Thermal Plasma Synthesis of Superparamagnetic Iron Oxide ...
Plasma Chem Plasma Process (2012) 32:519–531 DOI 10.1007/s11090-012-9364-1 ORIGINAL PAPER
Thermal Plasma Synthesis of Superparamagnetic Iron Oxide Nanoparticles Pingyan Lei • Adam M. Boies • Steven Calder • Steven L. Girshick
Received: 30 January 2012 / Accepted: 29 February 2012 / Published online: 16 March 2012 Ó Springer Science+Business Media, LLC 2012
Abstract Superparamagnetic iron oxide nanoparticles were synthesized by injecting ferrocene vapor and oxygen into an argon/helium DC thermal plasma. Size distributions of particles in the reactor exhaust were measured online using an aerosol extraction probe interfaced to a scanning mobility particle sizer, and particles were collected on transmission electron microscopy (TEM) grids and glass fiber filters for off-line characterization. The morphology, chemical and phase composition of the nanoparticles were characterized using TEM and X-ray diffraction, and the magnetic properties of the particles were analyzed with a vibrating sample magnetometer and a magnetic property measurement system. Aerosol at the reactor exhaust consisted of both single nanocrystals and small agglomerates, with a modal mobility diameter of 8–9 nm. Powder synthesized with optimum oxygen flow rate consisted primarily of magnetite (Fe3O4), and had a room-temperature saturation magnetization of 40.15 emu/g, with a coercivity and remanence of 26 Oe and 1.5 emu/g, respectively. Keywords
Iron oxide Nanoparticles DC thermal plasma Magnetic properties
Introduction Magnetic nanoparticles have applications in many areas, including data storage, catalysis, environmental remediation, magnetic fluids and biomedicine [1, 2]. At present there is P. Lei A. M. Boies S. Calder S. L. Girshick (&) Department of Mechanical Engineering, University of Minnesota, 111 Church St. S.E., Minneapolis, MN 55455, USA e-mail: [email protected] Present Address: A. M. Boies Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, UK S. Calder Laboratory of Organic Chemistry, Wageningen University, Dreijenplein 8, 6703 HB Wageningen, The Netherlands
123
520
Plasma Chem Plasma Process (2012) 32:519–531
particularly intense interest in the use of superparamagnetic iron oxide nanoparticles (SPIONs) for biomedical applications, including bacterial detection, protein purification, drug delivery, thermal therapies, and magnetic resonance (MR) imaging [3–7]. Ferrimagnetic iron oxide nanoparticles, including both the magnetite (Fe3O4) and maghemite (c-Fe2O3) phases, exhibit superparamagnetism when their crystallite size is smaller than about 20 nm [8], where their magnetic anisotropy energy is smaller than their random thermal energy, allowing their magnetization to flip spontaneously for temperatures greater than their blocking temperature, which is generally far below room temperature. SPIONs can serve as contrast agents to accelerate the spin–spin relaxation of water protons in MR imaging, resulting in contrast-enhanced images for tumor detection [9, 10]. Moreover SPIONs can be heated by applying an alternating magnetic field, and thus can be used for hyperthermia tumor destruction in cancer treatment [11]. Numerous studies have reported synthesis of SPIONs. The most common method is through wet chemistry by co-precipitation of Fe2? and Fe3? salts in alkaline solutions [4, 12–14]. While wet chemistry is the most studied synthetic approach, it has potential drawbacks. It involves batch processes that can take many hours or even days, and nanoparticle production rates are typically small. It requires management and disposal of hazardous solvents, and impurity residues are a potential problem. Surfactants are typically required to suppress agglomeration, and these surfactants may need to be removed from the nanoparticle surfaces before subsequent coating and/or functionalization for biomedical applications. An alternative approach, gas-phase synthesis, provides continuous (rather than batch) nanoparticle production with residence times on the order of milliseconds, and most gasphase processes are scalable to achieve high production rates. Solvents are not involved, and impurity residues are less of an issue than in wet chemistry. While for biomedical applications the final desired product is nanoparticles in aqueous dispersion, a number of recent studies have demonstrated gas-phase functionalization or coating of aerosol nanoparticles [15–17]. Thus the gas-phase synthesis could involve a sequential, continuous-flow process, involving SPION synthesis followed by surface functionalization or coating, followed by dispersion into liquid. A number of gas-phase methods have been utilized to synthesize iron oxide nanoparticles, the most common being combustion flames and plasmas. We here focus on the latter, and comment in the ‘‘Results and Discussion’’ section on a comparison of our results with those obtained with flame synthesis. Plasma synthesis of iron and iron oxide nanoparticles has been the subject of a number studies, dating to at least 1981. Several types of plasmas have been used, including radio frequency (RF) thermal plasmas [18, 19], microwave plasmas [20–28], transferred arcs [29–33], non-transferred arc direct current (DC) plasmas [29, 34], and low-pressure RF plasmas [35–39]. Iron-containing chemical precursors have included solid iron or steel anodes or wire electrodes [29–33], coarse iron powder injected into a plasma, where it evaporates and recondenses [18, 19, 34], and vapor-phase compounds including ferrocene (Fe(C5H5)2) [20, 35, 36, 39], iron trichloride (FeCl3) [21, 22], iron pentacarbonyl (Fe(CO)5) [23–29, 37, 38], and triiron dodecacarbonyl (Fe3(CO)12) [22]. Unfortunately only a few of these studies characterized the magnetic properties of the material produced. Vollath et al. [22] in their microwave synthesis experiments, reported superparamagnetic behavior when they used FeCl3 as precursor but not for the case where Fe3(CO)12 was used. In the former case the saturation magnetization Ms at room temperature measured *4 emu/g, too low to be useful for most applications. Kalyanaraman et al. [23], who also used a microwave plasma, also reported superparamagnetism, though
123
Plasma Chem Plasma Process (2012) 32:519–531
521
they noted that their magnetic measurements were preliminary, and they did not report values of specific saturation magnetization. Similarly, McIlroy et al. [36], who injected ferrocene into a low-pressure RF plasma without added oxygen, in an attempt to produce pure Fe nanoparticles, found only Fe3O4, and reported magnetic properties ‘‘reminiscent of superparamagnetic behavior,’’ though their sample sizes were too small to quantify this effect. Note that pure Fe nanoparticles are themselves superparamagnetic for sizes smaller than about 8 nm [40]. However it is difficult to produce pure Fe nanoparticles, because their high specific surface area, together with the high affinity of iron for oxygen—pure Fe nanoparticles are highly pyrophoric—makes it difficult to avoid oxidation either during or after particle synthesis. The group of Banerjee et al. [30, 31, 33], who used transferred arcs with iron anodes, performed extensive magnetic characterization of the material they produced. However their reported particle sizes were mostly too large ([20 nm) for superparamagnetism. They reported values of saturation magnetization as high as 88 emu/g for magnetite and 79 emu/ g for maghemite, close to the bulk values, but their high measured coercivities, *350 Oe, indicate that their material was not superparamagnetic. Panchal et al. [39] synthesized SPIONs using low-pressure argon RF plasmas into which they injected either ferrocene or iron pentacarbonyl [38]. In the ferrocene case no oxygen was added, the objective being to produce pure Fe particles, but, as in [36], they found only oxide particles, consisting of 25–40 nm agglomerates composed of 2–4 nm crystallites. Magnetic property measurements showed that the particles were superparamagnetic, although the apparent saturation magnetization at 300 K, *6 emu/g, is low. In the case of iron pentacarbonyl as precursor, they obtained particle diameters of 7–14 nm, and magnetic property measurements showed the particles to be superparamagnetic. The saturation magnetization at 300 K measured 21 emu/g, a more promising value for potential applications. However it should be noted that their method is a batch process, as the nanoparticles become negatively charged and trapped in the nonthermal RF plasma, and are collected when the plasma is turned off. The authors noted that their powder yields were low. In the present work, we report synthesis of SPIONs using a DC thermal plasma with injected ferrocene together with various flow rates of oxygen. For cases with sufficient oxygen flow, the powders produced were found to consist substantially of magnetite, possibly mixed with maghemite, and were superparamagnetic, with values of saturation magnetization at room temperature around 40 emu/g at a coercivity of 26 Oe. To our knowledge this is the best magnetic performance achieved to date for any plasma synthesis of SPIONs, and is comparable to the best reported results for flame synthesis.
