Microscopy Microanalysis
Microsc. Microanal. 15, 114–124, 2009 doi:10.1017/S1431927609090114
AND
© MICROSCOPY SOCIETY OF AMERICA 2009
A Transmission Electron Microscopy Study of Fe-Co Alloy Nanoparticles in Silica Aerogel Matrix Using HREM, EDX, and EELS Andrea Falqui,1,2 Anna Corrias,1 Mhairi Gass,3 and Gavin Mountjoy 1, * ,† 1
Dipartimento di Scienze Chimiche and INSTM, Università di Cagliari, S.P. Monserrato-Sestu Km 0.700, 09042 Monserrato, Cagliari, Italy 2 Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy 3 SuperSTEM, Daresbury Laboratory, Keckwick Lane, Daresbury, Cheshire WA4 4AD, UK
Abstract: Magnetic nanocomposite materials consisting of 5.5 wt% Fe-Co alloy nanoparticles in a silica aerogel matrix, with compositions FexCo1⫺x of x ⫽ 0.50 and 0.67, have been synthesized by the sol-gel method. The high-resolution transmission electron microscopy images show nanoparticles consisting of single crystal grains of body-centered cubic Fe-Co alloy, with typical crystal grain diameters of approximately 4 and 7 nm for Fe0.5Co0.5 and Fe0.67Co0.33 samples, respectively. The energy dispersive X-ray ~EDX! spectra summed over areas of the samples gave compositions FexCo1⫺x with x ⫽ 0.48 6 0.06 and 0.68 6 0.05. The EDX spectra obtained with the 1.5 nm probe positioned at the centers of ;20 nanoparticles gave slightly lower concentrations of Fe, with means of ^x& ⫽ 0.43 6 0.01 and ^x& ⫽ 0.64 6 0.02, respectively. The Fe 0.5Co0.5 sample was studied using electron energy loss spectroscopy ~EELS!, and EELS spectra summed over whole nanoparticles gave x ⫽ 0.47 6 0.06. The EELS spectra from analysis profiles of nanoparticles show a distribution of Fe and Co that is homogeneous, i.e., x ⫽ 0.5, within a precision of at best 60.05 in x and 60.4 nm in position. The present microscopy results have not shown the presence of a thin layer of iron oxide, but this might be at the limit of detectability of the methods. Key words: Fe-Co alloy, nanoparticles, HREM, EDX, EELS
I NTR ODUCTION A topic of great current technological interest is the synthesis of nanoparticles with controlled size and composition. Efforts to understand the physics of these smaller structures are combined with attempts to exploit their special properties. For example, metal nanoparticles have been used to improve the performance of catalysts, and thin films of metal nanoparticles have been developed for high density magnetic recording ~Li et al., 1999!. Magnetic nanocomposite materials consisting of ferromagnetic nanoparticles embedded in an insulating matrix have attracted attention because the matrix can stabilize nanoparticle size and dispersion ~Abeles, 1976!. The magnetic properties of the nanoparticles, for example, superparamagnetism, depend on the composition, size, and shape of the nanoparticle, and on the volume fraction of nanoparticles in the matrix. The matrix, via its density and porosity, strongly influences Received July 23, 2008; accepted December 1, 2008 *Corresponding author. E-mail:
[email protected] † Permanent address: School of Physical Sciences, University of Kent, Canterbury, Kent CT2 7NH, UK
the average size of, and distance between, the magnetic nanoparticles. Hence a complete structural characterization plays a fundamental role in understanding the physical properties of such nanocomposite materials. The present study uses advanced electron microscopy techniques for this goal. Recent advances in preparation methods have encouraged the preparation of magnetic nanocomposite materials using the sol-gel technique ~Brinker & Scherer, 1990!. This technique offers the advantages of accurate control over composition, purity, and homogeneity at a microscopic level ~e.g., Ennas et al., 2001; Moreno et al., 2002!. In particular, the well-developed sol-gel chemistry of silica has been widely applied to the preparation of silica-based nanocomposites with controlled textures ~e.g., Ennas et al., 2001; Moreno et al., 2002!. Silica aerogels offer attractive features such as chemical inertness, low density, transparency, and low dielectric constant, due to the fractal-like porous silica matrix ~Husing & Schubert, 1998; Pierre & Pajonk, 2002!. Innovative magnetic nanocomposites such as high coercivity NdFeB-SiO2 aerogels have been prepared ~Gich et al., 2003!. The present study examines magnetic nanocomposites of ferromagnetic Fe-Co alloy nanoparticles in silica aerogel matrix prepared by the sol-gel method.
