- Email: [email protected]
Total-reflection X-ray fluorescence: An alternative tool for the analysis of magnetic ferrofluids
Spectrochimica Acta Part B 63 (2008) 1387–1394
Contents lists available at ScienceDirect
Spectrochimica Acta Part B j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s a b
Total-reflection X-ray fluorescence: An alternative tool for the analysis of magnetic ferrofluids☆ R. Fernández-Ruiz a,⁎, R. Costo b, M.P. Morales b, O. Bomatí-Miguel b, S. Veintemillas-Verdaguer b a b
Servicio Interdepartamental de Investigación (SIdI), Facultad de Ciencias, Universidad Autónoma de Madrid, Cantoblanco, E-28049, Madrid, Spain Instituto de Ciencia de Materiales de Madrid, CSIC, Cantoblanco, E-28049, Madrid, Spain
a r t i c l e
i n f o
Article history: Received 8 August 2007 Accepted 3 October 2008 Available online 22 October 2008 Keywords: TXRF Medical image Iron ferrofluids contrast agent AD-TXRF Nanoparticles
a b s t r a c t This work presents the first application of the total-reflection X-ray fluorescence (TXRF) to the compositional study of magnetic ferrofluids. With the aims of validating the best analytical conditions and also, limitations of the TXRF in the compositional study of these materials, an alternative empirical method, based in the use of angle-dependence TXRF (AD-TXRF) measurements, is proposed. Three kinds of ferromagnetic nanoparticles, with different morphologies, have been studied. The techniques of inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma optical emission spectroscopy (ICP-OES) have been used to validate the TXRF results. In contrast with the plasma techniques, the developed TXRF procedure need not of previous chemical acid digestion. Additionally, two procedures of magnetic nanoparticles synthesis, coprecipitation and laser-pyrolysis, have been checked for the contaminants trace metals Zn, Mn and Cr. It has been found that the method of laser-pyrolysis produces nanoparticles of higher purity. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Suspensions of magnetic nanoparticles in aqueous and nonaqueous media are nowadays attracting much interest because of their potential for many applications such as biomedicine [1], including biomolecules separation, magnetic resonance imaging and drug delivery and also as recording media, by controlling the evaporation process leading to the self assembly of the nanoparticles into ordered regions, i.e. superlattices [2]. To optimise the suitability of a nanoparticle suspension for any specific application, the samples must be synthesized and characterized reproducibly. So, the nature, the size and the elemental composition of the magnetic nanoparticles must be carefully controlled and analysed. The best strategy to synthesize magnetic nanoparticles such as iron, cobalt, iron-platinum and iron oxide has been recently developed by decomposition of organometallic precursors in the presence of a surfactant [3]. The nanoparticles are isolated and stabilized in the solution by capping groups such as long-chain alkyl surfactants which keep the particle surface unreactive and repulsive toward other particles [4]. It is difficult to find out the chemical composition of such suspensions of nanoparticles as it is difficult to remove and replace ☆ This paper was presented at the 12th Conference on Total Reflection X-ray Fluorescence Analysis and Related Methods held in Trento (Italy), 18-22 June 2007, and is published in the Special Issue of Spectrochimica Acta Part B, dedicated to that conference. ⁎ Corresponding author. Universidad Autónoma de Madrid, Facultad de Ciencias, Servicio Interdepartamental de Investigación, Modulo C-9, Laboratorio de TXRF, Crta. Colmenar, Km 15. Cantoblanco-28049-Madrid, Spain. Fax: +34 914973529. E-mail address: [email protected] (R. Fernández-Ruiz). 0584-8547/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.sab.2008.10.017
such ligands for chemical analysis. This point is important for example in the case of iron oxide nanoparticle suspensions for intravenous injection. Total human iron stores amount of 3500 mg, while normal liver contains approximately 0.2 mg of iron/g wet weight of liver and chronic iron toxicity (cirrhosis) develops only after the liver iron concentration exceeds 4 mg Fe/g liver wet weight. The total amount of iron oxide proposed for diagnostic imaging is small compared to the normal iron store, that is 1-2 mg Fe/Kg bodyweight [5]. It is well-known that iron oxides are ferromagnetic and therefore, sensitive to the action of external magnetic fields. So, in the presence of a magnetic field, the constituent magnetic dipoles of the nanoparticles are aligned in the direction of the external field. On the other hand, the organic covering of dextrane or albumen is useful and necessary when these nanoparticles are used as image contrast agents in live organisms, because they prevent that the antibodies of the organism detects the presence of the anomalous magnetic nanoparticles in the organism. One of the main future research goals will be to obtain the functionalization of the organic core with specific radicals for each one of the cellular kinds that exist in an living organism. So, it would be possible to obtain selective contrast agents for imaging and treatment of any affected organ by a cancer. These features convert to these materials in a new precise diagnostic agents for tumors and for their treatment. The treatment would be possible by means of the heat liberation that takes place in the hysteresis cycle of the ferromagnetic nanoparticles when they are introduced under the effects of alternate external magnetic fields. Iron concentrations in suspensions, which are used as contrast agents for medical NMR imaging, are usually determined by
1388
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
spectroscopic techniques such as AAS or ICPS, requiring digestion of the samples in acids of high concentration. This fact requires considerable amount of time as well as high cost of the chemicals. TXRF can help to minimize the cost by means of the simple, precise and accurate procedure developed in this work. So, a research program to optimize the application of the TXRF in this materials was initiated. The study was centered in the following aspects. First, the particles average sizes of solid deposition were measured by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) to check the real morphologies of three different kinds of magnetic iron nanoparticles. Second, the method of deposition was evaluated for the following aspects: the kind of sample carrier; quartz and methacrylate, and the deposition procedure; heat and vacuum. Third, Fe AD-TXRF intensity measurements were obtained. The evaluation of the AD-TXRF profiles allows us to define the empirical conditions to validate the best TXRF analytical parameters for this kind of materials and, by extrapolation, for any other. Additionally, the behaviour of the Fe signal with respect to the sample concentration was evaluated. The purpose of this study was to obtain the lineal range in the Fe measurements by TXRF. Finally, the procedure was validated by comparison with ICP-MS and ICP-AES techniques for different sample morphologies and for two different synthesis procedures of the iron nanoparticles. TXRF main features and potential applications can be found in the illustrative paper of Prange [6] or the excellent book of Klockenkämper [7]. The previous works, carried out by Fernández-Ruiz et al. [8– 11], related with the direct quantitation of solid samples by means of the suspension method, were the point of departure for this investigation. In this way, the best approach to the ideal thin film condition for the deposition of solid particles [12] is that the following three fundamental requirements are fulfilled: (1) chemical homogeneity of the solid particles, (2) average particle sizes around 1 μm [13], and (3) homogeneous spatial distribution of the deposited particles on the sample carrier. Recently, an anomalous behavior for some elements in archaeological ceramics, with a complicated mineral composition, has been found by mean of AD-TXRF intensity profiles [14]. Therefore, the evolution of the Fe AD-TXRF profiles was also performed with the aims of checking the behaviour of the iron signal with respect to the sample mass transferred to the sample carrier. 2. Experimental 2.1. Instrumentation The TXRF measurements were made using an Atomika 8030C TXRF spectrometer (Cameca, Germany) equipped with a 3 kW Mo/W dual target X-ray tube and a W/C double monochromator multilayer. The instrument is also equipped for variation of incident angle of the Xrays by tilting the sample holder systematically. A Si(Li) detector with an active area of 80 mm2 and a resolution of 150 eV at 5.9 keV (Mn Kα) was used for detection and measurement of X-rays produced. TXRF