Experimental System The experimental system used to synthesize SPIONs is shown schematically in Fig. 1. The plasma was generated by a PRAXAIR SG-100 DC plasma torch. While this paper focuses on the effects of oxygen flow rate on properties of the powder produced, preliminary experiments were conducted to explore the effects of several other system parameters, including arc current and the flow rates of helium, plasma torch cooling water, and argon dilution gas that was mixed with the precursor vapor before injection into the plasma. The results indicated that higher arc currents and helium flow rates, both of which lead to higher plasma power, tend to favor the production of particles that are sufficiently small to exhibit superparamagnetic behavior, as determined by their measured magnetic coercivity. Consequently, in the experiments reported here, the torch was operated at its highest rated
123
522
Plasma Chem Plasma Process (2012) 32:519–531
Fig. 1 Schematic of experimental setup
current of 250 A, and 5 slm of helium was added to the main argon flow (30 slm) to further increase the power to the range of 6.7–7.0 kW. Gas flow rates were controlled by mass flow controllers (Sierra 810C). The torch cooling water flow rate and argon dilution flow were found to have little effect on the magnetic properties of the powder produced, although the use of argon dilution flow was found to reduce the formation of particle agglomerates. Thus argon dilution flow (1500 sccm) was used in the current study. Ferrocene (Sigma-Aldrich, F408-100G) was used as the iron precursor. Ferrocene is a stable powder at room temperature, and sublimes upon heating. It is a much safer reagent than iron pentacarbonyl, which is highly flammable and toxic. Ferrocene vapor was introduced into the plasma using a packed bed [41, 42] composed of a stainless steel tube filled with 3-mm-diameter glass beads (Sigma-Aldrich) over which the powder was dispersed. The packed bed was maintained at a total pressure of *81 kPa, and was heated to 120°C using a heating mantle, corresponding to a ferrocene vapor pressure, as determined using the Clark and Glew equation [43], of 1,165 Pa. The ferrocene vapor was entrained in 500 sccm of argon flowing through the packed bed. The ferrocene flow rate QFerr can then be estimated using QFerr ¼
PFerr QAr ; PAr
ð1Þ
where the partial pressure PAr of the argon carrier gas is the difference between the total pressure and the ferrocene vapor pressure. Based on the given conditions, one obtains a ferrocene flow rate of *7 sccm. Oxygen and the argon dilution flow discussed above were introduced downstream of the packed bed, and the resulting mixture was injected into the plasma at the upstream end of a converging nozzle made of boron nitride. All precursor delivery lines were heated to 125–135°C to avoid ferrocene condensation before injection into the plasma. The flow exiting the nozzle expanded into a 250-mm-diameter chamber, maintained at a pressure in
123
Plasma Chem Plasma Process (2012) 32:519–531
523
the range 40–53 kPa. A ceramic tube (32-mm OD, 25-mm ID, 356-mm length) was positioned 51 mm above the nozzle exit to provide a longer, more uniform high-temperature region for the nanoparticles that nucleate in the nozzle expansion [44]. Experiments were conducted with various oxygen flow rates, as this was expected to affect the composition and magnetic properties of the synthesized powder.
Materials Characterization The size distribution of plasma-synthesized particles was measured on-line by extracting aerosol into a sampling probe located in the reactor exhaust, using an ejector driven by high-pressure nitrogen, similar to the system described in [45]. The sampled aerosol was delivered to a scanning mobility particle sizer (SMPS) system, consisting of an electrical neutralizer (Po-210 bipolar charger), a differential mobility analyzer (TSI model 3085) and a condensation particle counter (TSI model 3025A). Additionally sampled particles were collected onto lacey carbon grids (Ted Pella 01890) located downstream of the neutralizer, using an electrostatic precipitator with an applied voltage of 3 kV. These samples were characterized by high-resolution transmission electron microscopy (HRTEM), conducted on a Tecnai G2 F30 electron microscope. Larger quantities of powder were collected on glass fiber filters mounted in the reactor exhaust line. Chemical and phase composition were investigated using X-ray diffraction (XRD), performed with a Siemens D-500 diffraction meter with a 2.2-kW sealed cobalt source. Magnetic properties were characterized with a Princeton micro vibrating sample magnetometer (VSM) for room temperature hysteresis loops, and with a Quantum Design magnetic property measurement system (MPMS), utilizing a Superconducting Quantum Interference Device (SQUID) magnetometer, for low temperature measurements. To obtain zero-field cooling/field cooling (ZFC/FC) curves, the sample was initially cooled to 5 K in zero field, and then the induced magnetization was measured with an applied field of 50 Oe while heating the sample to 300 K and then cooling back to 5 K. Saturation isothermal remanent magnetization (SIRM) curves were obtained by saturating the sample in a 2.5-T field at 300 K, and then recording the remanent magnetization in zero field while cooling from 300 to 20 K and then heating back to 300 K.
Results and Discussion Nanoparticle Size Distribution and Morphology Figure 2 shows particle mobility diameter distributions measured with SMPS, and TEM images of particles sampled for the same set of conditions, for an oxygen flow rate of 20 sccm. As seen in Fig. 2a, the measured size distributions for a given set of run conditions showed good reproducibility. For the case shown, the results indicate the mode in the size distribution of *8 to 9 nm, a maximum agglomerate size around 30 nm, and a geometric standard deviation of *1.49, close to the *1.45 value expected for a coagulating aerosol [46]. Figure 2b shows a corresponding low-resolution TEM image, and Fig. 2c and d show two HRTEM images. Note that the agglomerate structures seen in Fig. 2b and c were not necessarily formed in the aerosol phase, as loose agglomerates can form during deposition on the TEM grid. Nevertheless, as the HRTEM images show that individual crystallites are smaller than 10 nm, it is evident that the mobility diameter distributions seen in Fig. 2a
123
524
Plasma Chem Plasma Process (2012) 32:519–531
Fig. 2 Aerosol sampled from reactor exhaust at oxygen flow rate of 20 sccm: a particle size distributions measured by SMPS (different symbols correspond to different voltage scans of the differential mobility analyzer); b low-resolution TEM image; c, d high-resolution TEM images. Large structure in TEM images is TEM grid
represent a mixture of isolated crystallites together with small agglomerates. Figure 2d clearly shows the lattice fringes of a single iron oxide nanoparticle. The d-spacing measured in the image equals *0.30 nm, corresponding to the magnetite (022) plane or the maghemite (220) plane. Chemical and Phase Composition Figure 3 shows measured XRD patterns of powder collected on glass fiber filters for various oxygen flow rates. As the flow rate of oxygen increases from 0 to 30 sccm, the XRD patterns show a clear trend toward increasing oxidation. For the case where no oxygen was added to the plasma, no crystalline signal is detected in the XRD pattern. Instead the pattern in this case is mainly indicative of amorphous carbon. Although the thermal plasma is expected to break the Fe–C bonds in the ferrocene
123
Plasma Chem Plasma Process (2012) 32:519–531
525
Fig. 3 X-ray diffraction patterns of powder collected on glass fiber filters located in reactor exhaust, for various flow rates of oxygen. Patterns were analyzed using JADE 9 software (Materials Data, Inc.)
precursor, the presence of sufficient oxygen is evidently necessary to prevent the condensation of carbonaceous material. An HRTEM image for the zero-oxygen case is shown in Fig. 4. The material appears to be mainly amorphous, consistent with the XRD pattern, but one also sees isolated areas of lattice fringes, from crystallites that are only a few nm in size. The d-spacing of the crystallite highlighted in the center of the image measures *0.21 nm. Within the uncertainty of the measurement, it could represent either the (110) plane of Fe (0.2027 nm), the (004) plane of magnetite (0.2096 nm), or the (400) plane of maghemite (0.20869 nm). The small size of these crystallites, together with their being embedded in an amorphous carbon matrix, may explain the absence of evident peaks in the XRD pattern. For cases where oxygen was added to the gases injected into the plasma, at flow rates ranging from 10 to 40 sccm, crystalline peaks appear in the XRD patterns of collected powder, as seen in Fig. 3. At an O2 flow rate of 10 sccm, the XRD pattern shows peaks associated with pure Fe, wu¨stite (FeO) and magnetite (Fe3O4) and/or maghemite (c-Fe2O3). Note that magnetite and maghemite have the same cubic spinel structure and quite similar lattice parameters (magnetite a = 0.8396 nm, maghemite a = 0.83474 nm) [8], as a result of which their dominant XRD peaks are effectively indistinguishable. At an oxygen flow rate of 20 sccm, the pure Fe peak almost disappears, while the peaks associated with magnetite and/or maghemite show increased intensity. The wu¨stite peak remains (though much smaller relative to the magnetite/maghemite peaks than in the 10-sccm case) and a small peak associated with hematite, a-Fe2O3, appears. With further increases in the oxgyen flow rate, to 30 and 40 sccm, the wu¨stite peak disappears. The powder is predominantly magnetite and/or maghemite, with a small component of hematite. The hematite phase is antiferromagnetic, which decreases the average saturation magnetization of the products.
123
526
Plasma Chem Plasma Process (2012) 32:519–531
Fig. 4 HRTEM image of sample for case of zero oxygen flow. Circles highlight several regions with evident crystalline lattice fringes
Magnetic Properties Hysteresis loops measured by VSM at room temperature with a maximum applied field of 1.5 T are shown in Fig. 5a for cases that correspond to the XRD patterns shown in Fig. 3. Magnetic moments were normalized in each case by the total sample mass collected on the glass fiber filters. All the curves show nearly zero hysteresis, indicating that the particles are superparamagnetic [12, 47], and approach their saturation values rapidly, with the exception of the case with an oxygen flow rate of 10 sccm. The saturation magnetization (Ms) and coercivity (Hc) determined from these measurements are summarized in Fig. 5b and c.
