Carbon Nanocapsule-Mediated Formation of Ferromagnetic Fe5Si3

Jul 29, 2009 - Teddy M. Keller†. Chemistry DiVision, Material Science & Technology DiVision and SAIC, NaVal Research Laboratory,. Washington, D.C. ...
0 downloads 0 Views 3MB Size
J. Phys. Chem. C 2009, 113, 14663–14671

14663

Carbon Nanocapsule-Mediated Formation of Ferromagnetic Fe5Si3 Nanoparticles Manoj K. Kolel-Veetil,*,† Syed B. Qadri,‡ Michael Osofsky,‡ Ramasis Goswami,§ and Teddy M. Keller† Chemistry DiVision, Material Science & Technology DiVision and SAIC, NaVal Research Laboratory, Washington, D.C. 20375 ReceiVed: May 5, 2009; ReVised Manuscript ReceiVed: June 24, 2009

Magnetic and structural characteristics were investigated for the nanoparticles of Fe5Si3 grown by a carbonmediated, kinetically controlled pyrolysis of a thermosetted ferrocenylsiloxane polymeric network. Kinetics of the pyrolysis determined the volume fraction of the crystallographic phase of Fe5Si3 or bcc-Fe that was formed. Nanoparticles of Fe5Si3 were formed during pyrolysis of the network to 1000 °C at 10 °C/min, and silicon-doped bcc-Fe nanoparticles resulted when the network was ramped to 1000 °C at a slower rate of 1 °C/min. Transmission electron microscopy and X-ray diffraction studies revealed that Fe5Si3 nanoparticles were contained in carbon nanocapsules, and the bcc-Fe nanoparticles were associated with linear carbon fibers, an outcome of the distinct interfacial energies of the Fe-Si growth phase and carbon surface at the two pyrolysis rates. Fe5Si3 nanoparticles were found to be ferromagnetic with a Curie temperature of ∼375 K by magnetic measurements. Introduction Metal silicides have been typically synthesized by traditional solid-state reactions1 or by ion beam-induced mixing (IBM) of alternating metal and silicon layers.2 Among the known binary iron silicides, the iron-rich Fe5Si33 is a high-temperature silicide phase that is metastable with respect to FeSi and Fe3Si below 825 °C.4,5 It has a Curie temperature (TC) of 385 K3e and is known to possess impressive giant magnetoresistance (GMR) properties3 and the potential for applications in spintronics.6 In one instance, Fe5Si3 nanogranular particles formed at a Fe/c-Si interface on irradiation with 100 MeV swift heavy ions of Fe7+ were reported to exhibit a GMR up to 2400%.3d During solid-state reactions of the Fe-Si couple, among the possible iron silicides as evident in the Fe-Si phase diagram (Figure 1), FeSi has been predicted to form as the first silicide because of its favorable ∆Gv (change in free energy per unit volume or the driving force for nucleation) characteristics.1a This prediction was corroborated by the formation of FeSi as the first phase in amorphous deposited elemental layers of silicon and iron in a thin-film diffusion couple during thermal annealing.2 Other iron silicides such as Fe3Si, FeSi2, and Fe5Si3 were produced subsequent to the formation of FeSi. It was reasoned that metastable phases such as Fe5Si3 will be formed when they are easier to nucleate than the more thermodynamically stable phases.2 Interestingly, even in two recently reported vapor phase reactions involving Fe and Si components, FeSi has been reported to form as the first silicide.8,9 In the first report, FeSi nanowires were synthesized on silicon substrates covered with a thin layer of silicon oxide by the decomposition of the singlesource organometallic precursor trans-Fe(SiCl3)2(CO)4 in a CVD process.8 In the other report, FeSi or Fe5Si3 nanowires were produced by a phase-controlled vapor transport method from * To whom correspondence should be addressed. Manoj.kolel-veetil@ nrl.navy.mil. † Chemistry Division. ‡ Material Science & Technology Division. § Material Science & Technology Division and SAIC.

10.1021/jp904188f

FeI2 and SiI4. The nature of the product was found to depend on the concentration ratio of FeI2(g) and SiI4(g).9 Thus, it seems obvious that a means to produce an iron silicide other than FeSi, especially a metastable one, as the first phase from the solid-state reaction of the Fe-Si couple would be to introduce additional γ (interfacial free energy) into the system with an extraneous component so as to create a ∆Gv force scenario that is favorable for the formation of the targeted iron silicide. In such heterogeneous solid-state reactions, the rate of the heterogeneous nucleation, neglecting the strain energy, is defined as10

I ) I0 exp(-∆G*/kT) where I is the rate of the heterogeneous nucleation, I0 is the preexponential factor, k is Boltzman’s constant, T is temperature, and ∆G* (activation barrier for nucleation) ) 16πγ3f(θ)/3∆Gv2, where γ and ∆Gv are as defined above and f(θ) is the shape factor with θ being the wetting angle between the catalytic surface and the product phase. The formation of phases in such nucleations depends on the outcome of the interplay between ∆Gv, γ, and f(θ). Thus, for a given ∆Gv and γ, the nucleation rate will be higher with lower θ. In this report, we describe for the first time the use of carbon as an extraneous component during the solid-state reaction of the Fe-Si couple, which facilitates the formation of nanoparticles of metastable Fe5Si3. Additionally, the thermoset of the ferrocenylsiloxane polymer FS11 (Figure 2), which serves as the precursor for the Fe-Si couple and carbon of this study, yields nanoparticles of Fe5Si3 or silicon-doped body centered cubic (bcc)-Fe, respectively, during its faster or slower pyrolysis due to distinctive θ and γ between the evolving Fe-Si phase and carbon moieties. The products have been characterized by X-ray diffraction (XRD), transmission electron microscopy (TEM), and magnetic measurements.