measurements were made with the Mo Kα excitation condition by means of the adequate arrangement of the monochromator geometry. The SEM instrument used in this work was the Philips XL-30 equipped with a W source, detectors of secondary and backscattering electrons and a vacuum sample chamber with a vacuum of lower than 4 × 10− 4 Pa. The TEM instrument used in this work was a Jeol Jem 1010 of 100 kV (Tokyo, Japan) equipped with a Gatan digital camera model BioScan (Pleasanton, Canada). Finally, ICP-MS and ICP-OES instruments used were ELAN-6000 and PE-5500 respectively from Perkin Elemer Sciex (Ontario, Canada). 2.2. Sample preparation 2.2.1. Morphology of the samples Two kinds of aqueous colloidal suspensions of magnetic iron oxide nanoparticles (BKS25 and BKS31), synthesized by co-precipitation, and other synthesized by laser induced pyrolysis (TEOS16) [15], were evaluated. The samples BKS25 and BKS31 were an iron oxide nucleus covered by an organic layer of dextran and albumen respectively. The sample TEOS16 was an iron oxide nucleus covered by a quartz layer and, additionally, an organic matter layer. Fig. 1 shows the three different morphologies. 2.2.2. Sample preparation for TXRF analysis The quantitation procedure of the ferrofluid samples by TXRF was as follow: 100 μL of sample, 100 μL of Co with a certified concentration of 1000 ng/μL (Merck, Germany) and 1000 μL of Milli-Q water were added to a vial. After agitation, 2 μL of the dissolution were deposited on a quartz sample carrier. By mean of a ceramic hot plate, the samples were dried. All the process was accomplished in a class A-100 chamber. The consumed time was no more than 15 min/sample. Additionally and with the aims of evaluating the best condition related to the spatial distribution of the deposited nanoparticles on the sample carrier, a statistical method, based on the reproducibility of the Fe analytical values, was designed. Two different sample-carriers, quartz and methacrylate, and two different evaporation procedures, heat and vacuum, were investigated. Five microlitres of one representative colloidal ferrofluid (BKS25), previously diluted with water and mixed with cobalt internal standard, in such a way that the concentration of internal standard was 50 ng/μL, were deposited on five different sample carriers five times each one. In this way, the samples were analyzed by TXRF. An additional experiment to check the linearity in the quantitative Fe measurements was carried out. Ten dissolutions of the same sample (BKS31) with the approximate sample concentrations; 9000 ng/μL, 4500 ng/μL, 2250 ng/μL, 1000 ng/μL, 360 ng/μL, 180 ng/ μL, 120 ng/μL, 90 ng/μL, 18 ng/μL and 9 ng/μL, were prepared by means of successive dilutions. After a strong agitation, two microlitres of the dissolutions were deposited on a quartz sample-carrier and their TXRF spectra were acquired during 1000 s each one. The AD-TXRF measurements were made as follow: first, a diluted colloidal suspension of the sample BKS31 was prepared. The obtained
Fig. 1. Morphology of nanoparticles BKS25, BKS31 (synthesized by co-precipitation) and TEOS16 (synthesized by laser induced pyrolysis).
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
nominal sample concentration was around 9000 ng/μL. Then, the sample was diluted with ultrapure water (Milli-Q, 18.2 MΩ) to the following factors: 1, 2, 25, 50, 75, 100, 500 and 1000. Afterwards, two microlitres of each dissolution were transferred over the quartz sample-carrier, drying with a ceramic plate. As result, the following iron masses were transferred: 18,000 ng, 9000 ng, 720 ng, 360 ng, 240 ng, 180 ng, 36 ng, 18 ng. In this way, AD-TXRF measurements for each one of the depositions were carried out. The incidence angle was varied from 1.0 to 4.5 mrad, around the critical angle for quartz. Mo Kα excitation was used in this experiment. 2.2.3. Sample preparation for ICP-MS and ICP-OES analysis An acid mix of HNO3:HF:H2O2:HCl in the proportions 3:3:1:1, for the TEOS samples, or an acid mix of HNO3:HCl in the proportion 4:4, for the BKS samples, together with 1000 μL of ferrofluid sample were introduced in a teflon reactor. The samples were digested in a Millestone EthosSel microwave oven (Sorisole, Italy). A constant temperature of 20 min to 200 °C was programmed [17]. Afterwards, the digested samples were rise up to one liter with ultra-pure water and spiked with 500 μL of Rh with a certified concentration of 1000 ng/μL (Merck, Germany) for instrumental drifts correction. An adequate calibration curve, with five concentration points of Fe certified standard (Merck, Germany), was carried out before of iron quantification. In this case, the consumed time was about 2 h/sample. 3. Results and discussion 3.1. Morphology of the nanoparticles studied The samples were evaluated by SEM and TEM to check the real morphology of the nanoparticles. Fig. 2 shows the TEM/SEM images obtained for the BKS25, BKS31 and TEOS16 samples.
1389
The use of TEM persecuted the vision and measure of the internal structure of the nanoparticles taking advantage of the different mass absorption of the electrons by the matter. In our case, samples BKS25 (Fig. 2a) and BKS31 (Fig. 2b) do not present internal structure, when owing to their morphology the difference between the magnetite nucleus and the organic layer should be perceptible. This fact can be due to volatilization of the organic layer due to the thermal action of the electron beam of TEM on the sample. Ramifications, probably due to the magnetic interaction among nanoparticles, are observed. This observation could imply the presence of an absorption effect in the quantitative analysis of the samples by TXRF. Nevertheless, the agglomeration magnetic effect is less accused in sample TEOS16 due to the magnetic screening induced by quartz layer which is clearly visible in the Fig. 2d. Fig. 2c and d shows the SEM and TEM images of the TEOS16 nanoparticles. In this case, the SEM image give us information about the average sizes of the organic layer and TEM image the average sizes of the internal structure of quartz layer and magnetite nucleus. By means of the ImageTools v.3.0 software package (Texas University, Health Science Center, San Antonio, USA) the particle sizes of the nanoparticle were evaluated. For statistical analysis a sampling order around 50 was used. The normalized probability density graphs and its associated probability curve are showed in Fig. 3. The analysis of the size distribution of sample BKS25 is displayed in Fig. 3a. In this case, the greater density of particles has a mean spherical size of 4.6 ± 0.6 nm, which can be associated to the average diameter of its magnetic nucleus. The analysis of the size distribution of sample BKS31 is displayed in Fig. 3b. In this case, the average diameter of its magnetic nucleus has a mean spherical size of 5.6 ± 1.1 nm. On the other hand, for the TEOS16 sample, a deeper study was carried out. The measurements of the magnetite nucleus and the
Fig. 2. TEM images of (a) BKS25, (b) BKS31, and (c) SEM and (d) TEM images for TEOS16.
