POLYMERIC NANOFIBRES
In the textile industry polymeric fibres having a high degree of orientation along the fibre axis are formed from a melt or solution of the polymer through a variety of spinning or drawing methods involving a number of complex molecular processes. These processes are essentially mechanical and are realized through some extrusion and drawing technique. Involved in them are time and temperature dependent molecular motions phase transitions under high stress entanglement constraints are various intermolecular reactions. Hence the final state of molecular order in a fibre is a function of the process variables such as stress strain temperature and time and the length and length distribution of the molecules. The polymer fibres developed by these techniques have a diameter in the range of microns (1 100 µm).
Research on the structure of polymeric fibres has lead to the ability to control structure formation and has resulted in fibres of high tensile molecules and tensile strength. These high strength high modulus fibres are used in making ropes satellite tethers high performance sails and reinforcing polymer composites for application such as aircraft boats automobiles sporting goods and biomedical implants. Research has also brought about a number of remarkable fibre properties that include UV resistance electrical conductivity and biodegradability. This ability to engineer properties to meet specific requirements and resulted in demands from the industry for the production of nano diameter fibres with specific features.
Structured polymeric fibres with diameters in the submicron range are of considerable interest in a variety of applications. The reduction of the fibre diameter to the nanometer scale (1 × 103 ~ 100 × 1003 µm) results in a number of favourable properties such as increase in surface to volume ratio decrease in pore size and a drop in structural defects and superior mechanical characteristics. As a result these fibres are good candidates for use in the areas of tissue scaffolding high performance filtration chemical biological protective clothing and polymer composite reinforcement.
Polymeric nanofibres have undergone a surge in development over the last decade. This can be demonstrated by the literature publications shown in Fig. 1. However a systematic review of the various issues related to these fibres has not been published yet. The objective of this work is to provide a preliminary review on the processing methods characterization techniques modeling and applications of polymeric nanofibres. The following sections describe the processing characterization modeling and applications of nanofibres in some detail.
Processing Method
Polymeric nanofibres can be processed by a number of techniques mainly including Drawing Template Synthesis Phase Separation Self Assembly and Electrospinning methods which are briefly reviewed in the section. Some of the polymers that can be converted into nanofibres and a comparison of various issues relating to these processing methods can be found in Tables 1 & 2 respectively.
Drawing
Nanofibres have been fabricated with citrate molecules through the process of drawing. A freshly prepared aqueous solution of 50 ml. of chloroauric acid (0.01%w/v) was boiled under reflux and 1.75 ml. of an aqueous solution of sodium citrate (1% w/v) was added. A millimetric droplet of the chloroauric acid/sodium citrate solution was deposited on a SiO2 surface and allowed to evaporate. The solution became more concentrated at the edge of the droplet due to capillary flow after a few minutes of evaporation. A micropipette with a diameter of a few micrometers was dipped into the droplet near the contact line using a micromanipulator. The micropipette was then withdrawn from the liquid and moved at a speed of approximately 100 µm/s resulting in nanofibre being pulled. The pulled fibre was deposited on the surface by touching it with the end of the micropipette. Fibres were also drawn from a pipette containing the chloroauric acid/sodium citrate solution through the help of micropipette and the micromanipulator. The drawing of nanofibres was repeated several times on every droplet.
The viscosity of the material at the age of the droplet increased with evaporation. At the beginning of evaporation corresponding to part A of the curve in Fig. 2 the drawn fibre broke due to Rayleigh instability. During the second stage of evaporation corresponding to part B of the curve nanofibres were successfully drawn. In the final stage of evaporation of the droplet corresponding to part C of the curve the solution was concentrated at the edge of the droplet and broke in the cohesive manner. Thus drawing a fibre requires a viscoelastic material that can undergo strong deformations while being cohesive enough to support the stresses developed during pulling. The drawing process can be considered as dry spinning at a molecular level.
It was found that the gelation time increased with increasing gelation temperature for a given polymer concentration and decreased with increasing polymer concentration for a given gelation temperature. It was also established that with increasing polymer concentration nanofibre diameter remained unchanged pore structure became increasingly uniform average unit length decreased and mechanical properties (Young s modulus elongation at break and tensile strength) of the matrices improved.
Nanofibre matrices from PLLA and PLLA polycaprolactone (PCL) blends of blend ratios 80 20 and 50 50 have also been fabricated through the phase separation method. It was also concluded that the key to obtaining a get from a polymer solution is a restriction of a molecular movement in the polymer and can be achieved by cooling to a temperature lower than the Tg of the polymer in the solution.
Electrospinning
Electrospinning is a process that creates nanofibres through an electrically charged jet of polymer solution or polymer melt. Investigations of the process have been carried out by a number of researches.
The electrospinning process in its simplest form consisted a pipette to hold the polymer solution two electrodes and a DC voltage supply in the kV range Fig. 3. The polymer drop from the tip of the pipette was drawn into a fibre due to the high voltage. The jet was electrically charged and the charge caused the fibres to bend in such a way that every time the polymer fibre looped its diameter was reduced. The fibre was collected as a web of Fibres on the surface of a grounded target (Fig. 3).
Characterization of Structure and properties
The structure and properties of nanofibres and nanofibrous meshes and related characterization methods are discussed in is section. Table 3 lists the structure and properties of nanofibres or nanofibrous meshes and their characterization methods and instruments. It can be seen from Table 3 that most properties of nanofibres and their meshes can be investigated through conventional methods and instruments.
Structure
Molecular structure
The molecular structure of a nanofibre can be characterized by Fourier Transform Infra Red (FTIR) and Nuclear Magnetic Resonance (NMR) techniques. If two materials were blended together for nanofibre fabrication not only the structure of the two materials can be detected but also the inter molecular reaction can be tested. In the case of a collagen and PEO blend used for electrospinning nanofibres the NMR spectrum showed a new phase structure which was caused by the hydrogen bond formation between the ether oxygen of PEO and the protons of the amino and hydroxyl groups in collagen.
Supermolecular structure
Supermolecular structure describes the configuration of the macromolecules in the nanofibre and can be characterized by optical birefringence wide angle x ray diffraction (WAXD) small angle x ray scattering (SAXC) and differential scanning calorimeter (DSC). Fong and Reneker studied the birefringence of the styrenebutadiene styrene (SBS) triblock copolymer nanofibres and diameters around 100nm under an optical microscope. The occurrence of birefringence reflects the molecular orientation. Zong found that the electrospun PLLA fibres quenched below 0°C resulted in amorphous fibre structure. But after drying the electrospun nanofibres at room temperature melting point. transitions appeared in two peaks by differential scanning calorimeter (DSC) Fig. 4. It was explained that during electrospinning the polymer molecule had no time to crystallize and hence it can only have an amorphous supermolecular structure. Since the supermolecular structure changed in elecrospinning the transition points of the polymers also changed. One of them was lower than the normal melting point due to defects existing to crystallization during drying.
Geometric structure of nanofibre
The diameter diameter distribution fibre arrangements (random parallel and layer by layer with different angles) and fibre morphology (cross sectional shape and surface roughness) can be observed and measured using scanning electron microscopy (SEM) transmission electron microscopy (TEM) and atomic force microscopy (AFM). The use of TEM does not require the sample in a dry state as is SEM. Hence nanofibres processed through self assembly from a polymer solution can usually be directly observed under TEM. An accurate measurement of the nanofibre diameter with AFM requires a rather precise procedure. The fibres appear larger than their actual diameter because of the AFM tip geometry. For a precise measurement two fibres crossing each other on the surface are generally chosen. The upper horizontal tangent of the lower fibre is taken as reference and the vertical distance above this reference is considered to be the exact diameter of the upper nanofibre.
If the nanofibre has a smooth surface the surface area and the surface area to volume ratio can be easily calculated from the diameter. But in most cases the nanofibre surface is not smooth since their morphology is based on different processing methods/parameters. And in this case an instrument is needed to measure the surface area of the nanofibre.
Structure of nanofibrous mesh
The porous mesh structure in a nanofibrous mesh can be observed under SEM TEM and AFM. The pore diameter and pore distribution in the mesh are very important parameters for the mesh to be used as a filter or a tissue engineering scaffold. They can be measured by mercury porosimeter. Li have shown that the porosity of electrospun nanofibrous structure was 91.63% indicating that it is a highly porous structure. The total pore volume was 9.69 mL/g the total pore area was 23.54 m2/g. and the pore diameters ranged broadly from 2 to 456 µm.
Properties of Nanofibre and Nanofibrous Mesh
Surface properties of nanofibre
The surface properties of nanofibres include surface chemical properties surface compare morphology iopography and surface roughness. Surface chemical properties can be determined by X ray photoelectron spectroscopy (XPS) water contact angle measure ment and FTIR ATR analyses. Deitzel have measured the atomic percentage of fluorine in a PMMA TAN blend. It was shown that the atomic percentage of fluorine in the near surface region of the electrospun fibres is about double the atomic percentage in a bulk polymer. Surface chemical properties of nanofibre also can be evaluated by its hydrophilicity which can be measured by the water contact angle analysis of the nanofibre membrane surface. AFM can be used to measure the roughness of fibres. The roughness value is the arithmetic average of the deviations of height from the central horizontal plane given in terms of millivolts of measured current. Table 4 lists the roughness of different fibres and it can be seen that the roughness of nanofibres is much higher than that of wet spun fibres.
Air and vapour transport properties Protective clothing is an important research are under development at Natick Soldier Center (NSG). The air and vapour transport properties of nanofibrous mesh have been measured using an apparatus called the Dynamic Moisture vapour permeration cell (DMPC). This device has been designed to measure both the moisture vapour transport and the air permeability (convective gas flow) of continuous film fabrics coated textile and open foams and battings.
Mechanical properties Tensile properties of a nanofibrous mesh can be measured by conventional mechanical testing methods. Specimens of 10 × 60 × 1mm were vertically mounted on two 10 × 10mm mechanical gripping units of the tensile tester at their ends leaving a 40 mm gauge length for mechanical loading. Load deformation data are recorded at a deforming speed of 0.05 cm/s and the stress strain curve of the nanofibres structure was constructed from the load deformation curve. Fig. 5 shows the typical stress strain curve of a PLLA nanofibrous mesh. It has been found the tensile strength of nanofibrous mesh is similar to that of the natural skin.
Due to the nanometer or submicron scale dimensions of nanofibres characterizing the mechanical properties of individual nanofibre is a challenge for many existing testing and measuring techniques. The established methods and standards for determining mechanical behaviour of conventional fibres are inadequate in the case of manipulation or testing of nanofibres. This is probably one of the reasons that articles on the mechanical properties of nanofibres are rare in the literature.
Warner has described a cantilever technique to measure the tenacity of a single polymer ultrafine fibre. A cantilever consisting of a 30µM glass fibre was glued at one end onto a microscope slide and a 15µm nylon fibre was attached at the free end of the glass fibre. The electrospun test fibre was glued with epoxy resin to the free end of the nylon fibre. A part of the same fibre was cut and deposited on a SEM specimen holder for diameter measurement using SEM. As the sample fibre was stretched with a computer controlled Instron model 5569 the deflection of the cantilever was measured under light microscopy using calibrated eyepiece. A chart was used to convert deflection into actual values of fibre tenacity. The elongation to break of electrospun polyacrylonitrile (PAN) fibres was estimated using a caliper. With the said technique it was reported that the electrospun PAN fibres with a diameter of 1.25 µm and length of 10mm exhibited failure of 0.4 mm deflection at 41mg of force and the resulting tenacity was 2.9 g/d. The mean elongation at break of the same fibre was 190% with a standard deviation of 16%.
Xie described a method for tensile measurement of a bunch of carbon nanotubes with the parallel structure by pulling long (-2 mm) ropes containing ten thousands of the aligned nanotubes with a specially designed stress strain puller.
Biological properties
Biodegradability of nanofibrous structure The biodegrada bility of nanofibrous mesh is expected to be different from the biodegradability of other structural scaffolds. However publications related to the biodegradability of nanofibrous structures in vitro or in vivo are not currently available in the literature. The nanofibrous scaffold can be placed in PBS at 37°C in an incubator or implanted into animals and the biodegradability can be measured from the mass loss the strength loss and morphology change over a period of time.
Cell growth on nanofibrous structures Cell proliferation on nano fibrous scaffolds have been tested by seedling cells on the nanofibre member. It has been observed that the cell proliferation speed is higher than that on other membrane materials. Fertala seeded osteoblasts on carbon nanofibrous scaffolds with different nanofibre diameters and found that compared to larger diameter carbon fibres osteoblasts synthesized more alkaline phosphatase and deposited more extra cell calcium on nanometer diameter carbon fibres after 7 14 and 21 days of culture.
Modeling and Simulation
The purpose of modelling and simulation of a subject is to relate the macroscopic behaviour or properties of the subject with its state variables and then to design and optimize the subject by choosing suitable variables. The three major modeling related areas are the elecrospinning process the nanofibrous perform and the single nanofibre.
Modeling of electrospinning process
Most of the modeling work in the current literature has been concerned with the elecrospinning process. However when a polymer melt or solution is charged with an electric field superfine fibres with diameters smaller by two to three orders of magnitude compared to microfibres can be developed. Parameters that influence the electrospinning process include (a) polymer solution properties such as viscosity conductivity and surface tension (b) controlled variables such as hydrostatic pressure in the capillary tube electric potential at the tip and the gap (distance between the tip and the collecting screen) and (c) ambient parameters such as temperature humidity and air velocity in the electrospinning chamber. In order to control the property geometry and mass production of the nanofibres it is necessary to understand quantitatively how the electrospinning process transforms the fluid solution through a millimeter diameter capillary tube into solid fibres which are four to five orders smaller in diameter.
The electrospinning process is a fluid dynamics related problem. When the applied electrostatic force overcome the fluid surface tension the electrified fluid forms a jet out of the capillary tip towards a grounded collecting screen. The process consists of three stages (a) jet initiation and the extension of the jet along a straight line (b) the growth of whipping instability and further elongation of the jet which may or may not be accompanied with the jet branching and/or splitting and (c) solidification of the jet into nanofibres.
Jet initiation
The basic principles for dealing with the elecrospinning jet fluids were developed by Taylor. Taylor showed that fine jets of various monomeric liquids can be drawn from conducting tubes by electrostatic forces. As the potential of the conducting tube rises the originally planar fluid meniscus becomes nearly conical and fine jets are drawn from the vertices.
According to Taylor the formation of fine threads from viscous liquid drops in an electric field is due to the maximum instability of the liquid surface induced by the electrical forces. Taylor also has shown that a viscous fluid exists in equilibrium in an electric field when it has the form of a cone with a semi vertical angle 49.3°. In other words a fluid jet is developed when the semi vertical cone angle becomes 49.3°. Such a cone is now well known as the Taylor s cone. The Taylor s cone angle has been independently vertified by Larrondo and Menley who experimentally observed that the semi vertical cone angle just before jet formation is about 50°.
Another issue related with the initiation of the jet is the strength of the electrostatic field required. Taylor also showed that the critical voltage Vc (expressed in kilovolts) at which the maximum jet fluid instability develops is given by
where H is the distance between the electrodes (the capillary tip and the collecting screen) L is the length of the capillary tube R is the radius of the tube and is the surface tension of the fluid [Units H L and R in cm in dyn per cm]. In spinning the flow beyond the spinneret is mainly elongational.
Hendricks also calculated the minimum spraying potential of a suspended hemispherical conducting drop in air as
where r is the jet radius. If the surrounding medium is not air but a non conductive liquid immiscible with the spinning fluid drop distortion will be great at any given electric field and therefore the minimum spinning voltage will be reduced. This means that if the elecrospinning process is encapsulated in vacuum the required voltage will be lower.
SYNTHESIS OF NANOSTRUCTURE
Over the past few years semiconductor materials have been a focus point in material science due to their high application potentials in various devices such as light emitting diodes single electrons transistors and field effects thin film transistors. Recently with the rising interests in nanoscience and nanotechnology a lot of effort has been made in the direction of low dimensional nanostructure semiconductors. In particular many works have reported chain like nanostructures synthesized by the assembly of nanoparticles including CdTe Ti Ni and so on. On the other had ever since carbon nanotubes were synthesized many other tubular nanostrucutres have been synthesized also such as WS2 MoS2 BN etc. However there is no report on the synthesis of chain like nanotubes.
Nickel sulfide has been the subject of considerable interest because of its properties as a metal insulator a paramagnetic-antiferromagnetic (PM-AFM) phase changing material and its use as a hydrodesulfurization catalyst and for solar storage. Different morphologies and nanostructures of nickel sulfide nanomaterials including nanoparticles hollow nanospheres nanorods nanotubes and urchin like nanostructures have been synthesized in different ways. The research is even expanding into the production of nanomaterials in two dimensional (2D) and three dimensional (3D) ordered superstructures. Some special superstructures of nickel sulfides have been synthesized Xie s groups synthesized 3D NiS with flower like architectures with the assistance of polymer. Qian s group reported urchin like nanostructures of NiS by hydrothermal synthesis and layer rolled tubes by a wet chemical method. But there are no reports about the synthesis of chain like NiS tubes and echinus like Ni3S2 nanostructures.
Experimental details
By using Ni nanochains as a precursor we have successfully synthesized nickel sulfides of these special nanostructures via a reaction between nickel and sulfureted hydrogen decomposed from thiourea. The Ni nanochains were obtained by the reduction of NiCl2. 6H2O with hydrazine monohydrate (N2H4H2O) in the presence of polyvinylpyrrolidone (PVP molecular weight 40000) at the constant temperature. Then thiourea as the sulfide sources was added to the Ni nanochain precursors. Echinus like Ni3S2 nanostructures or chain like NiS tubes were obtained with different concentrations of thiourea.
