In recent year much interest has been shown for remote, versatile, fast as well as accurate elemental analysis. Analysis techniques such as laser-ablation inductively coupled mass spectrometry (LA-ICP-MS)  , X-ray fluorescence spectrometry (XRFS)  , and secondary ion mass spectrometry (SIMS)  are able to resolve smaller element concentration. However, they often require specific sample preparation and under-lab condition operation and have smaller analyzed material capacity per time. Laser-induced breakdown spectroscopy (LIBS), which is considered as a type of atomic emission spectroscopy is based on focusing a short laser pulse on the surface of the sample where the material is ablated via thermal or non-thermal mechanisms. A high density plasma is generated when the laser intensity reaches the breakdown threshold of ≈1010 Wcm−2. The optical emission is then collected and spectroscopically analyzed. It is the different energy levels of each atom that induces different and well quantized energies with narrowband emissions governed by Heisenberg uncertainty principle. When a material is ablated by a short laser pulse, the plume partly contains free electrons, excited atoms and ions, which are produced by strong ionization under an intense external electric field. The spatial free electron density i.e., plasma is produced by multiphoton absorption process where the electrons absorbing the energy of laser beam photons via inverse bremsstrahlung will result in a process known as avalanche or cascade of ionization. When the laser pulse terminates, a few microseconds later the plasma begin to slow down due to collisions with ambient gas and hence cool down during the expansion where the electrons of the atoms and ions at the excited electronic states decay to ground states, hence causing the plasma to emit light with discrete spectral peaks. Both the continuum and discrete line emission indicate the radiative relaxation of different energy levels of the ions and the neutral atoms and that the excited species are mainly generated by ion-electron recombination. The main features of discrete lines characterizing the material are: wavelength, intensity and the shape which depend on structure of the emitting atom and the environment. The plasma temperature can reach to thousands Kelvin. The major advantages of LIBS are: no sample preparation is required, fast, real-time measurements, depth profiling and spatial resolution and relatively simple set up compared to other conventional techniques    . LIBS has been widely used as a powerful technique for detection and monitoring of traces of elements of matter in all states of solid, liquid and gas in variety of applications such as industrial for metal detection in water   , analysis of minerals  , environmental for identification of bacteria   , and biomedical such as analysis of cancerous tissues   and dental caries   .
Interestingly, iron oxides are present in living organism where Fe (II) with an electron configuration of 1s2 2s2 2p6 3s2 3p6 3d6 4s is responsible for transport of oxygen and storage of oxygen by means of haemoglobin and myoglobin. Magnetite (Fe3O4) phase containing both Fe (II) and Fe (III) with electron configuration of 1s2 s2 2p6 3s2 3d5 has iron ions arranged in octahedral (Oh) and tetrahedral (Th) sites which form the face-cent red cubic unit cell in the presence of oxygen. The structure of this phase can be represented as [Fe]Th[Fe3+Fe2+] OhO4.
In the inverse spinel structure of magnetite, of the octahedral holes are occupied by Fe2+ ions while the Fe3+ ions are equally divided between of the tetrahedral holes and of the octahedral holes; electron spins of Fe3+ ions in
octahedral holes are aligned anti-parallel to those in tetrahedral holes while the Fe2+ ions tend to align their spins parallel with those of Fe3+ ions in adjacent octahedral sites, leading to a net magnetization. When the size of these nanoparticles becomes so small that their dimension can be considered as a single domain, they lack a hysteresis loop and a possession high field irreversibility, high saturation field and extra anisotropy contributions, it is called superparamagnetic nanoparticles (SPION)   . Since, SPIONs posses remarkable size and morphology-dependent physical and chemical properties and obey the Coulomb’s law and are easily controlled by an external magnetic field, thus they can be utilized for various biomedical applications such as hyperthermia   drug delivery   bioimaging   and as nanofluid for heat transfer   .
