The binary compound of one or more oxygen atoms with at least one metal atom that forms a structure ≤100 nm is classified as metal oxide (MO_x) nanoparticle. MO_x nanoparticles have exceptional physical and chemical properties (especially if they are smaller than 10 nm) that are strongly related to their dimensions and to their morphology. These enhanced features are due to the increased surface to volume ratio which has a strong impact on the measured binding energies. Based on theoretical models, binding or cohesive energy is inversely related to particle size with a linear relationship Equation 1.

where E_NP and E_bulk is the binding energy of the nanoparticle and the bulk binding energy respectively, c is a material constant and r is the radius of the cluster. As seen from Equation 1, nanoparticles have lower binding energies than bulk material, which means lower electron cloud density and therefore more mobile electrons. This is one of the features that have been identified to contribute to a series of physical and chemical properties.

Synthesis of metal oxide nanoparticles

Since today, numerous synthetic methods have been developed with the most common ones presented in Table 1. These methods have been successfully applied for the synthesis of a variety of materials with 0-D to 3-D complex structures. Among them, the solvothermal methods are by far the most popular ones due to their simplicity. Between the two classes of solvothermal methods, slow decomposition methods, usually called thermal decomposition methods, are preferred over the hot injection methods since they are less complicated, less dangerous and avoid the use of additional solvents.

Table 1: Methods for synthesizing MO_x nanoparticles

Method	Characteristics	Advantages	Disadvantages
Solvothermal a) Slow decomposition b) Hot injection	a) Slow heating of M-precursor in the presence of ligand/surfactant precursor b) Injection of M-precursor into solution at high Temp.	a) Safe, easily carried out, variety of M-precursors to use b) Excellent control of particle distribution	a) Poor control of nucleation/ growth stages – Particle size b) Hazardous, Reproducibility depends on individual
Template directed	Use of organic molecules or preexistent nanoparticles as templates for directing nanoparticle formation	High yield and high purity of nanoparticles	Template removal in some cases causes particle deformation or loss
Sonochemical	Ultrasound influence particle nucleation	Mild synthesis conditions	Limited applicability
Thermal evaporation	Thermal evaporation of Metal oxides	Monodisperse particle formation, excellent control in shape and structure	Extremely high temperatures, and vacuum system is required
Gas phase catalytic growth	Use of catalyst that serves as a preferential site for absorbing Metal reactants	Excellent control in shape and structure	Limited applicability

A general schematic diagram of the stages involving the nanoparticles formation is shown in Figure 1. As seen, first step is the M-atom generation by dissociation of the metal-precursor. Next step is the M-complex formulation, which is carried out before the actual particle assembly stage. Between this step and the final particle formulation, oxidation of the activated complex occurs upon interaction with an oxidant substance. The x-axis is a function of temperature or time or both depending on the synthesis procedure.

Figure 1: Stages of nanoparticle synthesis.

In all cases, the particles synthesized consist of MO_x nanoparticle structures stabilized by one or more types of ligand(s) as seen in Figure 2. The ligands are usually long-chained organic molecules that have one more functional groups. These molecules protect the nanoparticles from attracting each other under van der Waals forces and therefore prevent them from aggregating.

Figure 2: Schematic representation of a surfactant/ligand stabilized nanoparticle.

Even though often not referred to specifically, all particles synthesized are stabilized by organic (hydrophilic, hydrophobic or amphoteric) ligands. The detection and the understanding of the structure of these ligands can be of critical importance for understanding the controlling the properties of the synthesized nanoparticles.

Metal oxide nanoparticles synthesized via slow decomposition

In this work, we refer to MO_x nanoparticles synthesized via slow decomposition of a metal complex. In Table 2, a number of different MO_x nanoparticles are presented, synthesized via metal complex dissociation. Metal–MO_x and mixed MO_x nanoparticles are not discussed here.

Table 2: Examples of MO_x nanoparticles synthesized via decomposition of metal complexes.