Various processes were explored for the synthesis of nickel sulfides with different morphologies and nanostructures. In a typical procedure for the preparation of chain like NiS tubes (process 1) 0.119 g of NiCl2 6H2O and 1.66 g of PVP were dissolved into a 40 ml solution of ethylene glycol (EG). Then 2.6 ml of hydrazine monohydrate (80%) diluted by 4 ml of EG was added dropwise with simultaneous vigorous agitation at room temperature. Afterwards the solution was heated to 158°C with stirring at that temperature for 3 h. Then 0.114 g of thiourea dissolved in 16 ml of EG was added to the mixture with stirring at 158°C for another 7 h. Finally brown black precipitates (chain like NiS tubes) were formed. The resulting nickel sulfide was processed by contrifugation at 4000 rpm for 30 min after being diluted with absolute ethanol several times. The product was then dried in vacuum at 60°C for 2 h.
Compared to the synthesis of chain like NiS tubes the dosages of the agents for the synthesis of echinus like Ni3S2 were all increased. For a typical procedure of the echinus like Ni3S2 synthesis (process 2) 0.142 g of NiCl2. 6H2O and 2 g of PvP war dissolved into a 40 ml solution of EG. Then 3.2 ml of hydrazine monohydrate (80%) diluted by 4.8 ml of EG was added dropwise with simultaneous vigorous agitation at room temperature. Afterwards the solution was heated to 158°C and stirred for 3h. Then 0.167g of thiourea dissolved in 16 ml of EG was added to the mixture with stirring at 158°C. After 7th of mild stirring at 158°C the echinus like Ni3S2 precipitates were formed. The resulting metal sulfide was filtered and washed with absolute ethanol several times and then dried in vacuum at 60°c or 2 h. Other experiments were performance to investigate any possible mutation process of the two morphologies. For a typical process (process 3) the dosages of the reagents were the same as those in process 2 except that the dosage of the thiourea was 0.137 g.
X ay powder diffraction (XRD) patterns of the products were obtained on a Rigaku Dmax 2200 x ray diffractometer using Cu K radiation (0.154060 nm) with 2 scanning for 10° to 70°. further microstructural analyses were performed using scanning electron microscopy (SEM JSM 5800 with accelerating voltage of 15 kV) and analytical transmisson electron microscopy (TEM JEOL 2100F). The TEM samples were prepared by dispersing the powder products in alcohol by ultrasonic treatment dropping the suspension onto a holey carbon film supported on a copper grid and drying it in air. The magnetic properties of the sample were measured using a Quantum Design SQUID magnetometer.
The morphology of the chain like NiS tube was visualized by SEM and TEM. The diameter has a narrow size distribution ranging from 280 to 320 nm and the length are over 10 µm. The wall of the tubes is about 80-100 nm in thickness.
The inset reveals further information on the nanostructures obtained by high resolution TEM (HRTEM). It clearly shows that the tube has branches similar to those of the precursor Ni nanochains. The tube is shown to have a structure of connected hollow spheres. Most of the hollow spheres do not have dissepiments to separate themselves from the adjacent ones. The inset shows the well crystallized structure of the tube. The lattice fringe spacing was determined as 0.48 nm which is close to the lattice spacing of the bulk NiS. To the best of our knowledge such chain like NiS tubes with a highly branched structure have not been reported yet in the literature.
The product was formed in echinus like morphology with acicular crystallites radiating from the centre with a uniform size distribution. In contrast to the urchin like nanostructure of NiS reported by Qian s group our products have fewer slenderer feelers with a smaller core. The feelers are almost of uniform size reaching out from the centre. The root is about 40 nm and the tip about 20 nm in diameter. The length of the feelers is about 500 mn-1 µm.
The HRTEM images provide further insight into the nanostructures of the feelers. The feeler has a typical diameter of ~20 nm. Lattice fringes parallel to the axial direction of the feeler with spacing about 0.41 nm corresponding to the plane are well resolved. The feeler was also found to be flexural without any crack showing flexibility of the materials in this nanometer scale. The crystalline structure can be even slenderer. The fringes correspond to planes of Ni3S2 which gives the growth direction of the feeler. Some nanotwins or stacking faults can be found in the feelers.
Regarding the growth of chain like NiS tubes and echinus like Ni3S2 we believer that the hydrazine monohydrate PVP temperature and the concentrations of reagents play important roles. Generally the whole growth process of the crystals is separated into two steps. In the first step the Ni chain precursors were synthesized via deoxidizing Ni+ ions with N2H4.H2O. The formation of Ni chain precursors was likely to evolve here according to the following chemical reactions.
It is well known that the crystal growth stage is a kinetically and thermodynamically controlled process that can from 1D and other more complicated structures by changing the reaction parameters such as temperature reaction time concentration and so on. According to the second step mentioned above our results indicate that the temperature plays an important role. Compared to decomposition at the boiling point 197°C of EG the thiourea decomposes slowly at 158°C according to the reaction kinetics. In many of our previous works the mixed solution was heated to 197°C and refluxed. The thiourea decomposed quickly and only solid particles shown in figure 6 were produced.
The concentration of the reagents is another key factor for controlling the morphology of the nickel sulfides. After the formation of Ni nanochains the thiourea was added to the system. The Ni particles which assembled into necklace like chains were then sulfureted by the sulfureted hydrogen decomposed from the thiourea. According to the image revealed by the HRTEM in figure 3 the sulfureted product shows a structure of concentration by tens of hollow spheres with only a few dissepiments between adjacent spheres. This suggests that the structure of the chain like tube is not obtained by a self assembly process from the hollow spheres which form in advance. Otherwise dissepiments should be observed. According to the mechanism reported by Xia we propose that the chain like tubes from with the precursor Ni nanochains as template. With the Ni nanochains as sacrificial templates and the slow release of the sulfureted hydrogen to react with the Ni described by equation (4) NiS would grow on the surface of the Ni nanochains. This process leads to the formation of a regular chain like tubular sheath whose morphology was complementary to that of the Ni nanochains. And eventually with the Ni depleting by the reaction a chain like tube of NiS is obtained. This explains why only a few dissepiments are observed. A schematic representation for the formation of the chain like NiS tubes is shown in figure 2(a).
As the concentration of the regents increases with process 2 the morphology of the synthesized product turned into being echinus like. Additional experiments by process 3 were carried out to study the possible mutation process. The results show that when the dosage of the thiourea was 0.136 g the synthesized product results in two different irregular morphologies both chain like tubes and echinus like nanostructures. Therefore with lower concentration of the reagents by process 1 chain like tubes form. When the concentration of the reagents increases by process 3 feelers are observed to grow on the walls of the tubes. When the thiourea dosage increases to 0.167 g by process 2 all the tubes are disjoint and this leads to the formation of a full echinus like nanostructure. A schematic representation for the formation of the echinus like nanostructure is shown in figure 2(b).
Another mechanism to form the echinus structure of Ni3S2 is also possible. The precursor Ni nanochains may break up into particles by the force of stirring. Then with the slow continuous decomposition of the thiourea the echinus like nickel sulfide structure starts growing from the surface of Ni nanoparticles. This is similar to the formation process CdX (X = Te or Se) 3D nanocrystals. Some of the naoparticles are shown in figure 4(a) with diffraction peaks for a Ni phase in the XRD pattern shown in figure 1(b). These evidence support the conjecture about the disorganization of the Ni nanochains. A schematic representation describing this mechanism is shown in figure 2(c). So the formation of the echinus like nanostructure is attributable to two different mechanisms. Further studies to understand the details of the formation mechanism of chain like NiS tubes and echinus like Ni3S2 nanocrystals are in progress.
The temperature and field dependences of dc magnetization M(T ) and M(H ) of our samples have been investigated using a SQUID magnetometer (Quantum Design). The data have been corrected for the background diamagnetism of the capsule and the straw which hold the sample for the measurements. Zero field cooling (ZFC) and field cooling (FC) processes were applied to measure M(T ). To obtain MZFC(T) the samples were cooled under zero applied fields to 5 K first and then warmed up in a small recording field of 90 Oe. A similar procedure was applied in the FC measurement to study MFC(T) expected that the samples were cooled with the presence of an applied magnetic field of 20 kOe.
Figure 3(a) shows the MZFC(T) and MFC(T) curves for the NiS nanotubes. The magnetization increase with decreasing temperature in both ZFC and FC measurements showing a typical paramagnetic (PM) property. However a weak ferromagnetic (FM) component is present as shown by the slight separation between the MZFC(T) and MFC(T) curves. The separation evidence the presence of a magnetic potential barrier with an FM phase. The roughness on the measuring curves through it seems to be reproducible to some extent is believed to arise from the measurement instead of from the sample property. The magnetic property of the sample was further investigated by the M(H) measurement performed at T = 5 K as shown in figure 9(b). The magnetization shows a strong PM response in the applied field going up to 9 kOe. The saturated magnetization Ms of the FM component was determined by extrapolating of the Mversus H curve to the axis H = 0. The value was determined as Ms ~ 0.14 emu g-1. The coercivity was determined as 45 Oe from the hysteresis loop in the low field region as shown in the inset of figure 3(b).
ASSEMBLY OF NANOCRYSTALS
White light emitting diodes (WLEDs) are promising devices for their potential use in many lighting applications including in the display and automotive industries. Up to the present several approaches to generating white light have been exploited including multi layer monolithic fabrication multi chip combinations and color conversion using phosphor molecules commonly pumped by blue emitting nitride LEDs. Among these approaches the color conversion technique using phosphors has been the most successful one and has been commercialized. For phosphors however it is difficult to control the granule size and the mix and deposit uniform films which disadvantageously results in undesired visible color variations. Also the phosphors photoemission properties are not easy to adjust and consequently the color parameters of its resulting while light are not as easy to tune as desired. As an alternative light source colour conversion WLEDs that rely on nanocrystal (NC) emitters instead of phosphors have recently been demonstrated to overcome these disadvantages. In our previous research work such hybrid light sources using multiple nanocrystals pumped by blue nitride LEDs have also been shown to allow their colour parameters to be readily tuned as an additional advantage. Unlike using phosphors employing such a combination of semiconductor nanocrystals makes colour tuning possible because their peak emission wavelength can conveniently be controlled with crystal size enabling sensitive colour using as the crystals can be synthetically prepared each with a narrow size distribution for a comparatively narrow photoluminescence spectrum.
In such hybrid NC WLEDS though nanocrystal efficiency and hybridization significantly affect the device performance. In previous research work mostly blue LED platforms are used as the excitation sources. However ultraviolet (UV) emitting LEDs provide a better platform of nanocrystal based white light generation for a number of reasons. First at UV wavelengths the absorption of nanocrystals is higher than in blue and as a result with UV pumping thinner nanocrystal films are needed to achieve white light generation compared to blue pumping. Second with a UV pump source the white light emission results solely from the photoemission of the nanocrystals. This implies that the generated white light does not directly depend on the LED platform but only on the combination of NC emitters facilitating easier color tuning. Third in the near future UV LEDs are expected to reach significantly higher optical power levels as they are increasingly aggressively pushed forward by other industrial wide scale high power applications (e.g. photolithography inkjet printing resin curing etc.). In this regard the famous Japanese LED maker Nichia (where blue LEDs were invented has announced the production of UV LEDs with output optical powers up to 5 W in the short term. To leverage rapidly advancing UV technology into efficient lighting with nanocrystal based color conversion it is critical to develop and demonstrate hybrid light sources on UV pumping platforms. Very recently Ali showed white light generation on a commercial UV LED coated with blended CdSeS nanocrystals. This work is particularly impressive for the synthesis of high quality CdSeS NCs and white photemission from the NC mixture under UV pumping. In this work we inform the NC mixture under UV pumping. In this work we independently introduced and demons trated an alternative hybrid light source that relieves on the layer by layer assembly of multiple nanocrystals carefully hybridized of our near UV emitting nitride diodes in a controlled manner for adjustable white light parameters for the first time. Such layer by layer hybridization offers advantages of controlling precisely the individual NC film thicknesses and NC film order in addition to setting the concentrations and overall film thickness. Here we report the design growth fabrication integration and characterization of these hybrid NC LED sources.
In this work we use CdSe/ZnS core-shell nanocrystals (emitting at PL = 504 580 and 615 nm) integrated on InGaN/GaN n UV LEDs that we developed as a UV excitation sources at LED = 383 nm for our nanocrystals. Here rather than blending NCs into a single film we incorporate these nanocrystals of different size (1.9 3.2 and 5.2 nm) layer by layer in adjacent thin films of a few hundred nanometers in thickness. Our first hybrid device includes the layer by layer assembly of cyan and re emitting NCs (PL = 504 and 615 nm). For this dichromatic NC combination the tristimulus coordinates (x y) are (0.37 0.46) with correlated color temperature Tc = 4529 K and color rendering index Ra = 43.1. Our second device uses the layer by layer hybridization of a trichromatic combination of cyan yellow and red emitting NCs. (PL = 500 540 and 620 nm) with (x y). = (0.38 0.48) Tc= 4474 k and Ra = 67.6 respectively. The use of such nanocrystal combinations enables the color properties of the resulting emission such as tristimulus coordinates correlated color temperature and color rendering index to be controlled and adjusted.
For white light generation we use CdSe/ZnS core-shell nanocrystals of crystal size 1.9 3.2 and 5.2 nm (with a size distribution of 5%) from Evident. The emission colors of these nanocrystals are cyan yellow and red tuned using the quantum size effect across the visible spectral range with their corresponding photoluminescence (PL) peaks at 504 nm 580 nm and 615 nm respectively. These nanocrystals have high PL quantum yields ranging between 30 and 50%. Such core-shell NCs are shown to yield even higher efficiencies up to 66%. We used similar types of nanocrystals in our previous work for white light sources intergrated on blue nitride LEDs and nanocrystal based UV scintillators intergrated on Si detectors. In this work were prepare high concentration NC solutions to vortex mix into the host polymer of poly (methyl methacrylate) (PMMA). We evaporate NC films in micro droplets of 25µl by drop casting at 70-100°C for optimal film formation and complete the polymerization process for each layer in the assembly. Figure 1 shows the photoluminescence and absorption spectra of our CdSe/ZnS core shell nanocrystals in thin films of the host PMMA.
We use InGaN/GaN based LEDs with a peak wavelength of 383 nm in the near ultraviolet as the pump source for the entire hybrid devices present in this paper. Figure 2 shows the design of our LEDs. For epitaxial growth we use a GaN dedicated metalorganic chemical vapor deposition (MOCVD) system. First we begin with a 14 nm thick GaN nucleation layer. To increase the crystal quality of the subsequent epitaxial layers we then grow a 200 nm thick GaN buffer layer. For the n type contact we grow a 690 nm thick Si doped epitaxial layer. For the active layers we continue the epi growth with five 2 3 nm thick InGaN wells and GaN barriers at a growth temperatures of 720°C. We finish our growth with the p type layers that consist of a 50 nm thick Mg doped AlGaN layer and a 120 nm thick Mg doped GaN layer as the contact cap. Finally we activate the Mg dopants at 750°C for 15 min. We used similar growth steps for the development of our GaN based quantum electroabsorption modulators.
Our device fabrication follows standard semiconductor processing procedures as in our pervious work. These include photolithography thermal evaporation (metallization) reactive ion etch (RIE) and rapid thermal annealing. We use pholithography and reactive ion etching (down to 940 nm) to expose the n type layer and then apply thermal evaporation to deposit the metal contacts. We form the ohmic p contacts with metal deposition of 15 nm of Ni with a 100 nm overlayer of Au followed by rapid thermal annealing at 700°C for 30 s. We lay down a 10 nm layer of Ti with a 200 nm overlayer of Al for the n contacts followed by a rapid thermal anneal at 600°C for 1 min under nitrogen. Figure 3(a) shows the I V characteristics of our fabricated n UV LEDs with turn on voltages at approximately 5 V and figure 3(b) shows their electroluminescence (EL) spectra under different levels of current injection with the EL peak wavelength at 383 nm. In hybrid devices this high energy electroluminescence of the LED excites the integrated NC layers resulting in spontaneous emission from each NC layer.
ELECCTROSYNTHESIS OF CEO2 NANOTUBES
In the emerging field of nanotechnology the production of nanostructures having special physical and chemical properties with respect to those of bulk materials is an objective due to their limited size and high density of corner or edge surface sites. In recent years interest in nanostructured metal oxides production has increased because of their important role in many areas of chemistry physics and materials science. Indeed oxides are used in several advanced applications including microelectronic circuits sensors fuel cells anti corrosion coatings and catalysts. Furthermore the most active areas of the semiconductor industry involve the use of oxides. A variety of nanostructured oxides such as Cu2O SnO2 MgO ZnO TiO2 PbO2 MnO2 NiO and MoO3 have been fabricated. Cerium oxide has been widely studied owing to the technological relevance. For instance CeO2 is used in solid oxide fuel cells (as a solid state electrolyte or anode) oxygen gas sensors oxygen permeation membrane systems humidity sensors photo catalytic devices for corrosion protection and as a catalyst in numerous industrial processes. Its used in catalysis is due to its chemical and physical stability high oxygen mobility at moderate temperature and high oxygen vacancy concentration. The possibility of switching between reduced and oxidized states (cycling around the red O×h potential of the CeIV/CeIII couple) allows the reversible addition/removal of O2 to/from CeO2. All of the above mentioned features explain why this material in an active component of three way catalysts for the control of environmental pollution (to remove the CO NOx and SOx formed during the combustion of fossil fuels) and it is employed in several catalytic processes (such as the water gas shift reaction methane steam reforming and oxidation of different hydrocarbons).