Study of Fe-Co Alloy Nanoparticles Using HREM, EDX, and EELS
Fe-Co alloys play an important role in the technology of transformers and generators because of their soft ferromagnetic behavior, with high Curie temperatures and large saturation inductions, about 15% larger than pure iron ~MacLaren et al., 1999; Saad et al., 2004!. Bulk Fe-Co alloys form in the a body-centered cubic ~bcc! phase over a wide composition range ~from 25 to 100 at.% Fe! ~Guillermet, 1988!, and the magnetic properties of this phase have been widely investigated. The Fe-Co alloy with composition Fe0.5Co0.5 has the highest permeability and lowest coercivity, and it has a saturation magnetization close to the largest value that is obtained for composition Fe 0.67Co0.33 ~Paduani & Krause, 1999!. Recently, FeCo-SiO2 xerogel ~Falqui et al., 2003! and FeCo-Al2O3 aerogel ~Casula et al., 2005! nanocomposite materials have been successfully prepared by the sol-gel method. The present study provides advanced electron microscopy results for Fe-Co alloy nanoparticles in a highly porous silica aerogel matrix. These materials have been obtained by using a modified sol-gel method involving urea as a catalyst ~Carta et al., 2007!. The characterization of Fe-Co alloy nanoparticles in a SiO2 aerogel matrix using standard transmission electron microscopy ~TEM! and X-ray diffraction ~XRD! ~Casu et al., 2008! has shown the dispersed and nanocrystalline nature, and the typical sizes, of the nanoparticles. X-ray absorption spectroscopy ~Carta et al., 2007! has given detailed information about atomic structure, but this information is necessarily averaged over all nanoparticles. Information is lacking about structure and composition of individual nanoparticles, whereas such details are directly relevant to modeling the magnetic properties of these magnetic nanocomposite materials ~Casu et al., 2008!. The techniques of highresolution electron microscopy ~HREM! and spatially resolved electron energy loss spectroscopy ~EELS! have previously been used to study FeCo-Al2O3 aerogel nanocomposites ~Casula et al., 2005! and FeCo-SiO2 xerogel nanocomposites ~Falqui et al., 2003!. In the latter study, spatially resolved EELS of FeCo particles with an average size of around 20 nm, embedded in a silica xerogel matrix, clearly showed the presence of an 3 nm thick iron oxide rich layer ~although that study used a significantly different sample preparation method!. To our knowledge, the present study is the first time that HREM, energy dispersive X-ray ~EDX! spectrometry, and EELS have been used to study FeCo-SiO 2 aerogel nanocomposites.
M ATERIALS
AND
M ETHODS
The sample preparation followed the sol-gel procedure reported in ~Carta et al., 2007!. Tetraethoxysilane @Si~OC2H5 !4 , Aldrich 98%, TEOS#, and iron ~III! and cobalt ~II! nitrates @Fe~NO3 !3{9H2O, Aldrich, 98%, and Co~NO3 !2{6H2O, Aldrich, 98%#, were used as precursors for the silica and for the Co-Fe alloy phases, respectively; absolute ethanol ~pur-
115
chased from Fluka! was used as mutual solvent. The alcogel was obtained by a two-step procedure, including the mixing of the ethanolic solution of the iron and cobalt nitrates into the TEOS, prehydrolyzed under acidic conditions, followed by the addition of urea ~NH 2CONH 2 , Sigma-Aldrich, .99.0%! under reflux for 2 h at 858C. Urea enhances both the successful co-gelation of the silica and Fe-Co precursors and the formation of pore structure in the gel ~Casula et al., 2007!. Gelation was performed at 408C in a closed container and occurred in less than 2 days. The alcogels were submitted to high temperature supercritical drying in an autoclave ~manufactured by Parr, 0.3 L capacity!. The autoclave was filled with an appropriate amount of ethanol and flushed with N2 and then heated to take the solvent to the supercritical state. The autoclave was then vented and highly porous aerogel samples were obtained. After supercritical drying, the samples were powdered and calcined at 4508C in static air for 1 h to eliminate the organics. Samples containing nanocrystalline Fe-Co alloy in highly porous SiO 2 aerogel matrix were obtained by reduction in pure H2 at 8008C for 2 h. Samples with ~Fe⫹Co!:SiO2 ratios of 5.5:94.5 wt% and alloy compositions of Fe 0.67Co0.33 and Fe 0.5Co0.5 were prepared, hereafter referred to as Fe0.67Co0.33 and Fe0.5Co0.5 samples, respectively. The latter corresponds to the most useful composition for magnetic applications, and the former corresponds to the composition with the largest saturation magnetization. The XRD patterns were recorded on an X3000 Seifert diffractometer equipped with a graphite monochromator on the diffracted beam. The scans were collected within the range of 10–908 2u using Cu-Ka radiation. Electron microscopy observations were carried out using two different electron microscopes. Samples for electron microscopy were prepared by grinding powders, suspending in absolute ethanol, and dropping onto 3 mm 400 mesh copper microscope grids covered with holey carbon film ~SPI Supplies!. The first microscope was a 200 kV JEOL 2100F TEM, equipped with a field emission gun and an ultra-highresolution pole piece, located at TEMSCAN facility, Université Paul-Sabatier, Toulouse, France, and was used to collect HREM images. This microscope can also operate in scanning transmission electron microscope ~STEM! mode, where the electron beam forms a 1.5 nm diameter probe, and this mode was used with scanning of the probe to collect bright field ~BF! STEM images ~with an objective aperture!. Note that it was not convenient to simultaneously collect both BF and dark field ~DF! images with this microscope because it had a high angle annular dark field ~HAADF! detector that blocked the BF signal and would have required frequent changes between long and short camera length for BF and DF images, respectively. The STEM mode was also used to collect EDX spectra ~without an objective aperture!, both summed over areas of the sample ~with scanning! and from the centers of individual nanoparticles ~without scanning!. For EDX spectra the typical probe current was approximately 1 nA; the usual collection time was 30 s. The EDX
116
Andrea Falqui et al.