Metal oxide	Shape	Size (approx.)
Cerium oxide	dots	5-20 nm
Iron oxide	dots, cubes	8.5-23.4 nm
Manganese oxide	Multipods	>50 nm
Zinc oxide	Hexagonal pyramid	15-25 nm
Cobalt oxide	dots	~ 10 nm
Chromium oxide	dots	12 nm
Vanadium oxide	dots	9-15 nm
Molybdenum oxide	dots	5 nm
Rhodium oxide	dots,rods	16 nm
Palladium oxide	dots	18 nm
Ruthenium oxide	dots	9-14 nm
Zirconium oxide	rods	7x30 nm
Barium oxide	dots	20 nm
Magnesium oxide	dots	4-8 nm
Calcium oxide	dots, rods	7-12 nm
Nickel oxide	dots	8-15 nm
Titanium oxide	dots and rods	2.3-30 nm
Tin oxide	dots	2.0-5.0 nm
Indium oxide	dots	~ 5 nm
Samaria	Square	~ 10 nm

A significant number of metal oxides synthesized using slow decomposition is reported in literature. If we use the periodic table to map the different MO_x nanoparticles (Figure 3), we notice that most of the alkali and transition metals generate MO_x nanoparticles, while only a few of the poor metals seem to do so, using this synthetic route. Moreover, two of the rare earth metals (Ce and Sm) have been reported to successfully give metal oxide nanoparticles via slow decomposition.

Figure 3: “Periodic” table of MO_x nanoparticles synthesized using the slow decomposition technique.

Among the different characterization techniques used for defining these structures, transition electron microscopy (TEM) holds the lion’s share. Nevertheless, most of the modern characterization methods are more important when it comes to understanding the properties of nanoparticles. X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), nuclear magnetic resonance (NMR), IR spectroscopy, Raman spectroscopy, and thermogravimetric analysis (TGA) methods are systematically used for characterization.

Synthesis and characterization of WO_3-x nanorods

The synthesis of WO_3-x nanorods is based on the method published by Lee et al. A slurry mixture of Me₃NO∙2H₂O, oleylamine and W(CO)₆ was heated up to 250 °C at a rate of 3 °C/min (Figure 4). The mixture was aged at this temperature for 3 hours before cooling down to room temperature.

Figure 4: Experimental setup for synthesis of WO_3-x nanorods.

Multiple color variations were observed between 100 - 250 °C with the final product having a dark blue color. Tungsten oxide nanorods (W₁₈O₄₉ identified by XRD) with a diameter of 7±2 nm and 50±2 nm long were acquired after centrifugation of the product solution. A TEM image of the W₁₈O₄₉ nanorods is shown in Figure 5.

Figure 5: TEM image of WO_3-x nanorods.

Thermogravimetric analysis (TGA) is a technique widely used for determining the organic and inorganic content of various materials. Its basic rule of function is the high precision measurement of weight gain/loss with increasing temperature under inert or reactive atmospheres. Each weight change corresponds to physical (crystallization, phase transformation) or chemical (oxidation, reduction, reaction) processes that take place by increasing the temperature. The sample is placed into platinum or alumina pan and along with an empty or standard pan are placed onto two high precision balances inside a high temperature oven. A method for pretreating the samples is selected and the procedure is initiated. Differential scanning calorimetry (DSC) is a technique usually accompanying TGA and is used for calculating enthalpy energy changes or heat capacity changes associated with phase transitions and/or ligand-binding energy cleavage.

In Figure 6 the TGA/DSC plot acquired for the ligand decomposition of WO_3-x nanorods is presented. The sample was heated at constant rate under N₂ atmosphere up to 195 °C for removing moisture and then up to 700 °C for removing the oleylamine ligands. It is important to use an inert gas for performing such a study to avoid any premature oxidation and/or capping agent combustion. 26.5% of the weight loss is due to oleylamine evaporations which means about 0.004 moles per gram of sample. After isothermal heating at 700 °C for 25 min the flow was switched to air for oxidizing the ligand-free WO_3-x to WO₃. From the DSC curve we noticed the following changes of the weight corrected heat flow:

The heat flow increase during the WO_3-x to WO₃ oxidation is proportional to the crystal phase defects (or W atoms of oxidation state +5) and can be used for performing qualitative studies between different WO_x nanoparticles.

