- Email: [email protected]
Au bimetallic nanoparticles
Journal of Alloys and Compounds 649 (2015) 104e111
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
The study of magnetic properties and relaxation processes in Co/Au bimetallic nanoparticles k a, Adriana Zelen a kova a, *, Vladimir Zelen a k b, Jozef Kova Pavol Hrubov ca cc a rik University, Park Angelinum 9, Kosice, Slovakia Department of Condensed Matter Physics, P.J. Saf rik University, Moyzesova 11, Kosice, Slovakia Department of Inorganic Chemistry, P.J. Safa c Institute of Experimental Physics, SAS, Watsonova 41, Kosice, Slovakia a
b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 June 2015 Accepted 5 July 2015 Available online 15 July 2015
Co/Au bimetallic fine nanoparticles were prepared employing the method of microemulsion using reverse micelle as nanoreactor, controlling the particles size. Magnetic and structural properties of two different samples Co/Au1 and Co/Au2 with almost comparable size of Co core and different size of Au layer were studied. The investigation of magnetic relaxation processes present in the particles was carried out by means of ac and dc magnetization data obtained at different temperatures and magnitudes of magnetic field. We observed the existence of superspin glass state characterized by the strong interparticle interactions in the nanoparticle systems. In this paper, we discuss the attributes of novel superspin glass magnetic state reflected on various features (saturated FC magnetization at low temperatures, shift of the ColeeCole arc downwards) and calculated parameters (relaxation time, critical exponent zv ~ 10 and frequency dependent criterion p < 0.05). Comparison of the magnetic properties of two studied samples show that the thickness of diamagnetic Au shell significantly influences the magnetic interactions and change the relaxation dynamics. © 2015 Elsevier B.V. All rights reserved.
Keywords: Magnetic nanoparticles Superparamagnetism Relaxation process Magnetic susceptibility
1. Introduction Magnetic nanoparticles have been investigated from different aspects for more than three decades. Its unique properties are suitable for numerous applications. Their introduction in medicine (cancer diagnosing [1], cancer treating [2], drug delivery [3]) or environmental science [4] posed new trends and challenges in these fields for the incoming 21st century. Therefore, the understanding of processes and phenomena present in the magnetic nanoparticle systems is crucial. Unique properties of monodomain magnetic nanoparticles stem in existence of giant magnetic moment, superspin (several orders larger than the magnetic moment of single atom), in individual nanoparticle. The superspin can freely fluctuate along the magnetocrystalline axis [5]. If the relaxation time of superspin reversal is short, superparamagnetic state is recognized [5e9]. However, this behavior can be suppressed by the influence of temperature, external magnetic field as well as by the effect of magnetic interparticle interactions. Proper combination of these conditions slows the magnetic moment relaxation and may cause superspin blocking, superspin freezing or even superspin long range ordering * Corresponding author. a kov E-mail address: [email protected] (A. Zelen a). http://dx.doi.org/10.1016/j.jallcom.2015.07.044 0925-8388/© 2015 Elsevier B.V. All rights reserved.
