Magnetic characterization of noninteracting, randomly oriented, nanometer scale ferrimagnetic particles

Transcription

1 Click Here for Full Article JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115,, doi: /2009jb006855, 2010 Magnetic characterization of noninteracting, randomly oriented, nanometer scale ferrimagnetic particles Changqian Cao, 1,2,3 Lanxiang Tian, 1,2 Qingsong Liu, 1 Weifeng Liu, 4 Guanjun Chen, 4 and Yongxin Pan 1,2 Received 5 August 2009; revised 5 February 2010; accepted 19 February 2010; published 16 July [1] Studying the magnetic properties of ultrafine nanometer scale ferrimagnetic particles (<10 nm) is vital to our understanding of superparamagnetism and its applications to environmental magnetism, biogeomagnetism, iron biomineralization, and biomedical technology. However, magnetic properties of the ultrafine nanometer sized ferrimagnetic grains are very poorly constrained because of ambiguities caused by particle magnetostatic interactions and unknown size distributions. To resolve these problems, we synthesized magnetoferritins using the recombinant human H chain ferritin (HFn). These ferrimagnetic HFn were further purified through size exclusion chromatography to obtain monodispersed ferrimagnetic HFn. Transmission electron microscopy revealed that the purified ferrimagnetic HFn are monodispersed and each consists of an iron oxide core (magnetite or maghemite) with an average core diameter of 3.9 ± 1.1 nm imbedded in an intact protein shell. The R value of the Wohlfarth Cisowski test measured at 5 K is 0.5, indicating no magnetostatic interactions. The saturation isothermal remanent magnetization acquired at 5 K decreased rapidly with increasing temperature with a median unblocking temperature of 8.2 K. The preexponential frequency factor f 0 determined by AC susceptibility is (9.2 ± 7.9) Hz. The extrapolated M rs /M s and B cr /B c at 0 K are 0.5 and 1.12, respectively, suggesting that the ferrimagnetic HFn cores are dominated by uniaxial anisotropy. The calculated effective magnetic anisotropy energy constant K eff = J/m 3, which is larger than previously reported values for bulk magnetite and/or maghemite or magnetoferritin and is attributed to the effect of surface anisotropy. These data provide useful insights into superparamagnetism as well as biomineralization of ultrafine ferrimagnetic particles. Citation: Cao, C., L. Tian, Q. Liu, W. Liu, G. Chen, and Y. Pan (2010), Magnetic characterization of noninteracting, randomly oriented, nanometer scale ferrimagnetic particles, J. Geophys. Res., 115,, doi: /2009jb Introduction [2] Superparamagnetic (SP) particles of magnetite (Fe 3 O 4 ) or maghemite (g Fe 2 O 3 ) are of great interest in the fields of rock magnetism, environmental magnetism, and magnetic materials science as well as for medical and industrial applications [Moskowitz et al., 1997; Winklhofer et al., 1997; Wong et al., 1998; Hanesch and Petersen, 1999; Nolan et al., 1999; Egli, 2009;Galindo Gonzalez et al., 2009]. For example, ultrafine grained magnetite and/or 1 Paleomagnetism and Geochronology Laboratory, Key Laboratory of the Earth s Deep Interior, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China. 2 Also at France China Bio Mineralization and Nano Structures Laboratory, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China. 3 Also at Graduate University of Chinese Academy of Sciences, Beijing, China. 4 State Key Laboratory of Microbial Technology, School of Life Sciences, Shandong University, Jinan, China. Copyright 2010 by the American Geophysical Union /10/2009JB maghemite are significant sources of magnetic susceptibility enhancement in soils and paleosols [Zhou et al., 1990; Maher, 1998; Banerjee, 2006]. Nanometer scale magnetite particles (2 5 nm) in clusters contained in the upper beak skin of homing pigeons are assumed to form part of a biological magnetic field receptor [Winklhofer et al., 2001; Tian et al., 2007]. Furthermore, nanometer scale magnetites are able to pass biological barriers that are usually inaccessible to larger particles; thus, they can be developed into novel diagnostic and therapeutic tools [Uchida et al., 2006, 2009]. [3] Néel theory [Néel, 1949] is the foundation for the quantitative interpretation of superparamagnetic grains and serves as the starting point for any modern applications to rock magnetism and paleomagnetism. The theory of thermoremanence by Néel [1949] links superparamagnetism and thermoremanence as thermally activated and blocked state, respectively, of single domain magnetic particles [Dunlop and Özdemir, 1997; Banerjee, 2006]. However, classic Néel theory does not always perfectly describe the superparamagnetic behavior of natural and synthetic samples 1of10