However, synthesis and characterization of high-quality magnetic iron oxide nanoparticles (MNPs) has always been a major issue particularly when dealing with ultrafine particles to control their size at the nanometric scale due to the high surface energy. In this view, the elemental analysis of NPs could be very beneficial particularly at the stage of synthesis of NPs such as the ratio of elements or where the information about physicochemical forms of an element are highly regarded because the redox state of an element in solution can be toxic, adsorption behavior, and transport mechanism. Information about various oxidation states of the elements in variety of objects is crucial and challenging in analytical chemistry as for instance, natural waters contain iron in oxidation states +2 and +3 and Fe III content is significantly higher than Fe II content and it strongly affects the solubility of iron in water   . Also, the iron oxide present in soil could form from an Fe II chloride solution in ambient temperatures and close to neutral pH values  and since oxidation leads to the formation of minerals such as goethite, an iron oxyhydroxide containing ferric iron, knowledge of Fe II: Fe III ration is important in this case. The commonly applied analytical methods are normally time consuming and require reagents and often produce inaccurate results in the presence of high Fe III content. Therefore, using LIBS as a non-contact, fast processing, and accurate technique would be desirable. Some researchers have employed LIBS to analyze the oxidation state of wine to monitor its quality  , to measure the concentration of magnetite (Fe3O4) nanoparticles (MNPs) in powder form in biocompatible matrix using Nd:YAG laser  , or to analyze the graphene nanoparticles using 800 nm laser  . To best of our knowledge, no results are reported in the literature regarding the application of LIBS to magnetite nanoferrofluid synthesized at elevated temperature using Nd:YAG laser second harmonic generation wavelength (532 nm). Following our recent reports on MNPs      , our goal in this work is to use pulse 532 nm laser to perform LIBS: 1) to investigate the emission lines of synthesized magnetite nanoferrofluid, 2) to determine the electron density and the temperature of dominant species, 3) to compare the elements ratio obtained at room and elevated temperatures, and 4) to study the effect of laser pulse frequency on elements ratio at different temperatures.
2. Materials and Methods
2.1. Synthesis of Nanoferrofluid
Solutions of 0.012 M ferric chloride hexahydrate (FeCl3∙6H2O, 99%, Merck) and 0.006 M ferrous sulfate heptahydrate (FeSO4∙7H2O, 99%, Merck) were prepared as iron source in double distilled water. The mixture of ferric and ferrous solution was then deoxygenated by bubbling N2 gas and sonicated for 30 minutes to prevent it from oxygenation. An amount of 0.9 M concentration of ammonia solution was used as alkaline source and vigorously stirred under N2 bubbling at room temperature. The mixture of ferric and ferrous solution was drop wisely added to ammonia solution and magnetically stirred at 1500 rpm for 30 min at ambient temperature. The colour of reaction mixture was immediately changed to back which is characteristic of a typical black nanoferrofluid material. The molar ratio of Fe2+:Fe3+ = 1:2 and complete precipitation of Fe3O4 between 7.5 - 14 pH was maintained under a non-oxidizing environment since this would critically affect the physical and chemical properties of the nanosized magnetic particles. The reaction temperature was kept constant at 25˚C for 1 hour then the black precipitated mixture was purified using magnetic separation 5 times and sedimented by centrifugation. The resultant material was dried by freeze dryer for 24 hours.
The crystalline phase of SPIONs was determined using x ray diffraction using Cu Kα radiation (XRD, λ = 1.5406 Å, FK60-40 X-ray diffractometer). Particle size and morphology of these nanoparticles were determined by transmission electron microscopy (TEM, Philips CM-200-FEG microscope, 120 KV). The saturation magnetization of samples was measured by vibrating sample magnetometer (VSM-PAR 155) at 300 K under magnetic field up to 8kOe. The plasma was analyzed using a spectrometer (UV-Vis USB 2000, Ocean Optics) equipped with 2048-pixel linear silicon CCD array (14 μm × 200 μm) with a detector range of 200 - 1100 nm, optical resolution of 0.1 - 10 nm FWHM connected to computer.
3. Results and Discussion
Figure 1(a) shows the TEM of synthesized SPIONs where polydispersity and some degree of aggregation, which mainly depends on the surface to volume ratio and augmentation of magnetic dipole-dipole interactions can be seen  . It is known that at room temperature,300 K, the critical size which defines the superparamagnetic region is equal to 27 nm  which in our case as seen in Figure 1(b), the NPs distribution covers a range between 4 - 22 nm with most dominant size at 12 nm which satisfies the above condition.
