stable vesicles [17" I has provided insight into bottom-up selfassembly principles. These principles have led to an understanding of physical limitations on the thermodynamic stability of unilamellar and double-lamellar vesicles [18] and to the development of lamellar disks [19']. Discher et al. (this issue) provide an important current update on the status of vesicles derived from polymers. Such vesicles excitingly offer an expanded range of properties, including dramatically increased mechanical stability.

4.2. Drug delivery Applications in drug delivery have been drivers for the development of a variety of hard and soft particle technologies, including smart liposomes and nanocrystalline drug suspensions [20]. Yang and Alexandridis (this issue) provide updates on delivery utilizing micelles of block copolymers, liposomes, polymeric beads, and hydrogels.

4.4. Precipitation Homogeneous nucleation theory has been very well modeled and tested as applied to vapor-liquid condensation processes [22]. However, application of these same theories to particle precipitation has not been very effective from a practical point of view, mostly because of difficulties in distinguishing nucleation and growth phases experimentally. Leubner (this issue) presents a brief review of this earlier art, and then recasts the general problem as a dynamical evolution algorithm that provides quantifiable observables. This approach promises to provide advances in nucleation modeling, as formulations can be more rigorously evaluated theoretically and experimentally. Crystal habit has empirically been found to traverse a wide range, from essentially spherical crystals to very high aspect ratio platelets and needles. Recent work by Scaringe and co-workers [23',241 has given some insight into developing surface active growth modifiers that restrain growth in certain directions, thereby leading to anisotropy. A variety of physical aspects affect anisotropic growth, and Adair (this issue) reviews these aspects in examining directed anisotropy in several different important systems.

4.5. Quantum size effects The electronic properties of nanostructures have attracted a lot of interest in the past decade, because of quantum size effects. We've seen that the visible spectrum can be tuned by modifying the size and passivation of quantum dots. These and related materials are being scrutinized for new optical applications and products. Grieve et al. (this issue) summarize both the synthesis and electronic properties of such nanoparticles and quantum dots, including single particle spectroscopy. John Texter Strider Research Corporation, Rochester, NY 14610-2246, USA

4.3. Interparticle potentials

Painvise particle interaction potentials are very well defined under various environmental conditions, such as continuous phase electrolyte concentration [21]. The direct measurement of such interactions have been thwarted by physical limitations connected with manipulating individual particles. Particles in the micron size range can now be manipulated, and their interactions with surfaces experimentally measured. Bike (this issue) reviews many of these advances, focusing upon total internal reflection microscopy and related evanescent wave light scattering microscopy.

References and recommended reading oo

of special interest of outstanding interest

[l] Wang H. Second harmonic generation studies of chemistry at liquid interfaces. Ann Arbor: University Microfilms, 1996; Thesis, Columbia University, 1996. [2] Wang H, Yan ECY, Borguet E, Eisenthal KB. Second haroo monic generation from the surface of centrosymmetric particles in bulk solution. Chem Phys Lett 1996;259:15-20. [3] Asami K. Dielectric relaxation spectroscopy of biological cell suspensions. In: Hackley VA, Texter J, editors. Handbook on

Editorial overview

ultrasonic and dielectric characterization techniques for suspended particulates. Westerville, Ohio: American Ceramic Society, 1998:333-349. [41 Conboy JC, Messmer MC, Walker RA, Richmond GL. An investigation of surfactant behavior at the liquid/liquid interface with sum frequency vibrational spectroscopy. Prog Colloid Polym Sci 1997;103:10-20. Silverbrook K. Liquid ink printing apparatus. US Patent 5,880,759, 1999. Tarazona A, Kreisig S, Koglin E, Schwuger MJ. Adsorption properties of two cationic surfactant classes on silver surfaces studied by means of SERS spectroscopy and ab initio calculations. In: Texter J, editor. Amphiphiles at interfaces. Darmstadt: Steinkopff Verlag, 1997:181-192. [71 Manne S. Visualizing self-assembly: force microscopy of ionic surfactant aggregates at solid-liquid interfaces. Prog Colloid Polym Sci 1997;103:226-233. Rusling JF. Mediated electro-organic synthesis in microemulsions. In: Texter J, editor. Reactions and synthesis in surfactant systems. New York Marcel Dekker (in press). Texter J. Supramolecular equilibria in microemulsions. Colloids Surfaces A Physicochem Eng Aspects 2000;167:115-122. Greener J, Contestable BA, Bale MD. Interaction of anionic surfactants with gelatin: viscosity effects. Macromolecules 1987;20:2490-2498. Cabane B, Duplessix R. Organization of surfactant micelles adsorbed on a polymer molecule in water: a neutron scattering study. J Phys (France) 1982;43:1529-1542. Langevin D, editor. Light scattering by liquid surfaces and complementary techniques. New York Marcel Dekker, 1992. Povey MJW. The application of acoustics to the characterisation of particulate suspensions. In: Hackley VA, Texter J,

.. ..

73

editors. Handbook on ultrasonic and dielectric characterization techniques for suspended particulates. Westerville, Ohio: American Ceramic Society, 1998:3-23. Rosoff M, editor. Vesicles. New York Marcel Dekker, 1996. Allen TN. Liposomal drug delivery. Curr Opin Colloid Interface Sci 1996;1:645-651. Jaeger DA, Schilling I11 CL, Zelenin AK, Li B, Kubicz-bring E. Reaction of a vesicular functionalized surfactant with 2-chloroethyl phenyl sulfide, a mustard simulant. Langmuir 1999;15:71-85. Kaler EW, Murthy AK, Rodriquez BE, Zasadzinski JA. Spontaneous vesicle formation in aqueous mixtures of single-tailed surfactants. Science 1989;245:1371-1374. Kaler EW, Zasadzinski JA. Private communication, 1999. Zemb T, Dubois M, Deme B, Gulik-Krzywicki T. Self-assembly of flat nanodiscs in salt-free catanionic surfactant solutions. Science 1999;283:816-819. 1201 Liversidge GG, Cundy KC, Bishop JF, Czekai DA. Surface modified drug particles. US Patent 5,145,684, 1992. Ohshima H. Interaction of electrical double layers. In: Ohshima H, Furusawa K, editors. Electrical phenomena at interfaces, 2nd ed. New York Marcel Dekker, 1998:57-85. Abraham FF. Homogeneous nucleation theory: the pretransition theory of vapor condensation. New York Academic Press, 1974. 1231 Scaringe RP, Miller DD, Brick MC, Shuttleworth L, Helber MJ, Evans S. Solid particle dispersions for imaging elements. US Patent 5,750,323, 1998. [241 Bagchi P, Scaringe RP, Bosch HW. Co-microprecipitation of nanoparticulate pharmaceutical agents with crystal growth modifiers. US Patent 5,665,331, 1997.

.

ELSEVI E R


Non-linear vibrational sum frequency spectroscopy of atmospherically relevant molecules at aqueous solution surfaces H.C. Allen19”,E.A. Raymondb,G.L. Richmondc’ aDepartmentof Chemistty, Ohio State Univemity, 100 W. 18‘h Aue., Columbus, OH 43210, USA bDepartment of Physics, Universityof Oregon, Eugene, OR 97403, USA ‘Department of Chemistry, University of Oregon, Eugene, OR 97403, USA

Abstract

Surface vibrational sum frequency spectroscopy has been shown to be a powerful surface probe of molecules adsorbed at solid and liquid surfaces. Studies described herein apply this method to studying heterogeneous air/aqueous solution interfaces to understand surface adsorption and structure of several solute molecules adsorbed at aqueous surfaces. The molecules examined at aqueous solution surfaces include dimethyl sulfoxide (DMSO), methane sulfonic acid (MSA) and acetone. These results reveal that small soluble molecules such as these organize in different ways at the surface of aqueous solutions. This surface organization has implications for atmospheric chemical processes since adsorption at the surface of atmospheric aerosols affects bulk chemical concentrations. 0 2000 Elsevier Science Ltd. All rights reserved. Keywords: Vibrational sum frequency spectroscopy (VSFS); Air-water interface; Dimethyl sulfoxide (DMSO); Methane sulfonic acid (MSA)

1. Introduction

The chemistry of the atmosphere involves both homogeneous and heterogeneous processes. Whereas atmospheric molecules have been studied extensively in the gas phase, very few spectroscopic studies have investigated these molecules on liquid surfaces of environmental relevance. In this paper we discuss studies of the molecular structure of atmospherically important molecules at an aqueous surface completed

+

+

*Corresponding author. Tel.: 1-541-346-4635; fax: 1-541346-5859. E-mail addresses: [email protected] (H.C. Allen), [email protected] (E.A. Raymond), [email protected] (G.L. Richmond). ‘Postdoctoral fellow for the NOAA Postdoctoral Program in Climate and Global Change.

by this research group [l-41. Specifically, adsorption and structural information from surfaces of aqueous solutions of dimethyl sulfoxide, methane sulfonic acid and acetone are presented. These molecules are present as trace constituents in the atmosphere, and while water soluble have significant surface activities. Due to the abundance of water in the lower atmosphere, understanding the molecular interactions that occur at aerosol surfaces is of importance. Recently, dimethyl sulfoxide has been proposed as the heterogeneous precursor to atmospheric condensed phase MSA through an atmospheric cycle originating with dimethyl sulfide, a phytoplankton degradation product [5-71. Aerosol particles containing MSA are thought to contribute to the class of aerosols which effectively scatter radiation out of the atmosphere [6,8]. Acetone, ubiquitous in many regions of the atmosphere, has recently been shown in regional studies in the North-

1359-0294/00/$ - see front matter 0 2000 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 9 - 0 2 9 4 ( 0 0 ) 0 0 0 3 3 - 9

H.C. Allen et al. /Current Opinion in Colloid & Inteflace Science 5 (2000) 74-80

ern Hemisphere to be the second highest concentration organic trace-gas constituent next to methane and may play a role in the growth and surface chemistry of atmospheric aerosols [9]. There has also been an ongoing effort to understand the atmospheric heterogeneous chemistry of acetone on condensed phase sulfuric acid [lo]. The molecular similarities of these molecules also provide interesting comparisons for understanding molecular adsorption of solute molecules on aqueous surfaces in general. Heterogeneous processes occurring at the gas/liquid interface are intrinsically difficult to study at the molecular level. The surface specific secondorder non-linear optical technique of vibrational sum frequency spectroscopy (VSFS) has proved in the past 10 years to be a powerful method for making such measurements. Recent advances in this technique, including shorter pulses and novel methods for generating infrared radiation have made the weak signals generated by molecules at these types of interfaces accessible [ll-131. The theory of vibrational sum frequency spectroscopy (VSFS), has been extensively described previously [14-191, so only a brief description will be discussed here. VSFS probes only those molecules existing in an environment which lacks inversion symmetry, such as the air/water interface. The intensity of the generated sum frequency light, IsF,is proportional to the square of the second-order macroscopic susceptibility, x'~': ISF

ak c i+

'IIRIvis

(1)

where NR and R denote the non-resonant and resonant pieces of the second-order susceptibility, and ZIR and Ivis are the intensities of the incident infrared and visible pulses that give rise to the SF response. xf' is proportional to the number of orientationally averaged molecules in the interfacial region and is resonantly enhanced when the incident infrared beam is resonant with vibrational modes of the interfacial molecules. Since the probability of producing a SF photon from the two incident pules (IR and visible) is low, improved detection techniques have been employed to improve the signal to noise levels. All of the spectra shown are taken using a SSP polarization combination, where SSP denotes the polarization of the sum frequency, incident visible and incident infrared beams, respectively (S polarized light oscillates perpendicular to the plane of incidence, while P polarization is in the plane of incidence). This polarization combination probes the second-order susceptibility xuz, which contains information on the resonant IR transition moments perpendicular to the surface and the isotropic Raman response. Thus the molecules in the interfacial region whose transition

15

moment is perpendicular to the surface plane will significantly contribute to a SSP sum frequency spectrum. Due to the coherent nature of the sum frequency process, the phase of the SF response must be included in the full analysis of the spectra to deconvolve the contributing interference effects [4,17,20,21]. It has been shown that under the SSP polarization condition, for vibrational modes of either C,, or C,, symmetry, the symmetric and asymmetric stretch modes will destructively interfere [20,221. Recall that the SF intensity is the square of a sum of terms, thereby producing cross-terms in the expression for the SF intensity. These cross-terms produce dips and asymmetric peaks in the spectra that are not observed in conventional IR and Raman spectroscopies. In the studies discussed here, where the methyl vibrational modes are all of C, symmetry, a destructive interference is observed between the methyl symmetric and asymmetric stretch peaks. 2. Experimental The laser system used for these VSFS experiments has been previously described in detail [2,11,12]. Briefly, two coherent beams (2-ps pulses at a kHz rep rate), one in the visible at 800 nm, the other tunable through the O H and CH stretching regions (2750-4000 cm-l), are spatially and temporally overlapped on the solution surface. The resulting SF beam passes through a series of spatial and wavelength filters, and is focused onto a thermoelectrically cooled CCD camera. In addition to vastly improving the signal to noise ratio, the CCD array allows for more precise spatial positioning and characterization of the SF response. The recent improvements in the IR generation and the use of the CCD camera for detection have made possible the measurement of the inherently low SF response from small molecules at air-aqueous solution interfaces.