1390
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
Fig. 3. Probability density graphs and probability curve associated to the size distributions found in the TEM and SEM images for (a) BKS25, (b) BKS31 and (c) TEOS16.
layer of quartz were made from the images of TEM (Fig. 2d). Additionally, the sizes of the organic layer were made from the SEM image (Fig. 2c). We can observe that, the organic matter is eliminated in the process of measurement by TEM. This fact can be deduced because the average diameters measured by TEM are much smaller than the measured ones by SEM. The effect is showed in Fig. 3c. The average particle sizes obtained shows a spherical organic layer with an average diameter of 78 ± 7 nm. The quartz layer shows a spherical structure with an average diameter of 40 ± 6 nm. Finally, the magnetite nucleus also shows a spherical structure of average diameter 8.8 ± 1.3 nm. With these particle size distributions, the first and second fundamental requirement, mentioned in the introduction, were assured. So, the quantification by TXRF could be done in an adequate way. 3.2. Homogeneity of the depositions The quantitative results associated to the homogeneity study are showed in Table 1. Fig. 4 shows the three characteristic deposition structures obtained by observation with binocular. The evaluation of the statistical parameters (Table 1), combined with the visual inspection of the depositions (Fig. 4), allows us to conclude that the best deposition procedures are quartz, as samplecarrier, and heat or vacuum, as drying method. It can be observed that the increment in the roughness of the deposition is correlated with the increment of the associated uncertainty, but do not with variations
of the nominal value, or accuracy, of the iron concentration. The simple change of the sample-carrier reduce the coefficient of variation from 9.6% for methacrylate to 6.6% for quartz, whereas the iron concentration is analytically the same, 49.4 ± 4.7 ng/μL for methacrylate and 45.2 ± 3.9 ng/μL for quartz. Finally, an SEM-EDAX mapping for Fe and Co, used as reference, was obtained. Five microliters for each one of the ferrofluid particles suspensions were deposited on a quartz sample carrier. The sample carrier was dried on a ceramic hot plate and afterwards, it was analysed by SEM-EDAX, after gold metalisation. Fig. 5 shows the chemical homogeneity of the depositions with respect to Fe as main component and Co as internal standard.
Table 1 Results of the quantitative analysis of sample depositions Deposition
QH (ng/μL)
MV (ng/μL)
QV (ng/μL)
D1 D2 D3 D4 D5 Mean (ng/μL) σ (ng/μL)a CV (%)b
42.64 45.70 45.20 40.70 48.80 44.6 3.1 6.9
45.70 44.21 53.27 55.18 48.54 49.4 4.7 9.6
48.20 45.10 43.10 48.20 41.60 45.2 3.0 6.6
a Standard deviation, bCoefficient of Variation. Q and M indicate Quartz and Methacrylate as sample-carriers respectively and H and V indicate Heat and Vacuum as evaporation procedure respectively.
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
1391
Fig. 4. Binocular images of the three different depositions obtained in the study of the deposition procedure. (a) 5 μL of diluted BKS25 over quartz and dried with heat. (b) 5 μL of diluted BKS25 over methacrylate and dried with vacuum. (c) 5 μL of diluted BKS25 over quartz and dried with vacuum.
Fig. 5. SEM-EDAX mapping of one representative (5 μL of BKS25 colloidal sample deposited over a quartz sample support) deposition of the magnetic ferrofluids analyzed by TXRF.
3.3. AD-TXRF measurements After selecting the best deposition method i.e. heating the sample on quartz sample supports, an AD-TXRF study was designed to validate the best analytical conditions for TXRF [16]. Fig. 6 shows the
evolution of the profiles, associated to the Fe K-lines, with respect to the mass of sample transferred over the quartz reflector. Two empirical parameters can be evaluated, one, the angular position of the maximum intensity (αmax) and another one, the intensity ratio between the constant intensity, after the critical angle,
Fig. 6. Evolution of the normalized AD-TXRF intensity profiles, associated to the Fe K-lines, with respect to the mass of sample transferred over the sample carrier.
1392
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
and the maximum intensity, before the critical angle (Imin/Imax). Fig. 7 presents the behaviors found for these empirical parameters with respect to the sample mass transferred over the sample-carrier. Fig. 7a shows that αmax evolve with the sample mass transferred. In this case, if masses are lower than 700 ng, αmax value is approximately constant and it is lower that the critical angle used in the quantitative TXRF analysis. On the other hand, if masses are higher of 700 ng, αmax values increase and are higher of the critical angle used for quantitative TXRF analysis, which is 3 mrad in our case. If we keep in mind that the ideal intensity relation between XRF and TXRF is 1:2 for thin layer depositions [7], the results obtained in Fig. 7b can be used to validate the deposited mass limits, that is, when the TXRF condition is achieved and when it is lose. In this case, the interpolation between the spline curve of the experimental results and the line which defines de ideal condition Imin/Imax = 0.5 for the TXRF, gives us the top sample mass that can be deposited over the sample carrier to achieve the ideal TXRF condition. As we can observe in Fig. 7b, this value is found around 2500 ng for each one of the nanoparticle ferrofluids evaluated. Fig. 8 show the behavior of the Fe K-lines net intensity with respect to the nominal sample concentrations of the suspensions. As we can see, the sample concentration where the linear behavior is lost, it is finding around of 1200 ng/μL. This agree with our previous sample mass limit (2500 ng) associated with Imin/Imax = 0.5. As conclusion, the optimum region of concentration for the analysis of this kind of materials has been found for a sample mass transferred on sample carrier lower than 2500 ng. Therefore, the methodology proposed in this section of the work seems to be valid for the optimization of the best analytical conditions for the quantitative analysis by TXRF, independently of the kind of sample analyzed.
3.4. Validation of the analysis by TXRF
Fig. 7. Change in empirical parameters (a) αmax and (b) Imin/Imax with sample mass on the sample carrier for Fe K-lines in the sample BKS31.
Once optimized the sample preparation conditions and with the aims of validate the Fe quantitative measurements and additionally, the correctness of the developed procedure by TXRF, several kind of ferrofluids nanoparticles were compared by mean of ICP-MS and ICPOES measurements.
Fig. 8. Change in Fe K-lines net intensity with respect to nominal sample concentrations (CS) in suspensions of nanoparticle BKS31.
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394 Table 2 Analytical results and statistical parameters for Fe concentration obtained by TXRF, ICPMS and ICP-OES in the validation process for different BKS and TEOS samples Sample
TXRF ng/μL ± σn = 5 / (CV)
ICP-MS ng/μL ± σn = 5 / (CV)
ICP-AES ng/μL ± σn = 3 / (CV)
BKS25 BKS31 TEOS16 P903 (BKS) G1 (BKS) B2 (TEOS)
6930 ± 456 / (6.59%) 8944 ± 295 / (3.30%) 75.6 ± 5.3 / (6.94%) 22582 ± 1364 / (6.04%) 26940 ± 1472 / (5.46%) 3622 ± 232 / (6.41%)
6790 ± 356 / (5.24%) 7546 ± 543 / (7.19%) 82.2 ± 3.6 / (4.38%) -
24375 ± 1864 / (7.65%) 25600 ± 1986 / (7.76%) 3325 ± 354 / (10.6%)
The first notable difference was found in the sample preparation time, 15 min/sample by TXRF compared with 2 h/sample for both ICPS techniques. On the other hand, the differences in materials and reagents also were considerable. From a more scientific point of view, the analytical results and the more significant statistical parameters, for Fe concentration, obtained by TXRF, ICP-MS and ICP-OES, are presented in Table 2. The nominal values of Fe concentrations obtained for the six kinds of samples analyzed were analytically equivalents within error bars. With respect to the associated uncertainties, the TXRF values are more precise, between 3.30% and 6.94% of coefficient of variation, while the ICP-MS or ICP-OES measurements were found between 5.24% and 10.6% of coefficient of variation. As conclusion, the TXRF seems a more reproducible technique for this kind of materials. On the other hand, it is demonstrated that the morphological differences in the evaluated nanoparticle ferrofluids have not influence in the Fe analytical value. This fact was not obvious because the different layer of organic matter (BKS) or of quartz and organic matter (TEOS), together to the agglomeration magnetic effect, could produce an absorption effect which could imply a constant drift in the concentration value of the Fe. In this point of the research, the TXRF has proven to be an outstanding alternative method for the iron quantification in ferrofluids. The final developed TXRF methodology has several important advantages with respect to the ICPS techniques, i.e. simplicity, low cost, low time consuming, precise and accurate results. 3.5. Impurities analysis by TXRF Finally, a comparative study, of the trace contaminants induced in the synthesis process, was carried out. The BKS samples were synthesized by the co-precipitation method and the TEOS sample
1393
Table 3 A comparison of trace elements analysis of ferrofluids obtained by co-precipitation and laser pyrolysis synthesis procedures Co-precipitation synthesis
Elements Zn Mn Cr a
Pyrolysis laser synthesis
BKS25 (ng/μL ± σn = 5)
BKS31 (ng/μL ± σn = 5)
TEOS31 (ng/μL ± σn = 5)
TXRF 1.35 ± 0.08 37.0 ± 2.6 43.8 ± 3.7
TXRF b1 43.2 ± 1.9 65.4 ± 2.9
TXRF 1.29 ± 0.04 b0.3 b0.2
ICP-MS n.d.a 44.8 ± 3.5 49.4 ± 3.2
ICP-MS n.d. 31.6 ± 4.2 45.2 ± 6.1
ICP-MS 1.20 ± 0.05 0.08 ± 0.01 n.d.