in magnetic nanoparticles [5,6]. Several works concerning the influence of the strength of inter-particle magnetic interactions on magnetization relaxation have been reported [7e11]. These studies show that with increasing concentration of the particles in the assembly the dipoleedipole magnetic interactions between superspins became predominant and collective behavior appears. It is considered to be the origin of superspin glass state observed and discussed in the last years [9,10]. The concept of superspin glass state was introduced for the magnetic state of the nanoparticle assembly which shows magnetic characteristics similar to the atomic spin glasses. However, the origin of the interactions controlling the magnetic ordering is completely different in both cases. De Toro et al. [11,12] observed the ideal superspin glass behavior in the ensemble of uniform 8 nm maghemite nanoparticles after they had been pressed into the disc. On the other hand, Tadic et al. [13] presented non-compressed maghemite nanoparticles (4 nm) which exhibited superparamagnetic properties. These results are in accordance with the finding of Jimenez-Villacorta et al. [8] who were investigating the concentration effect on magnetic characteristics of the fine Fe nanoparticles embedded in non-magnetic matrix. They concluded the higher concentration of the particles affect the strength of inter-particle magnetic interaction and leads to the change of the superparamagnetic to superspin or even superferromagnetic state. Ebbing et al. [14] reported the influence
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
105
of platinum capping on fine cobalt nanoparticles and documented the enhanced magnetic anisotropy and superspin glass behavior for the particles with thicker Pt layer. Song et al. [15] observed slight decrease of blocking temperatures with increasing coating layer in core@shell Co@Au nanoparticle system. They assumed the effect was induced by enhancing the inter-particle distances what led to the decay of the interaction strength. Corresponding with the theory of superparamagnetism, Su et al. [16] found the decrease of the blocking temperature with diminishing diameter of the Co nanoparticles. Also the values of coercivity of the Co nanoparticles measured below the blocking temperature turned out to be proportional to the volume of the particles. Obviously, superspin glass state can be induced by various factors. Concerning the above mentioned facts we attempted to produce systems of cobalt/gold bimetallic nanoparticles with the aim to encounter the superspin glass state and to study the processes leading to its formation. In the presented paper we have studied magnetic state and relaxation process in cobalt nanoparticles coated by protective gold layer, prepared in two different samples with different size of gold shell. We have shown that the nanoparticles collectively behave like a superspin glass and exhibit different magnetic responses when the size of diamagnetic shell differs. The existence of superspin glass state in bimetallic Co/Au nanoparticles has not yet been published.
Cetyltrimethylammonium bromide (CTAB, 99%), sodium borohydride (NaBH4; 98%), were purchased from Aldrich and Sigma. Nanoparticles were prepared using microemulsion method in reverse micelles [17]. All reverse micelle solutions were prepared using CTAB as the surfactant with octane as the oil phase. 1butanol was used as co-surfactant, increasing the polarity of the surfactant and helping to stabilize the micelle solutions [17]. Aqueous reactants of CoCl2, NaBH4, and HAuCl4 were used to form the reverse micelle. The metal particles were formed inside the reverse micelle by reduction of a metal salt using sodium borohydride. At first the pure Co core were produced using a magnetic stirrer and under flowing argon. Co based nanoparticles were subsequently coated by Au shell also by using of reverse micelle solutions. After nanoparticle preparation, the micelles in the reaction mixture were disrupted using aceton causing nanoparticles to precipitate. Repeated washing using a 1:1 mixture of chloroform/methanol removed the surfactant. The size of prepared nanoparticles was controlled by the water to surfactant molar ratio w ([H2O]/[CTAB]). Two different samples of Co/Au1 and Co/Au2 were prepared, where the same molar ratio for preparation of Co core (wCo ¼ 8) and different ratio for Au shell (wAu1 ¼ 5 and wAu2 ¼ 10) were used.
2. Experimental
The high-energy powder X-ray diffraction (HE-PXRD) experiments were carried out at BW5 wiggler beamline of DORIS positron storage ring in DESY (Hamburg, Germany) using monochromatic synchrotron radiation with beam energy of 100 keV (l ¼ 0:12398 Å). The experiments were carried out at room temperature in the transmission mode. LaB6 standard was used to
2.1. Preparation All chemicals, namely: Chloroauric acid (HAuCl4, 99.9%), Cobalt chloride (CoCl2, 97%) octane, 1-butanol (99.8%),
2.2. Characterization
Fig. 1. High resolution transmission electron images of samples Co/Au1 (a) and Co/Au2 (c) samples and its size distributions (b), (d), respectively.