Fig. 5 Vibrating sample magnetometer measurements at room temperature of collected powder samples for various oxygen flow rates: a hysteresis loops; b saturation magnetization (Ms); c coercivity (Hc)
123
Plasma Chem Plasma Process (2012) 32:519–531
527
The results for the zero oxygen case, which show a surprisingly high saturation magnetization of 32.3 emu/g, with a coercivity measuring 45 Oe, suggest that the crystallites seen in Fig. 4 are primarily pure Fe. The saturation magnetization of bulk iron, 222 emu/g [48], is much higher than that of either magnetite (92–100 emu/g) or maghemite (60–80 emu/g) [8], and the crystallites seen in Fig. 4 are smaller than the 8-nm superparamagnetic critical size for Fe. This hypothesis is further supported by the Fe peak in the XRD pattern, Fig. 3, for the 10-sccm case, since if pure Fe particles are formed in the 10-sccm oxygen case, then they are even more likely to form in the zero-oxygen case. The much higher coercivity in the 10-sccm case, *90 Oe, can plausibly be explained by the fact that growth of the Fe particles in the zero-oxygen case is constrained by their encapsulation in an amorphous carbon matrix, whereas there is much less carbon inclusion with added oxygen, allowing the Fe particles to grow larger than 8 nm. Although magnetite and maghemite particles are superparamagnetic up to 20 nm, the presence of a relatively small fraction of Fe particles larger than 8 nm would cause a significant increase in coercivity. The highest measured saturation magnetization, 40.15 emu/g, was achieved at an oxygen flow rate of 20 sccm. The coercivity and remanence for this sample measured 26 Oe and 1.5 emu/g, respectively, meaning that the powder is almost completely superparamagnetic. The fact that the saturation magnetization is lower than the values for bulk magnetite and maghemite may be due to the presence of a small fraction of hematite, as indicated by the XRD pattern, hematite being antiferromagnetic [8], and/or to the frequently observed, and theoretically predicted, strong reduction in saturation magnetization for superparamagnetic particles as crystallite size decreases [49–52]. However a recent study [53], which involved wet chemical synthesis of *5-nm-diameter magnetite particles that were subsequently coated with a thin layer of organic film and then gold, reported a very high Ms of 81 emu/g, suggesting that the effect of particle size on Ms may actually be due to impurities on the particle surfaces, which become more important as particle size decreases. As the oxygen flow rate increases above 20 sccm, Fig. 5b shows that the saturation magnetization decreases, while the coercivity, shown in Fig. 5c, is basically unchanged. We hypothesize that this behavior is due to increasing oxidation of the superparamagnetic particles from magnetite, Fe3O4, to maghemite, c-Fe2O3. As noted above, the bulk saturation magnetization of maghemite is lower than that of magnetite. From the XRD patterns in Fig. 3, the small peak at 33.28° attributed to hematite, a-Fe2O3, is essentially unchanged as the oxygen flow rate increases from 20 to 40 sccm, indicating that the increased oxidation occurs between the two superparamagnetic phases, and therefore does not affect the coercivity. Also supporting this hypothesis, as the oxygen flow rate increases the color of the synthesized powder, as observed by the naked eye, changes from black at 0 sccm to dark brown at 20 sccm, to brown at 30 sccm, and to reddish brown at 40 sccm. In any case, the magnetic properties obtained in this study are far superior to those reported for any previous plasma synthesis of SPIONs, as reviewed in the Introduction. There are several possible explanations for these improved results compared to other studies of plasma synthesis of iron oxide nanoparticles. The precursors in our system are injected into the plasma at the upstream end of a hot-wall converging nozzle, producing a relatively uniform environment for vapor mixing and particle nucleation [54]. A long ceramic tube at the exit of the reactor additionally provides a relatively uniform hightemperature environment, and the total residence time, a few tens of ms, is too short to allow for crystallite growth beyond the superparamagnetic critical size. Further, our study shows that careful control of the oxygen-to-ferrocene ratio is crucial for control of the stoichiometry and phase of the particles produced.
123
528
Plasma Chem Plasma Process (2012) 32:519–531
A number of studies have also reported flame synthesis of SPIONs [15, 52, 55–60]. Among these, the best magnetic properties achieved appear to be the recent results of Kumfer et al. [60]. In a series of five experiments with various flame conditions, their measured values of room-temperature saturation magnetization increased monotonically with coercivity, with values of Ms ranging from 6.5 to 60 emu/g, as values of Hc ranged from 13 to 76 Oe. They attributed increases in Ms both to increases in particle size and to the presence of a significant fraction of pure Fe in the sample with the highest Ms, and increases in coercivity to an increasing fraction of particles above the critical size for superparamagnetism. The sample that lay in the middle of their results, with Ms = 42 emu/ g and Hc = 30 Oe, is virtually identical to the results for our case of 20-sccm oxgyen flow, for which we obtained Ms = 40.15 emu/g and Hc = 26 Oe. As in our case (see below), they concluded that their powder samples consisted primarily of Fe3O4. This is in contrast to the more typical result that flame synthesis, with its inherent abundance of oxygen, tends to produce c-Fe2O3 [15, 55–57]. To produce magnetite, with its higher saturation magnetization than maghemite, Kumfer et al. [60] ran under more reducing conditions than in the typical flame. Hysteresis loops of plasma-synthesized iron oxide nanoparticles measured at room temperature (by VSM) and at 10 K (by MPMS) for the case of 20-sccm oxygen are compared in Fig. 6. At room temperature the hysteresis loop is nearly reversible, with almost zero hysteresis. At 10 K, a small increase of the saturation magnetization is observed, to *43 emu/g compared with *40 emu/g at room temperature, while the coercivity dramatically increases, to *330 Oe compared with *26 Oe at room temperature. This behavior is expected, as the thermal energy at 10 K is not sufficient to reverse the magnetization when the field is reduced to zero. At 10 K these particles are blocked, and behave like ferrimagnetic materials, with an increased remanence and coercivity as seen in Fig. 6. To provide further information about the magnetic properties of the powder produced in the 20-sccm oxygen case, zero-field cooling/field cooling curves were measured by MPMS with a field of 50 Oe, shown in Fig. 7a, and room-temperature saturation isothermal remanent magnetization (SIRM) measurements were obtained, as shown in Fig. 7b. The
Fig. 6 Hysteresis loops measured at both room temperature and 10 K, for powder sample synthesized at oxygen flow rate of 20 sccm
123
Plasma Chem Plasma Process (2012) 32:519–531
529
Fig. 7 Saturation isothermal remanent magnetization measurements of powder sample synthesized at an oxygen flow rate of 20 sccm: a zero-field-cooling/field-cooling measurements; b zero-field cooling and rewarming measurements
abrupt change in the magnetic moments for both the cooling and warming curves in Fig. 7b, as well as the sharp change in the slopes of the ZFC/FC curves in Fig. 7a, both of which occur around 120 K, corresponds to the known Verwey transition of magnetite, whose equilibrium structure is cubic above the Verwey temperature TV & 120 K, and monoclinic below TV [61, 62]. As no corresponding phase transition occurs in maghemite, these measurements indicate that the sample is substantially composed of magnetite, although the additional presence of maghemite cannot be ruled out. Indeed, the fact that the transition in the curves seen in Fig. 7b is relatively broad rather than sharp could be due to some degree of maghemitization of the magnetite nanoparticles [63]. Additionally, the temperature where the ZFC and FC curves in Fig. 7a separate, around 250 K, can be equated to the blocking temperature TB of the largest crystallites in the sample [64]. At temperatures below TB the thermal energy is insufficient to overcome the magnetic anisotropy energy of the nanoparticles, preventing superparamagnetism. The crystallite volume V can be estimated by V¼
25kTB ; Keff
ð2Þ
where Keff is the effective magnetic anisotropy constant, and k is the Boltzmann constant [65]. Using Keff & 5 9 104 J/m for magnetite [8, 66], and assuming spherical crystallites, the largest crystallite diameter estimated by Eq (2) equals *15 nm. Note that most crystallites are considerably smaller than this, as evidenced by both the SMPS measurements (which include agglomerates) and the TEM images in Fig. 2. Thus essentially all of the crystallites are below the *20-nm critical size for superparamagnetism in magnetite and maghemite.
Conclusions A DC thermal plasma system was used to synthesize superparamagnetic iron oxide nanoparticles, using ferrocene vapor as the iron precursor. The chemical and phase composition of the nanoparticles produced depended on the flow rate of oxygen added to the argon-helium plasma. At zero oxygen flow rate, the powder produced consisted of 2–3 nm Fe crystallites in an amorphous carbon matrix. As the oxgyen flow rate increased, the iron was increasingly oxidized.