detector was a Princeton Gamma-Tech LS30135 Si~Li! detector with ultrathin window and had a uniform response in the 3 to 20 keV range and a resolution of 135 eV at 5.9 keV. In the 6.0 to 9.5 keV range, electron generated X-ray emission peaks occur in the following sequence: Fe Ka at 6.404 keV, Co Ka at 6.9303 keV, ~overlapping with! Fe Kb at 7.058 keV, and Co Kb at 7.649 keV. Copper is a background signal ~Cu Ka at 8.048 keV and Cu Kb at 8.904 keV! due to the use of Cu microscope grids. The EDX spectra were fitted quantitatively using the PYMCA software distributed by the European Synchrotron Radiation Facility. The EDX signals from Fe and Co originate from Fe-Co alloy nanoparticles with thickness not greater than approximately 10 nm, and so the absorption and fluorescence corrections in the Cliff-Lorimer method ~Cliff & Lorimer, 1975! can be neglected without significant error. The Cliff-Lorimer ratio of kCoFe ⫽ 1.04 6 0.19 ~relative uncertainty of 18%! was estimated from kFeSi ⫽ 1.35 6 0.16 and kCoSi ⫽ 1.41 6 0.20 for 200 keV electrons, with the relative uncertainties being added in quadrature ~Williams & Carter, 1996!. The Cliff-Lorimer k-factor k CoFe was used to convert X-ray intensity I to wt% concentration C using CCo CFe
⫽ k CoFe
ICo IFe
.
~1!
The values of ICo and IFe were typically 1,000–5,000 counts, with relative uncertainties being 4–13% ~three standard deviations!. The relative uncertainty in CCo /CFe is the sum in quadrature of the relative uncertainties in ICo , IFe , and kCoFe . Note that ICo and IFe are different for each measurement, but the same value of kCoFe is used for all measurements. A value of x for composition Fe xCo1⫺x was then obtained using x⫽
1 1⫹
* CCo
,
~2!
* CFe
where C * is in at.% concentration. From equation ~2! it can be seen that the relative uncertainty in x is equal to the * * /CFe !. The typical relative relative uncertainty in ~1 ⫹ CCo uncertainty in x is 10–13%, but this includes the contribution from the relative uncertainty in kCoFe that affects all values of x in the same way ~e.g., if kCoFe is too large, then all values of x will be too small!. The second microscope was a 100 kV VG501 STEM microscope, equipped with a cold field emission gun and a spherical aberration corrector, located at SuperSTEM, Daresbury Laboratory, Daresbury, UK. In this microscope the electron beam forms a 0.13 nm diameter probe that is scanned. This microscope was used to collect HAADF images and EELS spectra using a Gatan parallel EELS spectrom-
Figure 1. XRD pattern for ~bottom! Fe0.5Co0.5 and ~top! Fe0.67Co0.33 samples. Asterisks show Bragg peak positions for bcc ~110!, ~200!, and ~211! lattice planes.
eter. The EELS spectra were obtained using two different dispersions of 0.2 and 0.3 eV/channel, and an energy resolution of 0.5 eV ~the full width at half maximum of the zero-loss peak!. The energy ranges used included the O K-edge and Fe and Co L2,3-edges at 532, 708, and 780 eV, respectively, but not the Si L2,3-edge at 100 eV. Information about the concentrations of Fe and Co was obtained by quantitative analysis of the EELS spectra using the Gatan Digital Micrograph software distributed by Gatan Incorporated. This includes calculation of partial ionization cross sections, as described in Egerton ~1996! ~values of convergence semiangle and collection semiangle were 24 and 8 mrad, respectively!.