Not detected.
was synthesized by the laser pyrolysis method. In both cases, the qualitative analysis shows the presence of the following traces elements; Si, Cl, K, Cr, Mn, Zn, Pb and Br, such as Fig. 9 shows. We can see in Table 3 that only the elements Mn, Cr and Zn were quantified simultaneously by TXRF and ICP-MS due to comparative reasons. The other elements detected in the qualitative analysis were not analyzed for several reasons; first by time consumption in the analysis by ICP-MS and second, because the detection limits for some of the elements with health interest, as Pb, had a concentration lower than the quantification limit for both TXRF and ICP-MS techniques. Table 3 shows that the pyrolysis laser synthesis method is a cleaner procedure than the co-precipitation synthesis method. This fact has an important relevance from the medical point of view due to its intra-venous use as contrast agent in the diagnostic by means of NMR imaging. 4. Conclusions This work presents the first application of the TXRF technique to the compositional analysis of nanoparticle ferrofluids which are important materials used in the medical image improvements and have a high potential for its application for treatment of the cancer in a near future. The TXRF conditions for such applications have been optimized. The AD-TXRF method has been used for first time to optimize the TXRF conditions for analysis of such materials. This technique of optimization of TXRF conditions is also equally valid for other similar materials. The roughness of the depositions used for TXRF measurements has been found to affect the precision of measurements. However, it does not affect the accuracy of the measurements. One of the main goals achieved in this work has been that the TXRF procedure do not need previous acid digestion so that, it is quick (15 min/sample), simply and not expensive (only dilution and addition of Co as internal standard). Additionally, the quantitative results have an adequate uncertainty level for the ferrofluids evaluated in this work (CV lower than 7%). Finally, these results show that TXRF is an advantage quantitative alternative with respect to ICP-MS or ICP-AES in the analysis of ferrofluids. The trace analysis results indicate that the pyrolysis laser synthesis procedure obtain nanoparticles of highest purity than the co-precipitation synthesis procedure, which implies relevant medical consequences. Acknowledgements
Fig. 9. Representative TXRF spectrum of the sample TEOS16. Intensity scale is normalized and represented in logaritmic units.
Authors would like to thank Esperanza Salvador and Francisco Urbano, from SIdI-UAM laboratories, for their expert advice on the analysis by SEM and TEM respectively. Also, they wish to express their gratitude to Liquids Research Limited, North Wales, UK for supply BKS25 and BKS31 samples and to Autonomous Community of Madrid (S-0505/MAT/0194) and Spanish Ministers of Education (MAT200503179, NAN2004-08805-C04-01) for financial support. Additionally, the authors express their gratitude to Dirección General de Investigación from Comunidad de Madrid for TXRF instrumentation financial support. Finally, the authors wish to express their gratitude to Lola for
1394
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
her English revision of the manuscript and to the reviewer for their suggestions for the clarity of the manuscript. References [1] P. Tartaj, María del Puerto Morales, S. Veintemillas-Verdaguer, T. GonzálezCarreño, C.J. Serna, The preparation of magnetic nanoparticles for applications in biomedicine, J. Phys. D 36 (2003) R182. [2] C. Shouheng Sun, B. Murray, Dieter Weller, Liesl Folks, Andreas Moser, Monodisperse FePt nanoparticles and ferromagnetic FePt nanocrystal superlattices, Science 287 (2000) 1989–1992. [3] J. Park, Kwangjin An, Yosun Hwang, Je-Geun Park, Han-Jin Noh, Jae-Young Kim, JaeHoon Park, Nong-Moon Hwang, Taeghwan Hyeon, Ultra-large-scale syntheses of monodisperse nanocrystals, Nat. Mater. 3 (2004) 891. [4] A.L. Willis, N.J. Turro, S. O'Brien, Spectroscopic characterization of the surface of iron oxide nanocrystals, Chem. Mater. 17 (24) (2005) 5970–5975. [5] R. Weissleder, A. Bogdanova, Edward A. Neuwelt, Mikhail Papisova, Longcirculating iron oxides for MR imaging, Adv. Drug Deliv. Rev. 16 (1995) 321–334. [6] A. Prange, Total reflection X-ray spectrometry: method andapplications, Spectrochim. Acta, Part B 44 (1989) 437–452. [7] R. Klockenkämper, Total reflection X Ray Fluorescence Analysis, Wiley, New York, 1997.
[8] M. García-Heras, R. Fernández-Ruiz, J.D. Tornero, Analysis of archaeological ceramics by TXRF, J. Archaeol. Sci. 24 (1997) 1003–1014. [9] R. Fernández-Ruiz, J.P. Cabañero, E. Hernandez, M. León, Determination of the stoichiometry of CuxInySez by total-reflection XRF, Analyst 126 (10) (2001) 1797–1799. [10] R. Fernández-Ruiz, J. Capmany, Determination of the rare-earth/Nb mass ratio in doped LiNbO3 by the TXRF technique, J. Anal. At. Spectrom. 16 (2001) 867–869. [11] R. Fernández-Ruiz, M. Furió, F. Cabello Galisteo, M. López Granados, R. Mariscal, J.L.G. Fierro, Chemical analysis of used three way catalysts by total reflection X-ray fluorescence, Anal. Chem. 74 (2002) 5463–5469. [12] R. Klockenkämper, A. von Bohlen, Survey of sampling techniques for solids suitable for microanalysis by total-reflection X-ray fluorescence spectrometry, J. Anal. At. Spectrom. 14 (1999) 571–576. [13] R. Klockenkämper, A. von Bohlen, Determination of critical thickness and the sensitivity for thin film analysis by total reflection X-ray fluorescence spectrometry, Spectrochim. Acta Part B 44 (1989) 461–469. [14] R. Fernández-Ruiz, M. García-Heras, Study of archaeological ceramics by totalreflection X-ray fluorescence spectrometry: semiquantitative approach, Spectrochim. Acta Part B 62 (2007) 1123–1129. [15] S. Veintemillas-Verdaguer, M.P. Morales, C.J. Serna, Continuous production of γ-Fe2O3 ultrafine powders by laser pyrolysis, Mater. Lett. 35 (1998) 227–231. [16] A. Prange, U. Reus, H. Schwenke, J. Knoth, Optimization of TXRF measurements by variable incident angles, Spectrochim. Acta Part B 54 (1999) 1505–1511.