123
530
Plasma Chem Plasma Process (2012) 32:519–531
For the oxgyen flow rate where the best magnetic properties were achieved, the powder consisted primarily of magnetite (Fe3O4), with possible contributions of maghemite (c-Fe2O3), as well as a small fraction of hematite (a-Fe2O3). The mode in the particle mobility diameter distribution measured by SMPS equalled 8–9 nm. SMPS measurements and TEM images indicate that the material produced consisted both of single crystallites and small agglomerates, with a maximum agglomerate size of *30 nm and a maximum crystallite size, determined by SQUID magnetometry, of *15 nm. Superparamagnetism was observed in the collected powder for most of the oxygen flow rates tested. The best room-temperature saturation magnetization measured equalled 40.15 emu/g, with a coercivity of 26 Oe and a remanence of 1.5 emu/g. To our knowledge these results are the best yet reported, from the viewpoint of high saturation magnetization with negligible coercivity, for synthesis of SPIONs by any type of plasma, and are comparable to the best results reported to date for flame synthesis. Acknowledgments This research was primarily supported by the U.S. National Science Foundation under Award Numbers CBET-0730184 and CBET-1066343, and by the Minnesota Futures Grant Program. Parts of the characterization work were conducted at the College of Science and Engineering Characterization Facility and the Institute for Rock Magnetism at the University of Minnesota.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
Willard MA, Kurihara LK, Carpenter EE, Calvin S, Harris VG (2004) Int Mater Rev 49:125–170 Lu AH, Salabas EL, Schuth F (2007) Angew Chem Int Ed 46:1222–1244 Pankhurst QA, Connolly J, Jones SK, Dobson J (2003) J Phys D 36:R167–R181 Gupta AK, Gupta M (2005) Biomater 26:3995–4021 Laurent S, Forge D, Port M, Roch A, Robic C, Elst LV, Muller RN (2008) Chem Rev 108:2064–2110 Gao J, Gu H, Xu B (2009) Acct Chem Res 42:1097–1107 Hao R, Xing R, Xu Z, Hou Y, Gao S, Sun S (2010) Adv Mater 22:2729–2742 Cornell RM, Schwertmann U (2003) The iron oxides: structure, properties, reactions, occurrences and uses, 2nd edn. Wiley-VCH, Weinheim, Germany Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ (1999) Radiol 210:459–466 Kumano S, Murakami T, Kim T, Hori M, Okada A, Sugiura T, Noguchi Y, Kawata S, Tomoda K, Nakamura H (2003) Am J Roentgen 181:1335–1339 Rabias I, Tsitrouli D, Karakosta E, Kehagias T, Diamantopoulos G, Fardis M, Stamopoulos D, Maris TG, Falaras P, Zouridakis N, Diamantis N, Panayotou G, Verganelakis DA, Drossopoulou GI, Tsilibari EC, Papavassiliou G (2010) Biomicrofluid 4:024111 Jeong JR, Lee SJ, Kim JD, Shin SC (2004) Physica Stat Solid B 241:1593–1596 Goya GF, Morales MP (2004) J Meta Nanocryst Mater 20–21:673–678 Smolensky ED, Park HYE, Berquo TS, Pierre VC (2010) Contrast Media Mol Imaging 6:189–199 Teleki A, Suter M, Kidambi PR, Ergeneman O, Krumeich F, Nelson BJ, Pratsinis SE (2009) Chem Mater 21:2094–2100 Boies AM, Roberts JT, Girshick SL, Zhang B, Nakamura T, Mochizuki A (2009) Nanotechnol 20:295604 Calder S, Boies A, Lei P, Girshick S, Roberts J (2011) Chem Mater 23:2917–2921 Yoshida T, Akashi K (1981) Trans Jpn Inst Met 22:371–378 Girshick SL, Chiu C-P, Muno R, Wu CY, Yang L, Singh SK, McMurry PH (1993) J Aerosol Sci 24:367–382 Chou CH, Phillips J (1992) J Mater Res 7:2107–2113 Vollath D, Szabo DV, Taylor RD, Willis JO, Sickafus KE (1995) Nanostruct Mater 6:941–944 Vollath D, Szabo DV, Taylor RD, Willis JO (1997) J Mater Res 12:2175–2182 Kalyanaraman R, Yoo S, Krupashankara MS, Sudarshan TS, Dowding RJ (1998) Nanostruct Mater 10:1379–1392 Janzen C, Wiggers H, Knipping J, Roth P (2001) J Nanosci Nanotechnol 1:221–225 Li SZ, Hong YC, Uhm HS, Li ZK (2004) Jpn J Appl Phys 43:7714–7717
123
Plasma Chem Plasma Process (2012) 32:519–531
531
26. Zajickova L, Synek P, Jasek O, Elias M, David B, Bursik J, Pizurova N, Hanzlikova R, Lazar L (2009) Appl Surf Sci 255:5421–5424 27. David B, Pizurova N, Schneeweiss O, Kudrle V, Jasek O, Synek P (2011) Jpn J Appl Phys 50:08JF11 28. Synek P, Jasek O, Zajickova L, David B, Kudrle V, Pizurova N (2011) Mater Lett 65:982–984 29. Bica I (1999) Mater Sci Engg B 68:5–9 30. Balasubramaniam C, Khollam YB, Banerjee I, Bakare PP, Date SK, Das AK, Bhoraskar SV (2004) Mater Lett 58:3958–3962 31. Banerjee I, Khollam YB, Balasubramanian C, Pasricha R, Bakare PP, Patil KR, Das AK, Bhoraskar SV (2006) Script Mater 54:1235–1240 32. Chazelas C, Coudert JF, Jarrige J, Fauchais P (2006) J Eur Ceram Soc 26:3499–3507 33. Banerjee I, Khollam YB, Mahapatra SK, Das AK, Bhoraskar SV (2010) J Vac Sci Technol A 28:1399–1403 34. Subramanian V, Baskaran R, Krishnan H (2009) Aerosol Air Qual Res 9:172–186 35. McIlroy DN, Zhang D, Norton MG, O’Brien WL, Schwickert MM, Harp GR (2000) J Appl Phys 87:7213–7217 36. McIlroy DN, Huso J, Kranov Y, Marchinek J, Ebert C, Moore S, Marji E, Gandy R, Hong YK, Norton MG, Cavalieri E, Benz R, Justus BL, Rosenberg A (2003) J Appl Phys 93:5643–5649 37. Kouprine A, Gitzhofer F, Boulos M, Veres T (2006) Carbon 44:2593–2601 38. Panchal V, Neergat M, Bhandarkar U (2011) J Nanopart Res 13:3825–3833 39. Panchal V, Lahoti G, Bhandarkar U, Neergat M (2011) J Phys D 44:345205 ¨ zdemir O ¨ (1997) Rock magnetism: fundamentals and frontiers. Cambridge University 40. Dunlop DJ, O Press, Cambridge, New York 41. Powell QH, Fotou GP, Kodas TT, Anderson BM (1997) Chem Mater 9:685–693 42. Zhang B (2007) Thermal plasma synthesis and photoinduced coating of aluminum nanoparticles. PhD thesis, Univ Minnesota, Minneapolis 43. Monte MJS, Santos LMNBF, Fulem M, Fonseca JMS, Sousa CAD (2006) J Chem Eng Data 51:757–766 44. Jain R, Girshick SL, Heberlein JV, Mukherjee R, Zhang B, Nakamura T, Mochizuki A (2010) Plasma Chem Plasma Process 30:795–811 45. Wang X, Hafiz J, Mukherjee R, Renault T, Heberlein J, Girshick SL, McMurry PH (2005) Plasma Chem Plasma Process 25:439–453 46. Friedlander SK, Wang CS (1966) J Coll Interface Sci 22:126–132 47. Benitez MJ, Mishra D, Szary P, Confalonieri GAB, Feyen M, Lu AH, Agudo L, Eggeler G, Petracic O, Zabel H (2011) J Phys Condens Matter 23:126003 48. Danan H, Herr A, Meyer AJP (1968) J Appl Phys 39:669–670 49. Berkowitz AE, Schuele WJ, Flanders PJ (1968) J Appl Phys 39:1261–1263 50. Coey JMD (1971) Appl Phys Lett 27:1140–1142 51. Frenkel J, Doefman J (1930) Nature 126:274–275 52. Li D, Teoh WY, Selomulya C, Woodward RC, Munroe P, Amal R (2007) J Mater Chem 17:4876–4884 53. Smolensky ED, Neary MC, Zhou Y, Berquo TS, Pierre VC (2011) Chem Commun 47:2149–2151 54. Rao N, Girshick S, Heberlein J, McMurry P, Bench M, Jones S, Hansen D, Micheel B (1995) Plasma Chem Plasma Process 15:581–607 55. Zachariah MR, Aquino MI, Shull RD, Steel EB (1995) Nanostruct Mater 5:383–392 56. Janzen C, Knipping J, Rellinghaus B, Roth P (2003) J Nanopart Res 5:589–596 57. Li D, Teoh WY, Woodward RC, Cashion JD, Selomulya C, Amal R (2009) J Phys Chem C 113:12040–12047 58. Strobel R, Pratsinis SE (2009) Adv Powder Technol 20:190–194 59. Guo B, Yim H, Khasanov A, Stevens J (2010) Aerosol Sci Technol 44:281–291 60. Kumfer BM, Shinoda K, Jeyadevan B, Kennedy IM (2010) J Aerosol Sci 41:257–265 61. Dunlop DJ (2007) J Geophys Res Solid Earth 112:B11103 62. Verwey EJW (1939) Nature 144:327–328 63. Ozdemir O, Dunlop DJ (2010) J Geophys Res-Solid Earth 115:B02101 64. Maity D, Choo S-G, Yi J, Ding J, Xue JM (2009) J Magn Magn Mater 321:1256–1259 65. Cullity BD, Graham CD (2009) Introduction to magnetic materials, 2nd edn. IEEE/Wiley, Hoboken 66. Guardia P, Batlle-Brugal B, Roca AG, Iglesias O, Morales MP, Serna CJ, Labarta A, Batlle X (2007) J Magn Magn Mater 316:E756–E759
123
Thermal Plasma Synthesis of Superparamagnetic Iron Oxide Nanoparticles Pingyan Lei • Adam M. Boies • Steven Calder • Steven L. Girshick
Received: 30 January 2012 / Accepted: 29 February 2012 / Published online: 16 March 2012 Ó Springer Science+Business Media, LLC 2012