The chemical procedure for synthesis of Fe3O4 is as follow: The first step is the hydrolysis of iron salts as
In a reactive system, NH4OH can be written as
The interaction between the resultant materials in solid state will lead to the
Figure 1. (a) An example of TEM micrograph of MNPs; (b) Size distribution of synthesized. MNPs at room temperature (25˚C); and (c) At elevated temperature (70˚C).
formation of magnetite nanoparticles.
and the interaction of cations with the resultant anions from hydrolysis of iron salts gives
By increasing the solution of iron salt to NH4OH, the ions of Fe3+ will precipitate as ferrihydrite which is a hydrous ferric oxyhydroxide mineral and then react with Fe2+ to produce magnetite. It is noteworthy that Ostwald ripening effect (growth of larger particles at the expense of dissolving the smaller ones) and aggregation of NPs can affect the growth phenomena. The synthesis process that contains many intermediate products can influence the nucleation and growth phenomena. Once the alkaline species and iron salts are added together, some cations and anions such as , , and other intermediate products including , and are produced. The release and usage rate of these materials determine the rate of nucleation and growth of MNPs. Once the magnetite nuclei are produced, the ion-to-ion attachment, NPs aggregation or Ostwald ripening will cause the growth of NPs  . The XRD result of crystalline structure of MNPs is given in Figure 2(a) where it indicates the formation of highly purified magnetite phase of iron oxide with diffraction peaks at (111), (220), (311), (400), (422), (511), (440), (533), which are the
Figure 2. (a) XRD pattern of MNPs; (b) VSM results of samples S1 and S2 at (a) 25˚C and (b) 70˚C.
characteristic peaks of magnetite nanoparticles (Fe3O4) inverse spinel structure (JCPDS file No. 19-0629) without any interference with other phases of FexOy. Magnetic properties of MNPs were determined by VSM at room temperature. Figure 2(b) shows that the synthesized NPs exhibit strong magnetic properties of 63.1 emu/g. However, these values can be lower than the bulk material of magnetite due to breaking of a large number of exchange bonds at the surface   . Also, a complete reversibility of magnetization process confirms the superparamagnetic behaviour of prepared nanoparticles. Higher concentration of alkaline media increases the probability of non-magnetite layer production (i.e., magnetically dead layer) which in turn decreases the amount of saturation magnetization. Thus, there is a limit that alkaline media can have a positive effect on the saturation magnetization. Using Langevin’s equation and magnetic experimental data, the average magnetic particle diameter was calculated  :
where aM is the magnetic particle diameter, kB is Boltzmann’s constant, μ0 is the vacuum magnetic permeability, and Mb is magnetization of bulk magnetite.
Figure 3 shows the experimental setup where the output of Q-switched Nd:YAG laser (LG-LM1) operating at second harmonic generation wavelength of 532 nm with a pulse duration of 8 ns was focused by a l00 mm lens into a 500 μm spot size on the surface of nanoferrofluid. The generated plasma was then detected by a 600 μm core diameter optical fiber (LIBS-600-6-SR, Ocean Optics) connected to spectrometer (UV-Vis USB 2000, Ocean Optics) for analysis and the results were then displayed on computer. The pictures of LIBS at different laser energy were recorded by a fast CCD Camera (Panasonic Super Dynamic WV-CP450) connected to an optical microscope (Prior-UK). The examples of plasma recorded at different laser energy demonstrate that by increasing the input energy, the corresponding volume of plasma increases. Clearly, the emitted radiation in the visible spectral range originates from atomic and molecular
Figure 3. Experimental set up for LIBS.
electronic transitions. Therefore, the heavy particles of low temperature plasma, the neutrals and their ions basically characterize the observed colour of a plasma. In our case, a very bright white spark indicating the high temperature plasma with blue shade probably because of ionized oxygen present in the surrounding environment was observed.
It is well-known that the laser wavelength and the pulse energy influence the interaction process where the former governs the amount of energy absorbed by the material and the plasma-material interaction and the latter, mainly controls the amount of ablated mas and the ablation rate. The plasma formation by nanosecond pulses as in our case seen in Figure 3, is determined by two major mechanisms of inverse Bremsstrahlung where the free electrons absorb energy the laser beam during a collision and increases their kinetic energy. It is responsible for heating the plasma electrons and secondly, the photoionization where the excited species and the ground state atoms are excited with high energies. As seen in Figure 4(a), the main transitions at 860 mJ are Fe I, Fe II, Cl III, O II, Cl I, OV and the amplitude of the peaks are relatively small and by increasing the pulse number not a significant difference was observed. A significant saturation has occurred at excitation wavelength and some lines are very close to other transitions such as OV at 654.37 nm to that of hydrogen alpha at 656.28 nm, which might be due to hydrogen in nanoferrofluid. However, the line shown by spectrophotometer software is compared and denoted to the closest with minimum difference. At 1050 mJ, the transition lines are the same but the peaks between 250 - 280 nm seem to be more differentiated as delineated by circle in Figure 4(b). This is mainly because at higher temperatures, the ablated material dissociates (i.e., breaks down) into excited ions and atomic species, the amplitude of lines is also slightly increased. At even higher pulse energy of 1260 mJ,
Figure 4. LIBS spectra at different laser pulse energy and pulse number.