This article not subject to U.S. Copyright. Published 2009 by the American Chemical Society Published on Web 07/29/2009

14664

J. Phys. Chem. C, Vol. 113, No. 33, 2009

Kolel-Veetil et al.

Figure 1. Fe-Si phase diagram (reproduced from ref 7).

Figure 2. Diacetylene-containing ferrocenylsiloxane polymer FS and its conversion into a thermoset by heating.

Experimental Methods The linear polymer FS was synthesized following a previously reported procedure.11 All of the thermal conversions involving FS were performed using a SDT 2960 DTA-TGA analyzer under a nitrogen flow rate of 100 cc/min. For the formation of the Fe5Si3 nanoparticles, FS was thermally treated as follows: Heating at 10 °C/min to 250 °C, followed by an isothermal treatment for 60 min, and subsequent ramp at 10 °C/min to 400 °C, followed by an isothermal treatment for 60 min resulted in the formation of the thermoset. The thermoset was subsequently heated at 10 °C/min to 1000 °C to produce the Fe5Si3 nanoparticle-containing composition. The corresponding heating cycles during the formation of the silicon-doped bcc-Fe nanoparticles were as follows: For conversion to the thermoset, the thermal cycle was the same as that for the Fe5Si3 nanoparticles, except that the ramp rate was 1 °C/min. For conversion to the silicon-doped bcc-Fe nanoparticle-containing composition, the thermoset was slowly heated at 1 °C/min to 1000 °C.

X-ray analyses were performed using a Rigaku 18 kW X-ray generator and a high-resolution powder diffractometer. X-ray diffraction (XRD) scans of the samples were obtained using Cu KR radiation from a rotating anode X-ray source. Temperaturedependent ac magnetoresistance measurements were performed at 13 Hz in magnetic fields up to 6 T on a Quantum Design Physical Property Measurement System. Leads were attached to the bulk samples with indium solder, and measurements were performed with a 1 mA excitation current for the sample containing the Fe5Si3 nanoparticles and a 0.1 mA current for the sample containing the silicon-doped bcc-Fe nanoparticles. Curie temperatures of the samples were determined by measurements of the magnetic moment versus temperature (300 K < T < 600 K) of the samples in a 1 T magnetic field in a Digital Measurements Systems Technologies DMS-886 vibrating sample magnetometer (VSM). A JEOL-2200 analytical transmission electron microscope (TEM), operated at 200 kV, was used to obtain the microstructure and composition of the pyrolysis residues. Specimens for TEM analysis were made by crushing the material in ethanol into a fine powder and transferring a few drops containing the fine powder onto a carbon-coated copper grid. Thermocalc software using the Scientific Group Thermodata Europe (SGTE) database was used to calculate the Gibbs free energy versus composition diagram at different temperatures. Results and Discussion The Fe:Si mole ratio in FS is 1:4. With respect to the Fe-Si binary reaction couple, this constitutes a mole fraction of 0.8

Formation of Ferromagnetic Fe5Si3 Nanoparticles

Figure 3. XRD spectrum of the product containing silicon-doped bccFe nanoparticles obtained from FS thermoset by pyrolysis at a slower rate.

(or 80 atom %) for Si in FS, making it a highly silicon-rich system. On the basis purely of ∆Gv considerations during nucleation and as apparent from the Fe-Si phase diagram,7 a 0.8 mole fraction of Si in FS is expected to yield predominantly a mixture of FeSi and R-FeSi2 or β-FeSi2 to temperatures of 1000 °C at either the slower or faster pyrolysis rate used in this study. However, the presence of the graphitic surface (vide infra) derived from the catalytic interaction of Fe on C in FS was observed to inject variability in the product formation. Even though starting from ∼600 °C to 800 °C essentially bcc-Fe nanoparticles were formed at the two pyrolysis rates, XRD analysis of the products revealed that a divergence in the reaction pathways occurred above 800-900 °C at the two pyrolysis rates. These temperatures coincide with the temperature range in which carbon nanotube formation has been reported during pyrolysis of carbonaceous thermosets of ferrocenyl compounds.12 By 1000 °C, the differentiation in the reaction pathways was complete as silicon-doped bcc-Fe nanoparticles13 (Figure 3) and Fe5Si3