Abstract Superparamagnetic iron oxide nanoparticles were synthesized by injecting ferrocene vapor and oxygen into an argon/helium DC thermal plasma. Size distributions of particles in the reactor exhaust were measured online using an aerosol extraction probe interfaced to a scanning mobility particle sizer, and particles were collected on transmission electron microscopy (TEM) grids and glass fiber filters for off-line characterization. The morphology, chemical and phase composition of the nanoparticles were characterized using TEM and X-ray diffraction, and the magnetic properties of the particles were analyzed with a vibrating sample magnetometer and a magnetic property measurement system. Aerosol at the reactor exhaust consisted of both single nanocrystals and small agglomerates, with a modal mobility diameter of 8–9 nm. Powder synthesized with optimum oxygen flow rate consisted primarily of magnetite (Fe3O4), and had a room-temperature saturation magnetization of 40.15 emu/g, with a coercivity and remanence of 26 Oe and 1.5 emu/g, respectively. Keywords
Iron oxide Nanoparticles DC thermal plasma Magnetic properties
Introduction Magnetic nanoparticles have applications in many areas, including data storage, catalysis, environmental remediation, magnetic fluids and biomedicine [1, 2]. At present there is P. Lei A. M. Boies S. Calder S. L. Girshick (&) Department of Mechanical Engineering, University of Minnesota, 111 Church St. S.E., Minneapolis, MN 55455, USA e-mail: [email protected] Present Address: A. M. Boies Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, UK S. Calder Laboratory of Organic Chemistry, Wageningen University, Dreijenplein 8, 6703 HB Wageningen, The Netherlands
123
520
Plasma Chem Plasma Process (2012) 32:519–531
particularly intense interest in the use of superparamagnetic iron oxide nanoparticles (SPIONs) for biomedical applications, including bacterial detection, protein purification, drug delivery, thermal therapies, and magnetic resonance (MR) imaging [3–7]. Ferrimagnetic iron oxide nanoparticles, including both the magnetite (Fe3O4) and maghemite (c-Fe2O3) phases, exhibit superparamagnetism when their crystallite size is smaller than about 20 nm [8], where their magnetic anisotropy energy is smaller than their random thermal energy, allowing their magnetization to flip spontaneously for temperatures greater than their blocking temperature, which is generally far below room temperature. SPIONs can serve as contrast agents to accelerate the spin–spin relaxation of water protons in MR imaging, resulting in contrast-enhanced images for tumor detection [9, 10]. Moreover SPIONs can be heated by applying an alternating magnetic field, and thus can be used for hyperthermia tumor destruction in cancer treatment [11]. Numerous studies have reported synthesis of SPIONs. The most common method is through wet chemistry by co-precipitation of Fe2? and Fe3? salts in alkaline solutions [4, 12–14]. While wet chemistry is the most studied synthetic approach, it has potential drawbacks. It involves batch processes that can take many hours or even days, and nanoparticle production rates are typically small. It requires management and disposal of hazardous solvents, and impurity residues are a potential problem. Surfactants are typically required to suppress agglomeration, and these surfactants may need to be removed from the nanoparticle surfaces before subsequent coating and/or functionalization for biomedical applications. An alternative approach, gas-phase synthesis, provides continuous (rather than batch) nanoparticle production with residence times on the order of milliseconds, and most gasphase processes are scalable to achieve high production rates. Solvents are not involved, and impurity residues are less of an issue than in wet chemistry. While for biomedical applications the final desired product is nanoparticles in aqueous dispersion, a number of recent studies have demonstrated gas-phase functionalization or coating of aerosol nanoparticles [15–17]. Thus the gas-phase synthesis could involve a sequential, continuous-flow process, involving SPION synthesis followed by surface functionalization or coating, followed by dispersion into liquid. A number of gas-phase methods have been utilized to synthesize iron oxide nanoparticles, the most common being combustion flames and plasmas. We here focus on the latter, and comment in the ‘‘Results and Discussion’’ section on a comparison of our results with those obtained with flame synthesis. Plasma synthesis of iron and iron oxide nanoparticles has been the subject of a number studies, dating to at least 1981. Several types of plasmas have been used, including radio frequency (RF) thermal plasmas [18, 19], microwave plasmas [20–28], transferred arcs [29–33], non-transferred arc direct current (DC) plasmas [29, 34], and low-pressure RF plasmas [35–39]. Iron-containing chemical precursors have included solid iron or steel anodes or wire electrodes [29–33], coarse iron powder injected into a plasma, where it evaporates and recondenses [18, 19, 34], and vapor-phase compounds including ferrocene (Fe(C5H5)2) [20, 35, 36, 39], iron trichloride (FeCl3) [21, 22], iron pentacarbonyl (Fe(CO)5) [23–29, 37, 38], and triiron dodecacarbonyl (Fe3(CO)12) [22]. Unfortunately only a few of these studies characterized the magnetic properties of the material produced. Vollath et al. [22] in their microwave synthesis experiments, reported superparamagnetic behavior when they used FeCl3 as precursor but not for the case where Fe3(CO)12 was used. In the former case the saturation magnetization Ms at room temperature measured *4 emu/g, too low to be useful for most applications. Kalyanaraman et al. [23], who also used a microwave plasma, also reported superparamagnetism, though
123
Plasma Chem Plasma Process (2012) 32:519–531
521
they noted that their magnetic measurements were preliminary, and they did not report values of specific saturation magnetization. Similarly, McIlroy et al. [36], who injected ferrocene into a low-pressure RF plasma without added oxygen, in an attempt to produce pure Fe nanoparticles, found only Fe3O4, and reported magnetic properties ‘‘reminiscent of superparamagnetic behavior,’’ though their sample sizes were too small to quantify this effect. Note that pure Fe nanoparticles are themselves superparamagnetic for sizes smaller than about 8 nm [40]. However it is difficult to produce pure Fe nanoparticles, because their high specific surface area, together with the high affinity of iron for oxygen—pure Fe nanoparticles are highly pyrophoric—makes it difficult to avoid oxidation either during or after particle synthesis. The group of Banerjee et al. [30, 31, 33], who used transferred arcs with iron anodes, performed extensive magnetic characterization of the material they produced. However their reported particle sizes were mostly too large ([20 nm) for superparamagnetism. They reported values of saturation magnetization as high as 88 emu/g for magnetite and 79 emu/ g for maghemite, close to the bulk values, but their high measured coercivities, *350 Oe, indicate that their material was not superparamagnetic. Panchal et al. [39] synthesized SPIONs using low-pressure argon RF plasmas into which they injected either ferrocene or iron pentacarbonyl [38]. In the ferrocene case no oxygen was added, the objective being to produce pure Fe particles, but, as in [36], they found only oxide particles, consisting of 25–40 nm agglomerates composed of 2–4 nm crystallites. Magnetic property measurements showed that the particles were superparamagnetic, although the apparent saturation magnetization at 300 K, *6 emu/g, is low. In the case of iron pentacarbonyl as precursor, they obtained particle diameters of 7–14 nm, and magnetic property measurements showed the particles to be superparamagnetic. The saturation magnetization at 300 K measured 21 emu/g, a more promising value for potential applications. However it should be noted that their method is a batch process, as the nanoparticles become negatively charged and trapped in the nonthermal RF plasma, and are collected when the plasma is turned off. The authors noted that their powder yields were low. In the present work, we report synthesis of SPIONs using a DC thermal plasma with injected ferrocene together with various flow rates of oxygen. For cases with sufficient oxygen flow, the powders produced were found to consist substantially of magnetite, possibly mixed with maghemite, and were superparamagnetic, with values of saturation magnetization at room temperature around 40 emu/g at a coercivity of 26 Oe. To our knowledge this is the best magnetic performance achieved to date for any plasma synthesis of SPIONs, and is comparable to the best reported results for flame synthesis.