Nanostructured materials with high specific surface areas are desirable for catalytic applications. Recent studies showed that nanostructured cerium oxide can be produced by different techniques such as the solution phase route sol gel process and hydrothermal route. However many of these techniques are complicated and thermal treatment is needed in most cases. Moreover nanotubes are often more difficult to fabricate than nanowires.
In this work we fabricated CeO2 nanotubes (NTs) from a non aqueous electrolyte using template electrochemical deposition for the first time. Through this technique we will show that it is possible to produce cerium (IV) oxide nanotubes in a single step without any thermal treatment. In comparison with other technologies of fabrication template electrodeposition is very simple quick to control and cheap.
We used anodic alumina membranes (AAMs) as a template of the fabrication of CeO2 NTs. AAMs present unique structural properties and it is possible to control their morphological parameters (memb ane thickness pore diameter and density) by varying the anodizing conditions (voltage and time an odizing bath composition and temperature). For these reasons a variety of nanostructures (metals alloys semiconductors oxides and polymers) with different morphologies (tubules fibers or wires and rods) have been fabricated using AAM templates.
The chemical composition and morphology of CeO2 NTs wer e investigated by SEM EDS XRD FTIR and Raman spectros copy while the nanotube length was controlled by varying the total electrodeposition time.
Experimental details
Fabrication of nanotube arrays
Cerium oxide NTs were grown in the pores of commercially available anodich alumina membranes (Whatman Anodisc 47). Prior to use the morphological features of these membranes were characterized by SEM analysis and permeability measurements. The porosity was found to be about 30% with a surface pore population of the order of 10 pore m-2. The average pore diameter was found to be about 210 nm. A membrane thickness of 55 ± 5µm was measured. For the template electrodeposition a conductive layer had to be deposited on one side of the membrane and it acted as the working electrode for the subsequent electrodeposition of the metal oxide. In this work a thin Au film was sputtered onto the membrane surface using a conventional sputter coater to make the surface of the membrane electrically conductive. A small piece of AAM was mounted onto an aluminum support by means of a conductive paste and the active surface was delimited by means of an insulating lacquer (Lacomit UK) in order to expose a plain area of about 1 cm2 to the solution. Thus the gold layer was in direct contact with the electrolyte permeating the pores of the membrane and the electrodeposition process took place inside pores. Prior to electrodeposition samples were immersed for 2 h in the deposition bath in order to fill the AAM channels. This step is very important in order to achieve a homogeneous growth of nanostructures through the whole electrode area.
CeO2 potentiostatic deposition was preformed in a three electrode cell using a graphite counter electrode and a saturated calomel electrode (SCE) as the reference electrode this was connected with an electrochemical cell through a salt bridge. The plating solution was prepared by dissolving 0.3M of CeCl3.7H2O in absolute ethyl alcohol. The alcoholic solution was chosen because in aqueous solution the deposition of hydroxide is the prevailing process. Prior to use the electrolyte solution was stirred overnight at room temperature. A fresh solution was used for each experiment. During electrodeposition the anodic and cathodic compartments were separated by a sintered glass disk (porosity grade 1) in order to avoid mixing of catholyte with anolyte.
The electrodeposition was performed at -10 Volts vs. SCE (V(SCE)) at room temperature for different times (from 3 min to 1 h).
All experiments were carried out using an EG&G potentiostat/galvanostat (model 273 A). The deposition current was recorded by a desk top computer through an analogic interface using LABVIEW 7TM software. For comparison ceria powder was also produced under conditions identical to those adopted for the electrodepositon of CeO2NTs in AAMs. In particular a film of ceria elecroposited on a graphite sheet was mechanically removed and finely ground before characterization.
Nanotube characterization
After electrodeposition the crystallographic structure of the nanotubes was investigated by x ray diffraction using a Philips generator (model PW 1130) and PW goniometer (model 1050). All diffractograms were obtained in the 2 range from 10° to 100° with a step of 0.02° and a measuring time of 0.5 s for step using cooper K radiation ( 1.54Å). Diffraction patterns were analyzed by comparison with the ICDD database. The CeO2 NTs morphology was investigated using a scanning electron microscope (SEM Philips ESEM XL 30) equipped with an x ray energy dispersive spectometer (EDS). SEM analysis was carried out before and after dissolution of the AAM in 1 M NaOH aqueous solution at room temperature. Prior to SEM examination sample surfaces were sputter coated with a thin layer of conducting gold in order to avoid electrostatic charging under the electron beam. The length of NTs after different electrodeposition times was estimated from the SEM cross section pictures. The chemical compositon of CeO2 NTs was analyzed by energy dispersive spectrometry (EDS). A Shimadzu spectrophotometer (model 8400) was used for FTIR analysis. Raman spectra were obtained at room temperature using a Renishaw (in Via Raman Microscope) spectrometer equipped with a microprobe (50×) and a CCD detector. The excitation was provided by the 532 nm line of a Nd YAG laser. The power of the incident beam on the sample was 5 mW and the width of the analyzed spot for each sample was about 2 µm. The time of acquisition was adjusted according to the intensity of the Raman scattering. The wavenumber values reported in the spectra have 1 cm-1 accuracy. For each experiment Raman spectra were recorded at several points of the sample to ascertain its homogeneity. The specra presented in this paper are averaged over all samples.
Results and Discussion
Ceria NT arrays were fabricated by potentiostatic deposition at -10 V (SCE) and room temperature. The deposition voltage was chosen taking into account that NT growth needs a high current density. Under such conditions the preferential sites for the formation of nanotubes are contact areas between the cathode surface and the channel wall at the bottom of the template. In contrast small current densities may lead to deposition only on some selected pore tips and thus to inhomogeneous pore filling. For ceria deposition we have found that -10 V (SCE) is an appropriate voltage for both NT formation and homogeneous filling of AAM pores.
Figure 1 shows the current density versus time plot recorded during the electrodeposition of cerium oxide. Three different regions can be observed due to different deposition steps. In region I of figure 1 the current density decreases sharply at the beginning of the electrodeposition process while the applied potential is kept constant. This behaviour is an indication that the nucleation process of CeO2 is occurring. The decrease of the measured current is due to the increase of surface resistance as a result of cerium oxide formation. Ater the nucleation step region II is characterized by a gradual current decrease during this period the growth of CeO2 NTs in AAM channels occurs. The almost linear shape of current in this region suggests that the thickening of CeO2 NTs happens at an approximately constant rate. Complete filling of the pores is marked by a second slope change (region III of figure 1). In this region the current density decreases very slowly with the electrodeposition time because pore filling is complete and the deposition of a continuous and dense layer of CeO2 onto the external surface of AAM occurs. The formation of this flm is confirmed by the presence of a yellow deposit characteristic of cerium (IV) oxide on the surface of the sample after the electrodeposition process. A contribution to current decrease in region II of figure 1 could arise also from the depletion of cerium ions inside the pores. However this hypothesis was discarded because in such a case a sharp increase of the current would be observed at the beginning of region III when the channels are filled and the nanotubes make contact with the bulk concentration of ions. The decreasing trend of the current density also in region III of figure 1 supports the alternative hypothesis of a resistivity increase due to deposition of oxide. The low slope of the curve in this region can be attributed to the slow increase of the thickness of the deposit.
During electrodeposition the solution of the anodic compartment became yellow this was caused by the formation of Ce4+ ions at the counter electrode according to the reaction
The length of ceria NTs was dependent on the deposition time. After 30 min the length of NTs was about 35µm. After 1 h the length of ceria NTs was about 60 µm (figures 2-3) equal to the thickness of the alumina membrane. This indicates that NTs grow at a rate of about 1 µm min-1. The final aspect ratio (length to width ratio) was about 300.
Similar results have been obtained for the other deposition times. This parameter influence the length of the NTs but not their extenal diameter and the wall thickness which are similar those of figures 2-3.
Figure 2 shows a typical EDS spectrum of the NT array after dissolution of the template. This spectrum reveals that the nanotubes consist of pure ceria. Only peaks relative to Al (from the template) Au (sputtered layer) and Ce are present. Ce peaks are the most intense. In addition. FTIR analysis of the electrodeposited powder revealed the presence of only one peak at about 400 cm-1 which is assigned to the Ce-O stretching band. The same analysis on NTs showed the identical peak confirming that NTs consist of pure ceria.
Figure 3 reports the XRD pattern of CeO2 nanotubes grown in the AAM (dotted line) obtained after 1 h of deposition for comparison the diffraction pattern of CeO2 powder is also reported (solid line). In both samples pure CeO2 was electrodeposited. The XRD patterns showed only peaks corresponding with those of CeO2 powder reported in the ICDD data base (card number 81 792). All peaks correspond to the face centered cubic CeO2 phase. Importantly the intensities of XRD diffraction lines of CeO2 NTs are lower in comparison with those found for ceria powder. This effect is due to the presence of the alumina template. Moreover the diffraction peaks relative to CeO2 NTs are borader than those of the powder. This finding suggests that the grain sizes of CeO2 NTs and CeO2 powder were different. To confirm this assumption we evaluated the grain size using Scherrer s equation and we found average sizes of about 3 nm for the NTs and 15 nm for the powder.
Raman spectra for CeO2 NTs (dotted line) and powder (solid line) are reported in figure 4. The Raman spectrum of CeO2 NTs is very different from that of the CeO2 powder. Both spectra exhibit a main peak at an almost equal wavenumber of about 466 cm-1 that can be assigned to the ceria phase. This mode is attributed to a symmetrical stretching of the Ce O8 vibrational unit that is very sensitive to any disorder in the oxygen sublattice (resulting from non stoichiometry and/or different grain size). However it is worth nothing that the main peak for ceria powder (at 466 cm-1) is shifted with respect to the corresponding peak relative to NTs (463 cm-1). According to the XRD analysis this shift can be attributed to the smaller grain size of CeO2 NTs in comparison to ceria powder. The microstructure of ceria deposit also influences the shape of the Raman spectrum in fact in figure 6 it is possible to observe for the NTs (dotted line) a broadening of the main line and an increased asymmetry with respect to the main line of the powder. Both features are attributed to eduction of the phonon lifetime in the nanocrystalline CeO2. From the Raman line broadening it is possible to calculate the grain size of the ceria deposit. Table 1 shows that the grain sizes of CeO2 (NTs and powder) calculated from Raman and XRD experiments are in good agreement.
2D PATTERNED NANOCRYSTAL ARRAYS
One (1D) and two dimensional (2D) patterned nanocrystal arrays are currently generating enormous scientific and technological interest. They may range from elemental meta1lic semiconducting magnetic superconducting and optical nanocrystal arrays patterned on a non interacting substrate to more complex composites consisting of nanocrystals embedded in an interacting matrix. Such composites may find many applications in spintronics electronics and computing magnetic data storage magneto optics and biology. Therefore any costeffective simplified processing technique that can produce such nanocrystal arrays will likely impact a wide area of science and technology. We demonstrate here for the first time a simple processing technique that can produce 1D/2D magnetic nanocrystal arrays by heterogeneous nucleation and growth of nanocrystals during partial crystallization of an amorphous alloy deposited on a 2D array of catalytic nanocrystals. Amorphous alloys are particularly attractive as starting materials because their metastable nature allows a much wider solubility of the elements in the supercooled liquid than is possible with their thermodynamically stable crystalline counterparts.
The choice of the catalytic nanocrystals (Cu) and the amorphous alloy Fe64.5Cr10Si13.5B9Nb3 used in this work originates from earlier crystallization mechanism studies of these types of amorphous alloys. Amorphous Fe73.5Si13.5B9Nb3Cu1 known as FINEMET was discovered in 1988 and drew immediate attention for its extremely soft ferromagnetic properties (high permeability very low coercivity low magnetostriction and high saturation magnetization). When partially crystallized it produces a very high density (~1024 m-3) of extremely small (~10 nm) ferromagnetic Fe Si nanocrystals. The nanocrystals are surrounded by the remaining amorphous matrix which is also ferromagnetic. It is now well understood that atomic clusters of Cu atoms first precipitate from the amorphous alloy in high density at spatially random locations (homogeneous nucleation and growth) 100-120 K below the primary crystallization temperature. Subsequently Fe Si nanocrystals nucleate and grow heterogeneously on the Cu clusters near the primary crystallization temperature (790 K). It is believed that heterogeneous nucleation is favored by the low interfacial energy due to good lattice match between fcc Cu(111) and bcc Fe(011) surfaces. Growth of the Fe Si nanocrystals is hindered by the slow rejection (diffusion) of much larger Nb atoms (atomic radius of 0.147 nm compared to 0.127 nm for Fe from the amorphous matrix producing very small ferromagnetic nanocrystals. The important point to note is that such a unique microstructure is possible because of the catalytic activity of Cu slow diffusivity of Nb and the mutual immiscibility of Cu Nb and Fe under equilibrium conditions.
In this work we have successfully applied the above concept to produce patterned magnetic nanoarrays by providing the catalytic Cu nanoparticles externally in a 1D or 2D pattern using laser interference induced self organization of Cu thin films. The basic process consists of the following steps. First an array (1D or 2D) of nanocrystals of a suitably chosen catalyst (Cu in the present case) is produced on an appropriate substrate (such as SiO2) by nanosecond laser induced self organization. In this technique nanoparticle arrays with spatial order can be assembled due to a variety of mechanisms including a dewetting instability and/or thermocapillary convection with Rayleigh like instabilities. Second an amorphous film of a ferromagnetic alloy (Fe64.5Cr10Si13.5B9Nb3 in the present case) is deposited on top of this metal catalyst nanoarray by laser ablation. Third by appropriate heat treatment the desired nanocrystals (ferro magnetic FeCrSi nanocrystals in the present case) are allowed to nucleate and a grow on top of the catalytic nanoparticles from the amorphous phase by partial crystallization (devitrification). Compared to homogeneous nucleation which is a stochastic random process heterogeneous nucleation of the ferromagnetic nanocrystals occurs preferentially at the Cu catalyst sites as explained in the subsequent sections mimicking their 1D/2D pattern. The template provided by the nanocatalyst array produced by laser induced self organization in the present case can also be formed by more conventional methods such as nanolithography.
The ferromagnetic amorphous alloy Fe64.5Cr10Si13.5B9Nb3 was chosen instead of the FINEMET (Fe73.5Si13.5B9Nb3Cu1) because of its easier glass formability and the paramagnetic state of the residual amorphous matrix at room temperature after primary crystallization. The amorphous alloy films were deposited on the patterned Cu nanoarrays as well as the unpatterned Cu films. Magnetic force microscopy (MFM) measurements on thermally treated films confirmed that the resulting nanocrystal arrays consisted of ferromagnetic particles most likely of FeCrSi alloys and their spatial arrangement followed the periodicity of the Cu nanoarrays. On the other hand in the case of thermal treatment of the alloy films on the unpatterned Cu films only random spatial arrangements of ferromagnetic nanocrystals were produced. These results confirmed the heterogeneous nature of the devitrification process leading to ferromagnetic nanocrystal array formation. Although demonstrated here for a ferromagnetic system this processing method may be applied to other types of (semiconducting superconducting or optically active) materials by judicious choice of the amorphous alloy and the alloy specific catalytic nanocrystals. Additional advantages of this method are likely to come from (1) the ability to enhance or reduce magnetic interactions among the nanocrystal arrays by tuning the composition of the remaining amorphous matrix after partial crystallization (2) control of the aspect ratio (height/diameter) of the nanocrystals to produce columnar nanostructures through adjustment of film thickness annealing conditions and size of the catalytic nanocrystals and (3) the chemical protection provided to the nanocrystals from the ambient environment by the remaining amorphous matrix surrounding them.
Experimental details
The Cu films of thickness 2 10 nm were deposited by thermal evaporation or electron beam evaporation under high vacuum (10-7 Torr for thermal evaporation and 10-9 Torr for electron beam heating/on commercial optical quality SiO2/Si wafers consisting of a 400 nm thick thermally grown oxide layer on polished Si(100) wafers. Subsequent irradiation by laser interference patterns using a Spectra Physics Nd YAG laser (266 nm wavelength 10 ns pulse width 50 Hz repetition rate) at laser energies above the melt threshold of the Cu film for ~1500 pulses produced 1D or 2D patterned arrays of Cu nanocrystals. Two beam interference was performed at an angle of 37° to produce a 1D pattern with spacing = ~ 416 nm where is the interference angle and is the laser wavelength. Three beam interference was performed with three pairs of beams interfering at angles of 54° 46° and 34°. These resulted in a 2D pattern with the periodic particle spacings of 295 340 and 460 nm respectively.
The ferromagnetic amorphous alloy films of thickness 20 30 nm were deposited on top of the planar Cu films or patterned Cu nanocrystals by pulsed laser ablation using the same laser. The laser ablation target was made by are melting of high purity (3N and higher) elements in stoichiometric ratio. The laser beam energy density used for ablation was 160 mJ cm-2 and the deposition time was 20 30 min at a base pressure of 3 × 10-8 Torr. During the amorphous alloy film deposition the substrate temperature was maintained around l35-140 K by liquid nitrogen cooling to facilitate amorphous film formation. The thin films were subsequently annealed at various temperature (673-823 K) for various lengths of times (15 min to 1 h) under high vacuum (10-6 Torr) inside sealed quartz tubes. Because of the possibility of easy oxidation of such thin films it was important to remove even trace amounts of oxygen from the sample surroundings during annealing. To achieve this roughly gram quantities of Ti50Zr50 getter ingots situated 4-5 inches away from the alloy films were included inside the quartz tubes. The getter was RF heated to ~1300 K for about 10 min to remove any residual oxygen before thermal annealing of the samples. The resulting nanostructure of the annealed alloy films was studied by scanning electron microscopy (SEM) and energy dispersive x ray spectroscopy (EDS) using a Hitachi S 4500 field emission SEM ambient tapping mode atomic force microscopy (AFM) and MFM. The AFM and MFM system used in this work is a Digital Instruments Dimension 3000 Multimode IIIA scanning probe microscope. The probes used are magnetic etched silicon probes (MESP) manufactured by Veeco Instruments. The MFM tips arc coated with a few tens of nanometer thick CoCr magnetic alloy.