J. Phys. Chem. C, Vol. 113, No. 33, 2009 14665 nanoparticles14 (Figure 4) were formed as the major Fe species during the slower and faster pyrolysis, respectively. After the thermoset formation and subsequent conversion of FS into either Fe5Si3 or silicon-doped bcc-Fe nanoparticlecontaining composition at 1000 °C, the resulting product in each instance retained about 42% of the initial mass of FS. If the complete conservation of the available Fe and Si in FS is to be assumed in the product, it is then required that the observed product mass be derived by the loss of ∼92% of the C, H, and O present in FS during the product formation. In other words, either the Fe5Si3 or the silicon-doped bcc-Fe nanoparticlecontaining composition that is produced should be composed of up to ∼81% of Fe5Si3 or silicon-doped bcc-Fe nanoparticles, respectively. In fact, on the basis of the XRD and TEM analyses (vide infra) of the products, this was estimated to be the case and C was found as the only element present other than Fe and Si in the products. The XRD spectrum of the product obtained at 1000 °C from faster pyrolysis of the thermoset of FS clearly exhibited diffraction peaks of the hexagonal (D8) phase of η-Fe5Si3 [JCPDS 74-4745; space group, P63/mcm(193)] with a silicon atom content of 37.5% (Figure 4).14 In addition, peaks for graphitic carbon nanostructures (2θ ∼ 26.6) and silicon carbide (2θ ∼ 35.5) were also observed. The five most intense characteristic diffraction peaks of the produced iron silicide appeared at 2θ values between 40 and 50 and were found to match well in intensity and 2θ values of a simulated diffraction spectrum of η-Fe5Si3 (Figure 4). The presence of a minor amorphous background was discernible in the spectrum. Using a single-peak analysis of the XRD spectrum, the apparent crystallite size “D” can be related to the Couchy (βC) and Gaussian (βG) widths of the diffraction peak at the Bragg angle θ. The constituent Couchy and Gaussian components can be obtained from the ratio of the full width at half-maximum intensity (2ω) and integral width (β).15 Using these parameters, we estimated the grain size of the Fe5Si3 particles to be 10-15 nm. However, TEM analysis revealed that even though the majority of the particles existed in the 10-15 nm range, ∼20% of the larger Fe5Si3 nanoparticles present were as large as 40 nm (Figure 5b). The energy dispersive X-ray spectrum (EDS)

Figure 4. XRD spectrum of the product obtained from FS thermoset at a faster rate of pyrolysis, exhibiting the diffraction peaks of Fe5Si3 nanoparticles, SiC, and graphite (carbon nanocapsules) (top). The simulated diffraction pattern of Fe5Si3 is shown at the bottom.

14666

J. Phys. Chem. C, Vol. 113, No. 33, 2009

Kolel-Veetil et al.

Figure 5. TEM micrographs of the pyrolysis product from FS thermoset obtained (a) at a slower rate, showing the structure of bcc-Fe nanoparticles in a carbon matrix and (b) at a faster rate, exhibiting the structure of Fe5Si3 nanoparticles. (c) EDS of a bcc-Fe particle showing only Fe KR and Kβ lines. (d) EDS from a Fe5Si3 particle showing Fe KR, Fe Kβ, and Si KR lines. Cu KR and Cu Kβ lines in panels c and d originate from the Cu grid used in TEM measurements.

of a majority of the particles exhibited Fe KR, Fe Kβ, and Si KR peaks of the constituent elements of the Fe5Si3 nanoparticles (Figure 5d). The presence of a minor fraction of bcc-Fe nanoparticles was also observed by TEM. Thus, during faster pyrolysis, a complete exclusion of the formation of bcc-Fe nanoparticles may not be possible at temperatures near 1000 °C. In comparison, the XRD spectrum of the product obtained at 1000 °C from slower pyrolysis of the thermoset of FS exhibited essentially the peaks for Fe nanoparticles (with a major peak at 2θ ) 45) in its bcc phase (JCPDS 06-0696; space group, Im3m)13 imposed on an amorphous background containing carbon nanostructures and silicon carbide (Figure 3). Using the single peak analysis as described above, we calculated the grain size of the silicon-doped bcc-Fe nanoparticles to be about 15 nm. In contrast, TEM evaluation of the product revealed that there was a bimodal distribution of particles in the ranges of 10-40 and 75-100 nm (Figure 5a). The EDS of the majority of the particles in this sample exhibited Fe KR and Fe Kβ peaks for Fe in the bcc-Fe nanoparticles (Figure 5c). In addition, the presence of a minor fraction of Fe5Si3 nanoparticles was also observed in this product as evident from the XRD and TEM studies. Thus, it appears that during slower pyrolysis a small probability exists in this system for the formation of the metastable iron silicide Fe5Si3 at temperatures near 1000 °C. When heated above 600 °C, the formation of the silicondoped bcc-Fe nanoparticles as the first product, instead of the thermodynamically favored FeSi or R- or β-FeSi2, implied a scarcity of Si for reaction with Fe in the evolving reaction system