R ESULTS X-Ray Diffraction Figure 1 shows XRD patterns of the Fe0.5Co0.5 and Fe0.67Co0.33 samples, which were previously reported ~Casu et al., 2008!. These XRD patterns clearly show the dominant phase of the Fe-Co alloy nanoparticles is bcc. Estimates of the average crystal grain diameters were obtained from XRD patterns by using the Scherrer formula and were found to be 4.2 nm for Fe0.5Co0.5 and 7.7 nm for Fe0.67Co0.33 ~with uncertainty of 61 nm!. It has previously been assumed that nanoparticles consist of single crystal grains, so that these values correspond to nanoparticle diameters.
BF STEM Images Figure 2 shows BF STEM images of the Fe 0.5Co0.5 and Fe0.67Co0.33 samples obtained from the JEOL 2100F TEM microscope operating in STEM mode. The BF STEM images are dark due to diffraction contrast when there is strong scattering of electrons due to crystalline regions,
Study of Fe-Co Alloy Nanoparticles Using HREM, EDX, and EELS
117
Figure 2. BF STEM images of ~a–b! Fe0.5Co0.5 and ~c–d! Fe0.67Co0.33 samples.
which provides a method of detecting crystalline Fe-Co alloy nanoparticles within an amorphous silica aerogel matrix. Figure 2a,c shows a distribution of distinct nanoparticles within the silica aerogel matrix. The average size of nanoparticles in the Fe0.5Co0.5 samples is noticeably smaller than those in the Fe0.67Co0.33 sample, consistent with the results from XRD. Figure 2b,d shows that nanoparticles tend to be spherical in form. The BF STEM images also show the very high porosity of the silica aerogel matrix.
HREM Images Figures 3 and 4 show a selection of HREM images of nanoparticles in the Fe0.5Co0.5 and Fe0.67Co0.33 samples, respectively, which were obtained from the JEOL 2100F TEM microscope. In each inset the corresponding twodimensional ~2D! Fast Fourier Transform ~FFT! is reported. Figures 3a,b and 4a,b are HREM images of well-rounded single crystal nanoparticles in which bcc ~110! lattice planes are visible. Figures 3c,d and 4c,d are HREM images of nanoparticles in which double and triple bcc ~110! lattice
planes are visible, due to orientation of nanoparticles on a zone axis as mentioned in the figure captions. Note that in Figure 3d there are lattice fringes appearing outside the nanoparticle due to the effect of delocalization. The nanoparticles observed in HREM tend to be larger in the Fe0.67Co0.33 sample than in the Fe0.5Co0.5 sample, as noted previously for XRD and BF STEM images. Approximately 90% of the HREM images showed well-rounded nanoparticles consisting of single bcc alloy crystal grains, indicating that most nanoparticles in the Fe0.5Co0.5 and Fe0.67Co0.33 samples consist of single crystal grains. Finally, Figure 4d shows a large nanoparticle, with some additional lattice planes appearing at the lower edges, which have different interplanar distance compared to those of the alloy core of the nanoparticle ~see comments at the end of the Discussion section!.
EDX Spectra Figure 5 shows EDX spectra that were obtained from the JEOL 2100F TEM microscope. Figure 5a,b shows EDX
118
Andrea Falqui et al.
Figure 3. HREM images of Fe-Co alloy nanoparticles in the Fe 0.5Co0.5 sample. ~a–d! show bcc ~110! planes, ~c! shows @001# zone axis, and ~d! shows @111# zone axis. ~Inset shows the corresponding 2D FFT.!
spectra obtained with the 1.5 nm probe positioned at the center of individual nanoparticles in the Fe 0.5Co0.5 and Fe0.67Co0.33 samples, respectively. The first two peaks in the EDX spectra correspond to Fe K a at 6.404 keV and Co Ka at 6.9303 keV ~overlapping Fe Kb at 7.058 keV!, and the third peak corresponds to Co K b at 7.649 keV. The Fe and Co X-ray emission intensities IFe and ICo were extracted by quantitative fitting of the EDX spectra and were then used to calculate the value of x for concentration Fe xCo1⫺x ~as discussed in the Materials and Methods section!. The EDX spectra shown in Figure 5a,b for individual nanoparticles in the Fe0.5Co0.5 and Fe0.67Co0.33 samples give values of x ⫽ 0.42 6 0.05 and 0.66 6 0.05, respectively. Such EDX spectra were collected from 22 and 19 individual nanoparticles in the Fe0.5Co0.5 and Fe0.67Co0.33 samples, respectively. The mean ^x& and standard deviation sx of the values of x obtained were ^x& ⫽ 0.42 6 0.01 and sx ⫽ 0.05 for the Fe0.5Co0.5 sample, and ^x& ⫽ 0.63 6 0.02 and sx ⫽ 0.09 for the Fe0.67Co0.33 sample @where the error in the mean is taken as sx /~N ⫺ 1!1/2 #. In addition, EDX spectra were obtained
while the probe was scanned over a 220 ⫻ 180 nm 2 area of sample, and this gave values of x ⫽ 0.48 6 0.06 and 0.68 6 0.05 for the Fe 0.5Co0.5 and Fe0.67Co0.33 samples, respectively.