2 Figure 1. TEM analyses of the ferrimagnetic HFn. Bright field TEM images of (a) negative stained and (b) unstained ferrimagnetic HFn, (c) a size distribution histogram, and (d) an electron diffraction pattern. because of the effects of magnetostatic interactions between poorly dispersed grains [Maher, 1988; Muxworthy, 2001; Muxworthy et al., 2003; Blanco Mantecón and O Grady, 2006]. Therefore, a population of noninteracting, superparamagnetic grains with a well defined size distribution would help workers refine existing theories of thermal activation and superparamagnetism. [4] Ferritin is an iron storage protein that is ubiquitously distributed throughout the microbial, plant, and animal kingdoms. It is composed of a spherical cagelike protein shell with an outer (inner) diameter of nm (8 nm) and a solid iron oxide or iron oxy hydroxide core [Arosio et al., 2009]. The magnetoferritin was first successfully reconstituted with magnetic iron oxide (magnetite or maghemite) by using demetalized horse spleen ferritin (apoferritin) as a constrained reaction vessel under slow oxidation and relative high temperature (60 C 65 C) and ph (8.5) [Meldrum et al., 1992; Wong et al., 1998]. The artificial magnetoferritin potentially provides an ideal material for studying superparamagnetism. Moskowitz et al. [1997] carried out very detailed low temperature magnetization and Mössbauer spectroscopy analyses on SP maghemite particles in reconstituted magnetoferritin. They determined the preexponential frequency factor f Hz for the SP sample. However, their results could be biased by particle aggregation, as revealed by their transmission electron microscopy (TEM) images and the Henkel plot (Figures 1 and 7 in their paper). Therefore, to properly characterize SP particles, studies on magnetoferritin samples with narrow grain size distribution and without magnetic interactions are needed. [5] In this study, we successfully synthesized nanometerscale ferrimagnetic iron oxide (magnetite or maghemite) in the protein cavity using genetically engineered human H chain ferritin (HFn). These ferrimagnetic HFn were further purified to obtain monodispersed ferrimagnetic HFn. Systematic low temperature magnetic measurements were conducted on the noninteracting ferrimagnetic HFn sample to understand the ultrafine superparamagnetism. In particular, we focused on determinations of the preexponential frequency factor f 0, the effective energy barrier constant K eff, and the surface anisotropy per unit of surface area K s for the ferrimagnetic HFn. 2. Materials and Methods [6] The recombinant plasmid pet12b HFn containing the sequence coding the human H chain ferritin was transformed into Escherichia coli strain Rosetta. The protein expression and purification were carried out according to the procedure of Santambrogio et al. [2000]. The purity of the protein was examined by sodium dodecyl sulfate polyacrylamide gel electrophoresis and the protein concentration was determined by BCA Protein Assay Kit (Pierce) with bovine serum albumin as standard. The ferrimagnetic HFn 2of10

3 was synthesized according to the method of Uchida et al. [2006] with some modifications. All solutions were carefully deoxygenated with argon and transferred into an anaerobic chamber (COY Laboratory Products Inc.). The solution of 50 ml 100 mm NaCl with HFn (0.25 mg/ml) was added to the reaction vessel. The temperature of the vessel was kept at 65 C through an electric heating mantle and the ph was stabilized at 8.5 using 50 mm NaOH with a ph stat titrator (842 Titrando, Metrohm). Fe(II) (12.5 mm (NH 4 ) 2 Fe(SO 4 ) 2 6H 2 O) was added in a rate of 100 Fe/ (protein min) using a dosing device (800 Dosino) connected with 842 Titrando. Simultaneously, stoichiometric equivalents (1:3 H 2 O 2 :Fe 2+ ) of freshly prepared H 2 O 2 (4.17 mm) were added as an oxidant, consistent with the following reaction. After adding theoretical 2300 Fe/ protein cage to the reaction vessel, the reaction continued another 5 min before finishing. Finally, 200 ml of 300 mm sodium citrate was added to chelate any free iron. The magnetite forming reaction can be expressed as 3Fe 2þ þ H 2 O 2 þ 2H 2 O! Fe 3 O 4 þ 6H þ : To get rid of the aggregated ferrimagnetic HFn, the synthesized ferrimagnetic HFn was further purified through two steps (1) centrifugating the synthesized ferrimagnetic HFn sample for 20 min at 10,000 g to remove aggregated clusters and (2) passing the sample through size exclusion chromatography (Sepharose 6B, GE Healthcare). Size exclusion chromatography is a useful method for fractionating proteins, water soluble polymers, and water based ferrofluid based on size difference [Nunes and Yu, 1987]. The method is usually achieved with a column that consists of a hollow tube tightly packed with small porous polymer beads designed to have pores of different sizes. When the ferrimagnetic HFn particles were loaded into the size exclusion column, the aggregates of ferrimagnetic HFn were unable to pass through the pore spaces because of their larger size. They would elute as different bands to achieve successful separation of monodispersed ferrimagnetic HFn. Finally, the monodispersed ferrimagnetic HFn were concentrated using Microcon Ultrafilters (Millipore) with 100 kd cutoff and then freeze dried. These monodispersed ferrimagnetic HFn were used for later TEM and magnetic analysis. [7] For TEM analysis, ferrimagnetic HFn particles were deposited on Formvar coated copper grids. Analysis was performed using HRTEM JEOL The magnetic measurements were performed using a Magnetic Property Measurement System (MPMS 5XL, sensitivity Am 2, Quantum Design, Inc.). Low field magnetization versustemperature curves were measured between 5 and 300 K in a field of 1.5 mt. The zero field cooling (ZFC) curve was measured while the sample was warmed after zero field cooling from 300 to 5 K and the subsequent field cooled (FC) curve was measured after being field cooled from 300 to 5 K in a field of 1.5 mt. The thermal decay of isothermal remanence acquired at 5 K in a field of 2.5 