derived from FS. As Si is known to be a ferrite (bcc-Fe phase) stabilizer, it facilitates the formation of the silicon-doped bccFe nanoparticles due to its limited solubility in Fe at the apparently low Si mole fraction under a diffusion-controlled situation.16 In diffusion-controlled reactions of multilayered films containing the Fe-Si couple, below 10 atom % of Si, a solid solution of Si in bcc-Fe (B2) has been observed.17 However, under the slow rate of heating used in this study, the rate and extent of dissolution of Si in Fe does not seem to be sufficient enough to cause the formation of FeSi or a metastable silicide under ∆Gv considerations. A similar dissolution of a limited amount of Fe in the available Si to form a silicon-rich silicide can be ruled out as the solubility of Fe in crystalline Si has been established to be extremely small (less than 0.005 atom %) even at temperatures as high as 1300 °C.18 This inferred reduction in the Si mole fraction around the reaction sphere of Fe is proposed to be due to the sequestration of Fe and Si into exclusive regions in the evolving reaction matrix by the intervening C (graphite) reactant. In addition, the formation of austenite or cementite phases on C alloying in Fe can be ruled out on the basis of XRD and TEM evidence. The divergence in the reaction pathway and formation of distinct Fe species above 800-900 °C during pyrolysis at the two rates is believed to be an outcome of the different interfacial interactions that the evolving silicon-dissolved iron species have with the carbon component owing to the variation in the catalytic abilities of the two Fe species (Scheme 1).19 During faster pyrolysis, the evolving Fe entities with a greater amount of dissolved Si will preferentially etch the surrounding graphitic structure more than catalyze its linear growth, resulting in its

Formation of Ferromagnetic Fe5Si3 Nanoparticles

J. Phys. Chem. C, Vol. 113, No. 33, 2009 14667

SCHEME 1: Schematic Representation of Pathways to Form Nanoparticles of Silicon-Doped bcc-Fe and Fe5Si3, Respectively, at Slower and Faster Pyrolysis ratesa

a

The wetting angle, θ, of the (Fe, Si) growth boundary at the graphite surface in each pathway is dependent on the corresponding pyrolytic rate.

conversion to carbon nanocapsules. However, at slower pyrolysis, Fe catalyst entities, with a lower amount of dissolved Si, will preferentially catalyze the formation of linear graphitic nanofibers. Similar major modifications in graphitic nanofiber growth characteristics due to the presence of a metal or a nonmetal adatom such as Si in group 3d metal catalysts (Fe, Co and Ni) have been reported and are believed to be due to different wetting properties [γ, f(θ)] of the metal catalysts with graphite.19 Such adatoms manifest as perturbations to the gas-metal and metal-solid carbon interfaces resulting in a change in the catalytic activity of the carrier catalyst. Under such a scenario, there is presumed to be a competition between the etching rate of the evolving graphitic moiety and the catalytic efficiency of the metal catalyst to produce more of the graphitic moiety. It has been postulated that when the etching rate dominates, encapsulation of the catalyst by the graphitic moiety occurs, and conversely, when the catalytic efficiency dominates, the formation of linear graphitic nanostructures takes place.19c Thermodynamic calculations of the Gibbs free energy were performed in the 800-900 °C range, where divergence of the reaction pathway occurs, as a function of mole fraction of Si for predicting the different phases of the Fe-Si system (Figure 6). The results clearly indicate that there is a considerable driving force for the formation of Fe5Si3 from liquid Fe-Si below 50% Si. The Fe5Si3 phase is found to be in equilibrium with the bccFe(Si) phase. A common tangent between bcc-Fe(Si) and Fe5Si3 can be drawn at 900 °C, suggesting that bcc Fe(Si) with 30% Si can be in equilibrium with Fe5Si3. Even though FeSi is also favored to be formed at this temperature, the presence of the extraneous carbon component shifts the Si mole fraction into the 0.3-0.4 range possibly due to the separation of the Fe and Si reactants, thereby limiting the availability of Si for reaction with Fe and thwarting the formation of FeSi. The carbon-aided encapsulation appears to provide a means for accessing a metastable Fe-silicide via a kinetically active

Figure 6. Gibbs free energy calculation at 1173 K (900 °C) in relation to the mole fraction of Si, calculated using Thermocalc software, for the different phases of the Fe-Si system. For this calculation, the enthalpies of pure Fe and Si at 298 K were taken as standard states.

zone that is usually not achievable in diffusion-limited solidstate reactions.1 In fact, with a pure Fe-Si couple, these types of highly disordered and kinetically active zones are produced only in IBM reactions at the interfaces of the Fe-Si couple due to rapid but transient atomic motions that produce alloying of elements in neighboring layers under highly nonequilibrium conditions that result in the formation of amorphous or metastable crystalline alloys.20 In addition, the prevailing mediation by C seems to prevent the formation of the ∆Gvfavored FeSi in the system. Evidence for the structure of the carbon architectures associated with the Fe species in the two cases were obtained from high-resolution transmission electron microscopy (HRTEM) analyses of the products. Carbon is found to exist as fragmented bundles and fibers radiating from the surfaces of the bcc-Fe nanoparticles (Figure 7). This was further corroborated by a

14668

J. Phys. Chem. C, Vol. 113, No. 33, 2009

Kolel-Veetil et al.

Figure 7. (Top) HRTEM image showing bcc-Fe nanoparticles with protruding carbon fibers. (a) HAADF image of two bcc-Fe particles. (c) EDS line scan across the two particles showing the presence of Fe. (b,d) Fine probe EDS maps for Fe and Si (present dispersed in the matrix), respectively.