3. Results and discussion

The SSP VSF spectrum from the surface of a pure dimethyl sulfoxide solution has been measured [11 and is shown in Fig. 1. The large peak at 2914 cm-' is assigned to the CH, symmetric stretches (SS) of DMSO. The CH, asymmetric stretches (AS) of the two methyl groups of DMSO are assigned to a peak at 2990 cm-l. All frequencies from the VSF spectra have an associated uncertainty of k.5 cm-l. The CH,-AS assignment was taken from the fit to the spectrum, which is shown with the data in Fig. 1 and is broken down into composite peaks in Fig. 2. These

H. C. Allen et al. / Current Opinion in Colloid & Interface Science 5 (2000)74-80

76

-I\

-.-

4000

2

3000-

c

c

2 c .-

-e.g

',

"

2000-

c

i'

Q)

c

12

LL

rJY

1000-

N

4 0 3000

3200

3400

3600

3800

Wavenumber (cm-')

Fig. 1. SSP VSF spectrum of pure DMSO. The triangles represent data points, while the solid line is the best fit to the data. The fit includes the appropriate phase relationships between the contributing peaks.

frequencies are in close agreement with the frequencies found in bulk Raman and IR studies [23]. Due to improved laser and detection instrumentation, the frequency measured here for the CH,-SS peak is somewhat different and more accurate than previously measured in this laboratory [l]. The asymmetry of the CH,-SS peak, specifically the sharper SF intensity drop on the high energy side, and the following dip in the spectrum are attributed to the destructive interference between the CH,-SS mode and the CH,AS mode. As expected, there are no other readily discernable features in either the CH stretching region or the higher energy region shown. Fig. 2 shows more clearly the region assigned to the destructive interference between the CH,-SS and the 20001

CH,-AS, and the peaks which make up the fit to the spectrum shown in Fig. 1. The large intensity of the symmetric stretch relative to the asymmetric stretch indicates that the methyl groups are oriented predominantly out of the interface. Further studies with varied concentrations of DMSO and the other solutes have been conducted. For these spectra, water molecules at the surface also contribute. Therefore, a brief discussion of the SSP spectrum of neat water (Fig. 3) is warranted. The broad band from 3000 to 3600 cm-' is assigned to the broad distribution of OH hydrogen bonding stretching modes in which the oxygen is tetrahedrally coordinated [24-291. The energy region from 3000 to 3250 cm-' is attributed to strong intermolecular hydrogen bonds of water molecules which give rise to a highly correlated hydrogen bonding network, assigned to OH symmetric stretches, vl. The higher energy broad band region ( 3250-3600 cm-') is assigned to more weakly correlated hydrogen bonding stretching modes of molecular water that encompasses both vl (OH symmetric stretch), and to a lesser extent, v3 (OH asymmetric stretch). The peak at 3702 cm-' is assigned to the OH dangling bond, or free OH stretch, of 3 and 2 subsurface-coordinated molecules [26,301. On either side of the dangling OH are the symmetric and asymmetric stretches of vapor water molecules, oriented with their hydrogens towards the water surface [2]. These two modes have a well-defined phase relationship to each other [4], similar to that of the methyl symmetric and asymmetric stretches. In addition, the SF signal from the oriented vapor and that from the dangling OH interfere, producing the lack of intensity at 3650 cm-' and the shoulder at 3760 cm-'. A spectrum of a 0.1 mole fraction (mf) solution of DMSO in water is shown in Fig. 4. The CH,-SS is

I

-

N

N

I

50 -

J = J,,exp[ - 16y3K2/3k3T3(lnS)2] No Input parameter No No

Input parameter N (nuclei/s) No Maximum growth rate Output parameter Yes Yes

3. Other models

Klein-Moisar and related models were discussed by Leubner et al. [61.

3.1. Klein and Moisar model

3.2. The primitive model Klein and Moisar [15.], Kharitanova et al. [16], and Sugimoto [171 derived equations that attempted to correlate the stable number of crystals formed in controlled batch precipitations as a function of precipitation conditions:

Z

= KRRgT/yDVmC,

Another model for Z can be derived if one assumes that in a limited nucleation time, to,and crystal nuclei of the size ro are formed. Using the mass balance between material addition rate and the mass of nascent nuclei leads to Eq. (3):

(2)

Z K = 1 . 0 / 8 ~[15.] = 3 . 0 / 8 ~[16] = 1 . 0 / 5 . 9 ~[17] The new variables introduced are 2, the number of stable crystals; R , the reactant addition rate (dmol/dt); Rg, the gas constant; D, the diffusion coefficient; V,, the molar volume (cm3/mol, crystal); and C,, the solution equilibrium concentration, as defined above, of the rate determining reactant. Eq. (2) assumes diffusion controlled crystal growth and spherical morphology of the crystals. A comparison of these models with the classical and the BNG models is shown in Tables 1 and 2. Frequently, non-linear correlations between the experimental crystal number and the reaction variables, R, T , and C,, are observed, which in Eq. (2) need to be accommodated by arbitrary exponents. In addition, the temperature dependence of the number of crystals is inconsistently predicted to increase with temperature or to be temperature independent [6]. The balanced nucleation and growth (BNG) model developed by the author and co-workers overcomes these difficulties and leads to greater insights [5',6]. Eq. (2) is a limiting case of the BNG model, where r/r* is constant and independent of molar addition rate, R , temperature, T , and solubility, C,. Here, r is the average crystal size, and r* is the critical size, which has equal probability to grow or to dissolve under the precipitation conditions. The limitations of the

= RtoVm/kuri

(3)

Here, k,, is the volume constant which converts the crystal size into crystal volume. This equation does not explicitly include important reaction variables, like temperature and solubility that control the nucleation process. Generally, arbitrary assumptions must be made about either to or ro. Also, the nucleation rate depends on the supersaturation during nucleation and is not a linear process as assumed for Eq. (3). 4. Balanced nucleation-growth (BNG) model

The balanced nucleation and growth (BNG) model combines nucleation and growth in the nucleation phase for controlled precipitations. It is concerned with modeling of the initial phase of the formation of stable crystals in the controlled precipitation of organic and inorganic materials, and the relation of the number of crystals formed with the control variables of the precipitation. 4.1. Assumptionsfor modeling

In Table 3, a comparison of the modeling for the classical and the BNG model is shown. It is apparent,

I.H. Leubner / Current Opinion in Colloid & Interface Science 5 (2000)151-159

154

Table 4 The balanced nucleation-growth modela Time dt

Supersaturation SSI

Nucleation

N,

Growth G,

Supersaturation

Comments

s, 0 Only nucleation Growth and nucleation Growth and nucleation Growth and nucleation End of nucleation

aAbbreuiations:SS,, starting supersaturation; Si,ending supersaturation; N,, number of crystals formed; G,, maximum growth rate in time interval dt = i.

that they use some common initial conditions. The two models separate where the classical model introduces the Arrhenius equation to model the nucleation rate J , and where the BNG model introduces the (maximum) growth of the crystals during nucleation. The classical model may be used to model the nucleation rate in the BNG model; however, generally not enough information is available to solve the classical model for J [Eq. (01. Certain assumptions will be made to arrive at a model that can be mathematically evaluated: The molar addition rate, R,, is constant during the nucleation phase. During the first time interval, solely nucleation takes place. For convenience, first order kinetics is assumed for the chemical reaction leading to the crystallizing product. The nuclear cluster (nucleus) is modeled by the classical model. Until the end of nucleation, formation of new crystals and growth of existing crystals compete for the available unreacted material (supersaturation). During the nucleation period, all crystals grow at maximum growth rate. For the convenience of the present modeling a constant maximum growth rate, G,, will be assumed during the nucleation-growth phase. At the last step of nucleation, the formation of new crystals ends and the existing crystal population is at maximum growth rate. After nucleation stops the growth rate of the crystal population drops below the maximum growth rate. Several of the above assumptions may be modified to accommodate different systems. This includes different maximum growth rates, a knowledge of nucleation rates during nucleation kinetics of the reaction, knowledge of supersaturation, and any

other information that can be quantified in the model. The mathematical treatment to model the nucleation rates during the nucleation period is shown in Table 4 [MI. It should be noted that a similar nucleation and growth model was developed for vapor-to-liquid nucleation processes within the diffusion cloud chamber [19,20]. In that approach, both nucleation and growth processes were considered and the effect of both vapor depletion and latent heat effects were included. Eq. (1) was used as the basis of the experiments and for the evaluation of the experimental results. The surface energy of the droplets was used as the adjustable parameter and was experimentally determined. An important difference between the case considered here and the diffusion cloud chamber is that the latter permits a quantitative measurement of the actual nucleation rate [21]. This is generally not possible for liquid/solid nucleation. In the cloud chamber model the growth process was modeled after nucleation step while in the present model nucleation and growth compete simultaneously for added reactant. 4.2. Growth rate and maximum growth rate

Before the BNG model is derived, it is necessary to discuss the growth reaction during the nucleation process. The material balance of added material and consumption by growth generally determines the growth rate. This definition is also true for the maximum growth rate. G = dr/dt

= RVm/3.0k,rzZ

(4)

where G is the growth rate which is defined by the change in crystal size, dr, per time unit dt. R is the


molar addition rate (dmol/dt), V, is the molar volume (cm3/M), k, is the crystal surface factor, which converts crystal size, r, to surface area and Z is the number of crystals in the reactor. Since the growth rate G is derived from a material balance, it is independent of temperature, solubility, diffusion, convection, and other reaction conditions. In controlled double-jet precipitations, R is given by the addition rate of the reactants. In systems where the material is formed by chemical reactions, R is given by the reaction kinetics of the system. For systems where crystallization is initiated by temperature changes, R is given by the change of supersaturation with the rate of temperature change. Unlike the value of the growth rate G below maximum growth rate, the maximum growth rate, G,, is a function of temperature, solubility, material, and the mechanism of growth, i.e. diffusion or kinetically-controlled growth. Strong and Wey [22-241 developed a growth model for maximum growth rate for silver halides. This model includes the effects of growth mechanism and was used to derive the general BNG model and the special cases of nucleation under diffusion and kinetically-controlled growth conditions [5',61.