Figure 6: TGA/DSC plot for WO_3-x nanorods.

The detailed information about the procedure used to acquire the TGA/DSC plot shown in Figure 6 is as follows.

Fourier transform infrared spectroscopy (FTIR) is the most popular spectroscopic method used for characterizing organic and inorganic compounds. The basic modification of an FTIR from a regular IR instrument is a device called interferometer, which generates a signal that allows very fast IR spectrum acquisition. For doing so, the generatated interferogram has to be “expanded” using a Fourier transformation to generate a complete IR frequency spectrum. In the case of performing FTIR transmission studies the intensity of the transmitted signal is measured and the IR fingerprint is generated Equation 2.

Where I is the intensity of the samples, I_b is the intensity of the background, c is the concentration of the compound, ε is the molar extinction coefficient and l is the distance that light travels through the material. A transformation of transmission to absorption spectra is usually performed and the actual concentration of the component can be calculated by applying the Beer-Lambert law Equation 3.

A qualitative IR-band map is presented in Figure 7. The absorption bands between 4000 to 1600 cm^-1 represent the group frequency region and are used to identify the stretching vibrations of different bonds. At lower frequencies (from 1600 to 400 cm^-1) vibrations due to intermolecular bond bending occurs upon IR excitation and therefore are usually not taken into account.

Figure 7: Selected FTIR stretching and bending modes associated with the typical ligands used for nanoparticle stabilization.

TGA/DSC is a powerful tool for identifying the different compounds evolved during the controlled pyrolysis and therefore provide qualitative and quantitative information about the volatile components of the sample. In metal oxide nanoparticle synthesis TGA/DSC-FTIR studies can provide qualitative and quantitative information about the volatile compounds of the nanoparticles.

TGA–FTIR results presented below were acquired using a Q600 Simultaneous TGA/DSC (SDT) instrument online with a Nicolet 5700 FTIR spectrometer. This system has a digital mass flow control and two gas inlets giving the capability to switch reacting gas during each run. It allows simultaneous weight change and differential heat flow measurements up to 1500 °C, while at the same time the outflow line is connected to the FTIR for performing gas phase compound identification. Grand-Schmidt thermographs were usually constructed to present the species evolution with time in 3 dimensions.

Selected IR spectra are presented in Figure 8. Four regions with intense peaks are observed. Between 4000 – 3550 cm^-1 due to O-H bond stretching assigned to H₂O that is always present and due to due to N-H group stretching that is assigned to the amine group of oleylamine. Between 2400 – 2250 cm^-1 due to O=C=O stretching, between 1900 – 1400 cm^-1 which is mainly to C=O stretching and between 800 – 400 cm^-1 cannot be resolved as explained previously.

Figure 8: FTIR spectra of products from WO_3-x pyrolysis.

The peak intensity evolution with time can be more easily observed in Figure 9 and Figure 10. As seen, CO₂ evolution increases significantly with time especially after switching our flow from N₂ to air. H₂O seems to be present in the outflow stream up to 700 °C while the majority of the N-H amine peaks seem to disappear at about 75 min. C=N compounds are not expected to be present in the stream which leaves bands between 1900 – 1400 cm^-1 assigned to C=C and C=O stretching vibrations. Unsaturated olefins resulting from the cracking of the oleylamine molecule are possible at elevated temperatures as well as the presence of CO especially under N₂ atmosphere.

Figure 9: 3D representation of FTIR Spectra of the volatile compounds of WO_3-x.

Figure 10: Intensity profile of FTIR spectra of the volatile compounds formed from the pyrolysis of WO_3-x.

From the above compound identification we can summarize and propose the following applications for TGA-FTIR. First, more complex ligands, containing aromatic rings and maybe other functional groups may provide more insight in the ligand to MO_x interaction. Second, the presence of CO and CO₂ even under N₂ flow means that complete O₂ removal from the TGA and the FTIR cannot be achieved under these conditions. Even though the system was equilibrated for more than an hour, traces of O₂ are existent which create errors in our calculations.