Figure 8. (a) HRTEM image showing a Fe5Si3 particle encapsulated by a carbon nanocapsule. (b) High magnification of a portion of the carbon nanocapsule showing the (0002) lattice fringes. (c) HAADF image of two Fe5Si3 particles encapsulated by carbon. (e) EDS line scan across the two particles showing the presence of Fe and Si. (d,f) Fine probe EDS maps for Fe and Si, respectively.

high-angle annular dark-field (HAADF) image of two neighboring bcc-Fe particles (Figure 7a). HAADF imaging provides compositional information by taking advantage of the high atomic number (Z) contrast,21 i.e., the scattering cross-section in such a case is proportional to Z2. In this imaging mode, the brighter regions correspond to heavier atoms. The line scan across the two particles essentially revealed that they consisted of Fe (Figure 7c). The fine probe EDS maps for Fe and Si showed that Fe was concentrated in the two particles, and that Si was dispersed in the matrix in a diluted form (Figure 7b,d). In the case of Fe5Si3 nanoparticles, HRTEM of the particles revealed that carbon was present as nanocapsules encapsulating the Fe5Si3 particles (Figure 8a,b). A HAADF image of two neighboring Fe5Si3 nanoparticles corroborated the concentration of the heavier elements in the two particles (Figure 8c). The line scan across the two particles showed that the particles consisted of Fe and Si (Figure 8e). Furthermore, fine probe EDS maps for Fe and Si revealed that Fe and Si were concentrated in the two particles, and that additional Si was concentrated in regions around the Fe5Si3 nanoparticles (Figure 8d,f). Interestingly, even though the slower rate pyrolysis product contained a minor fraction of carbon-encapsulated Fe5Si3 nanoparticles, there was no discernible evidence of any carbon encapsulation of the bcc-Fe nanoparticles in the TEM studies. Examples of carbon-encapsulated bcc-Fe nanoparticles have been reported during an arc-discharge process using a graphite cathode and graphite anode filled with bulk iron metal,22a arcevaporation of a Fe-packed graphite rod in a modified fullerene generator,22b ultrasonic irradiation of a slurry of carbon spherules and Fe(CO)5 in diphenyl methane,22c detonation-induced pyrolysis of ferrocene,22d chemical vapor synthesis using iron acetyl acetonate,22e thermolysis of a sodium azide-cholorocarbon mixture in the presence of ferrocene,22f and laser-assisted chemical vapor decomposition of ferrocene.22g In light of such formations, it is interesting that in the current system the

encapsulation of an Fe silicide is favored over the encapsulation of Fe under the various ∆Gv and [γ, f(θ)] conditions prevalent in this study. Thus, in the present system, the results clearly suggest that Fe with a higher amount of Si is associated with the formation of carbon nanocapsules, while Fe with a lower level of Si facilitates the linear growth of graphitic nanofibers. It is important to consider whether the Fe5Si3 or the silicondoped bcc-Fe nanoparticles produced in either of the reactions could have originated from the ∆Gv-favored FeSi species. In this regard, the reported precipitation of R-Fe (silicon-doped bcc-Fe) and Fe3Si phases at an Fe-Si interface after exposures to 10 keV Xe+ ions and subsequent vacuum annealing at 700 °C is relevant.23 In this system, the product of the “asprepared” sample after synthesis by IBM was found to contain mainly FeSi as the iron silicide. However, on annealing the sample to 700 °C, the formation of Fe3Si and bcc-Fe was observed. The precipitation of Fe3Si at 700 °C was explained as being due to the disproportionation of the FeSi phase, which also resulted in the concomitant precipitation of the bcc-Fe phase. Because Fe5Si3 is known to be metastable with respect to FeSi and Fe3Si, FeSi cannot function as the precursor for Fe5Si3 formed in the current study. Similarly, the absence of Fe3Si in the silicon-doped bcc-Fe containing composition as deduced from XRD and Curie temperature measurements (vide infra) suggested that the silicon-doped bcc-Fe nanoparticles were not formed by the disproportionation of FeSi. The Curie temperatures (TC) of the products were determined by VSM measurements to further establish the identity of the products (Figure 9). The Fe5Si3-containing sample exhibited a TC of 375 K, which is near the reported TC of Fe5Si3 of 385 K.3e In comparison, the TC values of the other iron silicides, including that of Fe3Si at 799 K,24 Fe2Si at 500-508 K,24,25 and FeSi2 at 550 K,26,27 are higher than this determined TC value.

Formation of Ferromagnetic Fe5Si3 Nanoparticles

Figure 9. Measurements of the Curie temperatures of the compositions containing Fe5Si3 nanoparticles (bottom curve) and silicon-doped bccFe nanoparticles (top curve).