G,

= 2yVmC,K,(1.0- r*/r)/R,Tr*(l.O

E =

KJDV,

- E/r)

(6)

5. Crystal number in the BNG model Using the maximum growth rate, G,, and the mass balance of reactant addition leads to the fundamental correlation between molar addition rate, temperature, solubility and other reaction variables:

-

[ /(r2r*)n dr/(r + 1 / ~ )

1

/ n d r / ( r + 1 / ~ )/RgT

the crystal size distribution at the end of nucleation is known. This is generally not possible, and to solve this equation analytically, the following substitutions were made [5']:

Z = / n dr

(8)

Z r = /nr dr

(9)

where Z was defined earlier as the number of stable crystals formed in the reactor. Since growth is an essential part of the BNG model [Eqs. (5)-(7)] the nucleation step is a function of the growth mechanism. The equations for the two limiting cases, diffusion- and kinetically-controlled growth, were derived [5']. 5.1. Nucleation under difision-controlled growth conditions

(5)

Here, Ki is the surface integration (reaction) constant, and E represents the relative resistance of bulk diffusion to surface reaction. The other variables and constants were defined for Eq. (2). In Eq. (5), r is the average crystal size and r* is the critical crystal size, which is smaller than the average crystal size r. Crystals with the size r* have equal probabilities to grow and dissolve in the reaction system.

R = (2.Ok,yCSDV,)

155

With the appropriate substitutions [Eqs. (7)-(9)] one obtains:

Z =RRgT/2k,yV,C,D(r/r*

The integrals may be experimentally determined if

(12)

where Z = number of stable crystals; R = molar addition rate; Rg = gas constant; T = temperature (K); k, = surface shape factor; y = surface energy; D = diffusion coefficient of the reaction controlling reactant; V, =crystal molar volume; C, = sum of the solubility with regard to the reaction controlling reactant; r = average crystal size; r* = the critical crystal size at which a crystal has equal probability to grow or to dissolve by Ostwald ripening. The critical crystal size can be a function of molar addition rate, R, solubility, C, and temperature, T. If Z is constant during the precipitation, r/r* is constant. That is, r* changes with r. From the experimental determination of r*, the critical supersaturation ratio, S* , can be calculated:

S* (7)

- 1.0)

= C*/C, = 1.0

+ 2.0yVm/RgTr*

(13)

If the supersaturation can be determined experimentally or otherwise, it may be possible to calculate r*.


156

5.2. Crystal formation (nucleation) as a function of addition rate, solubility, and temperature for difision controlled growth /nucleation 5.2.1. Addition rate dlnZ/dlnR

= 1.0 -

(R/(r/r*

-

1.0))

x (d(r/r* )/dR)

(14)

5.2.2. Solubility dlnZ/dlnC,

=

-

[1.0

+ (C,/(r/r*

-

1.0))

x (d(r/r* )/dC,)I

(15)

5.2.3. Temperature dlnZ/dlnT

= - (T/(r/r*

X

6. Crystal formation (nucleation) in the presence of ripener

If a compound is added to the reactor before nucleation and a smaller number of crystals are obtained than without this compound, this compound is defined as a ripener. For otherwise equal precipitations, this results in larger crystals. These compounds generally accelerate Ostwald ripening of the crystals. 6.1. Corollary

- 1.0))

(d(r/r* )/dT)

(16)

These equations predict that plots of log2 vs. logR, loge, and logT will lead to linear correlations. For correct correlations, only R , C,, and T must be varied while at the same time keeping the other variables constant. These predictions were confirmed for silver bromide [6] and silver chloride [7]. 5.3. Nucleation under kinetically-controlledgrowth conditions

And with the appropriate substitutions [Eqs. (8)-(10) and (17) into Eq. (711 one obtains

(18)

2 =RRgT/2ksCsKi(ri/r* - r )

where ra is the area weighted average crystal size, and r is the number weighted averaged crystal size. For monodisperse systems or those close to monodispersity, it may be assumed that ra r, and the equation simplifies to:

-

Z

all reaction variables and r/r* are constant. This was experimentally confirmed [5']. From the experimental results where growth changes from kinetic to diffusion controlled, K, was determined for a precipitation of silver iodide [5'].

= RRgT/2k,C,K,.r(r/r* -

1.0)

(19)

If a compound is not present (removed) in the reactor before nucleation and a greater number of crystals is obtained, this compound acted as ripener. For otherwise equal precipitations, this omission of ripener leads to smaller crystals. This may mimic the effect of restrainers (see also definition of growth restrainers in Section 7). Ripeners increase the solubility of the precipitation reaction. For diffusion-controlled nucleation the following equation was derived [lo]: If

S, I So, then Z

If

S,

= 2,

(21)

+ log(aK)

(22)

> So, then:

log2 =

-

(nk,)logR,

where Z is the crystal number with ripener; Z, is the crystal number without ripener; S, is the solubility with ripener; So is the solubility without ripener; R, is the ripener concentration; n, k, a, K are defined as constants for particular reaction conditions. The condition (21) and Eq. (22) predict that at low So), the crystal number is ripener concentrations (S, I independent of ripener concentration. Above a limiting ripener concentration (S,>So) the number of crystals decreases logarithmically linear with the ripener concentration. This predicted correlation was experimentally confirmed [lo].

This can be further modified by multiplying both sides of the equation with the average crystal size, r: Zr =RRgT/2k,C,K,(r/r* - 1.0)

(20)

Eqs. (18)-(29) predict that the number of crystals decreases as the average crystal size increases so that the product Zr is constant under the conditions that

7. Crystal formation (nucleation) in the presence of growth restrainers

If a compound is added to the reactor before nucleation and a greater number of crystals is obtained than without this compound, this compound is defined


as a restrainer. For otherwise equal precipitations, this results in smaller crystals. These compounds generally retard growth of the crystals.

157

seed crystals, respectively. The values of b and c are defined as: a = RgT/2k,yVmC,D(r/r* - 1.0)

(26)

Z 1. Corollary

If a compound is not present (removed) in the reactor before nucleation and a smaller number of crystals is obtained, this compound acted as restrainer. For otherwise equal precipitations the omission of restrainer leads to larger crystals. This may mimic the effect of ripeners (see also the definition of the ripener in Section 6). These compounds adsorb to the crystal surface and retard growth. For Langmuir-type adsorption and nucleation under kinetically-controlled growth conditions, the following equation was derived [ll]:

where 2 is the total crystal number with restrainer; 2, is the total crystal number without restrainer; K is the defined nucleation factor without restrainer for the precipitation conditions; KL is the Langmuir adsorption constant; Kiis the kinetic growth constant; C is the growth restrainer concentration. Eq. (23) predicts that the number of crystals increases linearly with restrainer concentration, C. For the limiting condition without ripener, C equals zero, Z is equal to 2,. The predictions by this equation were experimentally confirmed [ 113.

b = 3.0kur~G,/V, c = 3.0kuG,S,/k,

V,

= 3.OGm/r,

(28)

The size r is the average size of the nucleated crystals, and r, is that of the seed crystals, S , is the molar surface area of the seed crystals. Eqs. (24) and (25) predict the conditions where nucleation does and does not occur. Where renucleation occurs, the number of crystals formed is linearly proportional to the molar addition rate. Negative values of Z, indicate the resistance of the system to renucleation. For a 2, value of zero, b and c can be determined and the maximum growth rate of the seed crystals calculated These predictions were experimentally confirmed [lll. 9. Non-seeded crystallization in the continuous stirred tank reactor (CSTR)

A comparison of the predictive potential of the Randolph-Larson and the BNG models are shown in Table 2.

9.1. The Randolph-Larson model 8. Renucleation

When crystals are grown by the addition of material, new smaller crystals are observed when the addition rate exceeds the maximum growth rate of the seed crystal population. This process of new crystal formation is defined as 'renucleation'. The BNG model was used to model the number of new crystals formed, 2, as a function of the seed crystal population (2,= crystal number, M, = mass in molar units) and addition rate, R. Two related equations were derived [ll]: Z,

= aR - abZ,

(24)

and 2, = aR - acM,

(25)

The values of a and b are related to the precipitation conditions and the maximum growth rate of the

The Randolph-Larson model was derived to model the crystal size distribution in a continuous stirred tank reactor (CSTR) under steady-state conditions. This reactor is also sometimes referred to as MSMPR (continuous mixed-suspension mixed-product removal) crystallizer. Eq. (29) was derived to model the number-frequency distribution, n, of the product size distribution in CSTR crystallizers [25',26',27].

Here, n is the population density at size L,, no is the population density, G is the growth rate, and T is the residence time. Newly introduced is the residence time, T , which is equal to the reaction volume divided by the total input (= output) flow rates. From the slope and the residence time, the growth rate G can be determined. The intercept at L, = 0 gives no. Eq. (29) predicts that a plot of log ( n ) vs. L, will give a linear correlation. It was shown that this model does not correctly describe the crystal size distribution for CSTR precip-


158

itations of silver halides. The experimentally-obtained size distribution was generally relatively symmetrically distributed around an average crystal size [13']. This crystal size distribution did not follow the linear log(n) vs. L, predicted by Eq. (29). Extrapolation to L, equal to zero to determine no became meaningless. Furthermore, Eq. (29) does not explicitly relate the crystal size distribution and average crystal size to reaction conditions like molar addition rate, R , solubility, C,, and temperature, T . Equally, this equation does not contain variables or constants that relate to the composition of the crystals. The growth rate, G , in Eq. (29) is a mass-balance variable and is not related to the precipitation conditions. Altogether, Eq. (29) did not yield information that could be meaningfully related to the results of precipitation of silver halides [13'1. 9.2. The BNG model

While the Randolph-Larson model [Eq. (2911 is concerned with the crystal size distribution, Eq. (30) obtained from the BNG model correlates the average crystal size at steady-state in the CSTR reactor [13']. k,r3RgT- 2k,-yDV~C,(r/r*- 1.0) - 3k,G,RgT /r2T = 1.0

(30)

Solving for the average crystal size, r, leads to an unwieldy third-order equation. Thus, for a set of experiments where the residence time, 7,was varied, the equation was solved for T : T = r/3Gm - 2ks-yDV,'Cs(r/r* -

/ 3 k, G , RgTr2

1.0) (31)

Similarly, the reaction can be solved for the temperature ( T ) and solubility (CJ. The unknown G, and r/r* are obtained by inserting the constants and experimental variables. Eq. (31) predicts among others that for zero residence time (plug-flow nucleation) r will be larger than zero. In contrast to the Randolph-Larson model, this equation predicts that the average crystal size is independent of reactant addition rate, suspension density, and reaction volume. These and other predictions were experimentally supported [13'1. From the experimental results, the ratio of nucleation to growth, the size of the nascent nuclei, the critical crystal size, r* , supersaturation, supersaturation ratio, and the maximum growth rate were calculated. Before the advent of this model, these data could not be obtained for controlled continuous crystallizations.