T after cooling in ZFC and a 2.5 T FC was measured between 5 and 300 K. [8] To evaluate magnetostatic interactions for ferrimagnetic HFn, the isothermal remanent magnetization (IRM) [I r (H)] acquisition and DC demagnetization remanence [I d (H)] curves were measured at 5 K up to ±600 mt. For each field step, a constant magnetic field was applied for 120 s, and the remanent moment was measured after removal of the field and a delay of 120 s. The hysteresis loops were measured at temperatures between 2 and 15 K as well as at 300 K, in fields of ±3 T. AC susceptibility measurements were made at frequencies of 30, 100, 330, 1000, and 1400 Hz in a weak field of 0.4 mt between 5 and 200 K. 3. Results 3.1. Transmission Electron Microscopy Analysis [9] The bright field TEM images of negative stained ferrimagnetic HFn (Figure 1a) show that those particles are monodispersed and each particle has an electron dense core with an intact protein cage. The outer diameter of the protein cage is 12 nm. No crystal is observed outside of the protein cage. The size distribution of ferrimagnetic HFn cores, fitted into lognormal distribution function, have uniform grain size with a mean diameter of 3.9 nm and a standard deviation of s V = 1.1 (Figures 1b and 1c). Selected area electron diffraction (SAED) of ferrimagnetic HFn (Figure 1d) displays d spacings of 2.54, 2.12, 1.48, and 1.12 Å, which are characteristic of the {311}, {400}, {440}, and {642} family of lattice planes for magnetite (and/or maghemite), respectively. Nanoparticles of magnetite and of maghemite of space group Fd3m cannot be distinguished by their electron diffraction patterns Magnetostatic Interactions [10] For an assembly of randomly oriented noninteracting single domain (SD) particles with uniaxial anisotropy, I r (H) and I d (H) follow the Wohlfarth equation: I d (H) =1 2I r (H) [Wohlfarth, 1958]. Figure 2a shows the IRM acquisition curves [normalized to the maximum saturation isothermal remanent magnetization (SIRM)] and the DC demagnetization curves [rescaled as 0.5(1 + I d (H)/SIRM)] [Cisowski, 1981] measured at 5 K. The R value of the Wohlfarth Cisowski test for the measured sample is 0.5, indicating a lack of magnetic interactions. The median destructive field is 35.7 mt, which is consistent with the B cr (35.5 mt). Figure 2b is the Henkel plot [Henkel, 1964] of the ferrimagnetic HFn, which displays a linear trend (r 2 = 0.99). Again, it demonstrates that the ferrimagnetic HFn sample has nearly no magnetostatic interactions Low Field Magnetization Curves [11] Figure 3a shows the temperature dependence of lowfield (1.5 mt) magnetization between 5 and 60 K after the ZFC and FC treatments. The pattern is generally similar to the observation of Moskowitz et al. [1997] except for the much lower unblocking temperature for our samples. In the low field, the ZFC curve is proportional to the initial DC susceptibility. As temperature increases, the susceptibility increases as SD particles are progressively unblocked. The susceptibility peaks at 9.2 K. In contrast, the FC curve decays continuously with increasing temperature, because the magnetization was already aligned with the field direction during cooling through the blocking temperature [Moskowitz et al., 1997]. For the measured sample, its ZFC and FC curves merge at about 50 K, which corresponds to the maximum unblocking temperature. The differences 3of10

4 Figure 2. Remanence data of ferrimagnetic HFn measured at 5 K. (a) Normalized IRM acquisition and DC demagnetization (DCD). The DCD curve was rescaled as 0.5(1 + IRM( H))/SIRM. (b) Henkel plot, I r (H) versus I d (H), normalized by SIRM. The straight solid line is the theoretical Wohlfarth equation for noninteracting systems. between the FC and ZFC curves equal the thermal remanent magnetization [Liu et al., 2004]. Above the maximum unblocking temperature (50 K), the reciprocal susceptibility (1/c i ) plotted as a function of temperature can be fitted to the Curie Weiss law (Figure 3b), i ¼ C= ðt Þ; ð1þ where C is the Curie constant and is an ordering temperature [Gittleman et al., 1974; Pedersen et al., 1994]. The reciprocal susceptibility plotted as a function of temperature above 50 K (Figure 3b) shows a good linear trend, which defines C = 6.3 and = 0.36 K Demagnetization of Saturation Isothermal Remanent Magnetization [12] The thermal demagnetization curves of SIRM acquired at 5 K show a rapid decrease of remanence with increasing temperature (Figure 4). Above 15 K, the remanence decreases to less than zero, and as temperature increases to 300 K, the remanence trends asymptotically approach zero (see inset in Figure 4). The negative values above 15 K could be caused by the presence of negative residual fields in the MPMS 5XL, which is unavoidable especially after the acquisition of the SIRM at 5 K, or because of a ferrimagnetic and an antiferrromagnetic phase coupling within ferritin cores. The 2.5 T ZFC and FC curves show very similar trends and are nearly superimposed. These data display classical SP behavior of ferrimagnetic particles: as temperature increases, the SD particles are unblocked into SP particles. The median unblocking temperature (the temperature where half of the remanence decayed) in the FC curve is T bm = 8.2 K. The remanence of ZFC and FC decreased to zero at 15 K, which means all the particles are unblocked into SP particles Hysteresis Loops [13] The hysteresis loops at different temperatures between 2 and 15 K are measured in fields of ±3 T (Figure 5). The Figure 3. (a) Low field (1.5 mt) susceptibility as a function of temperature measured after zero fieldcooling (ZFC) and field cooling (FC) treatments. (b) Reciprocal susceptibility (1/c i ) plotted as a function of temperature from the ZFC curve. The solid line is the best fit to equation (1) in the text. 4of10