Contents lists available at ScienceDirect
Spectrochimica Acta Part B j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s a b
Total-reflection X-ray fluorescence: An alternative tool for the analysis of magnetic ferrofluids☆ R. Fernández-Ruiz a,⁎, R. Costo b, M.P. Morales b, O. Bomatí-Miguel b, S. Veintemillas-Verdaguer b a b
Servicio Interdepartamental de Investigación (SIdI), Facultad de Ciencias, Universidad Autónoma de Madrid, Cantoblanco, E-28049, Madrid, Spain Instituto de Ciencia de Materiales de Madrid, CSIC, Cantoblanco, E-28049, Madrid, Spain
a r t i c l e
i n f o
Article history: Received 8 August 2007 Accepted 3 October 2008 Available online 22 October 2008 Keywords: TXRF Medical image Iron ferrofluids contrast agent AD-TXRF Nanoparticles
a b s t r a c t This work presents the first application of the total-reflection X-ray fluorescence (TXRF) to the compositional study of magnetic ferrofluids. With the aims of validating the best analytical conditions and also, limitations of the TXRF in the compositional study of these materials, an alternative empirical method, based in the use of angle-dependence TXRF (AD-TXRF) measurements, is proposed. Three kinds of ferromagnetic nanoparticles, with different morphologies, have been studied. The techniques of inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma optical emission spectroscopy (ICP-OES) have been used to validate the TXRF results. In contrast with the plasma techniques, the developed TXRF procedure need not of previous chemical acid digestion. Additionally, two procedures of magnetic nanoparticles synthesis, coprecipitation and laser-pyrolysis, have been checked for the contaminants trace metals Zn, Mn and Cr. It has been found that the method of laser-pyrolysis produces nanoparticles of higher purity. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Suspensions of magnetic nanoparticles in aqueous and nonaqueous media are nowadays attracting much interest because of their potential for many applications such as biomedicine [1], including biomolecules separation, magnetic resonance imaging and drug delivery and also as recording media, by controlling the evaporation process leading to the self assembly of the nanoparticles into ordered regions, i.e. superlattices [2]. To optimise the suitability of a nanoparticle suspension for any specific application, the samples must be synthesized and characterized reproducibly. So, the nature, the size and the elemental composition of the magnetic nanoparticles must be carefully controlled and analysed. The best strategy to synthesize magnetic nanoparticles such as iron, cobalt, iron-platinum and iron oxide has been recently developed by decomposition of organometallic precursors in the presence of a surfactant [3]. The nanoparticles are isolated and stabilized in the solution by capping groups such as long-chain alkyl surfactants which keep the particle surface unreactive and repulsive toward other particles [4]. It is difficult to find out the chemical composition of such suspensions of nanoparticles as it is difficult to remove and replace ☆ This paper was presented at the 12th Conference on Total Reflection X-ray Fluorescence Analysis and Related Methods held in Trento (Italy), 18-22 June 2007, and is published in the Special Issue of Spectrochimica Acta Part B, dedicated to that conference. ⁎ Corresponding author. Universidad Autónoma de Madrid, Facultad de Ciencias, Servicio Interdepartamental de Investigación, Modulo C-9, Laboratorio de TXRF, Crta. Colmenar, Km 15. Cantoblanco-28049-Madrid, Spain. Fax: +34 914973529. E-mail address: [email protected] (R. Fernández-Ruiz). 0584-8547/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.sab.2008.10.017
such ligands for chemical analysis. This point is important for example in the case of iron oxide nanoparticle suspensions for intravenous injection. Total human iron stores amount of 3500 mg, while normal liver contains approximately 0.2 mg of iron/g wet weight of liver and chronic iron toxicity (cirrhosis) develops only after the liver iron concentration exceeds 4 mg Fe/g liver wet weight. The total amount of iron oxide proposed for diagnostic imaging is small compared to the normal iron store, that is 1-2 mg Fe/Kg bodyweight [5]. It is well-known that iron oxides are ferromagnetic and therefore, sensitive to the action of external magnetic fields. So, in the presence of a magnetic field, the constituent magnetic dipoles of the nanoparticles are aligned in the direction of the external field. On the other hand, the organic covering of dextrane or albumen is useful and necessary when these nanoparticles are used as image contrast agents in live organisms, because they prevent that the antibodies of the organism detects the presence of the anomalous magnetic nanoparticles in the organism. One of the main future research goals will be to obtain the functionalization of the organic core with specific radicals for each one of the cellular kinds that exist in an living organism. So, it would be possible to obtain selective contrast agents for imaging and treatment of any affected organ by a cancer. These features convert to these materials in a new precise diagnostic agents for tumors and for their treatment. The treatment would be possible by means of the heat liberation that takes place in the hysteresis cycle of the ferromagnetic nanoparticles when they are introduced under the effects of alternate external magnetic fields. Iron concentrations in suspensions, which are used as contrast agents for medical NMR imaging, are usually determined by
1388
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
spectroscopic techniques such as AAS or ICPS, requiring digestion of the samples in acids of high concentration. This fact requires considerable amount of time as well as high cost of the chemicals. TXRF can help to minimize the cost by means of the simple, precise and accurate procedure developed in this work. So, a research program to optimize the application of the TXRF in this materials was initiated. The study was centered in the following aspects. First, the particles average sizes of solid deposition were measured by scanning electron microscopy (SEM) and transmission electron microscopy (TEM) to check the real morphologies of three different kinds of magnetic iron nanoparticles. Second, the method of deposition was evaluated for the following aspects: the kind of sample carrier; quartz and methacrylate, and the deposition procedure; heat and vacuum. Third, Fe AD-TXRF intensity measurements were obtained. The evaluation of the AD-TXRF profiles allows us to define the empirical conditions to validate the best TXRF analytical parameters for this kind of materials and, by extrapolation, for any other. Additionally, the behaviour of the Fe signal with respect to the sample concentration was evaluated. The purpose of this study was to obtain the lineal range in the Fe measurements by TXRF. Finally, the procedure was validated by comparison with ICP-MS and ICP-AES techniques for different sample morphologies and for two different synthesis procedures of the iron nanoparticles. TXRF main features and potential applications can be found in the illustrative paper of Prange [6] or the excellent book of Klockenkämper [7]. The previous works, carried out by Fernández-Ruiz et al. [8– 11], related with the direct quantitation of solid samples by means of the suspension method, were the point of departure for this investigation. In this way, the best approach to the ideal thin film condition for the deposition of solid particles [12] is that the following three fundamental requirements are fulfilled: (1) chemical homogeneity of the solid particles, (2) average particle sizes around 1 μm [13], and (3) homogeneous spatial distribution of the deposited particles on the sample carrier. Recently, an anomalous behavior for some elements in archaeological ceramics, with a complicated mineral composition, has been found by mean of AD-TXRF intensity profiles [14]. Therefore, the evolution of the Fe AD-TXRF profiles was also performed with the aims of checking the behaviour of the iron signal with respect to the sample mass transferred to the sample carrier. 2. Experimental 2.1. Instrumentation The TXRF measurements were made using an Atomika 8030C TXRF spectrometer (Cameca, Germany) equipped with a 3 kW Mo/W dual target X-ray tube and a W/C double monochromator multilayer. The instrument is also equipped for variation of incident angle of the Xrays by tilting the sample holder systematically. A Si(Li) detector with an active area of 80 mm2 and a resolution of 150 eV at 5.9 keV (Mn Kα) was used for detection and measurement of X-rays produced. TXRF