All models assume stirring conditions that result in a perfectly mixed reaction system. Stirring models that quantitatively connect with nucleation models are presently not available. 10. Future directions

The availability of several crystallization models opens a wide range of new research. The BNG model provides the most comprehensive range of correlations and predictions. It will stimulate research to determine surface energies, solubilities, and diffusion properties for research and industrial crystallization systems. It allows one to more accurately determine the maximum growth rate of crystals, which will result in improved control of precipitations. Crystallizations that have until now been fitted to the classical standard models will need to be evaluated considering the newer BNG model. New experiments will be needed to fully define precipitations for experimental and industrial control. The BNG model for continuous precipitations may be applicable to the formation of rain droplets, sleet, hail, and snow to improve meteorological predictions. The new models will result in efficiencies of research and product development. The predictive correlations will allow one to minimize the number of experiments that are needed to define number (size) correlations with experimental variables. Critical breaks in nucleation behavior, like the transition from octahedral to twinned AgBr crystals, may be predicted and may be used to optimize crystallizations. The author is presently modeling the CSTR crystallizer for seeded emulsions with and without spontaneous nucleation [28].Further extended modeling will be stimulated by experimental results and the need to model new precipitation systems. 11. Conclusion

The nucleation/growth phase of the balanced nucleation-growth process was modeled. Nucleation rate, maximum growth rate, and supersaturation were modeled as a function of time where molar addition rate, R,, initial nucleation rate, Ni, nucleation efficiency, F,, and growth rate G, were used as adjustable parameters. The BNG model correctly describes experimental results that many crystallization processes lead to a limited number of crystals during a nucleation period followed by growth. The model predicts that factors that affect the mechanism of the maximum growth rate will affect the nucleation outcome. This was confirmed for sepa-


rate nucleation models for diffusion and kineticallycontrolled growth process [YI. The inclusion of growth processes also predicts that experimental effects that affect growth during the nucleation period will affect the number of stable crystals formed. Thus, the effect of Ostwald ripening agents and growth restrainers was modeled in the BNG model. The predictions of the crystal number/concentration correlations of the models are in agreement with experimental results [10,11]. The linkage of nucleation and growth in the nucleation process leads to a model of the competition between heterogeneous and homogeneous nucleation as exemplified by renucleation in seeded systems. The modeling with the BNG model leads to improved experimental determination of the maximum growth rate of crystals. Balanced nucleation and growth model was also useful to model the results of continuous precipitation processes and led to new insights and mathematical correlations. The BNG model provides experimental results that cannot be obtained from the classical Randolph-Larson (R-L) model. The models complement each other with the potential to model both the fundamental variables (BNG) and crystal size distribution (R-L). The balanced nucleation and growth mechanism has led to many new correlations and predictions for crystallization processes and added new insights to the processes that affect the transient phase of the nucleation process. References and recommended reading 00


[l] Mullin JW. Crystallization. 3rd Oxford: Butterworth-Heinemann, 1993. [2] Hurle DTF, editor. Handbook of crystal growth, la. NorthHolland, Amsterdam-London-New York-Tokyo, 1993:187. [3] Mullin W. Crystallization. 3rd Oxford: Butterworth-Heinemann, 1993:172. [4] Mutaftschiev B. Nucleation theory. Handbook of crystal growth, Hurle DTF editors. North-Holland, AmsterdamLondon-New York-Tokyo, 1993:1a:187. A good reference for the derivation and extension of the classical nucleation model. [5] Leubner IH. Crystal formation (nucleation) under kinetically and diffusion controlled growth conditions. J Phys Chem 1987:91:6069. Contains a critical comparison with the classical nucleation model. [6] Leubner IH, Jagannathan R, Wey JS. Formation of silver bromide crystals in double-jet precipitations. Photogr Sci Eng 1980;24:26.

159

[7] Leubner IH. Formation of silver halide crystals in double-jet precipitations: AgCl. J h a g Sci 1985;29:219. [8] Leubner IH. Crystal formation (nucleation) of silver halides. Comparison of models. Proceedings of the Society for Imaging Science and Technology (IS&T) 45th Annual Conference, East Rutherford, NJ, 199213. Contains the information on the prediction of the transition from octahedral to twinned crystal morphology for AgBr. [9] Leubner IH. Number and size of AgBr crystals as a function of addition rate. J Imag Sci Technol 1993;37:268. [lo] Leubner IH. Crystal formation (nucleation) in the presence of Ostwald ripening agents. J Imag Sci 1987;31:145. [ l l ] Leubner IH. Crystal formation (nucleation) in the presence of growth restrainers. J Cryst Growth 1987;84:496. [12] Leubner IH. Crystal growth and renucleation. Theory and experiments. J h a g Sci Technol 1993;37:510. [13] Leubner IH. A new crystal nucleation theory for continuous precipitation of silver halides. J h a g Sci Technol 1998:42355. A critical comparison with the Randolph-Larson model in [23] is given here. [14] Katz JL, Donohue MD. Adv Chem Phys 1979;40: 137. A kinetic approach to the classical nucleation model for ionic solutions. [15] Klein E, Moisar E. Ber Bunsenges Phys Chemie 1963;67: 0 349. The limitations of the Klein-Moisar model and of ref. [16] were discussed by Leubner et al. [6]. Those limitations also apply to the model of Sugimoto [17]. [16] Kharitanova AI,Shapiro BI, Bogomolov KS. Z Nauchn Prikl Fotogr Kinematogr 1979;24:34. [17] Sugimoto T. Proceedings of the 11th Symposium on Industrial Crystallization. Garmisch-Partenkirchen,Germany, 1990. [18] Leubner IH. The balanced nucleation and growth model for crystallization. Private communication, submitted for publication, 2000. [19] Brito J, Heist RH. Chem Eng Commun 1982;15:133. [20] Heist RH, Kacker A, Brito J. Chem Eng Commun 198q28:117. [21] Heist RH, Kacker A. J Chem Phys 1985;82:2734. [22] Wey JS, Strong RW. Photogr Sci Eng 1977;21:248. [23] Wey JS, Strong RW. Photogr Sci Eng 1977;21:14. [24] Wey JS, Strong RW. Photogr Sci Eng 1979;23:344. [25] Randolph AD, Larson MA. AICHE J 1962;s: 639. The original derivation of the Randolph-Larson model for continuous crystallizations. [26] Bransom HS, Dunning J, Millard B. Disc Faraday SOC 0 1949;5:83. The original derivation of the Randolph-Larson model for continuous crystallizations. [27] Randolph AD, Larson MA. Theory of particulate processes, analysis, and techniques of continuous crystallization. 2nd San Diego, CA: Academic Press, 1991. A review and summary of the Randolph-Larson model. [28] Leubner IH. Growth and nucleation in the CSTR crystallizer for seeded precipitations without and with spontaneous nucleation. Private communication, manuscript in preparation.

ELSEVI E R


Morphological control of particles James H. Adair", Ender Suvaci Pennsylvania State University, University Park, PA 16801, USA

Abstract

The objective of this review is to highlight the theoretical and practical aspects of particle morphological control. Materials with directional properties are opening new horizons in material science. Structural, optical, and electrical properties can be greatly augmented by the fabrication of composite materials with anisotropic microstructures or with anisotropic particles uniformly dispersed in an isotropic matrix. Examples include structural composites, magnetic and optical recording media, photographic film, and certain metal and ceramic alloys. The new applications and the need for model particles in scientific investigations are rapidly outdistancing the ability to synthesize anisotropic particles with specific chemistries and narrowly distributed physical characteristics (e.g. size distribution, shape, and aspect ratio). Anisotropic particles of many compositions have been produced but only a few (y-Fe,O, and AgI) are produced with any degree of chemical and physical control. These two examples are the result of literally decades of study. Unfortunately, the science and technology (mainly the technology) that have evolved are maintained as proprietary information. Thus, while we generally know what systems yield single crystal, anisotropic-shaped particles, we do not know how to make powders of these crystals with the desired control of shape uniformity, aspect ratio and phase composition. Particle shape control is a complex process requiring a fundamental understanding of the interactions between solid state chemistry, interfacial reactions and kinetics, and solution (or vapor) chemistry. During synthesis of other than a large single crystal the parameters controlling crystal growth must be balanced with the requirements for anisotropic powder nucleation and growth. Although there has been considerable progress in large single crystal growth and the synthesis of powders composed of monodispersed, spherical particles, these efforts have not often been transferred to the synthesis of anisotropic particles. 0 2000 Elsevier Science Ltd. All rights reserved. Keywords: Particle morphology control; Crystal growth; Growth directed synthesis; Template directed syntheses

1. Introduction morphology

- approaches

to control particle

There are essentially two ways to approach particle shape control: growth directed syntheses typical of precipitation processes and template directed syntheses wherein the growth is directed by epitaxy via a pre-existing structure upon which nucleation and growth take place. In the past year, there have been a number of reports about processes. Much of the *Corresponding author. Tel.: + 1-814-863-6047; fax: 863-9704. E-mail address: [email protected] (J.H. Adair).

+ 1-814-

attention has been focused on growth directed processes with only a relatively few reports associated with template directed growth. Some of the relevant theoretical work will be reviewed, followed by a discussion of growth directed syntheses and template directed particle formation studies.

2. Theoretical aspects

There have been several studies focused on predicting morphological forms of particles [l-71 One of the most elegant is that of Noh et al. [l"] in which a numerical analysis of the role of flow of the supersaturated solution was analyzed. It was predicted that


J.H. Adair, E. Suvaci /Current Opinion in Colloid & Interjace Science 5 (2000) 160-167

uniaxial flow would produce oblate spheroids while biaxial flow would produce prolate barrel-like shapes. Wickham et al. [2'] showed that shape change for nanometer size CdSe particles could be monitored by X-ray diffraction. Another important theoretical analysis was conducted by Privman et al. [3"1, in which the formation of spherical particles via aggregation was modeled. It was demonstrated that the theoretical predictions were in reasonable agreement with experimental data in the formation of spherical gold particles. Templated growth was modeled by Mandel et al. [4"1 via an approach that employed molecular modeling between metal atom core and ligands. In addition to the theoretical papers directly related to modeling of morphological form, there have been several accounts related to a better understanding of the role of the solution and adsorbates [S-101. Wu and Nancollas have described ways to obtain interfacial free energy at particle surfaces via crystallization and dissolution data as well as provided new insights into the relationship between nanometer size particles and the corresponding solubility [S0',9']. The interfacial free energy enters into many models predicting morphology while the role of solubility for the nanometer size particles is important both for synthesis of such particles and understanding the supersaturation during nucleation. Criscenti and Sverjensky have compiled a voluminous list of bivalent adsorbates on a number of important solids including silica, alumina, and iron (hydrous) [lo'] oxide. Considering the importance of adsorbates in growth directed morphological control, the compilation by Criscenti and Sverjensky may be useful in providing a basis for morphological control in a number of important metal oxide or hydrous oxide systems. 3. Growth directed particle formation

Growth directed particle formation is broken down into three subsets related to the material precipitated; precipitation of a biomineral based on calcium carbonate [ll-141, precipitation of metals [15-171, precipitation of inorganic compounds including metal oxides and semiconductor compounds [19-371. Most of these papers are directed toward the synthesis of the materials based on the use of specific growth directing agents. Since the first recognition of growth directed precipitation and templated directed growth, there has been a strong interest in better understanding the physiological factors that direct morphologies, orientation, and phases for biominerals such as calcium carbonate. The morphological forms and phases of calcium carbonate are reported in particular for the mineral produced via urea decomposition that resulted in several morphologies and phases depending

161

on the synthesis conditions [ll'l. Acicular, prismatic, spherulitic, plate-like particles were produced for several different phases of calcium carbonate including aragonite, vaterite, and calcite. Several polymers were used in the syntheses that promoted different phases and morphologies. It was also shown that vaterite is promoted by the presence of sulfate ions. Vucak et al. have compiled a list of surface active polymers and their affect on calcium carbonate morphology [14']. Control over the size and shape of metallic particles is an important feature for their use as conductor pathways in microelectronics. Several reports have discussed the syntheses of submicron, monosized, or nearly so, metal particles [15-18]. Goia and Matijevic have provided a rather comprehensive review on the preparation of monodispersed metal particles which documents many of the synthesis conditions for a number of important metal systems [15'1. Zhou et al. [17] have demonstrated the use of ultraviolet radiation to promote the formation of nearly monodisperse gold particles of several different morphologies depending on the polymer present during the synthesis. Mayer and Antonietti [18'] have provided a compilation of various polymers that can be used as protective colloids during the synthesis of metal particles. There have been a number of reports directed toward morphological control during synthesis of inorganic compounds [19'-371. Some of the particles produced by Adair and co-workers are shown in Fig. 1. Most of these have been directed toward metal oxides. Bell and Adair [19'] have shown that the synthesis conditions can control the morphology alpha-alumina precipitated from glycol solution. In particular, several organic acids and bases were evaluated to determine their effect on alpha-alumina particle shape. Typical shapes were plate-like, but spindle-shaped alpha-alumina particles were produced in the presence of acetic acid. Sugimoto and co-workers reported in several publications the effect of various additives on the morphologies of alpha iron oxide (hematite) [21-231. Sugimoto et al. [21'] used several organic additives were used to produce a variety of equiaxed hematite morphologies. It was shown that adsorption of the organic is a necessary condition for shape modification in the hematite system. The mechanism associated with morphological control of hematite by sulfate ions was also presented [22"1. It was shown that adsorption of sulfate ion to specific habit planes on the hematite particles produces the morphological control. However, if the concentration of sulfate was too high, acicular goethite (alpha-Fe00H) was produced rather than the hematite. In an elegant modification of the usual precipitation conditions, Pascal et al. [24"] demonstrated that gamma-Fe,O, particles can be directly produced in

162

J.H. Adair, E. Suvaci /Current Opinion in Colloid & Interface Science 5 (2000) 160-167

Fine plate-lilre a-Al-&

from Bell, &41201.