EELS Spectra The VG501 STEM microscope was used to collect EELS spectra from nanoparticles in the Fe 0.5Co0.5 sample, using a 0.13 nm probe. Figures 6 and 8 show HAADF images of the nanoparticles that were studied using EELS. Figure 6 shows two nanoparticles that were studied by collecting EELS spectra over a rectangular area entirely encompassing the nanoparticle ~as illustrated!, and Figure 7 shows the EELS spectra obtained. Because Fe and Co are present only in the nanoparticles ~and not in the silica aerogel matrix!, these summed spectra show the average Fe and Co L2,3-edges summed over whole nanoparticles. Quantitative analysis of the EELS spectra in Figure 7a,b gave compositions Fe xCo1⫺x with x ⫽ 0.47 6 0.06.
Study of Fe-Co Alloy Nanoparticles Using HREM, EDX, and EELS
Figure 4. HREM images of Fe-Co alloy nanoparticles in the Fe 0.67Co0.33 sample. ~a–d! show bcc ~110! planes, ~c! shows also bcc ~200! planes ~arrow!, and ~d! shows @001# zone axis. ~Inset shows the corresponding 2D FFT. The dashed square and arrows in the inset of ~d! are discussed at the end of the Discussion section.!
Figure 5. Quantitatively analyzed EDX spectra obtained with the 1.5 nm probe positioned at the center of individual Fe-Co alloy nanoparticles in ~a! the Fe0.5Co0.5 sample and ~b! the Fe0.67Co0.33 sample.
119
120
Andrea Falqui et al.
Figure 6. HAADF images of two Fe-Co alloy nanoparticles in the Fe0.5Co0.5 sample. Rectangles show areas over which EELS spectra were collected.
Figure 8 shows another two nanoparticles that were studied by collecting EELS spectra from lines crossing the nanoparticles ~as illustrated!. Figure 9 shows a representative EELS spectrum from a single probe position at the center of the analysis line of Figure 8a. At each probe position on the line, the EELS spectra were analyzed quantitatively ~as in Fig. 7! to obtain the relative concentrations of oxygen, Fe, and Co, and Figure 10 shows the results along the lines crossing the two nanoparticles. As expected, reliable Fe and Co signals are obtained only from the nanoparticles, whereas the Fe and Co signals fluctuate around zero in the surrounding matrix where there is no Fe or Co. Oxygen is present in the surrounding silica aerogel matrix, and there is a dip in oxygen concentration at the position of the nanoparticle, corresponding to the displaced volume of silica aerogel matrix occupied by the nanoparticle. The EELS results show that Fe and Co are approximately equally and uniformly distributed within the nanoparticles, i.e., an approximate composition of Fe0.5Co0.5 at all positions within the nanoparticles. This observation is limited by the apparent statistical fluctuations, which have a precision of at best 60.05 in x and 60.4 nm in analysis position.
D ISCUSSION The electron microscopy results presented here for Fe0.5Co0.5 and Fe0.67Co0.33 samples are a valuable contribution to the structural characterization of these recently synthesized materials. Previous XRD results ~Casu et al., 2008! were interpreted as the nanoparticles consisting of bcc Fe-Co alloy and having average crystal grain diameters of 4.2 nm for
Figure 7. EELS spectra of Fe 0.5Co0.5 alloy nanoparticles from areas shown in Figures 6a and 6b, respectively. The sequence of edges is Fe L2,3 at 708 eV and Co L2,3 at 780 eV. The upper curves ~left axis! are the raw spectra, and the lower curves ~right axis! are the signals extracted at the Fe and Co edges for quantitative analysis ~the smooth black line is the background used!.
Study of Fe-Co Alloy Nanoparticles Using HREM, EDX, and EELS
121
Figure 8. HAADF images of two Fe-Co alloy nanoparticles in the Fe0.5Co0.5 sample. Straight lines show lines over which EELS spectra were collected.
Figure 9. Raw EELS spectrum from midpoint of line crossing Fe0.5Co0.5 alloy nanoparticle in Figure 8a. The sequence of edges is oxygen K at 532 eV, Fe L2,3 at 708 eV, and Co L2,3 at 780 eV. ~Note that Si is present due to the silica matrix, but the Si L2,3 edge at 100 eV is outside the range of the EELS spectrum.!