The TC of only FeSi at 28 K28 is below this value. This rules out the presence of any of these iron silicides and particularly that of the ∆Gv-favored FeSi in the sample obtained from faster pyrolysis. The observance of residual magnetism in this sample above 375 K is consistent with the presence of a small fraction of the silicon-doped bcc-Fe nanoparticles in the sample. The TC of the silicon-doped bcc-Fe nanoparticle-containing sample was found to be higher than the temperature limit of the measurement (573 K) (Figure 9), which is reasonable considering that the reported TC of bcc-Fe is 1043 K.29 However, an initial dip in the magnetization curve, consistent with the presence of a minority volume of material with TC ∼380 K, corroborated the presence of a small fraction of Fe5Si3 nanoparticles in this sample. Extrapolation of the magnetization curve (not shown) was found to place the TC of the silicon-doped bccFe sample at about 1055 K. Magnetic properties of the Fe5Si3- and silicon-doped bcc-Fe nanoparticle-containing products were of interest particularly in light of the reported GMR properties of Fe5Si3.3 Measurements of the temperature dependence of the resistivity of both of the samples exhibited insulating behavior. The magnetoresistance (MR) of both samples were measured between 0 and 6 T at temperatures of 100, 200, and 300 K, respectively. The MR values were estimated using the conventional definition, MR ) (R(H) - R(0))/R(0) × 100%, where, R(H) and R(0) are the resistance of a sample in the presence and absence of a magnetic field H, respectively. The GMR effect in heteroge-

J. Phys. Chem. C, Vol. 113, No. 33, 2009 14669 neous ferromagnetic insulator (FI) systems such as the Fe5Si3containing system of this study is proposed to occur at the boundary regions separating ferromagnetic and nonmagnetic materials rather than within the bulk of the ferromagnetic material.30 At room temperature, the Fe5Si3 nanoparticlecontaining product exhibited a negative MR, while the silicondoped bcc-Fe nanoparticle-containing product did not display any MR at all. Negative MR was also observed for the Fe5Si3 sample at 200 and 100 K, while positive MR was observed for the silicon-doped bcc-Fe sample at these temperatures. The MR of the Fe5Si3-containing product was determined to be -0.6, -0.7, and -0.9% at 300, 200, and 100 K, respectively (Figure 10). The corresponding values for the silicon-doped bcc-Fecontaining product was determined to be 0.0, +0.5, and +0.9%, respectively, (Figure 10). Thus, the highest MR observed for both samples was around 1%. The TEM images show that the magnetic Fe5Si3 nanoparticles are embedded in a carbon matrix of considerable volume with the particles rather distant from one another. Thus, it is likely that the MR of the Fe5Si3 nanoparticle-containing sample is far less than the 2400% GMR previously observed for the Fe5Si3 formed at the Fe/c-Si boundary3d because of weak coupling between the Fe5Si3 nanoparticles. This suggests that the coupling ability of carbon nanostructures is much lower than that of silicon with the Fe5Si3 nanoparticles. Furthermore, positive and negative magnetoresistance of this magnitude can be caused or influenced by several factors, including electronic band structure, weak localization, electron-electron interactions, and spin-orbit scattering. Because these phenomena are dependent on the microscopic properties of materials, it is not possible to determine the source of the MR without thorough chemical and structural studies of the interfacial regions of our samples. In addition, any quantitative treatment of the MR would require a more detailed analysis of the transport properties of the carbon matrix, which is beyond the scope of this work. The magnetization (M) of the samples was also studied as a function of the applied magnetic field (H) (Figure 11). The saturation moment of the sample with the Fe5Si3 nanoparticles was determined to be ∼7 emu/cm3, while the corresponding value for the silicon-doped bcc-Fe nanoparticle-containing sample was ∼16 emu/cm3. In comparison, the saturation moment for iron is 1714 emu/cm3.31 The small values are consistent with the large volume of the carbon matrix compared to the magnetic particles.

Figure 10. Magnetoresistance of the pyrolysis product of FS thermoset, at 100, 200, and 300 K, obtained at a faster rate (left), containing the Fe5Si3 nanoparticles, and at a slower rate (right), containing the silicon-doped bcc-Fe nanoparticles.

14670

J. Phys. Chem. C, Vol. 113, No. 33, 2009

Kolel-Veetil et al.

Figure 11. Magnetization of the Fe5Si3 nanoparticle-containing product (left) and silicon-doped bcc-Fe nanoparticle-containing product (right) as a function of the magnetic field. The hysteresis loops are depicted.