5 a noninteracting single domain particle with uniaxial anisotropy, the thermal relaxation process follows the Néel Arrhenius expression [Néel, 1949], ¼ 1=f 0 expðe a =ktþ; ð2þ Figure 4. Decay curves of saturation isothermal remanent magnetizationacquiredina2.5tfieldafterzero fieldcooling (ZFC) and field cooling (FC) treatments. value of the ratio of saturation remanence to saturation magnetization (M rs /M s ) at 2 K is 0.4, and the extrapolated value at 0 K is nearly 0.5, which indicates that the magnetization in the ferrimagnetic HFn is dominated by uniaxial anisotropy [Day et al., 1977]. The coercivity (B c ) of ferrimagnetic HFn at 2 K is 40.2 mt, which drops to 1.0 mt as temperature increases to 10 K. At 15 K, there is hardly any measurable coercivity. The remanence coercivity (B cr ) also decreases as temperature increases, from 60.6 mt at 2 K to 14.0 mt at 10 K. Owing to the SD unblocked into SP particles with increasing temperature, the hysteresis loops become more and more constricted (wasp waisted) and hysteresis disappears at 15 K, which shows typical SP unblocking behavior [Tauxe et al., 1996]. [14] Figure 6 shows the modified Day plot [Dunlop, 2002] for hysteresis parameters of the ferrimagnetic HFn measured at different temperatures (2 7 K) below its unblocking temperature (8.2 K). The ferrimagnetic HFn falls well within the SD + SP region. The decrease of M rs /M s and the increase of B cr /B c with elevated temperatures indicate the proportion of SP particles increases as the SD particles are unblocked accordingly. The extrapolated value of B cr /B c corresponding to M rs /M s = 0.5 is 1.12, which is slightly larger than the theoretical predicted value of 1.09 for randomly oriented uniaxial SD particles [Wohlfarth, 1958; Gittleman et al., 1974] AC Susceptibility [15] Figure 7a is the in phase (c ) susceptibility of the ferrimagnetic HFn measured between 5 and 200 K. The curves show peaks in the range K. The peak values of c decrease and the peak temperatures (T max ) increase with increasing frequencies (see inset in Figure 7a). Clearly, the c displays frequency dependence below 50 K, whereas above 50 K, c shows nearly no frequency dependence. For where t is the relaxation time of the magnetic moment of the particles, f 0 is the preexponential frequency factor, E a is the average effective energy barrier to be overcome for the magnetic moment of a grain to switch direction, k is Boltzmann s constant, and T is the temperature. For uniaxial anisotropy particles, E a = K eff V, where K eff is the effective magnetic anisotropy energy constant, and V is the particle volume [Gittleman et al., 1974; Kim et al., 2001]. The plot of lnt versus 1/T max (Néel Arrhenius plot; see Figure 7b) shows a nice linear trend, which means the ferrimagnetic HFn are noninteracting particles and follow the Néel Arrhenius expression. Thus, according to equation (2) and the Néel Arrhenius plot (Figure 7b), the average value of E a is J. The effective magnetic anisotropy energy constant K eff is estimated to be about J/m 3 ( erg/cm 3 ) and f Hz from the in phase susceptibility data of the ferrimagnetic HFn. [16] The T max of out of phase (c ) susceptibility of the ferrimagnetic HFn (Figure 7c) is nearly independent of the frequency. The c data approach zero when the temperature is increased above 50 K, which indicates that the sample is completely unblocked at higher temperature at the frequencies being measured and all particles in the sample behave superparamagnetically Determination of Median Blocking Temperature [17] The blocking temperature is the temperature at which the relaxation time is equal to the time scale of experimental measurement. For a typical DC remanence measurement with t = 100 s and H 0, setting t = t gives the following expression: T bm ¼ E a =k lnðf 0 tþ; ð3þ where T bm is the median blocking temperature corresponding to the median volume of the sample; the average effective energy barrier E a = K eff V J and the preexponential frequency factor f Hz were calculated from AC susceptibility data of ferrimagnetic HFn (see section 3.6). Therefore, we get T bm = 9.4 K, which is very consistent with the susceptibility peak around 9.2 K (Figure 3). [18] However, the value of T bm determined from the FC SIRM decaying curve (Figure 4) is 8.2 K, which deviates from the calculated T bm value and susceptibility peak temperature. The difference could be interpreted by effects from the size distribution. Theoretically, the T bm determined from the decay of high field saturation remanence (Figure 4) is less affected by size distribution and can be taken as true T bm ; the peak temperature of low field ZFC equals bt bm, where the value of b depends on the particle size distribution. Jiang and Mørup [1997] simulated b values for different lognormal volume distributions of ferromagnetic particle systems. They found the b value increased from 1 to 2 as the standard deviation s V increased from 0.1 to 1.3. From our sample, we can get a factor of b = of10