106
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
calibrate the sample-to-detector distance. The background intensity was subtracted directly from XRD patterns and the result was integrated using the software package FIT2D. The HRTEM micrographs were taken with a JEOL 2100F microscope. Copper grid coated with a holey carbon support lm was used to prepare samples for the TEM observation. The bright-field TEM image was obtained at 200 kV. The magnetic measurements were performed on a commercial SQUID-based magnetometer (Quantum Design MPMS 5XL) over a wide range of temperatures (2e300 K) and applied magnetic fields (up to 5 T). Zero field cooling (ZFC) and field-cooling (FC) measurements M(T) were carried out in dc mode. The samples were encapsulated into a plastic sample holder. The diamagnetic contribution of plastic capsule and plastic sample holder were measured and extracted and experimental data were corrected. The same instrument was used for the experimental study of dynamical properties by ac magnetic susceptibility. The complex ac magnetic susceptibility c0 (f)- i c00 (f), where c0 represents in-phase ac susceptibility (real) component and c00 out-of-phase (imaginary) component, was recorded in the temperature interval 2e300 K and in the frequency interval of 0.1e1000 Hz. 3. Results and discussion 3.1. Structural characterization Nearly spherical shape and crystallinity of individual nanoparticles were observed by TEM in both samples, Fig. 1. Sample Co/ Au1 is more homogenous and the particles are smaller in comparison with the Co/Au2 sample. Size distributions obtained from the TEM images with high number of particles (min. 100 particles) are displayed on Fig. 1 (b), (d) and yield the average particle diameter dTEM ~ 7 nm and dTEM ~ 11 nm for the Co/Au1 and Co/Au2, respectively. HRTEM micrographs of gold-coated cobalt nanoparticles show, that the gold shell can be identified in the majority of the particles and the cobalt core appears as dark black area in the center of the nanoparticles. The images of the nanoparticles appear different even at the same defocus because of unhomogeneous shapes and different orientations of the nanoparticles. The phase analysis of the composition carried out by X-ray diffraction is displayed on Fig. 2. The broadness of diffraction peaks refers to nanocrystalline character of the samples, but on the other hand, hampers the fine phase analysis. The FCC metallic gold structure is clearly visible on both diffraction patterns, but the cobalt signal is difficult to recognize. Numerous authors dealing with Co@Au core@shell nanoparticles reported the lack of cobalt signal in the XRD pattern [18e20]. Some of them presume this is the consequence of overlapping FCC gold and 3-cobalt diffraction peaks [19]. In our samples, we suppose the existence of magnetic Co core and its diffraction pattern is hidden under the pattern of gold due to the overlapping of their diffraction peaks. The dominant peak in diffraction pattern (111) was used for calculating the size of the nanoparticles employing the Scherrer formula [21]. The approximate values of dXRD ~ 10 nm and dXRD ~ 13 nm were obtained for the Co/Au1 and Co/Au2, respectively. In the sample Co/Au1, the cobalt oxide diffraction peaks can be also identified as a consequence of surface oxidation of the cobalt cores. Since the thickness of Au shell in Co/Au1 is smaller than in Co/Au2, there is a possibility of the surface oxidation of cobalt core in Co/Au1. 3.2. Ac magnetic susceptibility We have performed the ac susceptibility measurements as function of temperature (2e300 K) and frequencies (0.1e1000 Hz) of external alternating low magnetic field to study the dynamic
Fig. 2. The XRD pattern of Co/Au1 (a) and Co/Au2 (b).
behavior and relaxation processes in prepared samples. Both inphase (real) and out-of-phase (imaginary) components of ac susceptibility were recorded. As it is evident from Fig. 3 (a), (b), one sharp and narrow maximum corresponding to each attempt frequency was recognized for the Co/Au1 at the temperature around Tmax ~ 7 K. Above this temperature all the susceptibility vs. temperature curves are linear and they merge. Sample Co/Au2 displayed different ac susceptibility dependence, see Fig. 3 (c), (d). The sharp and narrow maximum at low temperature of about Tmax1 ~ 5 K was recognized, similar to the maximum of sample Co/Au1. Moreover the second very broad maximum at temperature interval Tmax2 ~ 50e90 K was observed. The distinct behavior of two maxima in the Co/Au2 predicts on different magnetization processes. The common features of ac