measurements were made with the Mo Kα excitation condition by means of the adequate arrangement of the monochromator geometry. The SEM instrument used in this work was the Philips XL-30 equipped with a W source, detectors of secondary and backscattering electrons and a vacuum sample chamber with a vacuum of lower than 4 × 10− 4 Pa. The TEM instrument used in this work was a Jeol Jem 1010 of 100 kV (Tokyo, Japan) equipped with a Gatan digital camera model BioScan (Pleasanton, Canada). Finally, ICP-MS and ICP-OES instruments used were ELAN-6000 and PE-5500 respectively from Perkin Elemer Sciex (Ontario, Canada). 2.2. Sample preparation 2.2.1. Morphology of the samples Two kinds of aqueous colloidal suspensions of magnetic iron oxide nanoparticles (BKS25 and BKS31), synthesized by co-precipitation, and other synthesized by laser induced pyrolysis (TEOS16) [15], were evaluated. The samples BKS25 and BKS31 were an iron oxide nucleus covered by an organic layer of dextran and albumen respectively. The sample TEOS16 was an iron oxide nucleus covered by a quartz layer and, additionally, an organic matter layer. Fig. 1 shows the three different morphologies. 2.2.2. Sample preparation for TXRF analysis The quantitation procedure of the ferrofluid samples by TXRF was as follow: 100 μL of sample, 100 μL of Co with a certified concentration of 1000 ng/μL (Merck, Germany) and 1000 μL of Milli-Q water were added to a vial. After agitation, 2 μL of the dissolution were deposited on a quartz sample carrier. By mean of a ceramic hot plate, the samples were dried. All the process was accomplished in a class A-100 chamber. The consumed time was no more than 15 min/sample. Additionally and with the aims of evaluating the best condition related to the spatial distribution of the deposited nanoparticles on the sample carrier, a statistical method, based on the reproducibility of the Fe analytical values, was designed. Two different sample-carriers, quartz and methacrylate, and two different evaporation procedures, heat and vacuum, were investigated. Five microlitres of one representative colloidal ferrofluid (BKS25), previously diluted with water and mixed with cobalt internal standard, in such a way that the concentration of internal standard was 50 ng/μL, were deposited on five different sample carriers five times each one. In this way, the samples were analyzed by TXRF. An additional experiment to check the linearity in the quantitative Fe measurements was carried out. Ten dissolutions of the same sample (BKS31) with the approximate sample concentrations; 9000 ng/μL, 4500 ng/μL, 2250 ng/μL, 1000 ng/μL, 360 ng/μL, 180 ng/ μL, 120 ng/μL, 90 ng/μL, 18 ng/μL and 9 ng/μL, were prepared by means of successive dilutions. After a strong agitation, two microlitres of the dissolutions were deposited on a quartz sample-carrier and their TXRF spectra were acquired during 1000 s each one. The AD-TXRF measurements were made as follow: first, a diluted colloidal suspension of the sample BKS31 was prepared. The obtained
Fig. 1. Morphology of nanoparticles BKS25, BKS31 (synthesized by co-precipitation) and TEOS16 (synthesized by laser induced pyrolysis).
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
nominal sample concentration was around 9000 ng/μL. Then, the sample was diluted with ultrapure water (Milli-Q, 18.2 MΩ) to the following factors: 1, 2, 25, 50, 75, 100, 500 and 1000. Afterwards, two microlitres of each dissolution were transferred over the quartz sample-carrier, drying with a ceramic plate. As result, the following iron masses were transferred: 18,000 ng, 9000 ng, 720 ng, 360 ng, 240 ng, 180 ng, 36 ng, 18 ng. In this way, AD-TXRF measurements for each one of the depositions were carried out. The incidence angle was varied from 1.0 to 4.5 mrad, around the critical angle for quartz. Mo Kα excitation was used in this experiment. 2.2.3. Sample preparation for ICP-MS and ICP-OES analysis An acid mix of HNO3:HF:H2O2:HCl in the proportions 3:3:1:1, for the TEOS samples, or an acid mix of HNO3:HCl in the proportion 4:4, for the BKS samples, together with 1000 μL of ferrofluid sample were introduced in a teflon reactor. The samples were digested in a Millestone EthosSel microwave oven (Sorisole, Italy). A constant temperature of 20 min to 200 °C was programmed [17]. Afterwards, the digested samples were rise up to one liter with ultra-pure water and spiked with 500 μL of Rh with a certified concentration of 1000 ng/μL (Merck, Germany) for instrumental drifts correction. An adequate calibration curve, with five concentration points of Fe certified standard (Merck, Germany), was carried out before of iron quantification. In this case, the consumed time was about 2 h/sample. 3. Results and discussion 3.1. Morphology of the nanoparticles studied The samples were evaluated by SEM and TEM to check the real morphology of the nanoparticles. Fig. 2 shows the TEM/SEM images obtained for the BKS25, BKS31 and TEOS16 samples.
1389
The use of TEM persecuted the vision and measure of the internal structure of the nanoparticles taking advantage of the different mass absorption of the electrons by the matter. In our case, samples BKS25 (Fig. 2a) and BKS31 (Fig. 2b) do not present internal structure, when owing to their morphology the difference between the magnetite nucleus and the organic layer should be perceptible. This fact can be due to volatilization of the organic layer due to the thermal action of the electron beam of TEM on the sample. Ramifications, probably due to the magnetic interaction among nanoparticles, are observed. This observation could imply the presence of an absorption effect in the quantitative analysis of the samples by TXRF. Nevertheless, the agglomeration magnetic effect is less accused in sample TEOS16 due to the magnetic screening induced by quartz layer which is clearly visible in the Fig. 2d. Fig. 2c and d shows the SEM and TEM images of the TEOS16 nanoparticles. In this case, the SEM image give us information about the average sizes of the organic layer and TEM image the average sizes of the internal structure of quartz layer and magnetite nucleus. By means of the ImageTools v.3.0 software package (Texas University, Health Science Center, San Antonio, USA) the particle sizes of the nanoparticle were evaluated. For statistical analysis a sampling order around 50 was used. The normalized probability density graphs and its associated probability curve are showed in Fig. 3. The analysis of the size distribution of sample BKS25 is displayed in Fig. 3a. In this case, the greater density of particles has a mean spherical size of 4.6 ± 0.6 nm, which can be associated to the average diameter of its magnetic nucleus. The analysis of the size distribution of sample BKS31 is displayed in Fig. 3b. In this case, the average diameter of its magnetic nucleus has a mean spherical size of 5.6 ± 1.1 nm. On the other hand, for the TEOS16 sample, a deeper study was carried out. The measurements of the magnetite nucleus and the
Fig. 2. TEM images of (a) BKS25, (b) BKS31, and (c) SEM and (d) TEM images for TEOS16.