Fig. 1. Metal oxide particles produced by precipitation from solution. Particle morphologies were controlled by the synthesis processing conditions including shear rate in the solution and organic additives.

electrochemical fields. While size scaled with applied electric field, morphological control was not demonstrated with this approach. In contrast, Kandori et al. [25'] demonstrated a variety of particle morphologies could be produced with dimethylformamide (DMF). It was shown that the morphology of the hematite particles changed with increasing DMF concentration from spherical to prismatic-shaped particles. Several different morphologies of indium hydroxide were produced by Wang et al. [27'] depending on the initial concentration of indium, the aging time, and other conditions of the syntheses. There has been a burgeoning literature in the complex metal oxides, particularly the perovskite, ferroelectric materials [31-371, because of interest in the piezoelectric properties of single crystals or polycrystalline ferroelectric materials with preferred grain orihave provided seventation. Moon et al. [31",32"1 eral reports specific to morphological control chiefly over lead titanate, but also on barium titanate, lead

lanthanum zirconate titanate solid solutions, and lead zirconate titanate. It has been demonstrated the combination of shear rate with a Ti-complexing agent, acetylacetonate, can produce plate-like morphologies while lower shear rates produce cubic-shaped particles. Peterson and Slamovich [33'] have also examined synthesis conditions leading to distinct morphologies in the lead titanate system. Morphological results similar to Moon et al. [31"1 were found, but the Peterson and Slamovich's work also shed additional light on the orthogonal growth planes often seen on the surface of the plate-like lead titanate. Gelabert et al. [35'] examined the solution chemistry, phase stability, and crystal growth of lead titanate for hydrothermal conditions up to 500°C. Potassium fluoride was used as an additive to promote crystallization in these experiments. A prismatic morphology with very large lead titanate particles were produced at the supercritical conditions. Bagwell et al. [36'] have reported the morphologies produced for barium ti-


163

tanate synthesized in the presence of polyacrylic acid or a co-block polymer of polyethylene oxide - coblock-polymethacrylic acid. Cubic morphologies were produced at a lower barium concentration of 0.64 M while more equiaxed particles were produced at higher barium concentrations as well as in the presence of the polymers.

4. Template directed syntheses The emphasis on the synthesis of nanoscale materials has resulted in a significant increase in ways to prepare nanometer size particles during the last two decades. Many of the synthesis techniques depend on use of microemulsions to template the growth of nanometer scale particle with a number of recent reports on template directed shape control of particles [38-501. For the most part, templates on self-assembled amphiphilic molecules are synthesized either in emulsions or via Langmuir monomolecular films. An excellent review has recently been written by Osseo-Asare [38"] that provides a compilation of many of the materials and conditions used to synthesize particularly metal oxides. Most of these particles are nanometer size because of the nature of syntheses in the nano-reactors provided by self-assembled systems. Most of the particles produced in such microreactors are spherical and often monodisperse. Adair and co-workers have reviewed applications of nanometer size particles and in particular discussed the synthesis of plate-like nanometer size particles such as those shown in Fig. 2 [39"1. The CdS particles in Fig. 2 as well as other CdS particles synthesized in octylamine-water bilayers are polycrystalline 3-5-nm spherical primary aggregates composed of particles. The aggregation template scheme hypothesized for the templating of the plate-like CdS particles is shown in Fig. 3. The review by Adair et al. also provides an extensive summary of optical data for nanometer size CdS, and silica particles with nanometer size core particles including CdS and Ag. Examples of Ag core composite silica particles are also shown in Fig. 2. Precipitation on Langmuir films have provided insight into the nature of the templating process not always available to those investigators precipitating particles in microemulsion systems. The role of the polar head group in binding the precipitating species and relative packing of the amphiphilic molecules can be examined to determine the epitaxial relationship between the template and the templated materials. Ravaine et al. [40'] have examined the formation of gold particles using several different amphiphilic

-

$'!

50 --witha

Agcae Ilddlica w.

Fig. 2. Particles produced by template-directed growth [39"]. The CdS plate-like particles were produced in a neat phase D bilayer of octylamine and water. The composite nanometer size particles were produced in a reverse micelle system composed of Igepal/cyclohexane/water with tetraethoxy silane used to synthesize the silica shell.

molecules for the Langmuir film. Ultraviolet radiation was used to provide photo-reduction of the gold chloride in the system. It was found that binding between the AuCl, ions and positively charged polar head groups results in templating with the formation of plate-like gold particles from 20 to 800 nm across the face. Polar head groups that were either negative or uncharged did not template the gold particles. A very elegant approach was taken to synthesize rigid, nanometer size polystyrene rods. Dendrimers of polystyrene consisting of dendritic side chains and a polymer core self-assembled into rod-like, rigid polystyrene particles by Stocker et al. [44']. Molecular dynamic calculations, small-angle neutron scattering, and surface force microscopy were used to show how the polystyrene dendrimers would packed and were verified by experiment. In another elegant approach to the synthesis of nanometer size anisotropicallyshaped particles, Peng et al. produced rod-like particles of CdSe for photovoltaic applications [45']. Although strictly not a templated process, CdSe rods were grown by injection of cadmium and sulfide con-

164


I

sp

I I I I I I I I I I I I I I 11'11 I I I I I I I I I I I I I I I

Oo~O=O 00~00*0&0QL00~0&00&00&00&

* I = = . = =

00~0~0~00.00000.00.00.000000000

IllIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

I

1 1I I I I I I I I I I I Id I M ~ ~ I b ~ l ~ b l i i I 000000000.000

0

0

0

Fig. 3. Hypothesized scheme for the formation of the templated aggregates shown in Fig. 1 [39"].

taining organic phosphine solutions into a hot trioctyl phosphine oxide solution. Manipulation of particle shape was obtained by manipulation of the growth kinetics via temperature reduction or elevation and the presence of hexyl-phosphine. The texture of mesoporous silica has been varied by Cheng et al. [47'] using cationic surfactants. The texture was varied from equiaxed particles to rod-like structures in the mesoporous materials. The constrained growth to promote nanometer size particles or structures has often been used. In a recent synthesis, Sidorov et al. [48'] used a cross-linked polystyrene matrix in which nanometer size cobalt particles were grown. Landau et al. [49'1 embedded a zeolite in a mesoporous alumina matrix to constrain the zeolite resulting in a higher activity than that found for bulk nanometer zeolite polycrystalline clusters. Li et al. [50'] have synthesized Ag core silica composite particles and shown the effect of synthesis conditions on the core size and silica shell thickness. In addition, four wave degenerate mixing was used to determine the optical properties of the composite particles and nanometer size silver spheres [39']. The third-order non-linear susceptibility of the Ag nanocomposite particles was approximately 200 times larger for the composite particles than the uncoated

silver particles consistent with an enhanced quantum confinement by the dielectric silica coating.

5. Conclusion Progress in morphological control will continue as the relatively new tools of molecular modeling and new characterization techniques are applied. However, it is expected that the relatively empirical approaches traditionally used in this area will continue particularly for precipitation processes with the complex interaction among supersaturation and speciation, complex ions, adsorbates, and crystal structure of the precipitating material. It is likely that the growing database on morphological as well as size control in precipitation processes will continue to outstrip fundamental efforts to better understand precipitation processes. However, template directed growth appears to be more amenable to molecular modeling because the nucleation and growth usually is constrained at an interface and the small dimensions of the particles lend themselves particularly well to molecular modeling. Thus, it is expected that molecular modeling will be a fundamental tool of the nanometer scale particle grower. However, in the future, the growing databases with respect to collat-


era1 phenomena to precipitation such as interfacial free energy compilations and lists of adsorbates on specific solids will contribute to a better understanding of morphological control in growth directed control of particle morphologies. References and recommended reading

DO


[l] Noh DS, Koh Y, Kang IS. Numerical solutions for shape evolution of a particle growing in axisymmetric flows of supersaturated solution. J Crystal Growth 1998;183(3):427. Numerical solutions of growth have been used to analyze the effect of flow on particle shape. It is shown that convectional currents near growing interface lead to higher local growth rates. It is shown that an initially spherical particle can become an oblate spherical shape or a prolate barrel-like shape in uniaxial or biaxial flow, respectively. [2] Wickham JN, Herhold AB, Alivisatos AP. Shape change as an indicator of mechanism in the high-pressure structural transformations of CdSe nanocrystals. Phys Rev Lett 2000;84(5):923. Shape change was monitored by X-ray diffraction as CdSe nanocrystals transform between fourfold and sixfold coordinated crystal structures. Analysis of the X-ray patterns reveled a shape change during the transformation from four to six fold coordination. [3] Privman V, Goia DV, Matijevic E, Park JS. Mechanism of DO formation of monodispersed colloids by aggregation of nanosize precursors. J Colloid Interface Sci 1999;213(1):36. Presents a kinetic model based on aggregation of nanosize precursors to explain the formation of spherical particles. Experimental results were in reasonable agreement to predicted values for the formation spherical gold particles. [4] Mandel A, Schmitt W, Powell AK et al. A bioinspired apDO proach to control over size, shape and function of polynuclear iron compounds. Coord Chem Rev 1999;190/1921067. The role of ligands in dictating the size and shape of nanoscale particles is developed. The effects of templating species are examined for Fe(II1) compounds and ligands based on iminodiacetate. [5] Li W-J, Shi E-W, Yin Z-W. Growth habit of rutile and alpha-Al,O,. J Crystal Growth 2000;208:546. [6] Sundberg DC, Durant IG. Thermodynamic and kinetic aspects for particle morphology control. NATO AS1 Ser E Appl Sci 1997;335:17712. [7] Stejskal J, Spirkova M, Prokes J. Polyaniline dispersions 8. The control of particle morphology. Polymer 1999;40(10):2487. [8] Wu W, Nancollas GH. Determination of interfacial tension DO from crystallization and dissolution data: a comparison with other methods. Adv Colloid Interface Sci 1999;79(2/3):229. Based on the importance of understanding and controlling interfacial tension for morphological control, a useful look at methods to determine interfacial tension. [9] Wu W, Nancollas GH. A new understanding of the relationship between solubility and particle size. J Solution Chem 1998;27(6):521. A new look at the relationship between solubility and size. Thermodynamic arguments are made that size and solubility may not follow the Freundlich-Ostwald relationship because of the effect of interfacial tension. It is also argued that the Kelvin and Gibbs-Thomp son equations while valid for solid-vapor phases may not be valid for solid-solution systems. DO