6 Figure 5. Hysteresis loops of the ferrimagnetic HFn measured at 2, 5, 7, 10, 15, and 300 K Surface Anisotropy [19] As the particle size decreases, the fraction of atoms lying at the surface with lower atomic coordination increases. This results in the increase of site specific surface energy, usually taken as a local uniaxial anisotropy normal to the surface. The surface anisotropy will lengthen the characteristic switching time of the particle magnetization and increase the effective energy barrier [Gilmore et al., 2005; Mazo Zuluaga et al., 2008; Pérez et al., 2008]. For a spherical magnetite particle with diameter of D, the effective 6of10

7 Figure 6. Modified Day plot, M rs /M s versus B cr /B c [Dunlop, 2002] for the ferrimagnetic HFn. Stars indicate the measured data at temperatures of 2, 4, 5, 6, and 7 K. The solid line is the best fit line to the equation M rs /M s = a(b cr /B c ) b with a = 0.55 and b = magnetic anisotropy energy per volume unit follows [Pérez et al., 2008]: K eff ¼ K v þ 6K s =D; where K eff is the effective magnetic anisotropy energy per volume unit, K v is the core anisotropy energy per volume unit, and K s is the surface anisotropy per unit of surface area. Assuming the magnetic mineral phase of ferrimagnetic HFn is magnetite, and the particle shape is spherical with no shape anisotropy, K v =K u = J/m 3 ( erg/cm 3 ) [Gilmore et al., 2005], then the value of K s extracted from equation (4) for D = 3.9 nm is erg/cm 2. The comparison of values of K eff, K s, and f 0 for the ferrimagnetic HFn in this study and other particles is listed in Table Discussion [20] Synthesized magnetoferritin is susceptible to proteinprotein aggregation, which causes broadening size distribution, increasing magnetostatic interactions, and blocking temperature [Moskowitz et al., 1997; Wong et al., 1998; Southern et al., 2007]. In this study, we obtained monodispersed ferrimagnetic HFn as shown by the TEM photographs, with intact protein cages and iron oxide cores (mean diameter of 3.9 ± 1.1 nm; Figure 1). It indicates that the centrifugation and size exclusion chromatography procedures removed aggregated magnetoferritins. Results of the IRM acquisition and demagnetization curves, hysteresis loops, and the d spacings indicate that the cores of the synthesized ferrimagnetic HFn in the present study are lowcoercivity ferrimagnetic minerals, magnetite or maghemite. However, it is difficult to determine the exact core composition at this stage, because (1) magnetite and maghemite have similar d spacings and (2) no measurable Verwey ð4þ transition may be attributable to the thermal unblocking of SP magnetite or maghemite. The Mössbauer spectroscopy may help but it requires large quantities of samples, which are unfortunately difficult to produce in the present study. [21] The crossing point of IRM and DC demagnetization for the ferrimagnetic HFn at 5 K in this study is 0.5 (Figure 2a), which indicates the ferrimagnetic HFn samples have nearly no magnetostatic interactions. This finding was further supported by the linearity of the Henkel plot (Figure 2b) and the Néel Arrhenius plot (Figure 7b). It should be noted that the ferrimagnetic HFn in this study is used after centrifugation and size exclusion chromatography treatments. Our ferrimagnetic HFn sample is monodispersed and magnetically noninteracting with narrow size distribution. Therefore, it provides us a good opportunity to estimate the f 0, K eff, surface anisotropy, hysteretic parameters, and other related magnetic behavior for the sample. [22] The preexponential frequency factor f 0 plays a crucial role in determining the relaxation time obtained from Néel Arrhenius expression and theoretically predicting AC magnetic, DC magnetic, magnetic resonance, and Mössbauer spectroscopy phenomena [Dickson et al., 1993; Tronc et al., 1995; Moskowitz et al., 1997; Egli, 2009]. However, the physically meaningful values of f 0 are hardly obtained, mainly because the magnetostatic interactions between particles are indeed difficult to avoid and they can influence superparamagnetic relaxation [D Amico et al., 1995]. The f 0 determined for the ferrimagntic HFn in this study is about Hz, which agrees with the reported value of Hz for quasi isolated maghemite particles with very weak interactions [Dormann et al., 1999]. However, the value of f 0 in this experiment is higher than f determined by Moskowitz et al. [1997]. This may be caused by different samples used in two experiments. The sample used in this earlier study was maghemite cores encapsulated in horse spleen ferritin with some interactions. Also, the f 0 of the ferrimagnetic HFn in this study is smaller than the f 0 of native horse spleen ferritin ( Hz) [Guertin et al., 2007]. We argue that this difference in the values of f 0 may be caused by the different types of magnetic spin ordering and anisotropy of the core in these two types of ferritins: ferrimagnetic in ferrimagnetic HFn and antiferromagnetic in native horse spleen ferritin. In addition, because the value of f 0 in this study was obtained by fitting to data between 12 and 16 K (Figure 7b) and extrapolating to 0 K, a potential error might be introduced by the extrapolation. Analyzing by a linear regression, we estimated the standard error for f 0 at 0 K to be Hz. [23] The calculated value of the effective magnetic anisotropy energy constant K eff J/m 3 is significantly larger than that of bulk magnetite ( or J/m 3 )[Dunlop and Özdemir, 1997; Gilmore et al., 2005] and bulk maghemite ( J/m 3 )[Birks, 1950]. Because the ferrimagnetic HFn are nearly spherical with weak shape anisotropy and no interaction, we attribute this larger K eff to surface anisotropy. The surface anisotropy per unit of surface area K s determined for ferrimagnetic HFn is erg/cm 2, which is larger than the value of other magnetoferritin ( erg/cm 2 )[Gilmore et al., 2005]. This may also be caused by different size distributions of these two samples: 3.9 nm for ferrimagnetic HFn and 7 nm for previous magnetoferritin. 7of10