Results nnd discussion
We have shown previously that irradiation of ultrathin metal films by ns pulsed lasers either during growth or following film deposition results in the formation of a variety of patterns. Typically irradiation by a uniform laser beam results in self organized dewetting patterns that eventually form a stable state of nanoparticles with spatial short range order. On the other hand irradiation by laser interference patterns eventually results in nanoparticles showing 1D or 2D spatial arrangements. Here we have applied two beam or three beam laser interference irradiation to ultrathin Cu films for the formation of 1D and 2D arrays respectively. The resulting 1D nanocrystal arrays are spatially periodic with spacings determined by the separation of the interference fringes and pattern forming instabilities.
Recently we have shown that for the case of ultrathin metal films pattern formation under two beam laser interference irradiation proceeds by the formation of elongated wire like structures during the early stages (small number of laser pulses) of laser irradiation. The length scale separating the wires is determined by the interference length scale and the dominant mass transport mechanism is thermocapillary flow under the periodic laser intensity distribution. For longer irradiations (large number of laser pulses) the wires break up into droplets by a Rayleigh like instability leading to the formation of lD patterns which show long range order due to fringes and short range order due to the Rayleigh like instability. However in the case of three beam irradiation the formation of the elongated wire like structures is suppressed and thermocapillary flow is presumed to be the primary mechanism of patterning. This results in patterns with 2D order.
For the 1D pattern the particles form along lines separated by about 416 nm consistent with the two beam interference conditions in this experiment. The average particle diameter was ~139 nm. The orientation and spacings of the periodic features were consistent with the geometry of the three beam patterns. The 2D pattern was better developed than the 1D pattern and the particle diameter followed a bimodal distribution (with averages of ~143 and ~65 nm a much smaller volume fraction of smaller particles was also present. The improved spatial order for the three beam over the two beam case is presumed to be due to the intrinsically different mechanisms in play for the two cases. As noted above the two beam patterns form as a result of thermocapillary flow and therom.
Rayleigh like instability. On the other hand the three beam appears to be primarily due to thermocapillary flow as is confirmed by the spatial length scales found in the patterns. However improved processing parameters including better control of the intensity of each incident laser beam optimized fluence irradiation time and initial film thickness are presently under investigation to determine if monomodal particle size distributions and improved patterns can be achieved for the case of Cu. While the periodic spacigs are determined by the interference beam geometry (for the three beam case) the diameter and separation of the particles along the lines appear to be controlled by the film thickness as demonstrated in figure l(c). Therefore the size and areal density of the Cu nanocrystals could also be potentially controlled by experimental variables. In the present experiments patterned areas are restricted to several regions of approximately 100 × 100 m2 or smaller in area. This limitation is primarily due to the spatial inhomogeneity of the laser beam intensity and is not an intrinsic limitation of the processing technique. By using laser sources that can provide higher spatial beam homogeneity along with spatial filtering techniques significant increase in the size of the patterned areas should be possible.
To produce ferromagnetic nanocrystal arrays thin films (20-30 nm thickness) of Fe64.5Cr10Si13.5B9Nb3 alloys were deposited by laser ablation onto as deposited Cu films on 1D and 2D patterned Cu nanocrystal arrays and on the SiO2 substrates. SEM micrographs of the as deposited films on SiO2 substrate or on unpatterned Cu films were devoid of contrast and failed to show any particles under the highest SEM magnification (500 000) although EDS measurements confirmed the presence of all the elements in the ferromagnetic alloy film. Corresponding AFM and MFM images also failed to show any particles. The absence of granular structures is indicative but not a confirmation that the as deposited alloy films are fully amorphous. Future studies are aimed at investigating the microstructure of the as deposited and devitrified alloy films using transmission electron microscopy.
The nanocrystals are most likely FeCrSi alloys as will be discussed in the next paragraph. Whatever the exact composition of the nanocrystals may be the most important point to note is that the nanocrystals have a random spatial distribution and have large particle size distribution (26-130 nm).
In contrast when the Fe64.5Cr10Si13.5B9Nb3 film was deposited on patterned Cu nanocrystal arrays a dramatically different microstructure was observed. A 2D pattern more developed in one direction is visible in all three images produced by the three different imaging techniques. The patterns are similar to those observed in the Cu films after laser irradiation (not shown but similar to that in figure 1(b)) before the deposition of the amorphous Fe64.5Cr10Si13.5B9Nb3 films. Most significantly the MFM image is unequivocal evidence that the nanocrystals are ferromagnetic not an image of the underlying nonmagnetic Cu nanocrystals. This clearly shows that the ferromagnetic nanocrystals have nucleated and grown on top of the Cu nanocrystals by heterogeneous nucleation and growth confirming the basic hypothesis outlined in the introduction. The exact composition of these nanocrystals is not known at the moment. However EDS measurements showed enhanced amounts of Fe Cr (comparable to Cu) and Si in these patterned regions. X ray diffraction studies of the partially devitrified amorphous ribbons of the same composition produced by melt spinning identified the nanocrystals as a bcc phase similar to FeSi with slightly larger lattice parameter. Magnetization measurements of the partially crystallized ribbons for the Fe64.5Cr10Si13.5B9Nb3 alloy showed a ferromagnetic Curie temperature of 470 K compared to 854 K for the corresponding alloy without Cr confirming the presence of significant amounts of Cr in these nanocrystals. All these data suggest that the nanocrystals are most likely FeCrSi alloys. Similar measurements of the partially crystallized thin films are however not possible because of the microscopic (a few micrograms) sample quantities.
When the same experiment was repeated for the Fe74.5Si13.5B9Nb3 alloy films the SEM and AFM images showed similar patterns but the corresponding MFM images were significantly blurred. We believe that since the remaining amorphous matrix after partial crystallization was also ferromagnetic (compared to paramagnetic in the presence of Cr5) the magnetic signal from the corresponding FeSi crystals became significantly distorted by the stray field of the surrounding amorphous matrix. Therefore the present experiment also demonstrates that the matrix surrounding the magnetic nanocrystals can be made ferromagnetic or paramagnetic by a proper choice of composition of the starting alloy film. This is an important advantage because the magnetic interactions among the nanocrystal arrays may be turned on/off or fine tuned for an application specific requirement.
ANALYSIS OF ELECTRON CURRENTS THROUGH NANOJUNCTIONS
As the integration density of semiconductor circuits increases and the size of devices decreases semiconductor technology approaches physical and technical limits. Thus Ihe impetus to develop techniques that can easily work with relatively small feature sizes. Chemists can synthesize unlimited numbers of organic molecules able to perform logical operations. Some molecules and clusters show negative differential resistance which allows us to design two terminal scenarios for electronics. Some molecules may present regions with well defined and different current voltage (I-V) characteristics still other molecules show hysteresis and allow us to design molecular memories. Other molecules show high conductance making molecular wires and others show huge resistance. Manipulation ofsingle molecules allowed the measurement of I-V characteristics of a single molecule. On the other hand methods developed in the last century for the treatment of crystal solids were successful for semiconductor devices using periodical boundary conditions a key point for the success of empirical and semiempirical solid state physics but these methods are not fitted to treat molecular devices as there are no periodical boundary conditions in molecular devices. However quantum chemistry also developed in the last century being able to calculate very precisely finite systems i.e. without periodic boundary conditions despite possible limitations for large systems. Furthermore electron transport systems require special considerations to merge the local chemistry of a molecule and the extended physics of the materials to which it is attached. The nature of the contacts attaching the molecule strongly depends on the scenario to be used for the implementation of molecular circuits.
Despite the great progress achieved challenges are still there for the practical application of molecular electronics. Experi mentally the manipulation of single molecules and the measurement of I-V characteristics are still in the infant stage of the technique. For example the I-V measurements for single molecule are in disagreement from experiment to experiment and from laboratory to laboratory. It is even unknown if we have successfully measured the I-V of any molecules.
From a theoretical point of view although the predicted results for I V gradually converge with experiment disagreement with the experiment is still a major concern and furthermore different calculation methods yield different results thus a comparison between theory and experiment is strongly needed. In the search for justifications of what makes theory and experiment disagree several works have been done. For instance to evaluate the influence of contact structures on the I-V characteristics Hu and Muller investigated molecular junctions of 4 4 bipyridine and 6 alkanedithiol contacted by gold electrodes with hundreds of different conformations and found that the zero voltage conductance depends strongly on the detailed atomic configuration of the electrodes and good agreement with conductance measurements for certain electrode configurations Pati et al have focused on several aspects of electron transport in organic molecules including the analysis of a molecular spin valve.
The conductances of molecular junctions are also very sensitive to the highest occupied molecular orbital (HOMO) and the lowest occupied molecular orbital (LUMO) energies of the molecule and the Fermi level of the electrodes however energy level alignment of the molecular junctions are determined by the intrinsic properties of the molecule and the electrodes and also by the interactions between the molecule and the electrodes which are hard to determine for both theory and experiment. The electronic structures of the conjugated molecules are strongly coupled to the conformation of the molecules which may be caused by the fabrication of the molecular junctions thus the conductance of the conjugated molecules may be very sensitive to the preparation process of the molecular junctions. Experi mentally one of the possible techniques that could conclusively tell us that the molecule contributes to the conduction is the inelastic electron tunneling spectroscopy (IETS) method since its spectrum as other spectroscopic techniques allows us to verify features such as bond stretching or angle bending. The theoretical methods employed in the calculation of I V characteristics are mainly based on quantum mechanics and combinations of classic and quantum transport theory. The quantum mechanics methods include empirical methods such as tight binding theory and first principles methods such as the Hartree Fock and the density functional theory methods.
In this work we calculate the current voltage characteristics of benzene naphthalene and anthracene addressed by a few gold nanoelectrode atoms representing different attachment geometries.
Methodology
Our procedure is based on a combination of two ab initio very precise density functional theory calculations one for an extended system using periodic boundary conditions (PBC) with the program Crystal 06 and another using a finite system a molecule or complex using the program Gaussian 03. The results from the PBC program are a set of partial density of states (DOS) for every atomic character participating in the contacts and the outputs from the molecular or complex calculation are a Hamiltonian and overlap matrices of the molecule attached to few atoms of the contact describing possible nanocontact structures. Thus the resulting interface is a matching of densities from the molecule or complex and the bulk nanocontacts. The Hamiltonian and overlap matrices are obtained self consistently for every voltage applied to the molecule. The partial DOS and the sets for each bias voltage of the Hamiltonian and overlap matrices are entered in our in situ developed program GENIP to calculate the electron transport characteristics using a Green s function approach that considers the local nature of the molecule as well as the nonlocal features of the contacts. For details of our procedure the readers are referred to.
Electronic structure calculations are carried out at the B3PW91/LANL2DZ level of theory. This in plies using the Becke s three parameter exchange functional and the Perdew Wang correlation functional. The quasi relativistic pseudopotential and basis set LANL2DZ is used for Au and the 6 31G basis set is used for C H and S. The molecules are optimized to a local minimum which is verified by a second derivative calculation whereby the Hessian matrix has non negative eigenvalues. Recent details of the formalism in the program GENIP are described elsewhere and details of its development can be found in.
Results and discussion
Description of the studied molecular junctions and nomenclatures
A molecular junction is usually composed of a molecule and two nanoelectrodes fabricated by various experimental techniques. In our nomenclature we represent the molecular junction by [AubulkAux]-[S-M-S]-[AuxAubulk] M is the molecule S is the sulfur atom [AubulkAux] and [AuxAubulk] represent the nanoelectrodes with Aux representing x gold atoms in direct contact with the molecule and Aubulk represents the bulk nanoelectrode.
In a self assembly process it is believed that the S H cleavage occurs at the on top site however a molecule in an on top conformation may migrate to a hollow site. On the other hand from theory the stable conformation is not exactly at the hollow site molecules are arranged in intermediate conformations between the hollow and bridge sites. Therefore all three local conformations may exist in the assembly of dithiol molecules to gold nanoelectrodes.
I-V Characteristics of benzene molecular junctions
Figure 2 shows the I-V characteristics of benzenedithiol assembled between two gold nanoelectrodes with one two or three gold atoms in direct contact with the molecule. The I-V characteristics have quite different features the one gold has the highest conductance while the three gold has the lowest conductance. Interestingly the two gold system matches the conductance of the one gold for low voltages (0.0-0.4 V) and of the three gold for high voltages (1.5-2.5 V). These small rearrangements of the gold atoms may yield effects such as negative differential resistance hysteresis and switching all of them important for the fields of electronics and sensors. Thus conductance decreases as the number of contact atoms increases. This seems to contradict common sense since the cross sectional area has a direct relation with the number of contact atoms and the larger the cross sectional area the larger the conductance. However at the atomistic scale this phe nomenon is due to orbital symmetry. For an on top configuration a bond (with rotational symmetry along the bond) is formed from the 3p orbital of S and the 6s orbital of Au. The bridge and hollow site conformations are slightly more stable than the on top site conformation but they are not symmetrically favored for electron tunneling because of the poor overlap with the 3p orbitals of S.
The one and two gold. structures show ohmic I V characteristics around zero voltages while the three gold shows a Schottky barrier of 0.5V. Comparing this to the first experimental I-V curve on a single molecule the shapes are similar when showing barriers but the theoretical conductance is almost 122 times larger than the experimental one. Calculations at several levels of theory also show large disagreements of for instance 110 fold and 20 or 400 fold according to the conformation. The experimental measurement for this molecule yielded ~0.1 A at 1 V bias. Several other calculations have been reported predicting a current of about 0.1 A at 1 V bias however several parameters were free to choose such as Fermi levels arbitrary geometries (no optimized no local minima verification) voltage distribution and HOMO LUMO setting with respect to Fermi metal level. On the other hand considering the discrepancies between experiments and the difficulty to characterize them it would not be very useful to perform fittings as the reasons for further agreements might be and are highly likely to be fortuitous. In addition the causes of disagreements with the theoretical methods could amount to disagreements of several orders of magnitude due to the use of a vast variety of precise ab initio methods semiempirical or empirical approaches. All these problems really call for a well defined methodology that considers the chemistry (bonding conformation electronic structure vibrational stability local effects) as well as the physics (extended nature Fermi levels nanosized contacts pinning of energy levels) of the junctions. On the practical or technological arena studies should be focused toward the practical utilization of the molecular devices as full systems that can be unambiguously tested and not just limited to reproduce experimental settings designed for single devices.
On the other hand an explanation of why the hollow site conformation has a barrier and the other two cases have no barrier could be explained by the surface states. The surface states are at about 0.5 eV below the Fermi level for the Au(111) surface. The surface states of the gold represent a barrier for electrons leaving the gold nanoelectrode however the S atom pulls electrons from the surface states and lowers the barrier height. It is shown that the charge on the on top S is about -0.7e compared to -0.4e for the hollow site thus the on top S pulls -0.3e charge from the surface states which is enough to eliminate the surface barrier for the on top site.
The center of the HOMO LUMO gap of the extended molecules shows interesting variation as the number of contact atoms increases (figure 3 center). The benzene molecular junctions with one or three direct contact gold atoms have similar Fermi levels at -4.7 and -5.0 eV respectively while the one with two gold atoms in direct contact has low Fermi level at -6.0 eV. This could be explained by the fact that for the molecular junction with two direct contact gold atoms the one of the Au 6s electron on each nanoelectrode is not paired since each S can only bond with one of the Au atoms. Therefore there is one unpaired Au 6s electron on each side of the molecular junction. The repulsion between these two unpaired electrons is quite low due to the physical separation and thus the Fermi level is lower than the other two cases. The HOMO LUMO gap also shows interesting variation as the number of direct contact atoms increases. For the molecular junction with only one direct contact gold atom the HOMO-LUMO gap is 2.8 eV while the HOMO LUMO gap of the one with two direct contact gold atoms is only 0.9 eV and that of the one with three direct contact gold atoms is 1.1 eV. Although the on top conformation is not as stable as the bridge site or hollow site conformations the HOMO-LUMO of the on top conformation is an S Au bond (antibond) compared to the weak Au-Au bonds of the bridge site or hollow site conformation thus the HOMO LUMO gap of the on top con formation is much wider than that of the bridge site or hollow site conformation.
The DOS of the molecular junctions shows systematic have changes as the number of direct contact atoms increases in the nanoelectrode. Figure 3 shows a small peak in the DOS for the molecular junction with three direct contact atoms (green curve) at -5.3 eV and this peak shifts to -6.0 eV for the one with gold atoms (red curve) in direct contact and to -6.8 eV for the one with two only one direct contact atom (blue curve). The other peaks are much farther from the Fermi level and are not as important as these peaks for the electron tunneling. The features of the transmission functions show significant variations. For the molecular junction with three gold atoms in direct contact a big peak appear at almost the same position as its DOS. However this peak is not coincident in energy to its Fermi level. For the molecular junctions with two gold atoms in direct contact a big peak also appears at the same position as its corresponding DOS. Since this peak location coincides almost exactly with its Fermi level the contribution to the electron tunneling probability is larger than the contribution from the molecular junctions with three gold atoms in direct contact. What is very different is the transmission of the molecular junction with one gold atom in direct contact its peak becomes a plateau perhaps due to the short distance between the electrodes so tunneling is highly probable even when no resonances are close by. The plateau shaped peaks of the transmission function have been observed for benzene especially with Ni Pd or Pt nanoelectrodes where the partially filled d orbital causes a strong interaction between the sulfur and the metal. From the position and height of the transmission peaks we can infer that the molecular junction with one gold atom in direct contact has the highest conductance and the one with three gold atoms has the lowest conductance since the transmission function around the Fermi level of the molecular junction with one gold atom is the largest and the one with three gold atoms is the smallest.