1390
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
Fig. 3. Probability density graphs and probability curve associated to the size distributions found in the TEM and SEM images for (a) BKS25, (b) BKS31 and (c) TEOS16.
layer of quartz were made from the images of TEM (Fig. 2d). Additionally, the sizes of the organic layer were made from the SEM image (Fig. 2c). We can observe that, the organic matter is eliminated in the process of measurement by TEM. This fact can be deduced because the average diameters measured by TEM are much smaller than the measured ones by SEM. The effect is showed in Fig. 3c. The average particle sizes obtained shows a spherical organic layer with an average diameter of 78 ± 7 nm. The quartz layer shows a spherical structure with an average diameter of 40 ± 6 nm. Finally, the magnetite nucleus also shows a spherical structure of average diameter 8.8 ± 1.3 nm. With these particle size distributions, the first and second fundamental requirement, mentioned in the introduction, were assured. So, the quantification by TXRF could be done in an adequate way. 3.2. Homogeneity of the depositions The quantitative results associated to the homogeneity study are showed in Table 1. Fig. 4 shows the three characteristic deposition structures obtained by observation with binocular. The evaluation of the statistical parameters (Table 1), combined with the visual inspection of the depositions (Fig. 4), allows us to conclude that the best deposition procedures are quartz, as samplecarrier, and heat or vacuum, as drying method. It can be observed that the increment in the roughness of the deposition is correlated with the increment of the associated uncertainty, but do not with variations
of the nominal value, or accuracy, of the iron concentration. The simple change of the sample-carrier reduce the coefficient of variation from 9.6% for methacrylate to 6.6% for quartz, whereas the iron concentration is analytically the same, 49.4 ± 4.7 ng/μL for methacrylate and 45.2 ± 3.9 ng/μL for quartz. Finally, an SEM-EDAX mapping for Fe and Co, used as reference, was obtained. Five microliters for each one of the ferrofluid particles suspensions were deposited on a quartz sample carrier. The sample carrier was dried on a ceramic hot plate and afterwards, it was analysed by SEM-EDAX, after gold metalisation. Fig. 5 shows the chemical homogeneity of the depositions with respect to Fe as main component and Co as internal standard.
Table 1 Results of the quantitative analysis of sample depositions Deposition
QH (ng/μL)
MV (ng/μL)
QV (ng/μL)
D1 D2 D3 D4 D5 Mean (ng/μL) σ (ng/μL)a CV (%)b
42.64 45.70 45.20 40.70 48.80 44.6 3.1 6.9
45.70 44.21 53.27 55.18 48.54 49.4 4.7 9.6
48.20 45.10 43.10 48.20 41.60 45.2 3.0 6.6
a Standard deviation, bCoefficient of Variation. Q and M indicate Quartz and Methacrylate as sample-carriers respectively and H and V indicate Heat and Vacuum as evaporation procedure respectively.
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
1391
Fig. 4. Binocular images of the three different depositions obtained in the study of the deposition procedure. (a) 5 μL of diluted BKS25 over quartz and dried with heat. (b) 5 μL of diluted BKS25 over methacrylate and dried with vacuum. (c) 5 μL of diluted BKS25 over quartz and dried with vacuum.
Fig. 5. SEM-EDAX mapping of one representative (5 μL of BKS25 colloidal sample deposited over a quartz sample support) deposition of the magnetic ferrofluids analyzed by TXRF.
3.3. AD-TXRF measurements After selecting the best deposition method i.e. heating the sample on quartz sample supports, an AD-TXRF study was designed to validate the best analytical conditions for TXRF [16]. Fig. 6 shows the
evolution of the profiles, associated to the Fe K-lines, with respect to the mass of sample transferred over the quartz reflector. Two empirical parameters can be evaluated, one, the angular position of the maximum intensity (αmax) and another one, the intensity ratio between the constant intensity, after the critical angle,
Fig. 6. Evolution of the normalized AD-TXRF intensity profiles, associated to the Fe K-lines, with respect to the mass of sample transferred over the sample carrier.
1392
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
and the maximum intensity, before the critical angle (Imin/Imax). Fig. 7 presents the behaviors found for these empirical parameters with respect to the sample mass transferred over the sample-carrier. Fig. 7a shows that αmax evolve with the sample mass transferred. In this case, if masses are lower than 700 ng, αmax value is approximately constant and it is lower that the critical angle used in the quantitative TXRF analysis. On the other hand, if masses are higher of 700 ng, αmax values increase and are higher of the critical angle used for quantitative TXRF analysis, which is 3 mrad in our case. If we keep in mind that the ideal intensity relation between XRF and TXRF is 1:2 for thin layer depositions [7], the results obtained in Fig. 7b can be used to validate the deposited mass limits, that is, when the TXRF condition is achieved and when it is lose. In this case, the interpolation between the spline curve of the experimental results and the line which defines de ideal condition Imin/Imax = 0.5 for the TXRF, gives us the top sample mass that can be deposited over the sample carrier to achieve the ideal TXRF condition. As we can observe in Fig. 7b, this value is found around 2500 ng for each one of the nanoparticle ferrofluids evaluated. Fig. 8 show the behavior of the Fe K-lines net intensity with respect to the nominal sample concentrations of the suspensions. As we can see, the sample concentration where the linear behavior is lost, it is finding around of 1200 ng/μL. This agree with our previous sample mass limit (2500 ng) associated with Imin/Imax = 0.5. As conclusion, the optimum region of concentration for the analysis of this kind of materials has been found for a sample mass transferred on sample carrier lower than 2500 ng. Therefore, the methodology proposed in this section of the work seems to be valid for the optimization of the best analytical conditions for the quantitative analysis by TXRF, independently of the kind of sample analyzed.
3.4. Validation of the analysis by TXRF
Fig. 7. Change in empirical parameters (a) αmax and (b) Imin/Imax with sample mass on the sample carrier for Fe K-lines in the sample BKS31.
Once optimized the sample preparation conditions and with the aims of validate the Fe quantitative measurements and additionally, the correctness of the developed procedure by TXRF, several kind of ferrofluids nanoparticles were compared by mean of ICP-MS and ICPOES measurements.
Fig. 8. Change in Fe K-lines net intensity with respect to nominal sample concentrations (CS) in suspensions of nanoparticle BKS31.
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394 Table 2 Analytical results and statistical parameters for Fe concentration obtained by TXRF, ICPMS and ICP-OES in the validation process for different BKS and TEOS samples Sample
TXRF ng/μL ± σn = 5 / (CV)
ICP-MS ng/μL ± σn = 5 / (CV)
ICP-AES ng/μL ± σn = 3 / (CV)
BKS25 BKS31 TEOS16 P903 (BKS) G1 (BKS) B2 (TEOS)
6930 ± 456 / (6.59%) 8944 ± 295 / (3.30%) 75.6 ± 5.3 / (6.94%) 22582 ± 1364 / (6.04%) 26940 ± 1472 / (5.46%) 3622 ± 232 / (6.41%)
6790 ± 356 / (5.24%) 7546 ± 543 / (7.19%) 82.2 ± 3.6 / (4.38%) -
24375 ± 1864 / (7.65%) 25600 ± 1986 / (7.76%) 3325 ± 354 / (10.6%)
The first notable difference was found in the sample preparation time, 15 min/sample by TXRF compared with 2 h/sample for both ICPS techniques. On the other hand, the differences in materials and reagents also were considerable. From a more scientific point of view, the analytical results and the more significant statistical parameters, for Fe concentration, obtained by TXRF, ICP-MS and ICP-OES, are presented in Table 2. The nominal values of Fe concentrations obtained for the six kinds of samples analyzed were analytically equivalents within error bars. With respect to the associated uncertainties, the TXRF values are more precise, between 3.30% and 6.94% of coefficient of variation, while the ICP-MS or ICP-OES measurements were found between 5.24% and 10.6% of coefficient of variation. As conclusion, the TXRF seems a more reproducible technique for this kind of materials. On the other hand, it is demonstrated that the morphological differences in the evaluated nanoparticle ferrofluids have not influence in the Fe analytical value. This fact was not obvious because the different layer of organic matter (BKS) or of quartz and organic matter (TEOS), together to the agglomeration magnetic effect, could produce an absorption effect which could imply a constant drift in the concentration value of the Fe. In this point of the research, the TXRF has proven to be an outstanding alternative method for the iron quantification in ferrofluids. The final developed TXRF methodology has several important advantages with respect to the ICPS techniques, i.e. simplicity, low cost, low time consuming, precise and accurate results. 3.5. Impurities analysis by TXRF Finally, a comparative study, of the trace contaminants induced in the synthesis process, was carried out. The BKS samples were synthesized by the co-precipitation method and the TEOS sample
1393
Table 3 A comparison of trace elements analysis of ferrofluids obtained by co-precipitation and laser pyrolysis synthesis procedures Co-precipitation synthesis
Elements Zn Mn Cr a
Pyrolysis laser synthesis
BKS25 (ng/μL ± σn = 5)
BKS31 (ng/μL ± σn = 5)
TEOS31 (ng/μL ± σn = 5)
TXRF 1.35 ± 0.08 37.0 ± 2.6 43.8 ± 3.7
TXRF b1 43.2 ± 1.9 65.4 ± 2.9
TXRF 1.29 ± 0.04 b0.3 b0.2
ICP-MS n.d.a 44.8 ± 3.5 49.4 ± 3.2
ICP-MS n.d. 31.6 ± 4.2 45.2 ± 6.1
ICP-MS 1.20 ± 0.05 0.08 ± 0.01 n.d.