165

[lo] Criscenti LJ,Sverjensky DA. The role of electrolyte anions (ClO-, NO;, and C1-) in the divalent metal adsorption on oxide and hydroxide surfaces in salt solutions. Am J Sci 1999;828. A voluminous compilation of adsorption data for a variety of divalent species on a number of important metal oxides including quartz, alpha-alumina, and hydrous ferric oxide. Considering the importance of adsorption in morphological development in precipitating systems, this compilation provides possible insight into species that should be evaluated for shape modification for some of the important inorganic compounds. [ l l ] Wang L, Sondi I, Matijevic E. Preparation of uniform needle-like aragonite particles by homogeneous precipitation. J Colloid Interface Sci 1999;218(2):545. Precipitation from homogeneous solution via thermal decomposition of urea was used to synthesize calcium carbonate of several different morphologies and phases depending on the synthesis conditions. [12] Rock ML, Tranchitella LJ, Pilato RS. Control of calcium carbonate particle size and shape by precipitation from CTAB/alcohol/hexadecane mixtures. Colloid Polym Sci Kolloid-Zeitschrift 1997;275(9):893. [13] Jung W-M, Kang S-H, Kim W-S, Choi CK. Particle morphology of calcium carbonate precipitated by gas-liquid reaction in a Couette-Taylor reactor. Chem Eng Sci 2000;55(4):733. The morphologies of calcium carbonate were controlled by C0,-Ca(OH), reaction precipitation. The experimental data was shown to fit a model based on monolayer adsorption. [14] Vucak M, Pons MN, Peric J, Vivier H. Effect of precipitation conditions on the morphology of calcium carbonate: quantification of crystal shapes using image analysis. Powder Techno1 1998;97(1):1. [15] Goia DV, Matijevic E. Preparation of monodispersed metal particles. New J Chem 1998;22(11):1203. A review of precipitation routes to prepare monodispersed metal, metal oxide, and composite particles. [16] Goia DV, Matijevic E. Tailoring the particle size of monodispersed colloidal gold. Colloids Surfaces 1999;46(1/3):139. [17] Zhou YCY, Zhu YR, Chen ZY. A novel ultraviolet irradiation technique for shape-controlled synthesis of gold nanoparticles at room temperature. Communications. Chem Mater 1999;11(9):2310,3. Ultraviolet radiation is used to prepare gold particles with a variety of shapes. [18] Mayer A, Antonietti M. Investigation of polymer-protected noble metal nanoparticles by transmission electron microscopy: control of particle morphology and shape. Colloid Polym Sci Kolloid-Zeitschrift 1998;276(9):769. A compilation of protective polymers used in the synthesis of noble metal particles. [19] Bell NS, Adair JH. Adsorbate effects on glycothermally produced alpha-alumina particle morphology. J Crystal Growth 1999;203(1/2):213. An analysis of the effect of various organic acids and bases on the morphology of alpha-alumina produced by precipitation in 1,4butanediol solutions. [20] Bell NS, Cho S-B, Adair JH. Size control of a-alumina particles synthesized in 1,Cbutaendiol solution by a-alumina and a-hematite seeding. J Am Ceram SOC1998;81(6):1411. [21] Sugimoto T, Itoh H, Mochida T. Shape control of monodisperse hematite particles by organic additives in the gel-sol system. J Colloid Interface Sci 1998;205(1):42. Control over the shape of hematite particles via organic additives is demonstrated in the gel-sol preparation.

166


[22] Sugimoto T, Wang Y. Mechanism of the shape and structure oo control of monodispersed alpha-Fe,O, particles by sulfate ions. J Colloid Interface Sci 1998;207(1):137. The mechanism of shape control of hematite by sulfate ions is discussed. [23] Sugimoto T, Wang Y, Muramatsu A. Systematic control of size, shape and internal structure of monodisperse alphaFe,O, particles. Colloids Surfaces A Physicochem E 1998;134(3):265. [24] Pascal C, Pascal JL, Payen C, Favier F, Elidrissi Moubtassim oo ML. Electrochemical synthesis for the control of gammaFe,O, nanoparticle size. Morphology, microstructure, and magnetic behavior. Chem Mater 1999;11(1):14. It is demonstrated that gamma-Fe,O, can be directly produced in a variety of shapes by electrochemical synthesis in N,N-dimethylformamide with a small amount of water. [25] Kandori K, Ohkoshi N, Yasukawa A, Ishikawa T. Morphology control and texture of hematite particles by dimethylformamide in forced hydrolysis reaction. J Mater Res 1998;13(6):1698. A variety of morphologies of alpha-Fe,O, were produced in the presence of dimethylformamide. [26] Hakuta Y ,Adschiri T, Arai K. Chemical equilibria and particle morphology of boehmite (AlOOH) in sub and supercritical water. Fluid Phase Equilibria 1999;158/160:733. [27] Wang L, Perez-Maqueda LA, Matijevic E. Rapid preparation of uniform colloidal indium hydroxide by the controlled double-jet precipitation. Colloid Polym Sci 1998;276(9):847. In(OH), particles of several different morphologies were prepared by double jet precipitation. [28] Zhong Q, Matijevic E. Preparation of uniform zinc oxide colloids by controlled double-jet precipitation. J Mater Chem 1996;6(3):443. [29] Bredol M, Merikhi J. ZnS precipitation: morphology control. J Mater Sci 1998;33(2):471. [30] Kang YC, Park SB, Okuyama K, Lenggoro IW. Morphology control of multicomponent oxide phosphor particles containing high ductility component by high temperature spray pyrolysis. J Electrochem SOC1999;146(7):2744. [31] Moon J, Carasso ML, Adair JH, Krarup HG, Kerchner JA. oo Particle-shape control and formation mechanisms of hydrothermally derived lead titanate. J Mater Res 1999; 14(3):866. Tabular lead titanate synthesized via hydrothermal synthesis in the presence of acetylacetonate. [32] Moon J, Kerchner JA, Krarup H, Adair JH. Hydrothermal 00 synthesis of ferroelectric perovskites from chemically modified titanium isopropoxide and acetate salts. J Mater Res 1999;14(2):425. Survey of a variety of ferroelectric materials and resulting morphologies synthesized by hydrothermal precipitation. [33] Peterson CR, Slamovich EB. Effect of processing parameters on the morphology of hydrothermally derived PbTiO, powders. J Am Ceram SOC1999;82(7):1702. Examination of the synthesis of the ferroelectric material, PbTiO,, by hydrothermal synthesis. [34] Roeder RK, Slamovich EB. Stoichiometry control and phase selection in hydrothermally derived Ba,Sr, -,TiO, powders. J Am Ceram SOC1999;82(7):1665. [35] Gelabert MC, Laudise RA, Riman RE. Phase stability, solubility and hydrothermal crystal growth of PbTiO,. J Crystal Growth 1999;197(1/2):195. Examination of the solution chemistry and phase stability of PbTiO, in supercritical growth conditions. A tabular morphology was produced.

[36] Bagwell RB, Sindel J, Sigmund W. Morphological evolution of barium titanate synthesized in water in the presence of polymeric species. J Mat Res 1999;14(5):1844. Morphology of barium titanate precipitated as a function of polyacrylic acid and a block copolymer were determined. [37] Kang YC, Park SB. Morphology control of Ba Mg All0 017: Eu particles: the use of colloidal solution obtained from alkoxide precursor in spray pyrolysis. J Electrochem Rev 2000;147(2):799. [38] Osseo-Asare K. Microemulsion-mediated synthesis of nanooo size oxide materials. In: Handbook of microemulsion science and technology, ch. 18. New York Marcel Dekker, Inc., 1999549-603. Comprehensive compilation and review of microemulsion syntheses of nanometer size particles. [39] Adair JH, Li T, Kido T et al. Recent developments in the oo preparation and properties of nm-size spherical and plateletshaped particles and composite particles. Mater Sci Eng Rep 1998;23(4/5):139. Review of the applications of nanometer materials and micellar conditions required to produce plate-like and composite nanometer size particles. There is also extensive data on optical properties of the nanometer size particles. [40] Ravaine S, Fanucci GE, Seip CT, Adair JH, Talham DR. Photochemical generation of gold nanoparticles in Langmuir-Blodgett films. Langmuir: ACS J Surfaces Colloids 1998;14(3):708. Synthesis of nanometer size gold platelets on a Langmuir film. [41] Whipps S, Khan SR, Talham DR, OPalko FJ, Backov R. Growth of calcium oxalate monohydrate at phospho-lipid. Langmuir monolayers. J Crystal Growth 1998;192(1/2):243. [42] Bonnani A, Seyringer H, Sitter H, Stifter D, Hingerl K. Control over the morphology changes in self-assembled Mnbased nanostmctures overgrown with mismatched material. Appl Phys Lett 1999;74(24):3732. [43] Kovtyukhova NI, Buzaneva EV, Mallouk TE, Waraksa CC. Ultrathin nanoparticle ZnS and ZnS: Mn films: surface sol-gel synthesis, morphology, photophysical properties. Mater Sci Eng B 2000;69/70:411. [44] Stocker W, Schurmann BL, Schluter AD. A dendritic nanocylinder: shape control through implementation of steric strain. Adv Mater 1998;10(10):793. Presents the preparation and analysis of rigid nanocylinders with a polymeric core. [45] Peng X, Manna L, Alivisatos AP et al. Shape control of CdSe nanocrystals. Nature 2000;404(6773):59. The preparation of tabular, nanometer size CdSe particles is presented. [46] Huynh WU, Peng X, Alivisatos AP. CdSe nanocrystal rods/ poly(3-hexylthiophene) composite photovoltaic devices. Adv Mater 199R11(11):923. [47] Cheng Y, Lin HP, Mou CY. Control of mesostructure and morphology of surfactant-templated silica in a mixed surfactant system. Phys Chem Chem Phys 1999;1(21):5051. The texture of mesoporous silica was varied by formation in mixtures of cationic surfactants. [48] Sidorov SN, Bronstein LM, Spontak RJ. Cobalt nanoparticle formation in the pores of hyper-cross-linked polystyrene: control of nanoparticle growth and morphology. J Info Chem Mater 1999;11(11):3210. A polystyrene matrix was used as a template to prepare nanometer size cobalt particles. [49] Landau MV, Tavor D, Mintova S. Colloidal nanocrystals of zeolite beta stabilized in alumina matrix. Chem Mater 1999;11(8):2030.


Nanocrystal of zeolite p were prepared and embedded in mesoporous structure of an alumina matrix. The activity of zeolite in cumene cracking was twice that of bulk nanometer zeolite clusters. [50] Li T, Morrone AA, Mecholsky JJ, Talham DR, Adair

the the size JH.

167

Preparation of Ag/SiO, Nanosize composites by a reverse micelle and sol-gel technique. Langmuir 1999;15(3):4228. Composite nanometer size particles of silica with a silver core were prepared in a variety of sizes by control over the precipitation conditions.