MULTI WALLED CARBON NANOTUBES DECORATED WITH TITANIUM
Since the discovery of carbon nanotubes (CNTs) in 1991 by Iijima CNTs have attracted a great deal of interest as light and relatively cheap materials for hydrogen storage. The spread of non gasoline fueled cars in the coming years depends mostly on the feasibility of carrying a suitable amount of H2 (5-10 kg of H2 depending on the vehicle to provide a 480 km range for a fuel cell/electric vehicle) in a reversible tank having volume and weight comparable to the tank of a gasoline fueled car. The initial claims of high absorption capacity by CNTs at high pressure and often at low temperatures proved irreproducible or unsuitable for practical applications. Several authors reported hydrogen storage data obtained by both theoretical and experimental studies that are close to the envisaged value of 6.5 wt% as required by automotive applications. However recent studies have shown that the hydrogen storage capacity on pristine CNTs is less than 0.01 wt% at room temperature. Therefore many efforts have been made to modify the electronic properties by doping to create defects and to change the chemical composition of CNTs. All this in order to enhance the absorption through chemisorption instead of physisorption. Theoretical calculations mainly based on DFT methods show that Ti coated CNTs exhibit remarkable absorption properties. More recently it has been found from these calculations that a single walled (SW) CNT decorated with Ti can absorb hydrogen up to 8 wt%. Only one paper is found in the literature concerning the preparation of Ti coated SWCNTs obtained by EB PVD (electron beam physical vapour deposition). A chemical way to obtain Ti attached to CNTs has not yet been reported in the literature.
The scope of the present work is the preparation of titanium decorated multi walled (MW) CNTs by using a synthetic chemical route and their characterization in order to determine the chemical nature of Ti as well as its spatial localization in the MWCNTs. This information has been obtained using a combined study with several techniques transmission electron microscopy (TEM) selected area electron diffraction (SAED) high resolution transmission electron microscopy (HREM) and electron energy loss spectroscopy (EELS). This is a very suitable combination of techniques because TEM and HREM provide information on the localization of Ti whereas SAED and EELS provide information on the chemical nature of Ti. Raman spectroscopy has also been utilized to show the differences induced by the functionalization of MWCNTs by Ti.
Experimental details
Titanium decorated MWCNTs were prepared by using the approach developed by Braga for the synthesis of transition metal intercalated graphite. MWCNTs were synthesized by catalytic decomposition of propane over a patented catalyst which through a peculiar growth mechanism can function at temperatures below 500 °C and without the formation of graphite and/or amorphous carbon. They were purified by removing the catalyst by two acid treatments always under magnetic stirring initially with hot concentrated HCl and then after filtering on a Gooch funnel with a 1 1 mixture of hot water and concentrated HCI. The suspension was again filtered on a Gooch funnel washed with deionized water until reaching pH ~ 7 and then with acetone. The solid was left to dry for some minutes under suction and then it was placed in an oven and held at 150°C to remove the adsorbed water. The SEM and TEM micrographs do not show other forms of carbon than CNTs. The distribution of outer and inner diameter and number of MWCNT walls was found almost uniform as expected from their growth mechanism. The average values of the inner and outer diameters and the length were found to be 4 nm 130 nm and 10 m respectively. Therefore the number of walls has to be considered very high being that the spacing between the walls is practically equal to the spacing between the graphene planes of graphite. This is generally recognized for MWCNTs of large outer diameter with open ends. The carbon content evaluated by thermogravimetric (TG) analysis (Netzsch STA 409 PC Luxx resolution 2 g) in pure oxygen is greater than 98.68 wt% as found by the TG weight loss. The residue is due to some traces of catalyst that are still present in the MWCNTs. Considering the catalyst weight gain during the oxidation its content is 1.06 wt%.
0.964 g of MWCNTs and a Tehon coated magnetic stirrer were placed in a 100 ml three necked Pyrex flask equipped with argon inlet/argon purity 99.9995% H2O < 2 ppm O2 < 1 ppm) a condenser and a pressure equilibrated dropping funnel. All the glassware was carefully flame heated before the experiment in order to remove the moisture. An amount of 0.495 g of freshly cut potassium metal (Aldrich 99.5%) weighed and stored under n heptane (distilled over CaH2) were added to the MWCNTs under flowing Ar. After the addition the flask was heated up to 200°C by a temperature controlled silicone oil bath and the MWCNTK mixture was kept under stirring. The reaction was quite difficult to start probably because of an oxide layer over the potassium pieces but after breaking the K pieces with a glass rod the reaction proceeded quickly and a red powder was formed. Then the product was allowed to cool to room temperature. Once cooled down the product was reacted with a solution of 1.164 g of Ti(IV) isopropoxide (Aldrich 99.999%) in 20 ml of THF (previously distilled over sodium benzophenone ketyl) added drop wise. During the addition of the solution the red powder turned black. After stirring for 2 h methanol was added to the suspension to destroy eventual K residues then the suspension was filtered with a Gooch funnel washed with 300 ml of methanol and dried in vacuum.
To determine accurately the Ti content in the TiMWCNTs inductively coupled plasma optical emission spectroscopy ICP OE (Varian Vista MPX) was utilized after dissolving a weighed amount in a hot 1 1 mixture of 65% HNO3-96% H2SO4. The Ti content was found to be 8.7 wt% (2.33 at. %).
The samples were characterized by powder XRD (Panalytical X Pert Pro Cu K radiation = 0.1542 nm) and Raman spec troscopy (Renishaw in Via Raman Microscope = 514.5 nm Ar+ UK) before and after the Ti decoration.
Bright field (BF) and dark field (DF) images and SAED patterns were obtained using a 200 kV JEOL 200CX trans mission electron microscope with a medium resolution pole piece. HREM images and EELS measurements were obtained using a 200 kV Philips FEI Tecnai F20 transmission electron microscope equipped with a field emission gun (energy resolution 0.7 eV) a spherical aberration corrector (point resolution 0.13 nm) and a Gatan spectrometer. EELS spectra were obtained in the range from 370 to 590 eV with a dispersion of 0.1 e V/channel and an energy resolution of 1 eV (the observed width of the zero loss peak). This range includes the following core edges C K edge at 284 eV N K edge at 410 eV Ti L3 2 edge at 455 eV and 0 K edge at 543 eV.
Samples for electron microscopy based measurements were prepared by dispersing powder in ethanol placing in an ultrasonic bath then putting droplets onto 3 mm copper grids coated with amorphous carbon film. Prior to measurements using the Tecnai microscope the sample was cleaned in a plasma cleaner for 10 s.
Electron energy loss spectroscopy
Spectra in figures 4 and 5 show respectively the EELS spectra of the Ti L3 2 edge and the O K edge collected over a wide area of the sample. The EELS spectra of the C K edge are not shown because the overwhelming majority of the sample is amorphous carbon (on the microscope grid) and MWCNTs and this will obscure the contribution from any additional C containing phase if there is one (i.e. TiC). There was no detectable N K edge which proves that no N containing phase (i.e. TiN) is present.
The Ti L3 2 edge EELS spectra from the sample shows only two peaks corresponding to the 2s 3p3/2(L3) and 2s 3p1/2 (L2) core electron excitations respectively at 455 and 461 eV. In oxides each of these peaks is split due to the crystal field effects each of these peaks is split due to the crystal field effects while no splitting is observed in Ti metal. The Ti L3 2 edge spectra can be analysed using the fingerprint method i.e. by comparing them with the spectra from known reference compounds. Figures 5 (b) and (c) show respectively the Ti L3 2 edge spectra from Ti metal and TiO2 anatase (the rutile spectrum is very similar to anatase). The comparison with the spectrum of the sample in figure 5 (a) indicates that the sample contains Ti metal.
However the presence of some oxide is indicated in the O K edge EELS spectra from the sample reported in Figure 6. Due to the high reactivity of the Ti nanoparticles they could be coated with an oxide layer. The O K edge spectra can also be analysed using the fingerprint method and figure 5 (b) shows the O K edge spectrum of TiO2 anatase (the rutile spectrum is very similar to anatase). The comparison with the spectrum of the sample in figure 6 (a) indicates that the oxide phase is not rutile or anatase but it remains that the Ti nanoparticles are coated with an amorphous oxide layer.
SYNTHESIS AND SELF ORGANIZATION OF AU NANOPARTICLES
Intense interest has over the last decade been focused on metal nanoparticles due to their potential for advanced applications in a variety of fields such as catalysis sensors diagnostics imaging therapeutics etc. The assembly of nanoparticles into well defined two dimensional (2D) and three dimensional (3D) superlattices has attracted considerable .attention. Especially when using metal nanoparticles for applications in optoelectronic devices it is essential that the fabrication of ordered assemblies consists of rapid and inexpensive processes. Several approaches based on Langmuir Blodgett (LB) techniques the electrophoretic deposition method and the use of block copolymers have so far been employed in order to obtain 2D and 3D structures.
For nanoparticle self assembly a very narrow size distribution is essential as well as an inherent van der Waals attraction between the particles. Polydispersion in the particle size prevents the formation of long ranged 2D and 3D structures. Moreover appropriate surface functionalization is required. So far most of the published studies on nanoparticle self assembly by solvent evaporation use mainly organosols and very rarely hydrosols.
Here we describe a straightforward chemical route for the large scale synthesis of Au nanoparticles virtually monodispersed which can form 2D superlattices by solvent evaporation methods. The as received particles possess a very narrow size distribution (
Typically in the experimental procedure 0.5 mmol of AuCl3 are dissolved in 20 ml of oleyl amine in the presence of 0.2 mmol of TOPO. The mixture is heated for ~10 min to 80°C whereupon a clear solution is formed. Then the temperature is raised to 200°C for approximately 30 min in order for the nucleation and growth of the nanoparticles to commence. The reaction mixture is allowed to cool down to room temperature and the capped nanoparticles formed are precipitated by ethanol collected by centrifugation washed with ethanol several times in order to remove any excess of unbound organic molecules and air dried by spreading on a glass plate. Finally the nanoparticles are dissolved in chloroform or a mixture of chloroform/octane solvents in order to create a stable colloidal suspension for the formation of 2D arrays upon solid substrates by solvent evaporation techniques.
The organophilic Au nanoparticles can easily be converted to a hydrophilic derivative by a relatively simple procedure a dispersion of 15 mg Au nanoparticles in 3 ml of CHCl3 is mixed with 10 ml of a 5% w/v. aqueous solution of cetyltrimethyl ammonium bromide (CTAB). The mixture is sonicated for ~10 min until a homogeneous emulsion is obtained. The emulsion is subsequently heated to ~50°C under constant magnetic stirring so as to slowly remove CHCl3 and then stirred overnight at room temperature. The nanoparticles are isolated by centrifugation at 12000 rpm for 30 min and redispersed in 5 ml of water. The hydrophilic particles coated with CTAB reveal positive surface charge. On the other hand by using sodium dodecyl sulfate (SDS) instead of CTAB the organophilic particles can be converted into hydrophilic ones bearing a negative surface charge. The slow mixing of a dispersion of negatively charged nanoparticles with a dispersion of positively charged ones under vigorous stirring leads to the formation of ID (worm like) and 3D nanoparticle structures in solution.
Zeta potential measurements were carried out in a Zeta Plus analyser (Brookhaven Instruments Corporation) equipped with a 35 mW solid state laser emitting at 660 nm with the dynamic light scattering method. A solution of nanoparticles with concentration 0.1 mg ml-1 was introduced into the instrument cell. Ten zeta potential measurements were collected for each experiment and the results were averaged. The statistical error was less than 1 %.
Results and discussion
A typical powder XRD profile of the Au nanocrystalline particles is given in figure l(a). It complies with fcc Au in both diffraction intensity and position. In addition the UV vis spectra presented in figure l(b) reveal a maximum at 523 nm which is characteristic of the gold surface plasmon band.
They demonstrate that a hexagonal close packed Au nanoparticle array is obtained when a drop of Au organosols is allowed to evaporate on a carbon coated copper grid under ambient conditions. From the TEM images it is evident that the particles have spherical shape and a very narrow size distribution with an average diameter of 10 nm. Specifically the evaporation of a chloroform solution with 10 mg ml-1 Au concentration leads to the deposition of a single Au nanoparticle layer (figure 2).
By changing the solvent medium in order to control the speed of the evaporation (use of mixtures of chloroform with n octane at various volume ratios) as well as the particle concentration various 3D assemblies of Au nanoparticles can be achieved on solid substrates. In the case of an n octane/chloroform solvent with volume ratio 0.3/0.8 and Au concentration about 20 mg ml-1 a 3D multilayer organization is achieved. Circular voids are formed by the slow evaporation of the 1/ln octane/chloroform solvent with 10 mg ml-1 particle concentration. From the above mentioned results it becomes evident that the alkylamine TOPO capped Au nanoparticles can straightforwardly form 2D and 3D selfassembled structures in a very fast and inexpensive manner.
Zeta potential measurements confirm that the water dispersible Au nanoparticles after treatment with CTAB have a positive surface charge (+45 mV) while the particles treated with SDS have a negative surface charge (-40 mV). In either case high colloidal stability is obtained. In the case of water soluble Au nanoparticles sponge like spherical structures form by the interaction of the positively charged particles with the negative ones. An essential condition for the formation of these structures is the complete charge neutralization which is based on the mixing ratio (1/1). In the case of an incomplete titration (1/0.5) 1D arrays are formed.
WATER SOLUBLE CARBON NANOTUBES
Carbon nanotubes (CNTs) have been considered as one of the most important building blocks for a rich variety of applications such as fabrication of nanodevices gene delivery vectors intracellular transporters and mass conveyors etc. However the lack of solubility as well as the intrinsic physical and chemical inertness blocks their applications in a wider range. The solubility and the functionalization on the sidewalls are essential for CNTs if they are to have potential applications in the nano biological field. Recently much attention has been paid to the fabrication of nanohybrids of Au nanoparticles and CNTs (Au@CNTs) in order toenhance the performance of CNTs as building blocks and to explore the exceptional properties of hybrids based on CNTs for potential applications in catalysts biosensors etc.
A variety of physical and chemical methods have been explored to produce hybrids of Au nanoparticles and CNTs. The main problems of the Au@CNTs hybrids reported in the literature are (1) the uniform distribution of Au nanoparticles along the surface of CNTs is still not so easy to be controlled (2) Au nanoparticles are prone to aggregation when their diameters are decreased to a limited range or when they suffer from a high temperature. To solve these problems several methods have been developed to produce nanohybrids of Au nanoparticles and CNTs by employing organics or polymers as interlinkers based on as prepared Au nanoparticles and by solution phase dispersion techniques using surfactants. Nevertheless good tuning of the size of Au nanoparticles to a smaller range as well as the uniform distribution of Au particles along CNTs still remains a great challenge. Besides the reported work on Au@CNTs hybrids has paid much attention to their fabrication but very few researchers have touched their properties and applications.
In this paper the nanohybrids of Au nanoparticles and multi walled carbon nanotubes (MWNTs) (Au@MWNTs) were obtained by reduction of gold ions where an organic optoelectronic active compound of N N bi (2 mercaptoethyl) perylene 3 4 9 10 tetracarboxylic diimide (MEPTCDI) was used as an interlinker between Au nanoparticles and MWNTs. The formation mechanism and optical properties of the Au@MWNT hybrids were investigated. One advantage of the method that is different from others is that it is an in situ fabrication in solution under very mild conditions so it is not necessary to prepare Au nanoparticles in advance. Meanwhile this method enables us to directly study the formation mechanism of the hybrids based on MWNTs and Au particles as well as to vividly investigate the properties of the hybrids. Moreover the introduced organic compound MEPTCDI served as both an interlinker and a stabilizer it is optoelectronic active and is expected to implant novel properties of the CNTs for a wider application.
Experimental section
Preparation of Au@MWNT hybrids
Materials. MWNTs were prepared by catalytic decomposition of acetylene and purified according to the published procedures. HAuCl4 was purchased from J & K Chemical. The other chemicals and reagents were commercially available.
MEPTCDI was synthesized and characterized in our laboratory. Perylene 3 4 9 10 tetracarboxylic dianhydride (PTCDA 490 mg. 1.25 mmol commercially available and recrystallized from acetone solution) and dehydrated Zn(Ac)2 (100 mg 0.55 mmol) were added into 20 ml of as purified quinoline solvent. The above solution was then heated to 120°C and stirred for 3 h under the nitrogen atmosphere. After cooling the solution was transferred to 500 ml of absolute ethanol followed by filtration. The filtrate was washed with hot 5% sodium hydroxide aqueous solution several times until the filtrate was colorless and then washed with deionized water until the pH value of the filtrate was neutral. The collected rufous solid was dried in vacuum at 70°C overnight and finally was sublimed under vacuum.
In a typical experiment 0.5 mg of MWNTs were added into 2 ml of deionized water consisting of 0.05 wt% sodium dedecyl benzene sulfonate (SDBS) and sonicated for 1 h then centrifuged at 5000 rpm for 10 min to collect a well dispersed suspension. Next 1 ml of 0.1 mM MEPTCDI solution in DMF was added into the above MWNT suspension and stirred for 30 min under the protection of nitrogen. Finally 1 ml of 1 mM HAuCl4 aqueous solution was dropped into the above mixture and stirred for another 12 h at room temperature giving a water soluble and well organized nanohybrid of Au nanoparticles and MWNTs (Au@MWNTs). The nanohybrid was further purified by centrifugation at 10000 rpm for 1 h and redispersed in water for further measurements.