Experimental System The experimental system used to synthesize SPIONs is shown schematically in Fig. 1. The plasma was generated by a PRAXAIR SG-100 DC plasma torch. While this paper focuses on the effects of oxygen flow rate on properties of the powder produced, preliminary experiments were conducted to explore the effects of several other system parameters, including arc current and the flow rates of helium, plasma torch cooling water, and argon dilution gas that was mixed with the precursor vapor before injection into the plasma. The results indicated that higher arc currents and helium flow rates, both of which lead to higher plasma power, tend to favor the production of particles that are sufficiently small to exhibit superparamagnetic behavior, as determined by their measured magnetic coercivity. Consequently, in the experiments reported here, the torch was operated at its highest rated
123
522
Plasma Chem Plasma Process (2012) 32:519–531
Fig. 1 Schematic of experimental setup
current of 250 A, and 5 slm of helium was added to the main argon flow (30 slm) to further increase the power to the range of 6.7–7.0 kW. Gas flow rates were controlled by mass flow controllers (Sierra 810C). The torch cooling water flow rate and argon dilution flow were found to have little effect on the magnetic properties of the powder produced, although the use of argon dilution flow was found to reduce the formation of particle agglomerates. Thus argon dilution flow (1500 sccm) was used in the current study. Ferrocene (Sigma-Aldrich, F408-100G) was used as the iron precursor. Ferrocene is a stable powder at room temperature, and sublimes upon heating. It is a much safer reagent than iron pentacarbonyl, which is highly flammable and toxic. Ferrocene vapor was introduced into the plasma using a packed bed [41, 42] composed of a stainless steel tube filled with 3-mm-diameter glass beads (Sigma-Aldrich) over which the powder was dispersed. The packed bed was maintained at a total pressure of *81 kPa, and was heated to 120°C using a heating mantle, corresponding to a ferrocene vapor pressure, as determined using the Clark and Glew equation [43], of 1,165 Pa. The ferrocene vapor was entrained in 500 sccm of argon flowing through the packed bed. The ferrocene flow rate QFerr can then be estimated using QFerr ¼
PFerr QAr ; PAr
ð1Þ
where the partial pressure PAr of the argon carrier gas is the difference between the total pressure and the ferrocene vapor pressure. Based on the given conditions, one obtains a ferrocene flow rate of *7 sccm. Oxygen and the argon dilution flow discussed above were introduced downstream of the packed bed, and the resulting mixture was injected into the plasma at the upstream end of a converging nozzle made of boron nitride. All precursor delivery lines were heated to 125–135°C to avoid ferrocene condensation before injection into the plasma. The flow exiting the nozzle expanded into a 250-mm-diameter chamber, maintained at a pressure in
123
Plasma Chem Plasma Process (2012) 32:519–531
523
the range 40–53 kPa. A ceramic tube (32-mm OD, 25-mm ID, 356-mm length) was positioned 51 mm above the nozzle exit to provide a longer, more uniform high-temperature region for the nanoparticles that nucleate in the nozzle expansion [44]. Experiments were conducted with various oxygen flow rates, as this was expected to affect the composition and magnetic properties of the synthesized powder.
Materials Characterization The size distribution of plasma-synthesized particles was measured on-line by extracting aerosol into a sampling probe located in the reactor exhaust, using an ejector driven by high-pressure nitrogen, similar to the system described in [45]. The sampled aerosol was delivered to a scanning mobility particle sizer (SMPS) system, consisting of an electrical neutralizer (Po-210 bipolar charger), a differential mobility analyzer (TSI model 3085) and a condensation particle counter (TSI model 3025A). Additionally sampled particles were collected onto lacey carbon grids (Ted Pella 01890) located downstream of the neutralizer, using an electrostatic precipitator with an applied voltage of 3 kV. These samples were characterized by high-resolution transmission electron microscopy (HRTEM), conducted on a Tecnai G2 F30 electron microscope. Larger quantities of powder were collected on glass fiber filters mounted in the reactor exhaust line. Chemical and phase composition were investigated using X-ray diffraction (XRD), performed with a Siemens D-500 diffraction meter with a 2.2-kW sealed cobalt source. Magnetic properties were characterized with a Princeton micro vibrating sample magnetometer (VSM) for room temperature hysteresis loops, and with a Quantum Design magnetic property measurement system (MPMS), utilizing a Superconducting Quantum Interference Device (SQUID) magnetometer, for low temperature measurements. To obtain zero-field cooling/field cooling (ZFC/FC) curves, the sample was initially cooled to 5 K in zero field, and then the induced magnetization was measured with an applied field of 50 Oe while heating the sample to 300 K and then cooling back to 5 K. Saturation isothermal remanent magnetization (SIRM) curves were obtained by saturating the sample in a 2.5-T field at 300 K, and then recording the remanent magnetization in zero field while cooling from 300 to 20 K and then heating back to 300 K.
Results and Discussion Nanoparticle Size Distribution and Morphology Figure 2 shows particle mobility diameter distributions measured with SMPS, and TEM images of particles sampled for the same set of conditions, for an oxygen flow rate of 20 sccm. As seen in Fig. 2a, the measured size distributions for a given set of run conditions showed good reproducibility. For the case shown, the results indicate the mode in the size distribution of *8 to 9 nm, a maximum agglomerate size around 30 nm, and a geometric standard deviation of *1.49, close to the *1.45 value expected for a coagulating aerosol [46]. Figure 2b shows a corresponding low-resolution TEM image, and Fig. 2c and d show two HRTEM images. Note that the agglomerate structures seen in Fig. 2b and c were not necessarily formed in the aerosol phase, as loose agglomerates can form during deposition on the TEM grid. Nevertheless, as the HRTEM images show that individual crystallites are smaller than 10 nm, it is evident that the mobility diameter distributions seen in Fig. 2a
123
524
Plasma Chem Plasma Process (2012) 32:519–531
Fig. 2 Aerosol sampled from reactor exhaust at oxygen flow rate of 20 sccm: a particle size distributions measured by SMPS (different symbols correspond to different voltage scans of the differential mobility analyzer); b low-resolution TEM image; c, d high-resolution TEM images. Large structure in TEM images is TEM grid
represent a mixture of isolated crystallites together with small agglomerates. Figure 2d clearly shows the lattice fringes of a single iron oxide nanoparticle. The d-spacing measured in the image equals *0.30 nm, corresponding to the magnetite (022) plane or the maghemite (220) plane. Chemical and Phase Composition Figure 3 shows measured XRD patterns of powder collected on glass fiber filters for various oxygen flow rates. As the flow rate of oxygen increases from 0 to 30 sccm, the XRD patterns show a clear trend toward increasing oxidation. For the case where no oxygen was added to the plasma, no crystalline signal is detected in the XRD pattern. Instead the pattern in this case is mainly indicative of amorphous carbon. Although the thermal plasma is expected to break the Fe–C bonds in the ferrocene
123
Plasma Chem Plasma Process (2012) 32:519–531
525
Fig. 3 X-ray diffraction patterns of powder collected on glass fiber filters located in reactor exhaust, for various flow rates of oxygen. Patterns were analyzed using JADE 9 software (Materials Data, Inc.)
precursor, the presence of sufficient oxygen is evidently necessary to prevent the condensation of carbonaceous material. An HRTEM image for the zero-oxygen case is shown in Fig. 4. The material appears to be mainly amorphous, consistent with the XRD pattern, but one also sees isolated areas of lattice fringes, from crystallites that are only a few nm in size. The d-spacing of the crystallite highlighted in the center of the image measures *0.21 nm. Within the uncertainty of the measurement, it could represent either the (110) plane of Fe (0.2027 nm), the (004) plane of magnetite (0.2096 nm), or the (400) plane of maghemite (0.20869 nm). The small size of these crystallites, together with their being embedded in an amorphous carbon matrix, may explain the absence of evident peaks in the XRD pattern. For cases where oxygen was added to the gases injected into the plasma, at flow rates ranging from 10 to 40 sccm, crystalline peaks appear in the XRD patterns of collected powder, as seen in Fig. 3. At an O2 flow rate of 10 sccm, the XRD pattern shows peaks associated with pure Fe, wu¨stite (FeO) and magnetite (Fe3O4) and/or maghemite (c-Fe2O3). Note that magnetite and maghemite have the same cubic spinel structure and quite similar lattice parameters (magnetite a = 0.8396 nm, maghemite a = 0.83474 nm) [8], as a result of which their dominant XRD peaks are effectively indistinguishable. At an oxygen flow rate of 20 sccm, the pure Fe peak almost disappears, while the peaks associated with magnetite and/or maghemite show increased intensity. The wu¨stite peak remains (though much smaller relative to the magnetite/maghemite peaks than in the 10-sccm case) and a small peak associated with hematite, a-Fe2O3, appears. With further increases in the oxgyen flow rate, to 30 and 40 sccm, the wu¨stite peak disappears. The powder is predominantly magnetite and/or maghemite, with a small component of hematite. The hematite phase is antiferromagnetic, which decreases the average saturation magnetization of the products.
123
526
Plasma Chem Plasma Process (2012) 32:519–531
Fig. 4 HRTEM image of sample for case of zero oxygen flow. Circles highlight several regions with evident crystalline lattice fringes
Magnetic Properties Hysteresis loops measured by VSM at room temperature with a maximum applied field of 1.5 T are shown in Fig. 5a for cases that correspond to the XRD patterns shown in Fig. 3. Magnetic moments were normalized in each case by the total sample mass collected on the glass fiber filters. All the curves show nearly zero hysteresis, indicating that the particles are superparamagnetic [12, 47], and approach their saturation values rapidly, with the exception of the case with an oxygen flow rate of 10 sccm. The saturation magnetization (Ms) and coercivity (Hc) determined from these measurements are summarized in Fig. 5b and c.