susceptibility maxima in both samples is so called frequency shift. Increasing the frequency the decrease of maximum value of inphase susceptibility and its slight shift towards higher temperature was observed. Out-of-phase susceptibility maximum also shifts towards higher temperatures but its value rises with frequency enhancement, see Fig. 3. According to numerous authors [5,6,13] such behavior of the ac susceptibility components is typical of systems in which blocking or freezing of magnetic moments occurs. Ac susceptibility measurements are useful tool for quantitative analysis of relaxation processes. Several theoretical and empirical models for ac susceptibility data evaluation were introduced and their results provide the information on character of magnetic
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
107
relaxation as well as on the strength of inter-particle interactions [5,6,12,13,22]. The assembly of magnetic nanoparticles can exhibit different behavior depending on the strength of possible interactions between particles. If mutual interactions are negligible (e.g. superparamagnetic state), the relaxation time t of single parl-Arrhenius law [23]. ticle moment follows the Nee EA
t ¼ t0 ekB T ;
(1)
where t0 denotes the pre-relaxation constant, EA ¼ KV the activation energy, K effective magnetocrystaline anisotropy constant, V the volume of the magnetic particle, kB Boltzman constant and T temperature. With increasing interaction strength the relaxation l-Arrhenius law, Eq. (1) and some time does not follow the Nee authors [12,13,24] employ the Vogel-Fulcher law for data evaluation. According to this model the relaxation time is determined by the equation [22]. E*
t ¼ t0 ekB ðTT0 Þ ;
(2)
where E* is the energy barrier (activation energy) modified by an effective contribution of the inter-particle interactions and T0 is the parameter corresponding to the strength of the interactions. For even stronger interactions the collective behavior of superspins predominate and so called superspin glass state below the critical temperature for superspin ordering TSSG can be observed [25]. In this case the relaxation time is given by equation [22].
t ¼ t*0
T TSSG TSSG
zv
;
(3)
with the critical exponent zv, where v is the critical exponent of the correlation length
xf
T TSSG v TSSG
(4)
and the exponent z relates the relaxation time with the correlation length via tfxz . The Eq. (3) is the modification of the equation derived from the dynamic scale hypothesis [26]. The denomination superspin glass comes from the similarity of the system behavior to the behavior of canonical spin glasses [27,28]. On the other hand there is fundamental difference in the nature of the ordering interactions. While the collective behavior of spins in atomic spin glass is induced mainly by exchange interaction and RKKY (RudermaneKitteleKasuyaeYosida) interaction the superspin glass state existence is caused predominately by the magnetic dipoleedipole interaction [23]. We have analyzed our experimental data in terms of above mentioned theory and the results we obtained are listed in Table 1. The interpretation of various parameters is introduced in following discussion. el-Arrhenius The fit of the data according to the Eq. (1) and Ne model [23] designed for mutually non-interacting particles provided physically non-reasonable small value of pre-relaxation constant t0 ¼ 3.77 1049 s for Co/Au1 maximum. According to several authors [5,6,22,29] typical values of this parameter are in the interval 1012 e 109 s. Due to this, we expected stronger inter-particle interactions in the systems and we employed Vogel-Fulcher model for the data analysis. The value t0 ¼ 1.17 1019 s established from
Fig. 3. Ac susceptibility of studied samples: (a), (c) in-phase and (b), (d) out-of-phase magnetic susceptibility dependence on temperature recorded at different frequencies of alternating magnetic field.
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
108
Table 1 Magnetic characteristics calculated from ac susceptibility data. Sample
Co/Au1 Co/Au2- 1. RP Co/Au2- 2. RP a b c
N-A theorya
VeF theoryb
SSG theoryc
p
t0 (s)
EA/kB (K)
t0 (s)
E*/kB (K)
T0 (K)
t0 (s)
TSSG (K)
zy
3.77 1049 4.47 1023 1.39 1012
794 264 1871
1.17 1019 1.03 1014 2.08 1012
115 97 1821
4.47 2.12 e
3.25 1012 9.68 107 e
6.57 4.04 e
10.7 10.2 e
0.022 0.046 0.092
el-Arrhenius theory using Eq. (1). Parameters were calculated in accordance with Ne Parameters were calculated in accordance with Vogel-Fulcher theory using Eq. (2). Parameters were calculated in accordance with dynamic scaling hypothesis using Eq. (3).