Synthesis of Au@MWNTs with SDBS. 0.5 mg of purified MWNTs and 0.1 wt% SDBS were added into 2 ml of deionized water and sonicated for 1 h then centrifuged at 5000 rpm for 10 min to collect the well dispersed suspension. 1 ml of 1 mM HAuCl4 was added to the above suspension and stirred for another 12 h at room temperature. The solution was centrifuged at 10000 rpm for 1 h to obtain Au@MWNTs with SDBS.
Synthesis of Au@MWNTs without organics. 0.5 mg of purified MWNTs were added into 1 ml of deionized water and sonicated for 1 h. Then 1 ml of 1 mM HAuCl4 was added to the suspension and stirred for another 12 h at room temperature. The solution was centrifuged at 10000 rpm for 1 h to obtain Au@MWNTs without organics.
Synthesis of MEPTCDI@MWNT hybrid. 0.5 mg of purified MWNTs was dissolved in water consisting of 0.1% SDBS and sonicated for 1 h. The mixture was then centrifuged at 5000 rpm for 10 min to collect a well dispersed suspension. Finally 1 ml of 0.1 mM MEPTCDI solution in DMF was added into the above MWNTs suspension and stirred for 12 h at room temperature to obtain the MEPTCDI@MWNT hybrid which was used directly for UV vis and photoluminescence characterization without further purification.
Characterization
Transmission electron microscopy (TEM) images and high resolution TEM (HRTEM) images were taken on transmission electron microscopes (JEM 200CX and JEM 1200EX) and a JEOL JEM 1230 high resolution transmission electron microscope respectively. Samples were dissolved in water and deposited onto holey carbon coated copper grids for TEM or HRTEM measurement. Field emission scanning transmission electron microscopy (FE SEM) images and the energy dispersive x ray (EDX) spectrum (25 kV) were obtained on a scanning electron microscope (SLR10N). The Au@MWNT samples were deposited on the silicon surface adhered by a layer of conducting paper in advance for FE SEM and EDX measurement. UV-vis and photoluminescence spectra were recorded on a Gary 100 Bio UV-vis spectrophotometer and a LS55 Luminescence spectrometer respectively. The atomic absorptions were conducted under an atom absorption spectrometer (180 50 Hitachi). All samples were dispersed in water for UV vis photoluminescence and atomic absorption measurements. For photoluminescence measurements all samples were recorded under the same conditions with a voltage of 775 mV and a slit of 5 nm. 1H NMR spectrum and FTIR for MEPTCDI were taken on an Avance 500 MHz spectrometer and a Vector 22 FTIR spectrometer respectively.
Results and discussion
Fabrication of Au@MWNT hybrids
Purified MWNTs were firstly dispersed by a small amount of SDBS in water and then mixed with MEPTCDI and with HAuCl4 aqueous solution successively which were incubated for hours to fabricate Au@MWNT hybrids. The formation of the Au@MWNT nanohybrid is clearly observed from TEM images in figure 1. Well distributed and isolated black nanoparticles were found to densely ornament the walls of MWNTs uniformly (see figures 1(A) and (B). The energy dispersive x ray (EDX) spectrum further demonstrated the black dots as Au particles (see figure 2). The TEM images revealed that the average diameter of Au nanoparticles was 2.73 + 1.8 nm which was determined by statistical calculation of 140 nanoparticles (figure 1 (E) and could be tuned by the concentration of MEPTCDI and HAuCl4 aqueous solutions. Furthermore the Au@MWNT hybrids were stable in water and have not suffered from precipitation and aggregation after a year.
In comparison if MEPTCDI was replaced by a surfactant of SDBS randomly distributed Au particles with a larger size and a much wider size distribution would have been coated on MWNTs. The Au@MWNT hybrid with SDBS was also quite stable in water due to the existence of a surfactant of SDBS. If neither MEPTCDI nor SDBS was added acid treated MWNTs were randomly decorated with Au particles with uneven sizes particularly in the ends of MWNTs. However the Au@MWNTs without organics was not stable in water which could be precipitated after the reaction. Additionally no nanoparticles were observed in a controlled experiment without adding HAuCl4 proving further that the formed particles in the mentioned hybrids were from Au instead of impurity.
Formation mechanism of Au@MWN hybrids
Since Au nanoparticles were formed from the acid treated MWNTs and HAuCl4 water solution it was proposed that the formation of Au particles resulted from the oxidization of MWNTs and the reduction of Au ions. The fact that CNTs could form redox pairs with Au3+ and could act as electron donors to metal ions or a reducing agent was demonstrated by Dai.
The affinity of CNTs with Au particles is thought to be a key factor in Au@MWNT hybrid fabrication. Chemical groups like carboxyl or hydroxyl grafted on the ends or the sidewalls of CNTs during the acid treatment could provide such a kind of affinity between CNTs and Au particles where Au nanoparticles could be selectively attached while other sites of CNTs are free of nanoparticle attachment due to the absence of chemical groups or organics as linkers and the intrinsic inert surfaces of CNTs. We indeed did not observe the Au@MWNT hybrid formation from the as prepared MWNTs without any chemical treatment and HAuCl4 water solution. The reason why more Au particles were coated on MWNTs under the existence of the surfactant SDBS was that SDBS could wrap the surfaces of MWNTs by van de Waals interaction and also act as weak linking molecules between MWNTs and Au nanoparticles through the static interaction. Employing surfactants to assist the formation of hybrids based on CNTs and metal particles has also been studied by other groups.
Various studies confirmed that organics or polymers with phenyl groups could interact with CNTs through stacking which is a stronger interaction than the van de Waals force. The organic compound MEPTCDI used in this work has two functions. One is as an interlinker. As seen from scheme 1 MEPTCDI could non covalently wrap MWNTs through stacking which could be demonstrated by the TEM images (figures 1 (A) and (B) where a layer of molecules with lower contrast could be observed. Meanwhile Au nanoparticles could be attached to the wall surfaces of MWNTs via the linking of a mercapto group of MEPTCDI. The other function of MEPTCDI is as a stabilizer. As shown in scheme 1 in addition to those MEPTCDI coated on MWNTs MEPTCDI non interacted with MWNTs (defined as the dissociated MEPTCDI here) may also bond to the surface of Au particles via a mercapto group which makes it possible to further stabilize Au nanoparticles and lower their surface energy. The distance between the active part of MEPTCDI (namely the mercapto group) and the rigid body part of MEPTCDI (namely the perylene group) allowed Au nanoparticles to keep each other isolated with a molecular space resulting in a uniform distribution and non aggregation even at a high temperature. The existence of the dissociated MEPTCDI in the Au@MWNT solution is proposed to play an important role in enabling Au atoms to nucleate and grow along the wall surfaces of MWNTs in the concept of the size distribution focusing mechanism on which various narrow sized and monodisperse inorganic nanocrystals have been synthesized. The function of MEPTCDI was demonstrated by the control experiment where MEPTCDI was added to the as prepared Au@MWNT hybrid without organics and allowed to stir for hours. As expected no well distributed Au nanoparticles along the surfaces of MWNTs were observed.
With the help of MEPTCDI an Au@MWNT hybri with more uniformity a higher loading of Au nanoparticIes and lower dimensions of nanoparticIes mostly reaching below 5 nm could be obtained in this way.
We measured the atomic absorption of Au3+ for three categories of as prepared hybrid samples (Au@MWNTs with MEPTCDI Au@MWNTs with SDBS and Au@MWNTs without organics) before and after purification to test the reduction degrees of Au3+ by MWNTs. The results revealed that all samples preserved unreacted Au3+. For Au@MWNTs with MEPTCDI about 50% of Au3+ has been reduced while the other part of Au3+ has been mostly adsorbed by MWNTs. The atom absorption results demonstrate that there are only (or even less than) 5% of free Au3+ existing in aqueous solutions containing Au@MWNT hybrids and the contents of free Au3+ are almost the same in purified and unpurified samples.
These results can be explained by the affinity of MEPTCDI with Au cations. Before the nucleation and growth of Au particles along the wall surfaces of MWNTs most of Au3+ could be adsorbed or surrounded by MEPTCDI since the strong nucleophilic mercapto group of MEPCDI could coordinate with AuCl4- to form a complex. Note that MEPTCDI was attached to the wall surfaces of MWNTs in advance so the Au cations were adsorbed along the surfaces of MWNTs. Even though those Au3+ surrounded by MEPTCDI were freely dispersed in aqueous reaction solution at first it was highly possible that they would attach to the wall surfaces of MWNTs as the reaction was going on under vigorous stirring. Consequently a great number of Au cations surrounded by MEPTCDI were transferred to the wall surface of MWNTs. Meanwhile the MWNTs for this study are structural impurities in other words they contain nanotubes with various diameters and different helixes which make MWNTs with various work functions. For those nanotubes with suitable work functions they can form Au3+/MWNTs redox pairs and result in spontaneous redox reactions to form Au@MWNT hybrids. Others that could not form redox pairs would not result in spontaneous redox reactions to form Au@MWNT hybrids leading to the large remains of free Au cations along the MWNTs. In addition if sidewalls of MWNTs were thickly wrapped by MEPTCDI it is possible that the redox reaction between Au3+ and MWNTs would not happen because they might not be able to form effective redox pairs. Wealso found that the remains of free Au cations could be decreased with increasing the reaction time the timedependent character was insignificant when incubated for more than 48h.
Both Au@MWNTs with SDBS and Au@MWNTs without organics contain about 30%-40% of free Au3+ before purification which decreases sharply to about 10%-15% after purification. This might be attributable to the weak affinity of SDBS with Au3+ or organic groups like carboxyl or hydroxyl with Au3+ as well as the weak affinity of SDBS with MWNTs and the scarcity of organic groups along the sidewalls of MWNTs. It is obvious that the TEM results (figures 1 (C) and (D)) could prop up this argument with less loading of Au nanoparticles on to these two hybrids.
This site specific characteristic adsorption of Au cations driven by organic linkers like MEPTCDI etc before reduction gives us the opportunity to control the nucleation and growth of nanoparticles along the sidewalls of carbon nanotubes and then to control the diameters and size distribution of Au particles. This explanation is consistent with the TEM results (figure 1). It further proves our assumptions that MEPTCDI plays a dual role (as a molecular interlinker and a stabilizer) during the formation of Au@MWNT hybrids based on which the nucleation and growth of nanoparticles along sidewalls of carbon nanotubes could be controlled.
NANOSPHERES FOR PHOTOLUMINESENCE
Semiconductor nanocrystals have attracted widespread attention owing to their distinct optical and electronic properties that can be tuned by the quantum size effect and surface chemistry. For instance significant changes in band gap energies and physical properties can be expected when the radius of the particles approaches the Bohr radius of the exciton. On the other hand particle size reduction generally gives rise to large surface areas that are beneficial to high activity. However it is not propitious to photoluminescence properties since a large amount of surface defects including dangling bonds are susceptible to oxidation coalescence are prone to yield other property instability and emission quenching problems from the nonradiative recombination sites of electron hole pairs. As a result the photo emission intensities of most nanocrystals are always weak putting severe limits on the applications of nanocrystals.
Improving the photoluminescence performance of nanomaterials is fundamentally important for many applications including flat panel displays photoluminescent/electroluminescent devices infrared windows and solar cells. ZnS is a typical wide gap II VI semiconductor with an energy gap of 3.7 eV. Many attempts have been made to modulate the optical properties of ZnS nanocrystals via the fabrication or encapsulated structures including ZnO/ZnS Mn esters/ZnS and ZnS/SiO2. Among all these coatings SiO2 is regarded as the best one and it shows many merits (i) SiO2 can act as the sheaths to prevent oxidation of the semiconductors or to avoid interferences in the building blocks of complex nanoscale circuits (ii) SiO2 coated on the surfaces of semiconductors could modulate the physical and chemical properties for improved performance of devices and (iii) the SiO2 network is a good binder that could be easily coated onto glass substrates for device fabrication. Up to now almost all ZnS/SiO2 nanostructures reported in the literature were prepared using sol gel methods. The coatings of SiO2 achieved by these methods are highly porous and unfavorable for improving the luminescence efficiency of ZnS nanocrystals because of the impurity introduction and significant light scattering. Consequently novel preparation methodologies are urgently required to achieve a low porosity SiO2 coating on ZnS nanocrystals which is expected to be fundamental for finding new strong luminescent materials.
In this work we explored a novel one step solvothermal method to produce low porosity ZnS/SiO2 encapsulated nanostructures. By carefully characterizing the structure morphology and optical properties or ZnS/SiO2 nanostructures we found an abnormal band gap shift of ZnS/SiO2 nanostructures compared to pure ZnS nanocrystals. The emission band of ZnS nanocrystals shifted from the green region to the blue region and simultaneously the luminescence intensity was significantly enhanced via the forming of a low porosity SiO2 shell.
Experimental details
The chemicals Zn(Ac)2.2H2O thiourea and tetraethyl orthosilicate (TEOS) were used as the starting materials. The preparation procedure of encapsulated structure ZnS/SiO2 can be described as follows. First 0.004 mol Zn(Ac)2.2H20 and 0.008 mol thiourea were dissolved in a mixed solution of 15 ml water and 24 ml ethanol under magnetic stirring then 3 ml TEOS was added dropwise to form a sol. After stirring for 0.5 h this sol was transferred into 25 ml Teflon lined stainless steel autoclaves which were allowed to react at 200°C for 30 h. The precipitate was centrifuged with distilled water several times and dried at 50°C in an oven then sample A was obtained.
Bare ZnS denoted sample B was prepared for comparison. The preparation procedure was similar to that of ZnS/SiO2 nanostructures. The only difference is that there was no TEOS involved during the preparation.
The phase structures of the samples were characterized by x ray diffraction (XRD) using a Rigaku DMAX2500 x ray diffractometer with a copper target. Average particle sizes were calculated using the Scherrer formula D = 0.9l/ (b cos) where l (=0.154 18 nm) is the wavelength of x ray employed is the diffraction angle of the diffraction peak (110) of wurtzite ZnS and b denotes the half height width after subtracting the apparatus broadening effect.
Sample morphologies were determined using transmission electron microscopy (TEM) on a JEM 2010 apparatus with an acceleration voltage of 200 kV. Samples for TEM were prepared by making a dispersion of the samples in ethanol and putting drops of them on carbon coated copper grids. The chemical compositions were examined with an energy dispersive x ray spectrometer (EDS) analysis. Infrared spectra of the samples were recorded on a Perkin Elmer IR spectrophotometer using a KBr pellet technique. UV-visible diffuse reflectance spectra of the samples were obtained using a Lamda 900 UV/VIS spectrometer in the wavelength region between 200 and 500 nm. BaSO4 was used as a reference material. Photoluminescence spectral measurements were carried out with a fluorescence spectrophotometer (Cary Eclipse) using an excitation wavelength of 309 nm at room temperature. Specific surface areas of the samples were measured by N2 adsorption isotherms at 77 K on a Micromeritics ASAP 2000 surface area and porosity analyzer after degassing the samples in a flowing nitrogen atmosphere at 100°C for 12 h in a separate degassing unit attached to the instrument.
Results and discussion
During the sample preparation thiourea played a very important role in controlling the phase purity. Low concentration of thiourea led to a mixture of nexagonal and cubic ZnS while as testified by a literature report excessive thiourea would result in hexagonal wurtzite ZnS. In order to obtain pure hexagonal ZnS nanocrystals the initial molar ratio of Zn (Ac) 2·2H2O to thiourea was carefully investigated and was selected to be 0.5 in this work. The phase purity and crystallinity of the as prepared samples were characterized by XRD. All diffraction data (figure 1) matched well those for ZnS with the hexagonal wurtzite structure (JCPDS No. 36 1450). For sample A a broad peak was clearly seen at 2= 23° which is associated with the amorphous silica. Based on the half height width of the (110) peak at 2= 47.5° the average particle size of the ZnS in sample A was calculated to be 27 nm using the Scherrer formula which is significantly larger than8.5 nm for ZnS nanocrystals in sample B prepared without the addition of TEOS. The ZnS/S O2 encapsulated nanostructures were formed by adding TEOS in the reaction.
It is clear that all particles were spherical with an encapsulated nanostructure. The SiO2 termination layers showed a lighter color compared to the core ZnS nanocrystals. These pattern spots can be readily indexed as the (011) (013) (111) planes for the hexagonal structure of ZnS. It should be noted that the core contained several spherical nanocrystals of ZnS with the particle sizes averaged in the range of 20 36 nm. Nevertheless. The particle size of the bare ZnS was only about 8 nm which was much closer to that calculated from XRD peak broadening. HRTEM analysis of bare ZnS nanocrystals showed that the interlayer spacing is 0.259 nm which is compatible with that for (013) planes of the ZnS wurtzite phase. The ring corresponded to the (013) plane of wurtzite ZnS consistent with our XRD and TEM analyses.
The chemical compositions of ZnS/SiO2 encapsulated nanostructures were examined by energy dispersive spectrometer (EDS) data (figure 2) in which Zn and S signals corresponded to ZnS while Si and O signals are from SiO2. The Cu element signal observed was due to the copper grid. In order to investigate the surface structure IR spectra were used to confirm the existence of SiO2 terminations. The complex set of infrared vibration bands observed at 468 801 968 1095 and 1233 cm-1 (figure 3) can be assigned separately as follows the broad high intensity band at 1095 cm-1 along with the accompanied shoulder at 1233 cm-1 was due to the asymmetric LO and TO stretching bonds Si-Q-Siº of the SiO4 tetrahedron. The band observed at 801 cm-1 was associated with the Si-O-Si symmetric stretch while the sharp one located at 468 cm-1 corresponded to the Si-O-Si or O-Si-O bending mode and the shoulder at 968 cm-1 corresponded to the Si-O-H stretching vibrations. These terminations were also highly hydrated as is indicated by the presence of an absorption band at 1626 cm-1 which was attributed to the bending vibration of O-H bonds. The surface of the encapsulated nanostructure is very clean since only infrared peaks related to the SiO2 surface layer were observed. The absence of the characteristic vibration band of ZnS (figure 3) confirmed that ZnS nanoparticles were encapsulated inside the SiO2 shell.