Fig. 5 Vibrating sample magnetometer measurements at room temperature of collected powder samples for various oxygen flow rates: a hysteresis loops; b saturation magnetization (Ms); c coercivity (Hc)
123
Plasma Chem Plasma Process (2012) 32:519–531
527
The results for the zero oxygen case, which show a surprisingly high saturation magnetization of 32.3 emu/g, with a coercivity measuring 45 Oe, suggest that the crystallites seen in Fig. 4 are primarily pure Fe. The saturation magnetization of bulk iron, 222 emu/g [48], is much higher than that of either magnetite (92–100 emu/g) or maghemite (60–80 emu/g) [8], and the crystallites seen in Fig. 4 are smaller than the 8-nm superparamagnetic critical size for Fe. This hypothesis is further supported by the Fe peak in the XRD pattern, Fig. 3, for the 10-sccm case, since if pure Fe particles are formed in the 10-sccm oxygen case, then they are even more likely to form in the zero-oxygen case. The much higher coercivity in the 10-sccm case, *90 Oe, can plausibly be explained by the fact that growth of the Fe particles in the zero-oxygen case is constrained by their encapsulation in an amorphous carbon matrix, whereas there is much less carbon inclusion with added oxygen, allowing the Fe particles to grow larger than 8 nm. Although magnetite and maghemite particles are superparamagnetic up to 20 nm, the presence of a relatively small fraction of Fe particles larger than 8 nm would cause a significant increase in coercivity. The highest measured saturation magnetization, 40.15 emu/g, was achieved at an oxygen flow rate of 20 sccm. The coercivity and remanence for this sample measured 26 Oe and 1.5 emu/g, respectively, meaning that the powder is almost completely superparamagnetic. The fact that the saturation magnetization is lower than the values for bulk magnetite and maghemite may be due to the presence of a small fraction of hematite, as indicated by the XRD pattern, hematite being antiferromagnetic [8], and/or to the frequently observed, and theoretically predicted, strong reduction in saturation magnetization for superparamagnetic particles as crystallite size decreases [49–52]. However a recent study [53], which involved wet chemical synthesis of *5-nm-diameter magnetite particles that were subsequently coated with a thin layer of organic film and then gold, reported a very high Ms of 81 emu/g, suggesting that the effect of particle size on Ms may actually be due to impurities on the particle surfaces, which become more important as particle size decreases. As the oxygen flow rate increases above 20 sccm, Fig. 5b shows that the saturation magnetization decreases, while the coercivity, shown in Fig. 5c, is basically unchanged. We hypothesize that this behavior is due to increasing oxidation of the superparamagnetic particles from magnetite, Fe3O4, to maghemite, c-Fe2O3. As noted above, the bulk saturation magnetization of maghemite is lower than that of magnetite. From the XRD patterns in Fig. 3, the small peak at 33.28° attributed to hematite, a-Fe2O3, is essentially unchanged as the oxygen flow rate increases from 20 to 40 sccm, indicating that the increased oxidation occurs between the two superparamagnetic phases, and therefore does not affect the coercivity. Also supporting this hypothesis, as the oxygen flow rate increases the color of the synthesized powder, as observed by the naked eye, changes from black at 0 sccm to dark brown at 20 sccm, to brown at 30 sccm, and to reddish brown at 40 sccm. In any case, the magnetic properties obtained in this study are far superior to those reported for any previous plasma synthesis of SPIONs, as reviewed in the Introduction. There are several possible explanations for these improved results compared to other studies of plasma synthesis of iron oxide nanoparticles. The precursors in our system are injected into the plasma at the upstream end of a hot-wall converging nozzle, producing a relatively uniform environment for vapor mixing and particle nucleation [54]. A long ceramic tube at the exit of the reactor additionally provides a relatively uniform hightemperature environment, and the total residence time, a few tens of ms, is too short to allow for crystallite growth beyond the superparamagnetic critical size. Further, our study shows that careful control of the oxygen-to-ferrocene ratio is crucial for control of the stoichiometry and phase of the particles produced.
123
528
Plasma Chem Plasma Process (2012) 32:519–531
A number of studies have also reported flame synthesis of SPIONs [15, 52, 55–60]. Among these, the best magnetic properties achieved appear to be the recent results of Kumfer et al. [60]. In a series of five experiments with various flame conditions, their measured values of room-temperature saturation magnetization increased monotonically with coercivity, with values of Ms ranging from 6.5 to 60 emu/g, as values of Hc ranged from 13 to 76 Oe. They attributed increases in Ms both to increases in particle size and to the presence of a significant fraction of pure Fe in the sample with the highest Ms, and increases in coercivity to an increasing fraction of particles above the critical size for superparamagnetism. The sample that lay in the middle of their results, with Ms = 42 emu/ g and Hc = 30 Oe, is virtually identical to the results for our case of 20-sccm oxgyen flow, for which we obtained Ms = 40.15 emu/g and Hc = 26 Oe. As in our case (see below), they concluded that their powder samples consisted primarily of Fe3O4. This is in contrast to the more typical result that flame synthesis, with its inherent abundance of oxygen, tends to produce c-Fe2O3 [15, 55–57]. To produce magnetite, with its higher saturation magnetization than maghemite, Kumfer et al. [60] ran under more reducing conditions than in the typical flame. Hysteresis loops of plasma-synthesized iron oxide nanoparticles measured at room temperature (by VSM) and at 10 K (by MPMS) for the case of 20-sccm oxygen are compared in Fig. 6. At room temperature the hysteresis loop is nearly reversible, with almost zero hysteresis. At 10 K, a small increase of the saturation magnetization is observed, to *43 emu/g compared with *40 emu/g at room temperature, while the coercivity dramatically increases, to *330 Oe compared with *26 Oe at room temperature. This behavior is expected, as the thermal energy at 10 K is not sufficient to reverse the magnetization when the field is reduced to zero. At 10 K these particles are blocked, and behave like ferrimagnetic materials, with an increased remanence and coercivity as seen in Fig. 6. To provide further information about the magnetic properties of the powder produced in the 20-sccm oxygen case, zero-field cooling/field cooling curves were measured by MPMS with a field of 50 Oe, shown in Fig. 7a, and room-temperature saturation isothermal remanent magnetization (SIRM) measurements were obtained, as shown in Fig. 7b. The
Fig. 6 Hysteresis loops measured at both room temperature and 10 K, for powder sample synthesized at oxygen flow rate of 20 sccm
123
Plasma Chem Plasma Process (2012) 32:519–531
529
Fig. 7 Saturation isothermal remanent magnetization measurements of powder sample synthesized at an oxygen flow rate of 20 sccm: a zero-field-cooling/field-cooling measurements; b zero-field cooling and rewarming measurements
abrupt change in the magnetic moments for both the cooling and warming curves in Fig. 7b, as well as the sharp change in the slopes of the ZFC/FC curves in Fig. 7a, both of which occur around 120 K, corresponds to the known Verwey transition of magnetite, whose equilibrium structure is cubic above the Verwey temperature TV & 120 K, and monoclinic below TV [61, 62]. As no corresponding phase transition occurs in maghemite, these measurements indicate that the sample is substantially composed of magnetite, although the additional presence of maghemite cannot be ruled out. Indeed, the fact that the transition in the curves seen in Fig. 7b is relatively broad rather than sharp could be due to some degree of maghemitization of the magnetite nanoparticles [63]. Additionally, the temperature where the ZFC and FC curves in Fig. 7a separate, around 250 K, can be equated to the blocking temperature TB of the largest crystallites in the sample [64]. At temperatures below TB the thermal energy is insufficient to overcome the magnetic anisotropy energy of the nanoparticles, preventing superparamagnetism. The crystallite volume V can be estimated by V¼
25kTB ; Keff
ð2Þ
where Keff is the effective magnetic anisotropy constant, and k is the Boltzmann constant [65]. Using Keff & 5 9 104 J/m for magnetite [8, 66], and assuming spherical crystallites, the largest crystallite diameter estimated by Eq (2) equals *15 nm. Note that most crystallites are considerably smaller than this, as evidenced by both the SMPS measurements (which include agglomerates) and the TEM images in Fig. 2. Thus essentially all of the crystallites are below the *20-nm critical size for superparamagnetism in magnetite and maghemite.
Conclusions A DC thermal plasma system was used to synthesize superparamagnetic iron oxide nanoparticles, using ferrocene vapor as the iron precursor. The chemical and phase composition of the nanoparticles produced depended on the flow rate of oxygen added to the argon-helium plasma. At zero oxygen flow rate, the powder produced consisted of 2–3 nm Fe crystallites in an amorphous carbon matrix. As the oxgyen flow rate increased, the iron was increasingly oxidized.