the fit of the data to the dependence ln(t) on Tmax derived from Eq. (2) matched better the expecting value, but still was out of acceptable interval. The best results t0 ¼ 3.25 1012 s and TSSG ¼ 6.57 K provided the data fit according to the dynamic scale hypothesis, Eqs. (3) and (4). Moreover, the dynamic critical exponent zn ¼ 10.7 is in good accordance with the values reported for superspins' freezing process [5]. All these facts are indicating the presence of very strong magnetic dipoleedipole interactions between nanoparticles leading to freezing of the superspins into the superspin glass state. Co/Au2 susceptibility data suggest on the presence of two distinct magnetization processes, therefore we evaluated both of the maxima separately. The character of the first maximum at low temperature Tmax1 ~ 5 K recommends the maximum recognized in Co/Au1. Similarly, we established physically non-reasonable values of preel-Arrhenius, but relaxation constant employing the model of Ne the Vogel-Fulcher model released t0 ¼ 1.03 1014 s. Dynamic scale hypothesis applied on the data established the values zn ¼ 10.2 and TSSG ¼ 4.04 K what is in good accordance with expectations. Only the pre-relaxation constant t0 ¼ 9.68 107 s was out of usually reported range. Assuming gained results, we infer the freezing of superspins is dominant process in the temperature region close to 5 K, however, the rate of collective behavior is lower in comparison with the Co/Au1. The second process (high temperature) in Co/Au2 at Tmax2 ~ 50e90 K is completely different. The pre-relaxation conel-Arrhenius model is stant t0 ¼ 1.39 1012 s obtained utilizing Ne on the edge of the mentioned acceptable interval and refers to blocking of superspins. Since the Vogel-Fulcher law, Eq. (2), provided negligible value of T0 parameter and the dynamic scale hypothesis, Eqs. (3) and (4), absolutely failed, we can conclude that the superspins blocking influenced by weak interactions dominates. Several authors [29] utilize frequency dependent criterion p to distinguish between superspin blocking and freezing process.
p¼
DTmax ; Tmax Dlog n
(5)
where Tmax denotes the average value of c0 (T) maximum temperatures in the range of experimental frequencies, while DTmax denotes the difference between the maximum and minimum value of Tmax. The calculated values p ¼ 0.022 and p ¼ 0.046 for the low temperature maxima in both samples Co/Au1, Co/Au2, respectively, were from the interval 0.005e0.05. Various superspin glass systems with p parameter from this interval were reported [12], thus the observed properties can be regarded as collective freezing of superspins. On the other hand, the value calculated for the second maximum in Co/ Au2 p ¼ 0.092 approximated the interval typical of pure superspin blocking process 0.1e0.13 [29]. Concerning this fact and previous el-Arrhenius and the Vogel-Fulcher models we results from the Ne assume the blocking process of superspins is affected by weak dipoleedipole inter-particle interactions in this case. Thus, the frequency dependent criterion independently confirmed that the particles' dynamics critically slows down at TSSG ¼ 6.57 K in Co/Au1. In opposite, the distribution of energy
barriers in the sample Co/Au2 is found to have two maxima corresponding to freezing and blocking process at TSSG ¼ 4.04 K and Tp ¼ 57.14 K, respectively. Additionally, we used ColeeCole plots (Argand diagrams) [5,8] (the dependence of out-of-phase susceptibility on in-phase susceptibility components) at temperatures close below the temperature of susceptibility maximum for the determination of magnetic interaction strength in the investigated systems. The shape of the plots reflects the distribution of relaxation times of superspins [5]. If the single relaxation time is characteristic for the process, the superparamagnetic relaxation is confirmed and the shape of ColeeCole plot is perfect semicircle [5]. One of the possible indications of collective behavior presence in the system is the wide range of relaxation times [8]. This reflects the shift of the ColeeCole plot arc downwards. Fig. 4 shows the ColeeCole plots constructed at the temperature of susceptibilities maxima in Co/Au1 Fig. 4(a) and Co/Au2 Figs. (b), (c). Since the ColeeCole plots constructed for low temperature region (a), (b) of both samples exhibited significant translation under the horizontal axis, we infer the presence of stronger magnetic interactions. Moreover, the arc for the Co/Au1 (a) exhibited higher flatness than the arc for low temperature relaxation process at Tmax1 ~ 5 K in Co/Au2, thus the interactions in Co/Au1 should be stronger. This result is in line with the conclusions we discussed previously and confirms the freezing of superspins. On the other hand, the ColeeCole arcs are tending to have a shape of semicircle while approaching the temperature to the Tmax2 ~ 50e90 K what is evident from Fig. 4 (c). This suggests the second maximum in the distribution of energy bariers in sample Co/Au2, which is connected to the blocking process with low influence of magnetic interactions. 