The surface areas of the encapsulated nanostructure were investigated by N2 adsorption-desorption isotherm analysis. As indicated in figure 5 the isotherm is characterized by a hybrid type between types I and IV in the Brunauer-Deming-Deming-Teller (BDDT) classification. The hysteresis loop resembles type B in de Boer s classification which corresponds to the adsorbents with slit shaped pores between the parallel layers. This loop is completely different from that observed for the counterparts prepared by sol-gel methods. On the other hand the specific surface area of the encapsulated nanostructure was 38 m2 g-l which is far smaller than that of 645 m2 g-1 for TiO2/SiO2 composites synthesized under sol gel conditions. For the latter case a large specific surface has been attributed to the porous SiO2 shell. Therefore the very small surface area for ZnS/SiO2 encapsulated nanostructures indicated that SiO2 layers could be less porous. Such encapsulated nanostructures with a low porosity SiO2 layer can be very useful for technological uses because light scattering from the porous layers could be eliminated to significantly enhance the photoluminescence emission.
Since the particle size of ZnS/SiO2 encapsulated nanostructures is much larger than that of bare ZnS nanocrystals (figures 1 and 2 for XRD and TEM) the larger band gap energy of ZnS/SiO2 nanostructures seems not to follow the quantum size effect. On the other hand the diameter of ZnS nanocrystals is about 8 nm which is much larger than the exciton Bohr radius of 2.5 nm for ZnS. Therefore the influence of the quantum size effect should be negligible for both bare ZnS and ZnS/SiO2 encapsulated nanostructures.
What are the reasons for the significant blue shift of ZnS/SiO2 encapsulated nanostructures? To answer this question we have to take into account the relevant surface band modi fications. Many investigations have shown that the properties of nanocrystals are highly dependent on the modifications of the band structures which are in turn strongly influenced by surface chemistry and quantum size effects. Here for bare ZnS and ZnS/SiO2 encapsulated nanostructures the surface chemistry should play the dominant role in modulating the band structures of ZnS nanocrystals because amorphous surface layers over many substances like ZrO2 and As2S3 nanocrystals can shift the optical absorption towards the lower energy side. For most semiconductor nanocrystals as the particle size decreases the surface to volume ratio increases which results in an increase in the number of the broken and dangling bonds. Therefore the microstructure of nanocrystals should be disordered and highly defected to give sub bands of defects in the gap of the crystals. These defect sub bands in the gap of nanocrystals would result in a red shift in optical absorption. We therefore propose that the abnormal band gap shift of bare ZnS nanocrystals compared to the ZnS/SiO2 encapsulated nanostructures could be mainly attributed to the amorphous surface layer and defect sub bands in the gap. According to the Tauc model localized band tail states and extended states exist in amorphous semiconductors and the optical absorptions might occur due to transitions such as from (1) extended states in the valence band to the localized states near the bottom of conduction band or (2) localized states near the top of the valence band to the extended states in the conduction band. The band gap energies of these transitions are generally smaller than that for the bulk. Furthermore defects such as vacancies and interstitials of Zn or sulfur would introduce defect energy levels into the gap leading to an optical absorption with smaller energies than the gap. It is well known that the surface capped organic molecules and high band gap shells on nanocrystals can reduce the surface defect densities to ameliorate the surface properties. From the TEM image it can be seen that the ZnS nanocrystals in ZnS/SiO2 nanostructures are fully encapsulated by the SiO2 layer which will decrease the number of broken and dangling bonds and reduce the influence of surface amorphous states on the optical absorption. Based on above mentioned analyses it can be concluded that the surface effects on ZnS/SiO2 nanostructures are limited by the SiO2 surface layer compared to that of bare ZnS nanocrystals. This is why ZnS nanocrystals show a significant red shift of optical absorption compared to ZnS/SiO2 encapsulated nanostructures.
The photoluminescence of ZnS nanomaterials is generally sensitive to the synthetic conditions crystal size and shape. The emission spectrum of ZnS/SiO2 encapsulated nanostructures is shown in figure 5 in which the emission spectrum of bare ZnS nanocrystals is also given for comparison. The broad emission band for ZnS/SiO2 nanostructures centered at about 430 nm is similar to the reported near UV emission band at 438 nm which is assigned to the transitions involving vacancy states in ZnS nanocrystals. Such emission bands have also been reported for ZnS nanowires and nanobelts. Herein we attribute the blue emission of ZnS/SiO2 nanostructures to self activated centers arising from the vacancies or interstitial atoms in lattice. In the case of interstitials when an atom is located into a site which is not occupied in the perfect crystal a rearrangement of the nearest neighbors would take place which gives rise to a lattice deformation. Therefore the extent of deformation depends on the size of the inserted atom. Sulfur ions are larger than zinc ions thus the interstitial sulfur ion would induce more strain in the lattice. Electronic levels originating from this interstitial site will have smaller binding energies. Therefore the energy levels induced by and interstitial sulfur will be closer to the valence band edge and that by interstitial zinc will be closer to the conduction band edge. Similarly the energy levels induced by the sulfur vacancies should be close to the conduction band compared to those induced by zinc vacancies. The localized donor sites produced by sulfur vacancies in the lattice of bulk ZnS are ionized at room temperature to populate the conduction band. This self activated blue emission of ZnSnanocrystals encapsulated by SiO2 is deviated from the band gap transition owing to the electron or hole shallow traps (sulfur vacancies or interstitial sulfur) that likely act as the recombination centers for photogenerated charge carriers. Upon photo excitation electrons and holes would be present in the conduction band and valence band respectively. Electron transitions between sulfur vacancies and the valence band or between the conduction band and sulfur interstitial may result in a blue emission. Interestingly no emission band appears in the green region for ZnS/SiO2 nanostructures. By contrast for bare ZnS nanocrystals as illustrated in figure 7 the emission peak was located at 525 nm though no emission band was observed at 430 nm. It is known that nanocrystals of ZnS doped with Cu+ Mn+ or rare earth ions exhibit visible emissions due to the doping levels. The strong green emission at 525 nm for bare ZnS nanocrystals has been thought to derive from some interstitial states zinc vacancies or surface defect centers. Kumbhojkar and co workers studied the photochemistry of ZnS clusters and assigned the green emission to the dangling orbitals of sulfur residing on the surface of the clusters. Dunstan et al systematically investigated the photoluminescence properties of ZnS colloids and assigned the green emission at 560 nm to sulfur species on the surface of ZnS nanocrystals. Ye and Jiang have attributed the green emission at 530 nm for hexagonal ZnS to the contribution of surface sulfur species. In order to understand the origin of the green emission observed in this work we first investigated the formation of ZnS and ZnS/SiO2 nanostructures. In the reaction process sulfur source is in double stoichiometric excess. Thus sulfur species are most likely present on the surface of ZnS nanocrystals in both samples. Are sulfur species really responsible for the green photoluminescence in bare ZnS nanocrystals? As it is known that the photoluminescence induced by sulfur species is highly sensitive to the change of surface structure (either the oxidation reaction or surface modification) any minor changes in electronic configuration can have a significant impact on the photoluminescence performance. Since the densities of surface trap centers were dramatically reduced when encapsulated by the SiO2 surface layer the green emission disappeared to show only the presence of blue emission. With regards to the absence of the emission at about 430 nm for ZnS nanocrystals it has been testified that the lifetime of the green emission induced by sulfur species is very much shorter than the blue emission induced by self activated centers. Therefore the self activated blue emission band could not compete with the elemental sulfur related emission. Consequently bare ZnS nanocrystals displayed only an emission band in the green region. Having these analyses in mind it can be concluded that sulfur species on the surface of ZnS nanocrystals contribute to the green photoluminescence band.
NANOCABLES FROM POLY (DIMETHYL SILOXANE)
One dimensional (1D) nanomaterials such as wires or rods belts tubes and cables have attracted tremendous attention recently due to their importance in basic scientific research and their unique applications in nanoscale devices. Up to now material systems including group IV and metallic elements oxides carbides and transition metal chalcogenides have been synthesized in 1D nanostructures by a variety of methods.
SiC is an important semiconductor that can be used for devices operating at high temperature high power high frequency and in a harsh environment. The mechanical properties of an individual SiC nanowire are much better than those of bulk SiC and SiC nanowires exhibit good field emitting properties. Silica (SiO2) nanotubes have many potential applications in bioanalysis catalysis and optic devices. SiC/SiO2 nanocables with crystalline SiC core and amorphous SiO2 shell have many potential application fields because they have the 1D features of both nanowires and nanotubes in the axial direction and are idea semiconductor insulator heterostructures in the radial direction. Therefore many groups have devoted their efforts to the synthesis and properties of SiC/SiO2 nanocables and many methods including laser ablation are discharge carbon thermal reduction of silica xerogels carbon/graphite thermal reduction of amorphous SiO heating NiO catalysed silicon substrate under a reductive environment using the carbothermal reduction of WO3 by C [30] treating a silicon substrate deposited on a thin Ni-C film at high temperature chemical vapor reactions using a mixture of milled Si and SiO2 powders and CH4 as raw materials and Ni(NO3)2 as a catalyst precursor exposure of an Au coated silicon substrate to carbon monoxide at high temperature heating an SiC Fe Co mixture under CO atmosphere and so on have been developed. However the length of nanocables prepared is usually shorter than 100 mm. Recently we developed a simple route to ultra long SiC/SiO2 nanocables (a few millimeters in length) by pyrolysis of poly(dimethyl siloxane) (PDMS) at 1050°C in a flowing Ar atmosphere. However the yield is still low and the nanocables are not so uniform (the diameter of the nanocables is in the range of about 15 110 nm).
Ferrocene has been demonstrated to be an efficient catalyst precursor for the synthesis of carbon nanotubes whereas it has not been used for the synthesis of SiC/SiO2 nanocables therefore in this work the pyrolysis of PDMS catalysed by ferrocene has been investigated. Not only ultra long (in millimeter scale) but also ultra thin (diameter of most nanocables in the range of 5-10 nm) SiC/SiO2 nanocables with much higher yield were obtained as we expected. Such thin and long SiC/SiO2 nanocables have not been reported previously. The microstructure and growth mechanism of the nanocables were studied.
Experimental details
The synthesis was carried out in the same system as that for the synthesis of SiC/SiO2 nanocables from pure PDMS. Briefly the system contained a quartz tube with one end sealed (inner diameter ~ 85 mm) and the sealed end was put in the hot zone of a box furnace. In a typical synthesis about 6 ml analytical pure PDMS and 0.1 g ferrocene were placed in a 10 ml corundum crucible and the crucible was put on a mullite firebrick (~150 × 80 × 12 mm3) and then the firebrick was carefully pushed into the hot zone of the quartz tube. A mullite stopper (diameter ~ 83 mm) was pushed into the tube until the furnace gate was reached to prevent the reactants from being directly carried out of the hot zone. A smaller quartz tube (inner diameter ~ 8 mm) for transporting Ar gas was passed through the stopper until it reached the sealed end of the big quartz tube. The open end of the big quartz tube was sealed with a rubber stopper for gas inlet and outlet (see schematic diagram of the system in) After completely eliminating the air in the tube furnace the furnace was heated up to ~1050°C at a rate of 10°C min-1 and held at 1050°C for 2 h in flowing Ar at a flow rate of ~8 ml min-1. After the furnace was cooled to room temperature a large quantity of white wool like product (up to 2 mm thick see figure 1 in supporting materials (available at stacks.iop.org/Nano/18/485601)) was observed on the surface of the firebrick. Compared with the nanocables from pure PDMS the product was shorter the yield however was increased several times.
The size structure and composition of the product were examined by powder x ray diffraction (XRD Rigaku D/max2550) scanning electron microscopy (SEM JSM5510) transmission electron microscopy (TEM H 800) and high resolution TEM (HRTEM JEM 2100F) equipped with energy dispersive x ray spectroscopy (EDS) respectively. A little of the product was stripped from the firebrick and stuck on the surface of a carbon film that was bonded on an aluminum pellet for conducting for SEM observation. For the TEM sample the product was gently ground with an agate mortar for a few seconds and then dispersed in ethanol by ultrasonic vibrations for 1 h before being dropped on a copper grid coated with carbon film.
Results and discussion
Figure 1 shows the XRD pattern of the product. Like the XRD pattern (lower pattern in figure 1) of the SiC/SiO2 nanocables from pure PDMS the XRD pattern (upper pattern in figure 1) of the product of this work has also a hillside background and peaks for SiC (JCPDS card file No. 29 1129) whereas the intensity for the (111) and (220) peaks becomes stronger. This roughly indicates that the product also contains SiC and SiO2. An additional weak peak (2q = ~33.7°) marked by an asterisk is considered to be from the carbon (JCPDS card file No. 89 8488) nanoparticles adhered to the bottom of the wire like product which will be explained in the following text.
SEM image (see figure 2 in supporting materials (available at stacks.iop.orglNano/18/485601)) shows that the product consists of nanowires. Unlike the entangled nanocables from pure PDMS the nanowires are mainly straight. The nanowires are dense very fine and too long to be determined under SEM. The length is estimated to be on the millimeter scale from the thickness of the product grown on the firebrick.
A stacking fault which is a typical fault characteristic in one dimensional SiC structures is occasionally observed. Besides stacking faults microtwinnings have also been observed in the core of some nanocables during the HRTEM observation. Figure 2(e) shows an HRTEM image of a thin nanocable with microtwinning and stacking faults.
It is found that the nanoparticles (darker contrast) are also covered by a thin sheath (lighter contrast) and that the sheath has the same contrast as the sheath of the nanocables suggesting the sheath is also amorphous SiO2.
In order to elucidate the growth mechanism of the nanocables the nanoparticles were examined in detail. Both the rod and nanoparticle show clear lattice planes. The spacing at the nanoparticle is ~0.2 nm which is close to the spacing of the planes of the cubic Fe (JCPDS card file No. 87 0722) suggesting that the nanoparticle is Fe (coated with a thin layer of amorphous SiO2). The SAED pattern (inset in figure 3(e) recorded on the nanoparticle confirms that the tip is single crystalline Fe.
Combining the XRD HRTEM SEAD and EDS results it can be deduced that the product consists of SiC core SiO2 shell nanocables with an Fe nanoparticle at their tips. The hillside background of the XRD pattern (figure 1) must be from the SiO2 shell.
Obviously the growth mechanism of the nanocables in the present work is different from that of the nanocables from pure PDMS. On the basis of the above results the growth process of the nanocables is proposed as follows. Ferrocene begins to vaporize at about 185°C and decomposes above 400°C. PDMS vaporizes above 200°C and then the cleavage of the Si-C bond occurs to produce methyl (-CH3) groups and the methyl groups decompose to form CH4 C2H6 H2 and C below 600°C. Some carbon nanoparticles are carried by the gas stream (Ar gas mixed with hydrocarbon gases from pyrolysis of PDMS) and deposited on the surface of the firebrick to form a carbon film that makes the firebrick black (see figure 1 in supporting materials (available at stacks.iop.org/Nano/18/485601)). The XRD and TEM samples could be contaminated by the carbon nanoparticles and therefore the XRD pattern (see upper pattern in figure 1) of the product has a peak for carbon and carbon nanoparticles are observed during the TEM observation (see figure 2(a)). The vaporized ferrocene is reduced by the H2 to form atomic iron which agglomerates into an iron cluster. At higher temperatures the cluster becomes an iron droplet and deposits on the carbon film. As all the methyl groups are separated from the PDMS and the temperature is high enough SiO vapor forms. The iron droplet serves as a preferential site for absorption of the hydrocarbon gases and SiO gas. The hydrocarbon gases are catalytically decomposed into carbon atoms and the carbon atoms diffuse into the iron droplet and then a liquid Fe-Si-O-C alloy forms. As I the Fe-Si-O-C alloy becomes supersaturated SiC and SiO2 concurrently segregate from the alloy droplet by the reaction 2SiO + C SiC + SiO2. Because the melting point of SiC is much higher than that of SiO2 SiC solidifies earlier an SiC nanorod is formed along the (111) direction and then amorphous SiO2 is nucleated on the outer surface of the newly formed SiC nanorod. This process continues as long as the catalyst alloy remains in a liquid state and the reactants are available and finally an ultra long nanocable with a crystalline Fe nanoparticle at its tip forms.
Because the PDMS decomposes step by step the growth of the nanocables lasts a long time to form long nanocables. Meanwhile the iron nanodroplets are very fine since they are formed by reducing ferrocene vapor and are collected as iron catalysts for a very short time at high temperatures the products are ultra thin.
As ferrocene (or another Fe containing reagent) is used as the catalyst precursor for the synthesis of carbon nanotubes hydrogen is always used for reducing ferrocene into Fe. In this work it is not necessary since hydrogen can be produced from the pyrolysis of PDMS. Like the growth of aligned carbon nanotubes we deduce that the SiC/SiO2 nanocables are also aligned because the nanocables growing parallel or nearly parallel to the firebrick surface will be terminated as they hit other nanocables or nanoparticles while those growing initially normal or nearly normal to the firebrick will continue to grow. This will be studied in future.