123
530
Plasma Chem Plasma Process (2012) 32:519–531
For the oxgyen flow rate where the best magnetic properties were achieved, the powder consisted primarily of magnetite (Fe3O4), with possible contributions of maghemite (c-Fe2O3), as well as a small fraction of hematite (a-Fe2O3). The mode in the particle mobility diameter distribution measured by SMPS equalled 8–9 nm. SMPS measurements and TEM images indicate that the material produced consisted both of single crystallites and small agglomerates, with a maximum agglomerate size of *30 nm and a maximum crystallite size, determined by SQUID magnetometry, of *15 nm. Superparamagnetism was observed in the collected powder for most of the oxygen flow rates tested. The best room-temperature saturation magnetization measured equalled 40.15 emu/g, with a coercivity of 26 Oe and a remanence of 1.5 emu/g. To our knowledge these results are the best yet reported, from the viewpoint of high saturation magnetization with negligible coercivity, for synthesis of SPIONs by any type of plasma, and are comparable to the best results reported to date for flame synthesis. Acknowledgments This research was primarily supported by the U.S. National Science Foundation under Award Numbers CBET-0730184 and CBET-1066343, and by the Minnesota Futures Grant Program. Parts of the characterization work were conducted at the College of Science and Engineering Characterization Facility and the Institute for Rock Magnetism at the University of Minnesota.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
Willard MA, Kurihara LK, Carpenter EE, Calvin S, Harris VG (2004) Int Mater Rev 49:125–170 Lu AH, Salabas EL, Schuth F (2007) Angew Chem Int Ed 46:1222–1244 Pankhurst QA, Connolly J, Jones SK, Dobson J (2003) J Phys D 36:R167–R181 Gupta AK, Gupta M (2005) Biomater 26:3995–4021 Laurent S, Forge D, Port M, Roch A, Robic C, Elst LV, Muller RN (2008) Chem Rev 108:2064–2110 Gao J, Gu H, Xu B (2009) Acct Chem Res 42:1097–1107 Hao R, Xing R, Xu Z, Hou Y, Gao S, Sun S (2010) Adv Mater 22:2729–2742 Cornell RM, Schwertmann U (2003) The iron oxides: structure, properties, reactions, occurrences and uses, 2nd edn. Wiley-VCH, Weinheim, Germany Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ (1999) Radiol 210:459–466 Kumano S, Murakami T, Kim T, Hori M, Okada A, Sugiura T, Noguchi Y, Kawata S, Tomoda K, Nakamura H (2003) Am J Roentgen 181:1335–1339 Rabias I, Tsitrouli D, Karakosta E, Kehagias T, Diamantopoulos G, Fardis M, Stamopoulos D, Maris TG, Falaras P, Zouridakis N, Diamantis N, Panayotou G, Verganelakis DA, Drossopoulou GI, Tsilibari EC, Papavassiliou G (2010) Biomicrofluid 4:024111 Jeong JR, Lee SJ, Kim JD, Shin SC (2004) Physica Stat Solid B 241:1593–1596 Goya GF, Morales MP (2004) J Meta Nanocryst Mater 20–21:673–678 Smolensky ED, Park HYE, Berquo TS, Pierre VC (2010) Contrast Media Mol Imaging 6:189–199 Teleki A, Suter M, Kidambi PR, Ergeneman O, Krumeich F, Nelson BJ, Pratsinis SE (2009) Chem Mater 21:2094–2100 Boies AM, Roberts JT, Girshick SL, Zhang B, Nakamura T, Mochizuki A (2009) Nanotechnol 20:295604 Calder S, Boies A, Lei P, Girshick S, Roberts J (2011) Chem Mater 23:2917–2921 Yoshida T, Akashi K (1981) Trans Jpn Inst Met 22:371–378 Girshick SL, Chiu C-P, Muno R, Wu CY, Yang L, Singh SK, McMurry PH (1993) J Aerosol Sci 24:367–382 Chou CH, Phillips J (1992) J Mater Res 7:2107–2113 Vollath D, Szabo DV, Taylor RD, Willis JO, Sickafus KE (1995) Nanostruct Mater 6:941–944 Vollath D, Szabo DV, Taylor RD, Willis JO (1997) J Mater Res 12:2175–2182 Kalyanaraman R, Yoo S, Krupashankara MS, Sudarshan TS, Dowding RJ (1998) Nanostruct Mater 10:1379–1392 Janzen C, Wiggers H, Knipping J, Roth P (2001) J Nanosci Nanotechnol 1:221–225 Li SZ, Hong YC, Uhm HS, Li ZK (2004) Jpn J Appl Phys 43:7714–7717
123
Plasma Chem Plasma Process (2012) 32:519–531
531
26. Zajickova L, Synek P, Jasek O, Elias M, David B, Bursik J, Pizurova N, Hanzlikova R, Lazar L (2009) Appl Surf Sci 255:5421–5424 27. David B, Pizurova N, Schneeweiss O, Kudrle V, Jasek O, Synek P (2011) Jpn J Appl Phys 50:08JF11 28. Synek P, Jasek O, Zajickova L, David B, Kudrle V, Pizurova N (2011) Mater Lett 65:982–984 29. Bica I (1999) Mater Sci Engg B 68:5–9 30. Balasubramaniam C, Khollam YB, Banerjee I, Bakare PP, Date SK, Das AK, Bhoraskar SV (2004) Mater Lett 58:3958–3962 31. Banerjee I, Khollam YB, Balasubramanian C, Pasricha R, Bakare PP, Patil KR, Das AK, Bhoraskar SV (2006) Script Mater 54:1235–1240 32. Chazelas C, Coudert JF, Jarrige J, Fauchais P (2006) J Eur Ceram Soc 26:3499–3507 33. Banerjee I, Khollam YB, Mahapatra SK, Das AK, Bhoraskar SV (2010) J Vac Sci Technol A 28:1399–1403 34. Subramanian V, Baskaran R, Krishnan H (2009) Aerosol Air Qual Res 9:172–186 35. McIlroy DN, Zhang D, Norton MG, O’Brien WL, Schwickert MM, Harp GR (2000) J Appl Phys 87:7213–7217 36. McIlroy DN, Huso J, Kranov Y, Marchinek J, Ebert C, Moore S, Marji E, Gandy R, Hong YK, Norton MG, Cavalieri E, Benz R, Justus BL, Rosenberg A (2003) J Appl Phys 93:5643–5649 37. Kouprine A, Gitzhofer F, Boulos M, Veres T (2006) Carbon 44:2593–2601 38. Panchal V, Neergat M, Bhandarkar U (2011) J Nanopart Res 13:3825–3833 39. Panchal V, Lahoti G, Bhandarkar U, Neergat M (2011) J Phys D 44:345205 ¨ zdemir O ¨ (1997) Rock magnetism: fundamentals and frontiers. Cambridge University 40. Dunlop DJ, O Press, Cambridge, New York 41. Powell QH, Fotou GP, Kodas TT, Anderson BM (1997) Chem Mater 9:685–693 42. Zhang B (2007) Thermal plasma synthesis and photoinduced coating of aluminum nanoparticles. PhD thesis, Univ Minnesota, Minneapolis 43. Monte MJS, Santos LMNBF, Fulem M, Fonseca JMS, Sousa CAD (2006) J Chem Eng Data 51:757–766 44. Jain R, Girshick SL, Heberlein JV, Mukherjee R, Zhang B, Nakamura T, Mochizuki A (2010) Plasma Chem Plasma Process 30:795–811 45. Wang X, Hafiz J, Mukherjee R, Renault T, Heberlein J, Girshick SL, McMurry PH (2005) Plasma Chem Plasma Process 25:439–453 46. Friedlander SK, Wang CS (1966) J Coll Interface Sci 22:126–132 47. Benitez MJ, Mishra D, Szary P, Confalonieri GAB, Feyen M, Lu AH, Agudo L, Eggeler G, Petracic O, Zabel H (2011) J Phys Condens Matter 23:126003 48. Danan H, Herr A, Meyer AJP (1968) J Appl Phys 39:669–670 49. Berkowitz AE, Schuele WJ, Flanders PJ (1968) J Appl Phys 39:1261–1263 50. Coey JMD (1971) Appl Phys Lett 27:1140–1142 51. Frenkel J, Doefman J (1930) Nature 126:274–275 52. Li D, Teoh WY, Selomulya C, Woodward RC, Munroe P, Amal R (2007) J Mater Chem 17:4876–4884 53. Smolensky ED, Neary MC, Zhou Y, Berquo TS, Pierre VC (2011) Chem Commun 47:2149–2151 54. Rao N, Girshick S, Heberlein J, McMurry P, Bench M, Jones S, Hansen D, Micheel B (1995) Plasma Chem Plasma Process 15:581–607 55. Zachariah MR, Aquino MI, Shull RD, Steel EB (1995) Nanostruct Mater 5:383–392 56. Janzen C, Knipping J, Rellinghaus B, Roth P (2003) J Nanopart Res 5:589–596 57. Li D, Teoh WY, Woodward RC, Cashion JD, Selomulya C, Amal R (2009) J Phys Chem C 113:12040–12047 58. Strobel R, Pratsinis SE (2009) Adv Powder Technol 20:190–194 59. Guo B, Yim H, Khasanov A, Stevens J (2010) Aerosol Sci Technol 44:281–291 60. Kumfer BM, Shinoda K, Jeyadevan B, Kennedy IM (2010) J Aerosol Sci 41:257–265 61. Dunlop DJ (2007) J Geophys Res Solid Earth 112:B11103 62. Verwey EJW (1939) Nature 144:327–328 63. Ozdemir O, Dunlop DJ (2010) J Geophys Res-Solid Earth 115:B02101 64. Maity D, Choo S-G, Yi J, Ding J, Xue JM (2009) J Magn Magn Mater 321:1256–1259 65. Cullity BD, Graham CD (2009) Introduction to magnetic materials, 2nd edn. IEEE/Wiley, Hoboken 66. Guardia P, Batlle-Brugal B, Roca AG, Iglesias O, Morales MP, Serna CJ, Labarta A, Batlle X (2007) J Magn Magn Mater 316:E756–E759
123