Fe0.5Co0.5 and 7.7 nm for Fe0.67Co0.33 . The BF STEM and HREM images presented here are consistent with this, indicating distinct, well-rounded nanoparticles. The HREM images show that most nanoparticles consist of single grain bcc nanocrystals, and hence the average crystal grain size is indicative of the average nanoparticle size, as previously assumed in the interpretation of XRD results. Due to the closeness in atomic number of Fe and Co, the methods of XRD, BF STEM, and HREM imaging are not sensitive to compositional effects, and hence additional chemically sensitive techniques are needed. The only previous detailed structural characterization of these materials was carried out using the chemically sensitive technique of X-ray absorption spectroscopy @ex-
Figure 10. Quantitative EELS measurement of relative concentration of oxygen ~dotted line!, Fe ~solid circles!, and Co ~empty circles! in cross sections of Fe 0.5Co0.5 alloy nanoparticles from lines shown in Figures 8a, and 8b, respectively.
tended X-ray absorption fine structure ~EXAFS! and X-ray absorption near edge structure spectroscopy ~XANES!# at the Fe and Co K-edges ~Carta et al., 2007!. The results
122
Andrea Falqui et al.
Figure 11. Two different scenarios for the microstructure of a 4 nm diameter Fe0.5Co0.5 alloy nanoparticle ~see the Discussion section!. ~A! Nanoparticle consists of a single homogenous alloy phase. ~B! Nanoparticle is covered with a 0.2 nm layer of iron oxide, and consequently there is a slightly reduced concentration of Fe in the alloy core of the nanoparticle. ~Vertical lines indicate relative probe sizes.!
confirmed that the nanoparticles were nm-scale bcc Fe-Co alloy crystals, with a larger diameter for the Fe 0.67Co0.33 sample compared to the Fe0.5Co0.5 sample. The size difference was indicated from the known effect of nanocrystal size on the Debye-Waller factors obtained from EXAFS. Interestingly, the EXAFS results also indicated that Fe is partially present in the form of an iron oxide compound, and that the Fe concentration in the bcc Fe-Co alloy is correspondingly reduced. It was concluded that the Fe-Co alloy nanoparticles are partially oxidized, with an iron oxide layer present at the surface. It is presumed that this occurs on exposure to air after reduction as a consequence of Fe being more easily oxidized than Co. This was more noticeable in the Fe0.5Co0.5 sample than the Fe0.67Co0.33 sample, which can be attributed to it having smaller nanoparticles with a higher surface to volume ratio than the Fe0.67Co0.33 sample. This scenario is illustrated for the Fe0.5Co0.5 sample in Figure 11 ~B!. Despite being a chemically sensitive technique, X-ray absorption spectroscopy nevertheless gives results representing the average of the entire sample. Additional techniques that are sensitive to composition on the scale of individual nanoparticles are especially useful for the study of these materials. The EDX and EELS results presented here are of great utility because they provide compositional information at the nm-scale. The Fe 0.5Co0.5 sample was studied using EELS, and the results showed x ⫽ 0.47 6 0.06 for single nanoparticles. The two elements Fe and Co are approximately equally and uniformly distributed within the nanoparticles to an
estimated precision of 60.05 in x and 60.4 nm in analysis position. EDX spectra obtained when the probe was scanned over a 220 ⫻ 180 nm 2 area of sample gave values of x ⫽ 0.48 6 0.06 and 0.68 6 0.05 for the Fe0.5Co0.5 and Fe0.67Co0.33 samples, respectively. When the EDX probe was positioned at the centers of ;20 individual nanoparticles, the mean values of x obtained were ^x& ⫽ 0.43 6 0.01 for the Fe0.5Co0.5 sample, and ^x& ⫽ 0.64 6 0.02 for the Fe 0.67Co0.33 sample. These mean values are lower than expected. It might be proposed that the values of x obtained from EDX have been underestimated if the value of the Cliff Lorimer sensitivity factor is too large ~kCoFe ⫽ 1.04 6 0.19! The lowest possible value is kCoFe ⫽ 1.00 ~because Co has a higher atomic number than Fe!. Using instead a value of kCoFe ⫽ 1.00 would give mean values of ^x& ⫽ 0.44 6 0.01 for the Fe0.5Co0.5 sample, and ^x& ⫽ 0.65 6 0.02 for the Fe0.67Co0.33 sample. So there seems a clear discrepancy for the Fe0.5Co0.5 sample, and this is difficult to reconcile with a scenario where nanoparticles consist of a single homogeneous alloy phase, as illustrated in Figure 11 ~A!. Before considering an alternative scenario in the following paragraph, we note that EDX measurements of ^x& may be affected if ~1! too few particles are measured limiting statistical accuracy or ~2! the nanoprobe has too high current density causing beam damage. Regarding the previous studies of these FeCo-SiO2 aerogel ~Carta et al., 2007! and similar FeCo-SiO2 xerogel ~Falqui et al., 2003! nanocomposite materials, it is important to consider an alternative scenario where the nanoparticles are coated with a thin iron oxide layer, as illustrated in Figure 11 ~B!. In the latter study, spatially