8 Figure 7. (a) In phase susceptibility (c ) of the ferrimagnetic HFn; (b) Néel Arrhenius plot, lnt versus 1/T max (the solid line is the best fit to the Néel Arrhenius equation): and (c) out of phase susceptibility (c ) of the ferrimagnetic HFn. [24] When the size of magnetite and maghemite get to the nanometer scale, the surface properties cannot be neglected [Kachkachi et al., 2000; Kim and Shima, 2007]. For example, the anisotropic energy of noninteracting or interacting maghemite of 3 10 nm is governed by surface anisotropy ( erg/cm 2 )[Tronc et al., 2000]. Iglesias and Labarta [2004] simulated surface anisotropy at 0 K and found that loops became elongated with a much higher closure field, increasing of coercive field and reducing the high field susceptibility. They explained that the surface anisotropy over exchange interactions created surface disordered states which became more difficult to reverse by the magnetic field. The observed hysteresis loop for the ferrimagnetic HFn at 2 K is elongated and closed at nearly 1500 mt, which supports the findings of Iglesias and Labarta [2004]. [25] For nanometer sized particles, the particle surface produces positive uniaxial anisotropy [Gazeau et al., 1998]. Table 1. Comparison of the Calculated Values of K eff, K s, and f 0 for the Ferrimagnetic HFn Used in This Study and Previous Reported Values for Magnetoferritin Ferrimagnetic HFn Magentoferritin Bulk magnetite Bulk maghemite a Horse spleen ferritin b K eff (J/m 3 ) c d d e K s (erg/cm 2 ) d f 0 (Hz) (9.2 ± 7.9) c a Birks [1950]. b Guertin et al. [2007]. c Moskowitz et al. [1997]. d Gilmore et al. [2005]. e Dunlop and Özdemir [1997]. 8of10

9 Recent models by Labaye et al. [2002] confirmed that the surface anisotropy could compete with bulk magnetocrystalline anisotropy to form a hedgehog type spin structure where all the spins were radially oriented, either outward or inward toward the center of the nanosphere. In this study, the extrapolated ratio M rs /M s is 0.5 at 0 K and is consistent with the study of Li et al. [2009], which supports that the ferrimagnetic HFn sample is dominated by uniaxial anisotropy. The extrapolated value of B cr /B c corresponding to M rs /M s = 0.5 is 1.12, which is slightly larger than the theoretical predicted value of 1.09 for randomly oriented uniaxial SD particles (Figure 6) [Wohlfarth, 1958]. There are several factors that may lead to increased values of B cr /B c, such as magnetostatic interactions, shape anisotropy, and surface effects [Wohlfarth, 1955; Moskowitz et al., 1997]. The shape anisotropy and magnetostatic interaction for our ferrimagnetic HFn sample can be excluded; thus, the large surface anisotropy may cause increasing values of B cr /B c. 5. Conclusions [26] We characterized the well dispersed, noninteracting, randomly oriented nanometer scale ferrimagnetic HFn particles using both TEM and low temperature magnetic analyses. This experimental data allowed us to better integrate Néel theory with direct observations of superparamagentic assemblages, which are widely distributed in geological samples and biological systems. Based on the preceding discussion, our conclusions are as follows: [27] 1. The synthesized ferrimagnetic HFn after purification have intact protein shells with iron oxides encapsulated. The SAED, IRM, and hysteresis analyses indicate that the core of the ferrimagnetic HFn is magnetite/maghemite. The ferrimagnetic HFn cores have an average size of 3.9 ± 1.1 nm. [28] 2. The ferrimagnetic HFn have nearly no magnetostatic interactions because of their intact protein shells. The observed median blocking temperature (8.2 K) is very consistent with our simulated results. [29] 3. The value of preexponential factor f 0 in the Néel Arrhenius equation can be obtained from the AC susceptibility data of the ferrimagnetic HFn. The value obtained is (9.2 ± 7.9) Hz. [30] 4. The extrapolated value of M rs /M s is nearly 0.5 and B cr /B c is 1.12 at 0 K, which suggests the ferrimagnetic HFn is dominated by uniaxial anisotropy. Owing to the surface anisotropy, the effective magnetic anisotropy energy constant K eff determined for ferrimagnetic HFn is J/m 3, which is larger than that of bulk magnetite, bulk maghemite, and magnetoferritin. [31] Acknowledgments. We thank Kuan Wang at Shandong University for kind help in genetic engineering and protein expression. We are grateful to Paolo Arosio for providing us the vector pet12b encoded cdna of human H chain ferritin. We thank Rixiang Zhu and Chenglong Deng from the Institute of Geology and Geophysics, Chinese Academy of Sciences; Lisa Tauxe from the Scripps Institution of Oceanography; and Subir K. Banerjee and Bruce Moskowitz from the Institute for Rock Magnetism at