ELSEVI E R


Synthesis and electronic properties of semiconductor nanoparticles/quantum dots Karen Grieve, Paul Mulvaney, Franz Grieser * ParticulateFluids Processing Centre, School of Chemistry, UniversiQ of Melbourne, Parkville 3010, Australia

Abstract This review examines recent work on the synthesis, characterisation and potential applications of semiconductor nanoparticles (quantum dots). Recent advances in single quatum dot spectroscopy is also reviewed. 0 2000 Elsevier Science Ltd. All rights reserved. Keywords: Nanoparticles; Quantum dots; Semiconductor particles; Colloids; Particle synthesis; Single-particle spectroscopy

1. Introduction

Investigations into the preparation and characterisation of nanocrystallites continues to remain an active area of research. The potential for exploitation of these fine particles has produced a number of start-up companies producing quantum dots (QDs) designed for specific applications, primarily in biotechnology [I]. There have been a variety of developments in QD synthesis and characterisation over the past 2 years since the last 'Current Opinion' review of the area appeared [2]. Particular trends to note are advances in the nucleation and growth of monodisperse samples as well as improved luminescence quantum yields. Disappointing is the basic lack of data on surface passivation. While observations of quenching or luminescence enhancement are regularly reported, systematic investigations of dielectric effects, hydrogen bonding and ligand-particle interactions are sorely missed. The exception is core-shell synthesis for sur-

Abbreuiations: QD, quantum dot; EL, electroluminescence; TOPO, trioctylphosphine oxide; SPS, single particle spectroscopy. * Corresponding author. Tel.: 61-3-9344-6476; fax: 61-39344-6233. E-mail address: [email protected] (F. Grieser).

+

+

face passivation, whereby a capping layer is used to modify the luminescence characteristics of the core. This remains an intensively investigated field. The present review comments on a number of recent publications that have advanced our knowledge of nanoparticles and their special properties. Reviews of research prior to mid-1998 can be found elsewhere, especially in the areas of single particle spectroscopy [3,4'] and structural effects on electronic properties [5']. To provide a boundary to an otherwise enormous field of research, this review is confined to quantum dots produced by wet chemical methods.

2. Synthesis

Several groups have begun exploring variations to the two most popular synthetic methods: the TOPO synthesis of CdSe developed by Bawendi and Alivisatos, and the aqueous preparation technique pioneered in Berlin by Henglein, Weller and collaborators using CdS capped by polyphosphate. Green and O'Brien [6'] have examined the use of air-stable precursors for metal chalcogenide nanoparticles, and have reported a number of syntheses that produce luminescent particles. Unfortunately, the preparation of the precursors often requires vacuum


K. Grieve et al. /Current Opinion in Colloid & Interface Science 5 (2000)168-1 72

or Schlenk line procedures and scale up may be difficult. However, this approach is very promising for improving the reproducibility of the nucleation step, which is still poorly understood in T O P 0 based systems. Many combinations of semiconductors have been used for core-shell particles, of particular interest are the combination of 11-VI and 111-V semiconductor materials in InAs @ CdSe particles [7] and layered quantum dots of Cd and Hg chalcogenides [8,9]. A novel approach for the stabilisation of semiconductor particles is silica coating. Correa-Duarte et al. [lo] have reported that silica coating leads to a 100-fold reduction in photocorrosion rates for CdS nanoparticles. In addition, the silica coating opens up the possibility for controlled synthesis of photonic crystals [ll']. A number of groups are synthesising well-characterised QDs on electrode surfaces. This technique allows the possibility of several alternative characterisation methods (e.g. XPS, ellipsometry) and also permits controlled removal of the particles from the solvent. Synthesis of nanoparticle films is facilitated by direct nucleation on the surface of the substrate, and practical electrochromic and photochromic devices are likely to be realised through this synthetic approach. The most active group in this domain is that of Hodes and coworkers [12]. An elegant review of the use of hybrid electrochemical and chemical synthesis and the application of a variety of solution and surface spectroscopies has been presented by Penner [13']. 3. Nucleation, growth, shape control and etching

Refinements of the CdSe preparative route have been published by Bawendi and by Alivisatos. In particular, the role of Ostwald ripening and monomer addition on the size and dispersity of CdSe has been examined [14'1. A surprisingly effective approach has been photoetching. van Dijken et al. [15] reported that ZnO, CdS, PbS and ZnS could all be prepared as monodisperse quantum dots with sharply structured spectra by the careful photoetching of larger colloids. While several protocols have appeared demonstrating shape control of metal particles, there has been less success with semiconductors until now. However, Peng et al. [14'] have recently communicated a procedure for synthesising CdSe rods whilst Li et al. [16] have provided a striking demonstration of template control involving the synthesis of BaCrO, nanorods in AOT. Another extension to colloid synthesis is the demonstration of a non-hydrolytic route to metal oxides by Trentler et al. [17]. Whilst template routes are beyond the scope of this survey, extensive reviews have ap-

169

peared quite recently, notably by Pileni [18]. The assembly of particles is an important aspect of particle control and the technique of coating small inert spheres with fluorescing nanoparticles, which can then be assembled into larger three-dimensional crystals, offers a promising route to more complex nanostructures [19]. 4. Charge separation and transfer

There is a gradual exploration of newer techniques for QD characterisation, especially fast spectroscopies to examine relaxation kinetics [8,20-221. A novel method for determining the band gap of semiconductor powders has been developed by Gal et al. [23] using surface photovoltage spectroscopy. This method appears to circumvent the difficulties of applying the standard Kubelka-Munk equations to determine the bandgap of powders. New photoelectrochemical methods (e.g. intensity modulated photoelectron spectroscopy, IMPS) have also been used to investigate excited state dynamics and decay in CdS nanoparticles [241. A more traditional approach to CdSe quantum dot films has confirmed that surface effects dominate electron transfer and, therefore, photoresponse characteristics [25]. The importance of surfaces in the trapping of charge carriers is constantly being reinforced, with recent studies showing the occupation by electrons of CdS/HgS interfaces in quantum dot/quantum well structures [26], and the contrast between the fates of electrons and holes arising from excitation. The electrons move to the surface and are liberated while the trapped holes are retained within the particle [27'1. A thorough examination of the electron and hole recombination processes in ZnO has been carried out by van Dijken et al. [28']. They identify the hole trapping and electron trapping sites on ZnO from the direct measurement of recombination kinetics. Another interesting observation is the up-shifting of photoluminescence in CdSe and InP, which is thought to result from coupling between an exciton and a phonon [29'1. 5. Theory Comprehensive surveys of the band structure and ab initio approaches to QDs are provided in a recent monograph by Gaponenko [30"]. At present, theory remains well in advance of experiment, because spectral broadening and polydispersity hamper comparisons with theory. Important papers covering some neglected aspects of QD spectroscopy have appeared. In particular the role of the solvent has not been

170

K Grieve et al. /Current Opinion in Colloid & Interface Science 5 (2000)168-1 72

tackled since Brus' original papers on the size dependence of the exciton position. However, he and his colleagues have now provided results of a detailed study on the effects of a dielectric medium on CdSe emission [31].

6. Applications

Applications for nanoparticles have been surprisingly slow to emerge, given the prolific range of particle applications in conventional colloid chemistry. This has been partially due to the difficulty and expense of synthesising large quantities of unagglomerated material, and the problem of finding markets where nanoparticles could function more ably than dyes. Two areas of increasing focus are electroluminescence (EL) [32,331 and bioattachment [34',35',36',371. EL work has centred on the incorporation of luminescent dots into polymer matrices. EL from Q-CdS in electrolytes showing band gap shifts has also been reported [381. Leeb et al. [391 observed EL from Mn-ZnS doped films. The stability of these systems is poor due to the need to maintain hole and electron transport within the diode structure. A novel application of QDs is in biology [34',35',36',37]. QDs have several potential advantages over dyes. The photodegradation of dyes under laser excitation (e.g. in flow cytometry), the hydrophobicity of many useful dyes, and the solventdependent quantum yields of dyes are issues that can be circumvented by replacing dyes with QDs as reporter molecules in a variety of bioassays. QD-based assays have already been reported [34',35',36'1. Attempts to combine semiconducting polymers and nanoparticles in useful photovoltaic devices have continued, with relatively high conversion efficiencies being noted for systems utilising nanorods [40']. Solar cell application of nanoparticles based on the 'Gratzel cell' rely on a combination of dyes as sensitizers and nanoparticle electrodes for efficient charge separation. The most frequently studied cells contain colloidal TiO, and recent insights into the electronic processes at work in the cells [41] and the bonding between the constituents [421 may lead to further development of these photoelectric cells. The development of ZnO as an alternative material for nanoporous solar cells is also promising [43]. The use of semiconductor nanoparticles for photocatalysis has also received recent attention, an advance of note being the photooxidation of organic molecules using MoS, particles, which absorb at visible wavelengths [441. A further avenue of research remains the extension of the QD luminescence into the NIR and IR regions.

This would facilitate commercialisation of several applications that utilise the full spectral width of standard Si photodiode detectors. Several groups have extended the synthesis of metal chalcogenides to include smaller band gap materials such as HgTe that emit in the NIR, but the chemical instability of small band gap materials remains problematic. Harrison et al. [45'] have obtained 1.2-pm emission from HgTe/CdS colloids and Mais et al. [46] have obtained 1.55-pm emission from ZnO/Er.

7. Single particle spectroscopy and STM studies

Perhaps the most exciting direction to emerge recently is single particle spectroscopy (SPS). SPS permits the direct measurement of QD optical properties as a function of particle size, and allows a direct comparison with theory for basic parameters such as emission lifetime and spectral width of excitonic states. However, several experimental results have raised new questions. In particular, the observations of spectral diffusion [4'] and intermittency [47] whereby individual QDs switch-off and recover with timescales ranging from milliseconds to seconds, have raised new problems concerning the primary steps in carrier recombination. The lifetime of the dark state appears to be independent of temperature, which excludes an activation limited process as an explanation for the phenomenon. At present the data seem to be consistent only with an Auger process, whereby individual QDs are turned off when a second photon interacts with the excited state leading to trapping of an Auger electron [47,481. The electron then tunnels back into the conduction band. Alperson et al. [49] have recorded individual tunnelling spectra and single electron charging of CdSe QDs from which the band gap of individual particles was measured. The observation of a transition to atomic like orbital structures has been claimed by Banin et al. [50] using STM on InAs dots. Both studies were carried out at 4.2 K. Tunnel diodes have been fabricated by Kim et al. [51] using CdSe nanocrystals.

8. Summary

The significant progress in the synthesis of quantum dots and the final emergence of nanoparticle technologies is likely to attract significant intellectual investment into nanoparticles for much of the decade ahead.