resolved EELS of FeCo particles with an average size around 20 nm, embedded in a silica xerogel matrix, clearly showed the presence of a 3 nm thick iron oxide rich layer ~although that study used a significantly different sample preparation method!. The presence of an iron oxide layer in the Fe 0.5Co0.5 sample in the present study would consequently mean a reduction of Fe content in the alloy core of the nanoparticle and would provide an explanation for why positioning the EDX probe in the center of nanoparticles gives a slightly lower than expected mean value of ^x& ⫽ 0.42 6 0.01 for the Fe 0.5Co0.5 sample. This scenario would also predict that the distribution of Co is more concentrated in the core of the nanoparticles, and the distribution of Fe extends over a wider range to include the iron oxide layer. There is a slight suggestion from the EELS results in Figure 10a of increased concentration of Co in the center of the nanoparticle, and in Figure 10b of a slightly wider distribution of Fe; however, these observations are within the error of the measurement. This may mean that a very thin oxide layer is beyond the detection capabilities of the present study @see Fig. 11 ~B!#. Assuming a 4 nm diameter nanoparticle with overall composition Fe0.5Co0.5 , then a 0.2 nm thick layer of iron oxide would contain 25% of the Fe, and the core of the nanoparticle would have a composition Fe0.43Co0.57 . Such a scenario would be detectable by EXAFS and XANES but possibly not
Study of Fe-Co Alloy Nanoparticles Using HREM, EDX, and EELS
by EDX and EELS. For the Fe0.67Co0.33 sample, assuming a 7 nm diameter nanoparticle, then a 0.2 nm thick iron oxide layer would contain 10% of the Fe, and the core of the nanoparticle would have a composition Fe 0.63Co0.37 . It is also difficult for HREM to detect the presence of a very thin ~e.g., 0.2 nm! iron oxide layer, especially because such a layer may be poorly crystalline, and the nanoparticle is superimposed on the amorphous silica background from the aerogel matrix. An exception is found in the HREM image in Figure 4d, which shows a large nanoparticle consisting of a Fe 0.67Co0.33 alloy bcc crystal grain oriented on the @001# zone axis, with the appearance of some additional lattice fringes at the lower edges of the particle. The corresponding FFT shows diffraction spots from the ~110! and ~200! planes. These diffraction spots lie on a square in this zone axis, and this is indicated by a dashed square that is drawn over the FFT in the inset. Inside the square, there is evidence of additional diffraction spots, indicated by the arrows, which correspond to estimated planar spacings of 2.12 6 0.05 Å and 2.43 6 0.06 Å ~which cannot be caused by delocalization of the lattice planes due to defocus!. As would be expected for an oxide structure, these are larger than the planar spacing of 2.03 Å for ~110! in the Fe-Co alloy bcc structure. The estimated planar spacings are quite similar to the planar spacings of 2.14 and 2.47 Å for the strongly scattering ~200! and ~111! planes, respectively, in FeO. They are also similar to the planar spacings of 2.09 and 2.53 Å for the strongly scattering ~400! and ~311! planes, respectively, in Fe 3O4 and g-Fe2O3 ~which have the same spinel structure!. They are not similar to planar spacings of strongly scattering planes in a-Fe2O3 .
C ONCLUSIONS The BF STEM and HREM images confirm that there are distinct well-rounded nanoparticles present with typical diameters of 4 nm in the Fe 0.5Co0.5 sample and larger typical diameters of 7 nm in the Fe 0.67Co0.33 sample. The HREM images show that most nanoparticles consist of single bcc crystal grains. The EDX spectra summed over areas of the samples gave compositions FexCo1⫺x with x ⫽ 0.48 6 0.06 and 0.68 6 0.05 for Fe 0.5Co0.5 and Fe0.67Co0.33 samples as expected. The EDX spectra obtained with the 1.5 nm probe positioned at the centers of ;20 nanoparticles gave somewhat lower concentrations of Fe, with mean values of ^x& ⫽ 0.43 6 0.01 and ^x& ⫽ 0.64 6 0.02, respectively. The Fe 0.5Co0.5 sample was studied using EELS. The EELS spectra summed over whole nanoparticles gave x ⫽ 0.47 6 0.06. The EELS spectra from analysis profiles of nanoparticles show a distribution of Fe and Co that is homogeneous, i.e., x ⫽ 0.5, within a precision of at best 60.05 in x and 60.4 nm in position. The present study has provided the first compositional information on the nmscale for these materials but has not definitely shown if
123
nanoparticles are surrounded by an iron oxide layer. This contrasts with the only previous study of a similar material ~Falqui et al., 2003! when EELS was used to identify a ;3 nm thick iron oxide rich layer on a 20 nm diameter FeCo alloy nanoparticle in a silica xerogel matrix. In the present study of much smaller nanoparticles, a very thin iron oxide layer ~e.g., 0.2 nm! may be at the limits of detectability of EELS or EDX, but it could nevertheless contain enough Fe to influence EXAFS and XANES results ~see the Discussion section!.