The M-H curves showed that both samples exhibited hysteresis at room temperature. The existence of hysteresis suggested the presence of magnetically coupled nanoparticles in the samples.32 The coercive field of the Fe5Si3-containing sample was found to be around 50 Oe. In reported studies of the magnetic properties of nanocrystalline Fe1-xSix alloys (0.15 < x < 0.34), the lowest coercive field of 2.9 Oe was found for a stoichiometry of Fe0.67Si0.33, which is the metastable Fe2Si (B2) phase.24,33 On annealing this metastable phase above 800 K, a decomposition into the neighboring stoichiometric Fe5Si3 and Fe3Si phases was induced, and the coercive field was observed to be enhanced by nearly an order of magnitude (from 2.9 to 20.8 Oe). This increase in coercivity was attributed to the formation of the hexagonal Fe5Si3 phase.24,33,34 Furthermore, a relationship between the coercive field and film thickness has been reported in in-plane hysteresis loops of sputtered polycrystalline films of Fe5Si3.35 For a 65 nm thick film, the coercive field at room temperature was found to be about 200 Oe. However, on increasing the film thickness to 500 nm, the room temperature coercive field was found to decrease to 97 Oe. Thus, the coercivity of our Fe5Si3 nanoparticle-containing sample seems to be in the range of reported coercivity values of Fe5Si3containing compositions and films. In comparison, the coercive field of the silicon-doped bcc-Fe nanoparticle-containing sample was found to be around 200 Oe, indicating its existence as a moderately hard magnetic system. Conclusions In summary, we describe the facile formation of nanoparticles of the metastable Fe5Si3 phase during pyrolysis of the thermosetted network of FS by the introduction of a carbon component in the Fe-Si couple. It is apparent that the metastable Fe5Si3 phase becomes stabilized due to the beneficial impact of carbon on the ∆Gv and γ parameters and, consequently, on the thermodynamic barriers associated with the diffusion and nucleation of the Fe and Si reactants. It is, thus, remarkable that the metastability of a phase in a binary system can be effectively mitigated, if favorable interfacial interactions can be brought into the system by an extraneous component. It is reasonable to expect that in reactions involving Fe and other alloying elements, carbon could be used as a mediating sequestering agent to facilitate the desired formation of specific phases due to the catalytic reactivity of Fe with carbon that results in the formation of various architectures of carbon around Fe in addition to its dissolution to form the ferrite phase. This should make the targeted synthesis of potential spintronic

candidates involving iron such as Fe1-xCoxSi,36 with unusual positive magnetoresistance and a large anomalous Hall Effect, a possibility upon a judicious utilization of this demonstrated ability of carbon. Acknowledgment. The authors thank the Office of Naval Research for financial support of this work. References and Notes (1) (a) Walser, R. M.; Bene, R. W. Appl. Phys. Lett. 1976, 28, 624. (b) Borders, J. A.; Picraux, S. I. Proc. IEEE 1974, 62, 1224. (2) Novet, T.; Johnson, D. C. J. Am. Chem. Soc. 1991, 113, 3398. (3) (a) Hines, W. A.; Menotti, A. H.; Budnick, J. I.; Burch, T. J.; Litrenta, T.; Niculescu, V.; Raj, K. Phys. ReV. B 1976, 13, 4060. (b) Lecocq, Y.; Lecocq, P.; Michel, A. C. R. Acad. Sci. 1964, 258, 5655. (c) Sawatzky, E. IEEE Trans. Magn. 1971, 374. (d) Srivatsava, P. C.; Tripathi, J. K. J. Phys. D: Appl. Phys. 2006, 39, 1465. (e) Langer, G.; Brown, P. Philos. Mag. 1967, 16, 521. (4) Okamoto, H. Desk Handbook: Phase Diagrams for Binary Alloys; ASM International: Materials Park, OH, 2000. (5) Massalski, T. B., Bennet, L. H., Murray, J. L., Baker, H., Eds. Binary Alloy Phase Diagrams; ASM International: Materials Park, OH, 1990; Vol. 2, p 1772. (6) Rouff, R. S.; Lorents, D.; Chan, B.; Malhotra, R.; Subramoney, S. Science 1993, 259, 346. (7) Brandes, E. A., Ed. Smithells Metals Reference Book, 6th ed.; Butterworths & Co.: London, 1983; pp 11-257. (8) Schmitt, A. L.; Bierman, M. J.; Schmeisser, Himpsel, F. J.; Jin, S. Nano Lett. 2006, 6, 1617. (9) Varadwaj, K. S. K.; Seo, K.; In, J.; Mohanty, P.; Park, J.; Kim, B. J. Am. Chem. Soc. 2007, 129, 8594. (10) Christian, J. W. In The Theory of Transformation in Metals and Alloys; Pergamon Press: Oxford, U.K., 1975. (11) Houser, E. J.; Keller, T. M. PMSE 1997, 76, 595. (12) (a) Keller, T. M.; Qadri, S. B. Chem. Mater. 2004, 16 (6), 1091. (b) Keller, T. M.; Laskoski, M.; Qadri, S. B. J. Phys. Chem. C., 2007, 111, 2541. (13) Basinski, Z. S.; Hume-Rothery, W.; Sutton, A. L. Proc. R. Soc. London, Ser. A. 1955, 229, 459. (14) Santamaria-Perez, D.; Nuss, J.; Haines, J.; Jansen, M.; Vegas, A. Solid State Sci. 2004, 6, 673. (15) Keljser, Th. H. de; Langford, J. I.; Mittemeijier, E. J.; Vogels, A. B. P. J. Appl. Crystallogr. 1982, 15, 308. (16) Raghavan, V. J. Alloy Phase Diagrams 1986, 2 (2), 97. (17) Bertotti, G.; Fiorillo, F. Magnetic Properties of Metals, LandoltBornstein New Series; Springer: Berlin, 1994; Vol. 19. (18) Struthers, J. D. J. Appl. Phys. 1956, 27, 1560. (19) (a) Chambers, A.; Rodriguez, N. M.; Baker, R. T. K. J. Phys. Chem. 1995, 99 (26), 10581. (b) Chambers, A.; Rodriguez, N. M.; Baker, R. T. K. J. Mater. Res. 1996, 11 (2), 430. (c) Tsai, S. H.; Lee, C. L.; Chao, C. W.; Shih, H. C. Carbon 2000, 38, 775. (20) Mayer, J. W.; Lau, S. S. In Surface Modification and Alloying by Laser, Ion, and Electron Beams; Poate, J. M., Foti, G., Jacobson, D. C., Eds.; Plenum: New York, 1983; pp 241-258. (21) (a) Pennycook, S. J.; Boatner, L. A. Nature 1988, 338, 565. (b) Browning, N. D.; Chisholm, M. F.; Pennycook, S. J. Nature 1993, 366, 143.