NANOROD ARRAYS AND THEIR FIELD EMISSION PROPERTIES
Zinc oxide (ZnO) nanostructures have attracted much attention due to their versatile properties and potential applications in electronic and optoelectronic devices chemical and bio sensors ultraviolet laser diodes and light emitting diodes solar cells photocatalysts photodetectors etc. However to develop such devices a simple growth technique and a precision control of morphology alignment and position are required. With this in view a large number of papers about growth methods have been reported such as physical vapor deposition metal organic chemical vapor deposition chemical vapor deposition thermal evaporation and the solution growth method. Among them the solution method is one of the most attractive techniques because it is simple cost effective and scalable to large areas. A selective area growth of well aligned ZnO nanostructures has also been achieved using various techniques like nanosphere lithography polystyrene-microsphere based self assembly monolayers silanebased self assembled monolayers and conventional photolithography. However these patterning methods except the conventional photolithography require relatively high temperatures expensive masks complex multi step processes and metal catalysts in some cases. Conventional photolithography is an easy and economical approach to achieve the required patterns on the substrates but it is not easy to position the nanostructures with nanoscale precision. To solve such a drawback of photolithography and grow ZnO nanostructures selectively on pre patterned substrates we examined the hybrid approach of combining electronbeam lithography (top down) and the solution growth method (bottom up ).
In this paper we report the selectively controlled growth of ZnO nanorod arrays on the desired areas of substrates by the hybrid approach. Prior to the growth of ZnO nanorods ZnO buffer layers were prepared on Si substrates using atomic layer deposition (ALD). The ZnO/Si substrates were then patterned either by electron beam lithography (EBL) for nanoscale patterning or by conventional photolithography for microscale patterning. Finally the ZnO nanorod arrays were grown selectively on wanted areas in solution at 70-90°C and they were subject to characterizations including field emission properties.
Experimental details
The ZnO seed layer having a thickness of ~100 nm was grown on an Si(100) substrate by the ALD technique at 250°C using diethyl zinc ((C2H5)2Zn 99.999%) and oxygen (99.999%) as the zinc and oxygen sources respectively. Details are described elsewhere. For patterning the ZnO/Si substrates with different dimensions and shapes in the nanoscale scanning electron microscopy equipped with a field emission electron gun and integrated with a nanopattern generator was employed. Polymethylmethacrylate (PMMA) was used as the electron beam resist material and was spin coated over the ZnO/Si substrate. After writing the patterned substrates were developed in a solution of methyl isobutyl ketone and isopropyl alcohol (volume ratio of 1 3) for 30 s. For the purpose of comparison we also patterned the ZnO/Si substrates on the micrometer scale by conventional photolithography with the help of an inexpensive photopolymer (polyethylene terephthalate) mask. The photo resist (PR) was spun on the substrates at 3500 rpm and baked at 90°C for 10 min. The baked substrates were exposed to ultraviolet (UV) light with the help of an inexpensive photopolymer mask and developed by (AZ300) followed by inductively coupled plasma etching with CH4/H2/Ar discharges to transfer a pattern on the ZnO/Si substrates. For the growth of ZnO nanostructures on the prepatterned Si substrates in solution zinc nitrate hexahydrate (Zn(NO3)26H2O Sigma Aldrich) and water soluble hexamethylene tetrammine (C6H12N4 Sigma Aldrich) were used as reagents. In a typical reaction the pH of the solution was kept at 6-7. The pre patterned ZnO/Si substrates were immersed in the solution and the temperature of the flask was kept constant in the range of 70-90 °C for 6h. After completion of the synthesis the system was cooled down to room temperature. The samples were rinsed with distilled water and then dried at room temperature. Water purified in a Millipore Milli Q Plus purification system (resistivity ~18.2 Mī cm) was used in all experiments. After the successful periodic growth of ZnO nanorod arrays isopropyl alcohol (IPA) and acetone lifted off the electron beam resist materials.
Results and discussion
Structural and optical properties of selectively grown ZnO nanorod arrays
The distance between arrays was periodically spaced and the lengths and diameters of as grown nanorods were 1.0-1.5 µm and ~50 nm respectively. These selectively grown nanorod arrays exhibited a remarkable uniformity in terms of density diameter and length. These results show that controlling the pattern size of the substrate through the EBL process readily controls the dimensions and density of nanorod arrays. It is worthwhile noting that the successful synthesis of periodic single or a couple of ZnO nanorods was achieved by decreasing the pattern size of the substrate to sub 100 nm (d). The occasionally missing spots in the 100 nm patterns were most likely caused by either poor electron beam patterning or during the removal of electron beam resist material. A possible growth mechanism for nanorod arrays on ZnO/Si substrates by the solution method was described in detail in our previous work.
Figure 2 shows XRD patterns and TEM images of the ZnO nanorods selectively grown on pre patterned substrates. A sharp strong and dominant peak at 34.4° assigned as ZnO(0002) was observed which is higher than any other peaks (a) indicating that the synthesized nanorod arrays are single crystalline grown along the direction. Moreover the strong intensity of the (0002) reflection with narrow width also shows that the ZnO nanorods are well oriented along the normal direction of the substrate surface which is almost consistent with the FESEM observations. The detailed structural characterization of the as grown ZnO nanorod arrays was performed by transmission electron microscopy (TEM). The typical diameter of the as grown hexagonal nanorod is ~50 nm which exhibits a clean and smooth surface passim to its length. The high resolution TEM (HRTEM) image of the corresponding nanorod shows a distance of 0.52 nm between two lattice fringes which represents the plane of the wurtzite hexagonal ZnO. The corresponding selected area electron diffraction (SAED) pattern indicates that the nanorod is single crystalline and grown along the (0001) direction (inset in (b)).
Figure 3 show photoluminescence spectra of selectively grown nanorods. The PL spectra of ZnO nanorods and the ZnO buffer layer are compared. It indicates that the selectively grown ZnO nanorods have a strong UV emission at 381 nm and a broad deep level visible emission at 580 nm compared to the ZnO film. It is worthwhile noting that the intensity of UV emission of ZnO nanorod arrays was independent of the shape of the array pattern. The UV emission is the band edge emission resulting from the recombination of excitonic centers. The green emission results from the recombination of electrons in singly occupied oxygen vacancies with photoexcited holes. We observed that the full width at half maximum (FWHM) values of UV peaks of the selectively grown ZnO nanorods and the ZnO buffer films are ~16 and ~68 nm respectively. The sharp excitonic UV emission peak of the nanorods compared to that of ZnO films indicates that the as grown ZnO nanorods are better single crystalline than the ZnO buffer layer. Interestingly the intensity of the visible emission peak of the as grown ZnO nanorods is also high as compared to ZnO buffer films indicating the presence of oxygen defects in the crystalline ZnO nanostructures. However it is worthwhile noting that the as grown ZnO nanorod arrays show a stronger UV emission with much suppressed green emission compared to the previously reported results.
Field emission properties of selectively grown ZnO nanorod arrays
Field emission a unique quantum mechanical effect is of great interest in displays and other electronic devices. As the one dimensional ZnO nanostructures have structural properties similar to carbon nanotubes recently they have attracted much attention as a good alternative for field emission cathodes. Figure 5 shows the emission current density (J) versus macroscopic electric field (E) plot obtained from the spherically shaped ZnO nanorod arrays grown selectively on Si substrates with an anode sample separation of 150 m. In addition a uniform fluorescence image was obtained (inset in (a). Here the macroscopic field is calculated from the external voltage (V) applied to the device divided by anode spacing d (obtained from a spherometric arrangement in the field emission apparatus). Theoretically the emission current density (J) is related to the macroscopic electric field E by.
OXIDATION OF OTS MONOLAYERS
Oxidative probe lithography has been demonstrated to be one of the most promising methods for the preparation of functional nanodevices. Originally applied to (semi) conducting substrates an electrically biased AFM tip can be used to locally oxidize underlying substrates. The local oxidation process can also be used on self assembled octadecyl trichlorosilane monolayers on silicon supports. In this process patterns of carboxylic acid groups can be formed at the air monolayer interface. Using self assembly techniques the created patterns can subsequently be utilized as templates in additional chemical modification steps for the introduction of chemical functionalities on spatially defined spots on a substrate. Recently a number of publications have been presented in which for example progress is reported towards the preparation of multi layer structures nanoscale electrical circuits and multi component nanoparticle assemblies. The possibility to use not only wetting or electrostatic interactions but also the ability to utilize the carboxylic acid groups in hydrogen bonding as well as for covalent interactions and end group conversions makes the presented method a versatile platform for the fabrication of a broad range of functional structures with nanometer dimensions. However for an industrially viable application of these techniques the patterning speed has to be still further improved. At the same time processes have to be extended to cover large areas on the substrate i.e. to provide faithful copies of the formed structures with high accuracy. In addition and scientifically more importantly the ability to functionalize large surface areas would allow for the first time the application of spectroscopic techniques to characterize the various stages of surface modification.
To obtain large area surface modification the use of macroscopic (patterned) electrodes has been reported. Conductive objects such as copper TEM grids imprint stamps compact disks or even a water droplet may be used as the electrode. Because the oxidative lithography critically depends on the formation of a water meniscus between the substrate and the stamp even these relatively rigid stamps can be used. The first disadvantage of the application of stamps (such as TEM grids) is that for each desired pattern a special electrode is required. The second is the fact that the local oxidation of OTS to form carboxylic acid end groups only occurs in a small window of process conditions. The large surface areas of the stamp and inhomogeneities lead to difficulties in the control of the oxidation process and (partial) over oxidation (i.e. degradation of the monolayer and formation of silicon oxide may occur). In addition the pressure exerted by a sharp tip on a surface is considerably larger. Therefore formed patterns cannot be compared directly to tip inscribed patterns.
To overcome these limitations we have chosen to use a combinaticn of automation and parallel tip oxidation to perform large scale oxidations. Examples of both methods and limitations will be discussed as well as examples of subsequent surface modification.
Results and discussion
Automation
The first strategy we employed to achieve large scale oxidation involves the use of an automated AFM. The automated AFM we applied for automated oxidation experiments is a modified automated atomic force microscope from NT MDT (Solver LS7) which is equipped with a programmable -y stage (capable of 360° rotation and 20 em translation) with about 1 m precision. The equipment is capable of autonomous tip landing and withdrawal and the utilized system was equipped with an external closed loop equivalent scanner. Additionally an optical microscope (1.5 m resolution) was used for visual surface inspection and a scratch on the surface was used as a reference point. To facilitate the automated oxidation experiments the software was adapted to allow the assignment of a list of coordinates which can be addressed sequentially. For each of these coordinates a specific action (i.e. oxidation pattern) may be assigned. The possible automated oxidation processes are not limited to oxidation experiments and may also be applied to dip pen nanolithography or force based patterning processes (dynamic plow lithography). Here we choose to apply the automated oxidation process on selfassembled monolayers of OTS on silicon substrates and the generated patterns of carboxylic acid end groups are used as chemically active surface templates in subsequent surface modification reactions. After the oxidation and/or the modifications steps the list of coordinates can be addressed again to image the results.
The most straightforward method to obtain large area electro oxidation is to perform automated raster imaging with a biased tip. In this way a biased tip is approached to and rastered over an OTS coated substrate resulting in the formation of square oxidation patterns that are spaced by movements and rotations of the sample stage. In a preliminary study we prepared in this way a 10 × 10 array of square areas of 4 m2 each separated by 15 m. Each of the oxidized squares could subsequently be decorated with cationic gold nanoparticles (d = 20 nm) the presence of the gold nanoparticles was confirmed by XPS (Au 4f7/2 84.4 eV). For a single modified area insufficient signal intensity would be available. For 100 areas however the total surface coverage amounts to approximately 2% of a 135 × 135 m2 area providing sufficient signal intensity for XPS measurements.
A disadvantage of oxidation by raster scanning of a biased tip is that the conductive tip coating degrades quickly due to the long contact between the tip and the surface. Tip degradation depends on the quality and nature of the coating of the tip itself but the lifetime is additionally decreased by working at high contact forces (i.e. cantilever high stiffness) and by the total distance the tip travels during or before the oxidation process.
In addition to raster scanning a biased tip over the surface the automated system also allows the writing of structures according to vector based designs. This has the additional advantage that the required time for lithography is reduced and that the contact times of the tip are significantly reduced thus expanding the potential lifetime of the tip. To enable the vector based oxidation the standard software was custom modified to allow vector oxidation without pre imaging the desired target area. Since the OTS surfaces are flat skipping the imaging step adds negligible risk of tip crashes by for example dust particles while the tip life is enhanced.
The pattern consists of 25 sets of four circles. The displayed structures were copied onto the substrate in a 5 × 2 array each area is separated 50 m from the next area leading to a total of 1000 identical circles. After the automated lithography process (bias voltage of -8.0 V 0.1 ms pulse duration) all patterned areas were imaged in contact mode using the same tip and the same coordination points stored within the software. The images were obtained without repositioning the piezo demonstrating the accuracy of the sample stage within 2 m. The conductive coating of the used tip did not degrade during writing as indicated by the constant line width and friction intensity for these areas (compare areas indicated 1 10 in figure 1).
After performing the lithography process the sample was removed from the AFM stage and the oxidized areas were used as templates in the subsequent assembly of positively charged fluorescent nanoparticles. Therefore the sample was dipped in a dilute solution of pre formed CdSe/ZnS core/shell nanoparticles with an amine containing stabilization shell followed by thorough rinsing with water and drying in a stream of nitrogen. The sample was repositioned on the AFM stage and the oxidized areas were imaged in intermittent contact mode using a different tip.
Successful assembly of the nanoparticles can be observed by a consistent increase in observed height for the patterned areas by 10-15 nm which corresponds to the diameter of these particles (see also the provided cross section). Successful oxidation was demonstrated by the complete and dense coverage with nanoparticles for all of the oxidized templates and the absence of nanoparticles on the non patterned areas indicates the specificity of the assembly process.
Parallel cantilever arrays
A second approach towards large scale surface patterning without losing obtainable lateral resolution is the simultaneous operation of a large number of AFM tips i.e. parallel patterning techniques. The most famous example of parallel tip operation is undoubtedly the IBM Millipede system in which an array of up to 4096 individually addressable tips is used in the thermal patterning and read out of thin polymer films. Tip arrays have also been applied for the imaging of millimeter sized samples as well as in the parallel surface patterning for both the DPN as well as the oxidative probe lithography techniques on semiconducting substrates.
Whereas for DPN cantilever arrays consisting of up to 50000 cantilevers have been reported due to the relative insensitivity to variations in the applied contact force the electro oxidation of 0TS monolayers is force dependent high contact forces reduce the efficiency of the oxidation and require long oxidation times. In an independent study the local electro oxidation of 0TS monolayers was found to be dependent on the applied force. Keeping the oxidation time and bias voltage constant the line width scales inversely with the applied contact force. If a single tip is approached to the surface additionally (by several micrometers) after landing (i.e. bringing the tip in hard contact just before crashing the cantilever) oxidation times need to be increased by a factor 10-20 to obtain reliable oxidation results. In addition oxidation at high force causes flexing of the cantilever resulting in an additional displacement of the resulting oxidized structures. A more detailed report on the oxidation of OTS under high force conditions is in preparation. The dependence on applied force implies that for a multiple cantilever setup the cantilevers should be identical in length and force constant and the array should be parallel to the substrate to avoid excessive differences in contact forces. For the experiments reported here we used a custom designed array of four cantilevers with a length of 110 m and a force constant of 4-5 N m-1.
These parameters allow the use of cantilevers in contact mode as well as in intermittent contact mode. Because in our system only one of the cantilevers can be in active control whereas the rest is in passive contact with the surface an elaborate approach method was used to ensure good horizontal alignment of the tips with the surface. For this method one of the outer cantilevers was approached to the surface in intermittent contact mode. Upon landing the feedback loop was disabled and the laser position switched to the tip on the other extreme of the array. For horizontally aligned cantilevers this cantilever was also found to be in feedback without the need for additional z displacement of the z piezo. If the cantilever was found not to be in contact the array was withdrawn from the surface and the legs of the AFM head were adjusted to obtain the correct tilting. With this iterative process it was generally possible to obtain good alignment of the cantilevers (less than 100 nm tilt) in less than five steps. After adjusting the cantilever array the operation mode was switched from intermittent contact to contact mode. Active control was adjusted on the cantilever which is at the furthest distance from the surface to ensure passive contact with the surface for the other cantilevers.
To enhance the tip life time no imaging was performed with the array. After oxidation the tip was exchanged for a normal single cantilever which was aligned using a pre recorded imaged of the cantilever array and a scratch marked substrate.
The imaging of this substrate was performed in intermittent contact mode using a single cantilever whereas the oxidation was performed using the four tip array. The structures observed after oxidation and patterning with nanoparticles look identical and all display an observed increase in height of 10-20 nm corresponding to the diameter of the particles. Therefore these images clearly demonstrate the successful use of a parallel cantilever array for oxidation of OTS monolayers and the subsequent assembly of the nanoparticles onto these patterns.
In an additional oxidation experiment the process was taken one step further by using the same cantilever array in a subsequent step to oxidize an additional area close to an existing assembly of nanoparticles.
Using the same cantilever array the area was imaged after particle assembly. After locating the desired area the array was switched to operate in contact mode and a small dot was oxidized in the center and at the bottom right corner of the structures. In contrast to contact mode imaging the additional dot appears as a small protrusion (1-2 nm) both in the trace as well as in the re trace direction. In the next step the surface was dipped in a solution containing smaller (d ~ 15 nm) CdSe/ZnS nanoparticles. The resulting surface pattern was imaged in intermittent contact mode after replacing the parallel cantilever array by a single cantilever.