K. Grieve et al. /Current Opinion in Colloid & Interface Science 5 (2000)168-1 72

References and recommended reading

00


[l] Rotman D. Quantum Dot Com. Techno1 Rev 2000;103:50-57. [2] Weller H. Quantum size colloids: from size-dependent properties of discrete particles to self-organized superstructures. Curr Opin Colloid Interface Sci 1998;3:194-199. [3] Empedocles S, Bawendi M. Spectroscopy of single CdSe nanocrystals. ACCChem Res 1999;32:389-396. [4] Empedocles SA, Neuhauser R, Shimizu K, Bawendi M. Pho0 toluminescence from single semiconductor nanostructures. Adv Mater 1999;11:1243-1256. This article and Empedocles et al. [3] review recent results in the spectroscopy of single CdSe nanocrystal QDs. The elimination of ensemble averaging is shown to reveal interesting spectroscopic properties of nanoparticles. [5] Heath JR, Shiang JJ. Covalency in semiconductor quantum dots. Chem SOCRev 1998; 2765-71. The preparation and differences in the electronic structure and electronic dynamics of quantum dots of 111-V (InP, InAs, GaAs etc.) and 11-IV (ZnS, CdS, CdSe etc.) semiconductors is reviewed, and an attempt is made to explain the differences in terms of lattice covalency. [6] Green M, O’Brien P. Recent advances in the preparation of semiconductors as isolated nanometric particles. Chem Commun 1999;9:2235-224 1. A summary of the methods developed by this group for producing nanoparticles from single mecursors. Cao Y-W, Banin U. Synthesis and characterization of InAs/InP and InAs/CdSe core/shell nanocrystals. Angew Chem Int Ed 1999;38:3692-3694. Mews A, Eychmiiller A. Quantum wells within quantum dots, a CdS/HgS nanoheterostructure with global and local confinement. Ber Bunsenges Phys Chem 1998;1021343-1357. Kershaw SV, Burt M, Harrison M, Rogach A, Weller H, Eychmiiller A. Colloidal CdTe/HgTe quantum dots with high photoluminescence quantum efficiency at room temperature. Appl Phys Lett 1999;75:1696-1698. Correa-Duarte MA, Giersig M, Liz-MarzLn LM. Stabilization of CdS semiconductor nanoparticles against photodegradation by a silica procedure. Chem Phys Lett 1998;286:497-501. Mulvaney P, Liz-Marzh LM, Giersig M, Ung T. Silica encaptulation of quantum dots and metal clusters. J Mater Chem 2000;10:1259-1270. Alperson B, Demange H, Rubinstein I, Hodes G. Photoelectrochemical charge transfer properties of electrodeposited CdSe quantum dots. J Phys Chem B 1999;103:4943-4948. Penner RM. Hybrid electrochemical/chemical synthesis of auantum dots. ACCChem SOC200033:78-86. Overview of a three-step process for producing chalcogenide or halide semiconductor particles on an electrode surface. [14] Peng X, Wickham J, Alivisatos AP. Kinetics of 11-VI col0 loidal semiconductor nanocrystal growth 111-V focussing of size distributions. J Am Chem SOC1998;1205343-5344. Concludes that precise control and constant monitoring are necessary if kinetics are to be used as a fine control of particle size distribution. [15] van Dijken A, Janssen AH, Smitsmans MHP, Vanmaekelbergh D, Meijerink A. Size-selective photoetching of nanocrystalline semiconductor particles. Chem Mater 1998; 103513-3522. 1161 Li M, Schnablegger H, Mann S. Coupled synthesis and selfassembly of nanoparticles to give structures with controlled organization. Nature 1999;402393-395. [17] Trentler TJ, Denler TE, Bertone JF, Agrawal A, Colvin VL.

- -

171

Synthesis of TiO, nanocrystals by nonhydrolytic solutionbased reactions. J Am Chem SOC1999;121:1613-1614. Pileni MP. Optical properties of nanosized particles dispersed in colloidal solutions or arranged in 2D or 3D superlattices. New J Chem 1998;22693-702. Rogach A, Susha A, Caruso F et al. Nano- and microengineering: three-dimensional colloidal photonic crystals prepared from submicron-sized polystyrene latex spheres precoated with luminescent polyelectrolyte/nanocrystal shells. Adv Mater 2000;12:333-337. Burda C, Link S , Green TC, El-Sayed MA. New transient absorption observed in the spectrum of colloidal CdSe nanoparticles pumped with high-power femtosecond pulses. J Phys Chem B 1999;103:10775-10780. Nagamura T, Inoue H, Grieser F, Urquhart R, Sakaguchi H, Furlong DN. Ultrafast dynamics of transient bleaching of surface modified cadmium sulphide nano-particles in Nafion films. Colloids Surf A 1999;146:265-272. Sengupta A, Jiang B, Mandal KC, Zhang JZ. Ultrafast electronic relaxation dynamics in PbI, semiconductor colloidal nanoparticles: a femtosecond transient absorption study. J Phys Chem B 1999;103:3128-3137. Gal D, Mastai Y, Hodes G, Kronik L. Band gap determination of semiconductor powders via surface photovoltage spectroscopy. J Appl Phys 1999;86:5573-5577. Bakkers EPAM, Reitsma E, Kelly JJ, Vanmaekelbergh D. Excited-state dynamics in CdS quantum dots adsorbed on a metal electrode. J Phys Chem B 1999;103:2781-2788. Ginger DS, Greenham NC. Charge injection and transport in films of CdSe nanocrystals. J Appl Phys 2000;87:1361-1368. Lifshitz E, Porteanu H, Glozeman A, Weller H, Pflughoefft M, Eychmiiller A. Optically detected magnetic resonance study of CdS/HgS/CdS quantum dot quantum wells. J Phys Chem B 1999;103:6870-6875. Dimitrijevic NM, Henglein A, Meisel D. Charge separation across the silica nanoparticle/water interface. J Phys Chem 1999;103:7073-7076. This paper describes the use of silica nanoparticles to determine the movement of charge carriers, both holes and electrons. It is shown that electrons move quantitatively to the surface and escape into the aqueous phase, whereas holes remain trapped in the particle. This observation is frequently made. [28] van Dijken A, Meulenkamp EA, Vanmaekelbergh D, Meijerink A. The kinetics of the radiative and nonradiative processes in nanocrystalline ZnO particles upon photoexcitation. J Phys Chem B 2000;104:1715-1723. [29] Poles E, Selmarten DC, Mick 01, Nozik AJ. Anti-stokes 0 photoluminescence in colloidal semiconductor quantum dots. Appl Phys Lett 1999;75:971-973. Relatively high up-conversion of low intensity photons was observed from CdSe and InP quantum dots. It has been proposed that a coupling of surface states and phonons is responsible for the blue shift in the photoluminescence. [30] Gaponenko SV. Optical properties of semiconductor 00 nanocrystals. Cambridge: Cambridge University Press, 1998. Monograph analysing electronic and spectroscopic characterization of semiconductor nanoparticles until the beginning of 1998. Rabani E, HetCnyi B, Berne BJ, Bms LE. Electronic properties of CdSe nanocrystals in the absence and presence of a dielectric medium. J Chem Phys 1999;110:5355-5369. Mattoussi H, Radzilowski LH, Dabbousi BO, Thomas EL, Bawendi MG, Rubner MF. Electroluminescence from heterostructures of polfiphenylene vinylenel and inorganic CdSe nanocrystals. J Appl Phys 1998;83:7965-7974. Mattoussi H, Radilowski LH, Dabbousi BO et al. Composite thin films of CdSe nanocrystals and a surface passivating/ electron transporting block copolymer: correlations between

172

K Grieve et al. /Current Opinion in Colloid & Interface Science 5 (2000)168-1 72

film microstructure by transmission electron microscopy and electroluminescence. J Appl Phys 1999;86:4390-4399. [34] Mitchell GP, Mirkin CA, Letsinger RL. Programmed assem0 bly of DNA functionalized quantum dots. J Am Chem SOC 1999;121:8122-8123. Quantum dots have been functionalized with single-stranded DNA and assembled into matrices. [35] Bruchez MJ, Maronne M, Gin P, Weiss S, Alivisatos AP. semiconductor nanoparticles as fluorescent biological labels. Science 1998;281:2016-2113. [36] Chan WCW, Nie S. Quantum dot bioconjugates for ultrasen0 sitive nonisotopic detection. Science 1998;281:2016-2018. This and the paper by Bruchez et al. [35] are significant papers that independently demonstrated the use of QDs as biolabels for the first time. [37] Sondi I, Siiman 0, Koester S, Matijevic E. Preparation of aminodextran-CdS nanoparticle complexes and biologically active antibody-aminodextran-CdS nanoparticle conjugates. Langmuir 2000;16:3107-3118. [38] Tanno K, Ashokkumar M, Mulvaney P. Quantized electroluminescence from Q-CdS films immersed in aqueous electrolytes. Colloids Surf A 2000;146:293-298. [39] Leeb J, Gebhardt V, Miiller G et al. Colloidal synthesis and electroluminescence properties of nanoporous (MnZnS)-ZnI1 films. J Phys Chem B 1999;103:7839-7845. [40] Huynh WU, Peng X, Alivisatos AP. CdSe nanocrystal rods/pol~3-hexylthiophene) composite photovoltaic devices. Adv Mater 1999;11:923-927. Continuing the development of polymer/nanocrystal photovoltaic devices, initial investigations into the use of nanorods rather than spherical particles has led to devices with conversion efficiencies significantly greater than observed in other similar heterostructures, due to efficient electron transport along the rod. [41] Cahen D, Hodes G, Gratzel M, Guillemoles JF, Riess I. Nature of photovoltaic action in dye-sensitized solar cells. J Phys Chem B 2000;104:2053-2059. [42] Lemon BI, Hupp JT. EQCM investigations of dye-functional-

ized nanocrystalline titanium dioxide electrode/solution interfaces: does luminescence report directly on interfacial electron transfer kinetics? J Phys Chem B 1999;103: 3797-3799. [43] Keis K, Vayssieres L, Lindquist SE, Hagfeldt A. Nanostmctured ZnO electrodes for photovoltaic applications. Nanostructured Mater 1999;12487-490. [44] Thurston TR, Wilcoxin JP. Photoxidation of organic chemicals catalyzed by nanoscale MoS,. J Phys Chem B 1999; 103:ll-17. [45] Harrison MT, Kershaw SV, Rogach AL, Kornowski A, Ey0 chmiiller A, Weller H. Wet chemical synthesis of highly luminescent HgTe/CdS core/shell nanocrystals. Adv Mater 2000;12123-125. Robust particles luminescing at 1200 nm have been produced by wet chemical means. [46] Mais N, Reithmaier JP, Forchel A, Kohls M, Spanhel L, Muller G. Er doped nanocrystalline ZnO planar waveguide structures for 1.55 p m amplifier applications. Appl Phys Lett 1999;75:2005-2007. [47] Banin U, Bruchez M, Alivisatos AP,Ha T, Weiss S, Chemla DS. Evidence for a thermal contribution to emission intermittency in single CdSe/CdS core/shell nanocrystals. J Chem Phys 1999;110:1195-1201. [48] Klimov VI, Mikhailovsky AA, McBranch DW, Leatherdale CA, Bawendi MG. Quantization of multiple auger rates in semiconductor quantum dots. Science 2000;287:1011-1013. [49] Alperson B, Rubinstein I, Hodes G, Porath D, Millo 0. Energy level tunneling spectroscopy and single electron charging in individual CdSe quantum dots. Appl Phys Lett 1999;75:1751-1753. [50] Banin U, Cao Y, Katz D, Millo 0. Identification of atom-like electronic states in indium arsenide nanocrystal quantum dots. Nature 1999;4000:542-544. [51] Kim S-H, Markovich G, Rezvani S, Choi SH, Wang KL, Heath JR. Tunnel diodes fabricated from CdSe nanocrystal mnolayers. Appl Phys Lett 1999;74:317-319.

Physicochemical aspects of colloid and interface

Physicochemical aspects of colloid and interface

Suggest Documents

Journal of Colloid and Interface Science Magnesium

process standardization and physicochemical aspects

Physicochemical Aspects of Food Engineering and Processing.pdf ...

Physicochemical Aspects and Efficiency of ... - Semantic Scholar

Journal of Colloid and Interface Science Thermo-responsiveness of ...

Colloid-interface interactions initiate osmotic flow

Journal of Colloid and Interface Science Cationic agents for ... - Core

Impact of atomic force microscopy on interface and colloid science

Current Opinion in Colloid & Interface Science ...

Sugar profile, physicochemical and sensory aspects

Analytical Sciences Anal. Sci. Journal of Colloid and Interface ...

14th Conference of the European Colloid and Interface ... - Google Sites

14th Conference of the European Colloid and Interface ... - Google Sites

Research Article Physicochemical Aspects of the ...

Advances in Colloid and Interface Science, 30 - ScienceDirect ...www.researchgate.net › file.PostFileLoader.html

1 Colloid Aspects of Cosmetic Formulations with ... - Wiley-VCH

Colloids and Surfaces A: Physicochemical and Engineering Aspects ...

Colloid and surface.pdf

Colloid Chemistry - MDPIwww.researchgate.net › publication › fulltext › Colloid-C

Architecture and Interface Aspects of Scheduling Decision ... - CiteSeerX

Aspects of Interface Agents: Avatar, Assistant and Actor - CiteSeerX

Vitreoretinal Interface Aspects after Macular Hole ...

The Interface Presented to Aspects - CiteSeerX

Migrating interface implementations to aspects - Semantic Scholar