Formation of Ferromagnetic Fe5Si3 Nanoparticles (22) (a) Jiao, J.; Seraphin, S.; Wang, X.; Withers, J. C. J. Appl. Phys. 1996, 80, 103. (b) Masuda, M.; Maeda, K.; Kobayashi, T.; Shiomi, S.; Fujiwara, Y.; Saito, Y. Jpn. J. Appl. Phys. 2000, 39, L733. (c) Pol, V. G.; Motiei, M.; Gedanken, A.; Calderon-Moreno, J.; Mastai, Y. Chem. Mater. 2003, 15, 1378. (d) Lu, Y.; Zhu, Z.; Liu, Z. Carbon 2005, 43, 369. (e) Enz, T.; Winterer, M.; Stahl, B.; Battacharya, S.; Miehe, G.; Foster, K.; Fasel, C.; Hahn, H. J. Appl. Phys. 2006, 99, 044306. (f) Bystrzejewski, M.; Huczko, A.; Lange, H.; Cudzilo, S.; Kicinnski, W. Diamond Relat. Mater. 2007, 16 (2), 225. (g) Elihn, K.; Lanstrom, L.; Alm, O.; Boman, M.; Heszler, P. J. Appl. Phys. 2007, 101, 034311. Kudryavtsev, Y. V. L. K.; Lee, Y. P.; Dubowik, J.; Szymanski, B.; Rhee, J. Y. Phys. ReV. B. 2002, 65, 104417. (23) Ogale, S. B.; Joshee, R.; Godbole, V. P.; Kanetkar, S. M.; Bhide, V. G. J. Appl. Phys. 1985, 57 (8), 2915. (24) Varga, L. K.; Mazaleyerat, F.; Kovac, J.; Kakay, A. J. Magn. Magn. Mater. 2000, 215-216, 121. (25) Yelsukov, E. P.; Konygin, G. N.; Voronina, E. V.; Korolyov, A. V.; Ulyanov, A. I.; Godovikov, S. K.; Zagainov, A. V. J. Magn. Magn. Mater. 2000, 214, 258. (26) Kudryavtsev, Y. V.; Lee, Y. P.; Dubowik, J.; Szymanski, B.; Rhee, J. Y. Phys. ReV. B. 2002, 65, 104417. (27) Desimoni, J.; Sa´nchez, F. H.; Ferna´nzed van Rapp, M. B.; Lin, X. W.; Bernas, H.; Clerc, C. Phys. ReV. B 1996, 54 (18), 12787.

J. Phys. Chem. C, Vol. 113, No. 33, 2009 14671 (28) Walterfang, M.; Keune, W.; Trounov, K.; Peters, R.; Ru¨cker, U.; Westerholt, K. Phys. ReV. B 2006, 73 (21), 214423. (29) (a) Keffer, F. Handbuch der Physik; Springer-Verlag: New York, 1966, Vol. 18, Part 2. (b) Heller, P. Rep. Progr. Phys. 1967, 30, 731, Part II. (30) Levy, P. M. Solid State Physics; Academic Press: New York, 1994, Vol. 47, pp 367-463 and references therein. (31) (a) Gerstenberg, D. Ann. Physik. 1958, 2, 236. (b) Mattis, D. C. The Theory of Magnetism, 2nd ed.; Springer Verlag: Berlin, 1988. (c) Herring, C. In Magnetism; Rado, G. T., Suhl, H., Eds.; Academic Press: New York, 1966, Vol. 4. (32) Allia, P.; Coisson, M.; Knobel, M.; Tiberto, P.; Vinai, F. Phys. ReV. B. 1999, 60 (17), 12207. (33) Velez, M.; Morales, R.; Alameda, J. M.; Briones, F.; Martin, J. I.; Vincent, J. L. J. Appl. Phys. 2000, 87, 5654. (34) Park, J. S.; Kim, C. O.; Lee, Y. P.; Kudrayavtsev; Dobowik, J.; Szymanski, B.; Rhee, J. Y. J. Appl. Phys. 2002, 91 (10), 7194. (35) Sawatzky, E. IEEE Trans. Magn. 1971, 374. (36) (a) Manyala, N.; Sidis, Y.; DiTusa, J. F.; Aeppli, G.; Young, D. P.; Fisk, Z. Nature 2000, 404, 581. (b) Manyala, N.; Sidis, Y.; DiTusa, J. F.; Aeppli, G.; Young, D. P.; Fisk, Z. Nat. Mater. 2004, 3, 255.

JP904188F