Not detected.
was synthesized by the laser pyrolysis method. In both cases, the qualitative analysis shows the presence of the following traces elements; Si, Cl, K, Cr, Mn, Zn, Pb and Br, such as Fig. 9 shows. We can see in Table 3 that only the elements Mn, Cr and Zn were quantified simultaneously by TXRF and ICP-MS due to comparative reasons. The other elements detected in the qualitative analysis were not analyzed for several reasons; first by time consumption in the analysis by ICP-MS and second, because the detection limits for some of the elements with health interest, as Pb, had a concentration lower than the quantification limit for both TXRF and ICP-MS techniques. Table 3 shows that the pyrolysis laser synthesis method is a cleaner procedure than the co-precipitation synthesis method. This fact has an important relevance from the medical point of view due to its intra-venous use as contrast agent in the diagnostic by means of NMR imaging. 4. Conclusions This work presents the first application of the TXRF technique to the compositional analysis of nanoparticle ferrofluids which are important materials used in the medical image improvements and have a high potential for its application for treatment of the cancer in a near future. The TXRF conditions for such applications have been optimized. The AD-TXRF method has been used for first time to optimize the TXRF conditions for analysis of such materials. This technique of optimization of TXRF conditions is also equally valid for other similar materials. The roughness of the depositions used for TXRF measurements has been found to affect the precision of measurements. However, it does not affect the accuracy of the measurements. One of the main goals achieved in this work has been that the TXRF procedure do not need previous acid digestion so that, it is quick (15 min/sample), simply and not expensive (only dilution and addition of Co as internal standard). Additionally, the quantitative results have an adequate uncertainty level for the ferrofluids evaluated in this work (CV lower than 7%). Finally, these results show that TXRF is an advantage quantitative alternative with respect to ICP-MS or ICP-AES in the analysis of ferrofluids. The trace analysis results indicate that the pyrolysis laser synthesis procedure obtain nanoparticles of highest purity than the co-precipitation synthesis procedure, which implies relevant medical consequences. Acknowledgements
Fig. 9. Representative TXRF spectrum of the sample TEOS16. Intensity scale is normalized and represented in logaritmic units.
Authors would like to thank Esperanza Salvador and Francisco Urbano, from SIdI-UAM laboratories, for their expert advice on the analysis by SEM and TEM respectively. Also, they wish to express their gratitude to Liquids Research Limited, North Wales, UK for supply BKS25 and BKS31 samples and to Autonomous Community of Madrid (S-0505/MAT/0194) and Spanish Ministers of Education (MAT200503179, NAN2004-08805-C04-01) for financial support. Additionally, the authors express their gratitude to Dirección General de Investigación from Comunidad de Madrid for TXRF instrumentation financial support. Finally, the authors wish to express their gratitude to Lola for
1394
R. Fernández-Ruiz et al. / Spectrochimica Acta Part B 63 (2008) 1387–1394
her English revision of the manuscript and to the reviewer for their suggestions for the clarity of the manuscript. References [1] P. Tartaj, María del Puerto Morales, S. Veintemillas-Verdaguer, T. GonzálezCarreño, C.J. Serna, The preparation of magnetic nanoparticles for applications in biomedicine, J. Phys. D 36 (2003) R182. [2] C. Shouheng Sun, B. Murray, Dieter Weller, Liesl Folks, Andreas Moser, Monodisperse FePt nanoparticles and ferromagnetic FePt nanocrystal superlattices, Science 287 (2000) 1989–1992. [3] J. Park, Kwangjin An, Yosun Hwang, Je-Geun Park, Han-Jin Noh, Jae-Young Kim, JaeHoon Park, Nong-Moon Hwang, Taeghwan Hyeon, Ultra-large-scale syntheses of monodisperse nanocrystals, Nat. Mater. 3 (2004) 891. [4] A.L. Willis, N.J. Turro, S. O'Brien, Spectroscopic characterization of the surface of iron oxide nanocrystals, Chem. Mater. 17 (24) (2005) 5970–5975. [5] R. Weissleder, A. Bogdanova, Edward A. Neuwelt, Mikhail Papisova, Longcirculating iron oxides for MR imaging, Adv. Drug Deliv. Rev. 16 (1995) 321–334. [6] A. Prange, Total reflection X-ray spectrometry: method andapplications, Spectrochim. Acta, Part B 44 (1989) 437–452. [7] R. Klockenkämper, Total reflection X Ray Fluorescence Analysis, Wiley, New York, 1997.
[8] M. García-Heras, R. Fernández-Ruiz, J.D. Tornero, Analysis of archaeological ceramics by TXRF, J. Archaeol. Sci. 24 (1997) 1003–1014. [9] R. Fernández-Ruiz, J.P. Cabañero, E. Hernandez, M. León, Determination of the stoichiometry of CuxInySez by total-reflection XRF, Analyst 126 (10) (2001) 1797–1799. [10] R. Fernández-Ruiz, J. Capmany, Determination of the rare-earth/Nb mass ratio in doped LiNbO3 by the TXRF technique, J. Anal. At. Spectrom. 16 (2001) 867–869. [11] R. Fernández-Ruiz, M. Furió, F. Cabello Galisteo, M. López Granados, R. Mariscal, J.L.G. Fierro, Chemical analysis of used three way catalysts by total reflection X-ray fluorescence, Anal. Chem. 74 (2002) 5463–5469. [12] R. Klockenkämper, A. von Bohlen, Survey of sampling techniques for solids suitable for microanalysis by total-reflection X-ray fluorescence spectrometry, J. Anal. At. Spectrom. 14 (1999) 571–576. [13] R. Klockenkämper, A. von Bohlen, Determination of critical thickness and the sensitivity for thin film analysis by total reflection X-ray fluorescence spectrometry, Spectrochim. Acta Part B 44 (1989) 461–469. [14] R. Fernández-Ruiz, M. García-Heras, Study of archaeological ceramics by totalreflection X-ray fluorescence spectrometry: semiquantitative approach, Spectrochim. Acta Part B 62 (2007) 1123–1129. [15] S. Veintemillas-Verdaguer, M.P. Morales, C.J. Serna, Continuous production of γ-Fe2O3 ultrafine powders by laser pyrolysis, Mater. Lett. 35 (1998) 227–231. [16] A. Prange, U. Reus, H. Schwenke, J. Knoth, Optimization of TXRF measurements by variable incident angles, Spectrochim. Acta Part B 54 (1999) 1505–1511.