the first pulse caused much stronger lines amplitudes but then again are reduced at the consequent pulses, Figure 4(c). In principle, the loss of line intensity is mainly explained by the interaction of trailing portion of pulse with the plume and ejected particles which acts a shield hence a considerable fraction of energy is attenuated. The Fe III emission lines lie between 400 - 500 nm, which are due to 5D-3P, 5D-3F, 5D-3G and 5D-3H transitions among levels of the 3d6 configuration (i.e., once it loses three electrons to form 3+, it becomes 1s2s2 2p63s2 3p63d5). These close emissions are mainly: He I (400.92 nm), Fe III (400.83 nm), O II (466.16 nm), Fe III (465.80 nm), Fe III (527.04 nm), and Fe II (527.33 nm). Thus, although there are some Fe III lines, only limited and most probable observable lines are indicated in the Figure 4 due to lack of sufficient space. The details of spectra analysis are given in Table 1  .
The central wavelength of line emission λ0 is defined by the photon energy difference ΔE = E2 − E1 where E1 and E2 correspond to the energy of photons at longer and shorter detected wavelengths respectively (i.e., lower and upper energy levels):
where h ≈ (6.63 × 10−34 Js) is Planck constant Js and c = 3 × 108 ms−1 is the speed
Table 1. Spectral data of identified elements during LIBS.
of light. Since, the energy of a transition is characteristic of an element thus, the central wavelength represents the radiating species. The line intensity is quantified by the line emission coefficient in the unit of W (m2 sr)−1.
where n(p) is the species population or density in the excited state, A21 is the transition probability and 4π represents the solid angle dΩ i.e., isotropic radiation, measured in sr. Since, the line profile correlates the emission coefficient with the spectral line emission coefficient we have:
The full width half maximum (FWHM) of the intensity is a characteristic of the line profile, which in turn depends on the broadening mechanism such as Doppler broadening where the profile has a Gaussian shape.
Since, the evaluation of plasma temperature is difficult due to its short lifetime and partly because of high temperature but one can use the Saha-Boltzmann equation for two or more consecutive spectral lines of the same species where the maximum and minimum amplitudes for each element can be used to measure the corresponding temperature 
Using the Equation (8) to determine ΔE and the values given in Table 1 are substituted in Equation (11) to determine the temperature of the species. Using the spectral data at 1050 mJ as the medium energy value, the plasma temperature was determined between 3881 ± 200 K (or 0.33 eV) for Fe I and the maximum of 26,047 ± 200 K (or 2.24 eV) for OV elements. Assuming that all the laser pulse energy is expended for ionization of MNPs, the number of free electrons corresponding to the element can be approximately calculated as Ne− ≈ (Ep /Pi)/eV where Ep and Pi are laser pulse energy and the element ionization potential respectively  . The results in Figure 5 show a range of electron density spanning between (0.47 - 6.80) × 1017, (0.58 - 8.30) × 1017 and (0.69 - 9.96) × 1017 at 860, 1050 and 1260 mJ respectively where in each range OV and Fe I correspond to highest and lowest values. However, a rough estimate of the plasma volume ≈ 1000, 1500 and 2000 μm3 can be obtained from magnification of pictures taken by CCD camera in Figure 1, which effectively changes the electrons number to an order of 1020 cm−3 in each case.
Generally, the angular distribution of the ablated material from the target can be represented by the equation I(θ) = I0cosnθ, where I(θ) is the flux intensity along a direction forming an angle θ with the normal to the target surface. I0 is initial intensity corresponding to θ = 0 and n is parameter related to the anisotropy of the distribution. When n = 1, it corresponds to a perfectly spherical distribution and n > 1 implies a more directional plume shape as in our case in
Figure 5. Calculated electron density of elements as a function of ionization potential.