Minnesota for their very beneficial discussions. We also thank Joshua Feinberg, an anonymous reviewer, and the Associate Editor for their comments on the manuscript. This work was supported by the Chinese Academy of Sciences projects and the National Natural Science Foundation of China grant and the CAS/SAFEA International Partnership Program for Creative Research Teams. Q. Liu and Y. Pan also received support from the 100 Talent Program of the CAS. References Arosio, P., R. Ingrassia, and P. Cavadini (2009), Ferritins: A family of molecules for iron storage, antioxidation and more, Biochim. Biophys. Acta, 1790, Banerjee, S. K. (2006), Environmental magnetism of nanophase iron minerals: Test the biomineralization pathway, Phys. Earth Planet. Inter., 154, , doi: /j.pepi Birks, J. B. (1950), The properties of ferromagnetic compounds at centimetre wavelengths, Proc. Phys. Soc., 63, 65 74, doi: / /63/2/301. Blanco Mantecón, M., and K. O Grady (2006), Interaction and size effects in magnetic nanoparticles, J. Magn. Magn. Mater., 296, , doi: /j.jmmm Cisowski, S. (1981), Interacting vs. non interacting single domain behavior in natural and synthetic samples, Phys. Earth Planet. Inter., 26, 56 62, doi: / (81) D Amico, L., F. Dorazio, J. L. Dormann, D. Fiorani, F. Lucari, and E. Tronc (1995), AC susceptibility measurements on nanometric g Fe 2 O 3 particles, Mater. Sci. Forum, 195, , doi: / net/msf Day, R., M. Fuller, and V. A. Schmidt (1977), Hysteresis properties of titanomagnetites: Grain size and compositional dependence, Phys. Earth Planet. Inter., 13, , doi: / (77)90108-x. Dickson, D. P. E., N. M. K. Reid, C. Hunt, H. D. Williams, M. Elhilo, and K. Ogrady (1993), Determination of f 0 for fine magnetic particles, J. Magn. Magn. Mater., 125, , doi: / (93)90109-f. Dormann, J. L., D. Fiorani, and E. Tronc (1999), On the models for interparticle interactions in nanoparticle assemblies: Comparison with experimental results, J. Magn. Magn. Mater., 202, , doi: / S (98) Dunlop, D. J. (2002), Theory and application of the Day plot (M rs /M s versus H cr /H c ) 1. Theoretical curves and tests using titanomagnetite data, J. Geophys. Res., 107(B3), 2056, doi: /2001jb Dunlop, D. J., and Ö. Özdemir (1997), Rock Magnetism: Fundamentals and Frontiers, 573 pp., Cambridge Univ. Press, Cambridge, U. K. Egli, R. (2009), Magnetic susceptibility measurements as a function of temperature and frequency I: Inversion theory, Geophys. J. Int., 177, , doi: /j x x. Galindo Gonzalez, C., J. M. Feinberg, T. Kasama, L. C. Gontard, M. Pósfai, I. Kósa, J. D. G. Duran, J. E. Gil, R. J. Harrison, and R. E. Dumin Borkowski (2009), Magnetic and microscopic characterization of magnetite nanoparticles adhered to clay surfaces, Am. Mineral., 94, , doi: /am Gazeau, F., J. C. Bacri, F. Gendron, R. Perzynski, Y. L. Raikher, V. I. Stepanov, and E. Dubois (1998), Magnetic resonance of ferrite nanoparticles: Evidence of surface effects, J. Magn. Magn. Mater., 186, , doi: /s (98) Gilmore, K., Y. U. Dzerda, M. T. Klem, M. Allen, T. Douglas, and M. Young (2005), Surface contribution to the anisotropy energy of spherical magnetite particles, J. Appl. Phys., 97, doi: / Gittleman, J. I., B. Abeles, and S. Bozowski (1974), Superparamagnetism and relaxation effects in granular Ni SiO 2 and Ni Al 2 O 3 films, Phys. Rev. B, 9, , doi: /physrevb Guertin,R.P.,N.Harrison,Z.X.Zhou,S.McCall,andF.Drymiotis (2007), Very high field magnetization and AC susceptibility of native horse spleen ferritin, J. Magn. Magn. Mater., 308, , doi: /j.jmmm Hanesch, M., and N. Petersen (1999), Magnetic properties of a recent parabrown earth from Southern Germany, Earth Planet. Sci. Lett., 169, 85 97, doi: /s x(99)00076-x. Henkel, O. (1964), Remanenzverhalten und wechselwirkungen in hartmagnetischen teilchenkollektiven, Phys. Status Solidi, 7, , doi: /pssb Iglesias, O., and A. Labarta (2004), Role of surface disorder on the magnetic properties and hysteresis of nanoparticles, Physica B, 343, , doi: /j.physb Jiang, J. Z., and S. Mørup (1997), Correlation between peak and median blocking temperatures by magnetization measurement on isolated ferromagnetic and antiferromagnetic particle systems, Nanostruct. Mater., 9, , doi: /s (97) Kachkachi, H., W. T. Coffey, D. S. F. Crothers, A. Ezzir, E. C. Kennedy, M. Nogues, and E. Tronc (2000), Field dependence of the temperature at the peak of the zero field cooled magnetization, J. Phys. Condens. Matter, 12, , doi: / /12/13/316. Kim, D. K., Y. Zhang, W. Voit, K. V. Rao, and M. Muhammed (2001), Synthesis and characterization of surfactant coated superparamagnetic 9of10

10 monodispersed iron oxide nanoparticles, J. Magn. Magn. Mater., 225, 30 36, doi: /s (00) Kim, T., and M. Shima (2007), Reduced magnetization in magnetic oxide nanoparticles, J. Appl. Phys., 101, 09M516, doi: / Labaye, Y., O. Crisan, L. Berger, J. M. Greneche, and J. M. D. Coey (2002), Surface anisotropy in ferromagnetic nanoparticles, J. Appl. Phys., 91, , doi: / Li, H., M. T. Klem, K. B. Sebby, D. J. Singel, M. Young, T. Douglas, and Y. U. Idzerda (2009), Determination of anisotropy constants of protein encapsulated iron oxide nanoparticles by electron magnetic resonance, J. Magn. Magn. Mater., 321, , doi: /j.jmmm Liu, Q. S., J. Torrent, Y. J. Yu, and C. L. Deng (2004), Mechanism of the parasitic remanence of aluminous goethite [a (Fe, Al)OOH], J. Geophys. Res., 109, B12106, doi: /2004jb Maher, B. A. (1988), Magnetic