A CKNOWLEDGMENTS We thank D. Loche for preparation of samples and the EPSRC UK for access to the SuperSTEM facility. This work was supported by a European Community Sixth Framework Programme Marie Curie Intra-European Fellowship ~Contract MEIF-CT-2005-024995!. We thank C.E. Lyman for much helpful guidance in improving this manuscript.
R EFER ENCES Abeles, B. ~1976!. Granular metal films. In Applied Solid State Science, Wolfe, R. ~Ed.!, pp. 1–117. New York: Academic Press. Brinker, C.J. & Scherer, G.W. ~1990!. Sol-Gel Science. San Diego, CA: Academic Press. Carta, D., Mountjoy, G., Gass, M., Navarra, G., Casula, M.F. & Corrias, A. ~2007!. Structural characterization study of FeCo alloy nanoparticles in a highly porous aerogel silica matrix. J Chem Phys 127, 204705. Casu, A., Casula, M.F., Corrias, A., Falqui, A., Loche, D., Marras, S. & Sangregorio, C. ~2008!. The influence of composition and porosity on the magnetic properties of FeCo-SiO 2 nanocomposite aerogels. Phys Chem Chem Phys 10, 1043–1052. Casula, M.F., Concas, G., Congiu, F., Corrias, A., Falqui, A. & Spano, G. ~2005!. Near equiatomic FeCo nanocrystalline alloy embedded in an alumina aerogel matrix: Microstructural features and related magnetic properties. J Phys Chem B 109, 23888–23895. Casula, M.F., Loche, D., Marras, S., Paschina, G. & Corrias, A. ~2007!. Role of urea in the preparation of highly porous nanocomposite aerogels. Langmuir 23, 3509–3512. Cliff, G. & Lorimer, G.W. ~1975!. The quantitative analysis of thin specimens. J Microsc 103, 203–207. Egerton, R.F. (1996!. Electron Energy-Loss Spectroscopy in the Electron Microscope, pp. 210–216. New York: Plenum Press. Ennas, G., Casula, M.F., Falqui, A., Gatteschi, D., Marongiu, G., Piccaluga, G., Sangregorio, C. & Pinna, G. ~2001!. Nanocrystalline iron-cobalt alloys supported on a silica matrix prepared by the sol-gel method. J Non-Cryst Solids 293–295, 1–9. Falqui, A., Serin, V., Calmels, L., Snoeck, E., Corrias, A. & Ennas, G. ~2003!. EELS investigation of FeCo/SiO2 nanocomposites. J Microsc 210, 80–88.
124
Andrea Falqui et al.
Gich, M., Casas, L., Roig, A., Molins, E., Sort, J., Surinach, S., Baro, M.D., Munoz, J.S., Morellon, L., Ibarra, M.R. & Nogue, S. ~2003!. High-coercivity ultralight transparent magnets. Appl Phys Lett 82, 4307–4309. Guillermet, A.F. ~1988!. Critical evaluation of the thermodynamic properties of the Fe-Co system. High Temp High Press 19, 477–499. Husing, N. & Schubert, U. ~1998!. Aerogels airy materials: Chemistry, structure, and properties. Angew Chem Int Ed 37, 22–45. Li, J.S., Mirzamaani, M., Bian, X.P., Doerner, M., Duan, S.L., Tang, K., Toney, M., Arnoldussen, T. & Madison, M. ~1999!. 10 Gbit/in.2 longitudinal media on a glass substrate. J Appl Phys 85, 4286–4291. MacLaren, J.M., Schulthess, T.C., Butler, W.H., Sutton, R. & McHenry, M.J. ~1999!. Electronic structure, exchange interactions, and Curie temperature of FeCo. J Appl Phys 85, 4833–4835.
Moreno, E.M., Zayat, M., Morales, M.P., Serna, C.J., Roig, A. & Levy, D. ~2002!. Preparation of narrow size distribution superparamagnetic gamma-Fe2O3 nanoparticles in a sol-gel transparent SiO2 matrix. Langmuir 18, 4972–4978. Paduani, C. & Krause, J.C. ~1999!. Electronic structure and magnetization of Fe-Co alloys and multilayers. J Appl Phys 86, 578–583. Pierre, A.C. & Pajonk, G.M. ~2002!. Chemistry of aerogels and their applications. Chem Rev 102, 4243–4265. Saad, A.M., Mazanik, A.V., Kalinin, Y.E., Fedotova, J.A., Fedotov, A.K., Wrotek, S., Sitnikov, A.V. & Svito, I.A. ~2004!. Structure and electrical properties of CoFeZr-aluminium oxide nanocomposite films. Rev Adv Mater Sci 8, 152–157. Williams, D.B. & Carter, C.B. ~1996!. Transmission Electron Microscopy, vol. 4, pp. 608–611. New York: Plenum Press.