Figure 1 particularly at higher energy levels. Indeed, the plasma plume is a combination of hot electrons, colder heavier ions and neutral atoms, which results in an inhomogeneous temperature distribution. The interaction of hot plasma with cold region at the interface with the surrounding colder air may destroy the shape of the emitted lines which can appear as a decrease of the spectral line intensity and increase of the width of line at FWHM   . Also, upon rapid expansion of the plasma both its temperature and recoil pressure decrease which reduce the rate of heat transfer to the surface. Also, according to Serov and Richardson  an intense magnetic field due to laser-induced plasma displays a 1/r2 radial dependence and that such fields (109 Wcm−2) would significantly affect the electron thermal conduction process. Thus, it has a crucial role in determining the temperature and density distribution within the plasma. It is also interesting to note that the plasma temperature and hence its possible consequent effects can influence the quality of the material surface during the interaction process. For example, the high pressure exerted on the surface by the plasma plume can suppress the vaporization of the material by raising the boiling point of the material well beyond its normal value and as the plasma expands it generates an impulse reaction on the surface thus, radiates part of its thermal energy back to the surface. Consequently, a high temperature is maintained for a short time after the laser pulse has ended. It is expected that the high temperature is transferred to the surface by high velocity electrons via both direct conductivity and radiative transfer   .
Figure 6 demonstrates the variation of elements ratio with energy for different temperatures at constant frequency of 1Hz where the higher pulse numbers produced a ratio with higher signal intensity. A characteristic behaviour of the results is their repeating systematic zig-zag feature at almost the same values of energy. The ratio of elements for samples synthesized at room temperature (S1) is higher than those at elevated temperature (S2) except in the case of Fe III/OII where the results are almost superimposed.
At higher frequency (3 Hz) shown in Figure 7, the ratio of elements remained
Figure 6. Variation of elements ratio for S1 and S2 with pulse energy at 1 Hz.
Figure 7. Variation of elements ratio for S1 and S2 with pulse energy at 3 Hz.
almost at the same values for S1 samples but it was increased from an average value of 0.4 to 0.6 for Fe II/Fe III and Fe II/ Fe/OII. Again, the similar behaviour was obtained in the case of Fe III/OII i.e., no significant difference. Interesting enough, the ratio for Fe II/Fe II + Fe III was reduced from 1.4 to 0.6 which is very close to desirable value. A similar to results to (3 Hz) was obtained at 6 Hz for both S1 and S2 at 6 Hz shown in Figure 8.
The highlights of the above observations are summarized as plots shown in Figure 9 where it demonstrates that at constant pulse energy (i.e., 860 mJ), the ratio increases with frequency for both samples of S1 and S2 with 1.1 lowest at 1 Hz and highest about 1.3 at 3 Hz for S1 (Figure 9(a1)) and 0.2 at 1 Hz and 0.6 at 3 Hz for S2 (Figure 9(a2)) respectively. As in the case of Fe II/Fe II + Fe III, significantly lower values were achieved for both S1 and S2 i.e., about 0.55 for S1 (Figure 9(b1)), and 0.1 at 1 Hz for S1 (Figure 9(b2)) and about 0.3 at 3 and 6 Hz respectively (Figure 9(b2)). Despite the similar behaviour at higher energy (1260 mJ), a noticeable change was observed at 6 Hz both for S1 and S2 (Figure 9(c2) and Figure 9(d2)) where the ratio showed an increasing trend up 1 and 0.5 for S1 and S2 respectively whereas in the case of 860 mJ it reached the plateau at 6 Hz indicating that at higher pulse energy more transition lines corresponding to Fe II is produced. However, no firm comment can be made about its linear extrapolation as the experiment was not performed beyond this value. It is suggested for the future research to consider the effects of temperature and plasma formation on the quality of nanoferrofluid composition to ensure its
Figure 8. Variation of elements ratio for S1 and S2 with pulse energy at 6 Hz.
Figure 9. Variation of Fe II:Fe III and Fe II:Fe II + Fe III for S1 and S2 with frequency at different energy.
structural integrity with respect to its applications to avoid any possible damage.
Elemental analysis of magnetite (Fe3O4) nanoferrofluid was investigated at two different temperatures using 532 nm-based LIBS technique. Using Saha-Boltzmann relation, the transition lines temperature and the corresponding electron density were calculated. It was shown that the constituting elements and their ratio can be identified depending on the synthesizing and optical conditions. The most important outcome of the experiment is that the elemental ratio is significantly lower at elevated temperature than those at room temperature at all laser energy. Equally, at higher frequency and room temperature no considerable change in the ratio was observed, however, at elevated temperature and 860 mJ, the ratio was increased and reached its maximum value at 3 Hz. This was not observed at 1260 mJ. One practical application where such an approach can be beneficial and plays a key role in number of scientific, engineering and biomedical applications is when one requires to determine and adjust the ratio of Fe II:Fe III or Fe II/Fe II + Fe III during the synthesis process.
The authors are thankful to AUT for the support and Faculty Funding for this research work.