properties of some synthetic sub micron magnetites, Geophys. J., 94, 83 96, doi: /j x tb03429.x. Maher, B. A. (1998), Magnetic properties of modern soils and Quaternary loessic paleosols: Paleoclimatic implications, Palaeogeogr. Palaeoclimatol. Palaeoecol., 137, 25 54, doi: /s (97)00103-x. Mazo Zuluaga, J., J. Restrepo, and J. Mejia Lopez (2008), Effect of surface anisotropy on the magnetic properties of magnetite nanoparticles: A Heisenberg Monte Carlo study, J. Appl. Phys., 103, , doi: / Meldrum, F. C., B. R. Heywood, and S. Mann (1992), Magnetoferritin: In vitro synthesis of a novel magnetic protein, Science, 257, , doi: /science Moskowitz, B. M., R. B. Frankel, S. A. Walton, D. P. E. Dickson, K. K. W. Wong, T. Douglas, and S. Mann (1997), Determination of the preexponential frequency factor for superparamagnetic maghemite particles in magnetoferritin, J. Geophys. Res., 102, 22,671 22,680, doi: / 97JB Muxworthy, A. R. (2001), Effect of grain interactions on the frequency dependence of magnetic susceptibility, Geophys. J. Int., 144, , doi: /j x x. Muxworthy, A. R., W. Williams, and D. Virdee (2003), Effect of magnetostatic interactions on the hysteresis parameters of single domain and pseudo single domain grains, J. Geophys. Res., 108(B11), 2517, doi: /2003jb Néel, L. (1949), Théorie du traînage magnétique des ferromagnétiques en grains fins avec application aux terrescuites, Ann. Geophys., 5, Nolan, S. R., J. Bloemendal, J. F. Boyle, R. T. Jones, F. Oldfield, and M. Whitney (1999), Mineral magnetic and geochemical records of late Glacial climatic change from two northwest European carbonate lakes, J. Paleolimnol., 22, , doi: /a: Nunes, A. C., and Z. C. Yu (1987), Fractionation of a water based ferrofluid, J. Magn. Magn. Mater., 65, , doi: / (87) Pedersen, M. S., S. Mørup, S. Sethi, M. Hanson, and C. Johansson (1994), Magnetic interaction between ultrafine amorphous Fe 1 x C x alloy particles in ferrofluids, Hyperfine Interact., 93, , doi: / BF Pérez, N., P. Guardia, A. G. Roca, M. P. Morales, C. J. Serna, O. Iglesias, F. Bartolomé, L. M. García, X. Batlle, and A. Labarta (2008), Surface anisotropy broadening of the energy barrier distribution in magnetic nanoparticles, Nanotechnology, 19, doi: / /19/47/ Santambrogio, P., A. Cozzi, S. Levi, E. Rovida, F. Magni, A. Albertini, and P. Arosio (2000), Functional and immunological analysis of recombinant mouse H and L ferritins from Escherichia coli, Protein Expr. Purif., 19, , doi: /prep Southern, P., A. P. Robinson, O. I. Kasyutich, B. Warne, A. Bewick, and W. Schwarzacher (2007), The effect of dipolar interactions on the thermal relaxation of magnetoferritin, J. Phys. Condens. Matter, 19, , doi: / /19/45/ Tauxe, L., T. A. T. Mullender, and T. Pick (1996), Potbellies, wasp waists, and superparamagnetism in magnetic hysteresis, J. Geophys. Res., 101, , doi: /95jb Tian,L.X.,B.Xiao,W.Lin,S.Y.Zhang,R.X.Zhu,andY.X.Pan (2007), Testing for the presence of magnetite in the upper beak skin of homing pigeons, Biometals, 20, , doi: /s x. Tronc, E., P. Prene, J. P. Jolivet, F. Dorazio, F. Lucari, D. Fiorani, M. Godinho, R. Cherkaoui, M. Nogues, and J. L. Dormann (1995), Magnetic behavior of g Fe 2 O 3 nanoparticles by Mössbauer spectroscopy and magnetic measurements, Hyperfine Interact., 95, , doi: /bf Tronc, E., A. Ezzir, R. Cherkaoui, C. Chaneac, M. Nogues, H. Kachkachi, D. Fiorani, A. M. Testa, J. M. Greneche, and J. P. Jolivet (2000), Surface related properties of g Fe 2 O 3 nanoparticles, J. Magn. Magn. Mater., 221, 63 79, doi: /s (00) Uchida, M., et al. (2006), Targeting of cancer cells with ferrimagnetic ferritin cage nanoparticles, J. Am. Chem. Soc., 128, 16,626 16,633, doi: /ja Uchida, M., D. A. Willits, K. Muller, A. F. Willis, L. Jackiw, M. Jutila, M. J. Young, A. E. Porter, and T. Douglas (2009), Intracellular distribution of macrophage targeting ferritin iron oxide nanocomposite, Adv. Mater., 21, , doi: /adma Winklhofer, M., K. Fabian, and F. Heider (1997), Magnetic blocking temperatures of magnetite calculated with a three dimensional micromagnetic model, J. Geophys. Res., 102, 22,695 22,709, doi: / 97JB Winklhofer, M., E. Holtkamp Rotzler, M. Hanzlik, G. Fleissner, and N. Petersen (2001), Clusters of superparamagnetic magnetite particles in the upper beak skin of homing pigeons: Evidence of a magnetoreceptor? Eur. J. Mineral., 13, , doi: / /2001/ Wohlfarth, E. P. (1955), The effect of particle interaction on the coercive force of ferromagnetic micropowders, Proc. R. Soc. London Ser. A, 232, , doi: /rspa Wohlfarth, E. P. (1958), Relations between different modes of acquisition of the remanent magnetization of ferromagnetic particles, J. Appl. Phys., 29, , doi: / Wong, K. K. W., T. Douglas, S. Gider, D. D. Awschalom, and S. Mann (1998), Biomimetic synthesis and characterization of magnetic proteins (magnetoferritin), Chem. Mater., 10, , doi: /cm970421o. Zhou, L. P., F. Oldfield, A. G. Wintle, S. G. Robinson, and J. T. Wang (1990), Partly pedogenic origin of magnetic variations in Chinese loess, Nature, 346, , doi: /346737a0. C. Cao, Q. Liu, Y. Pan, and L. Tian, Paleomagnetism and Geochronology Laboratory, Key Laboratory of the Earth s Deep Interior, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing , China. (yxpan@mail.iggcas.ac.cn) G. Chen and W. Liu, State Key Laboratory of Microbial Technology, School of Life Sciences, Shandong University, Jinan , China. 10 of 10