3.3. Dc magnetization Fig. 5 presents the dependence of magnetization of the samples on temperature recorded in both ZFC (zero-field-cooling) and in 100 Oe and 500 Oe FC (field-cooling) protocols. According to Tadic [29] the maximum of ZFC curve signifies the presence of relaxation process and the peak temperature Tp represents the characteristic temperature at which the superspins of the particles' majority become blocked or frozen. We determined the Tp ~ 6 K for the Co/ Au1 and Tp1 ~ 4 K, Tp2 ~ 57 K (see Table 2) for the first and the second maximum in Co/Au2, respectively. The character of FC magnetization vs. temperature curve below the Tp is the reliable feature commonly used for distinguishing between superparamagnetic and superspin glass state [30]. The FC magnetization monotonically increases with decreasing T for superparamagnet, while it tends to saturate to a constant value or even tends to decrease with temperature decay for superspin glass. The FC magnetization shows the tendency to saturate at very low temperatures what suggests the superspin glass state presence at this temperature region in the sample Co/Au1. Such effect cannot be documented responsibly in the Co/Au2 sample because the transition temperature is too low. According to [31], the temperature of irreversibility Tirr
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
109
Fig. 5. ZFC and FC magnetization of the samples Co/Au1 (a) and Co/Au2 (b). Inset the determination of blocking temperatures for the system Co/Au1 (a) and Co/Au2 (b).
Fig. 4. ColeeCole plots for relaxation process in Co/Au1 (a) and for the first (b) and the second (c) relaxation process in Co/Au2.
(temperature, at which the ZFC and FC curves begin to separate) can be associated with the blocking temperature of the largest nanoparticles in the assembly. The difference between Tirr and Tp is related to the width of activation energy distribution, thus, to the width of the size distribution of the particles [31]. The Tirr ~ 10 K close to Tp ~ 7 K was found for Co/Au1 confirming quite narrow size distribution. The existence of two ZFC M(T) maxima in sample Co/ Au2 we attributed to the existence of two maxima in energy barrier distribution corresponding to freezing and blocking processes at different temperature. One of the most relevant parameter typical of superparamagnetic system is blocking temperature TB. For T < TB the thermal fluctuations are insufficient for the superspin reversal over the activation energy barrier and the superspins will appear blocked or frozen, while for T > TB they will appear free comparable to a paramagnetic system. The blocking temperature is not an intrinsic temperature of the system, but depends strongly on the probing time scale of a measurement. It is defined as the temperature at which the relaxation time of the superspin equals the probing time scale of a measurement. This is the reason why some authors [22] establish its value from the relaxation time t vs. 1/T plot. The typical time scale of a standard SQUID magnetometry experiment is 10 s. It represents the dashed line in Fig. 5 (a) inset and intersects the curve t vs. 1/T at TB* ~ 7 K thus defining the blocking temperature of the Co/Au1 system. Similarly, we determined the TB1* ~ 5 K, TB2* ~ 60 K for the first and the second process in Co/Au2, Fig. 5 (b) inset.
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
110
Table 2 Structural and magnetic characteristics of Co/Au1 and Co/Au2 samples. Sample
Tp (K)
TB* (K)
mp (mB)
MS (2 K) (emu/g)
HC (2 K) (Oe)
dTEM (nm)
dXRD (nm)
dMAG (nm)
Co/Au1 Co/Au2
6.71 4.37
7.2 4.77
86.33 144.94
40.62 16.93
1183 543
7.03 11.22
10.12 13.29
5.36 6.90
57.14
62.46
Fig. 7. Magnetization M of Co/Au1 (a) and Co/Au2 (b) expressed as a function of H/T. Fig. 6. Magnetization on magnetic field dependence measured at various temperatures for Co/Au1 (a) and Co/Au2 (b). Inset the Langevin fit of magnetization vs. field at 280 K.
Typical superparamagnetic behavior of the particles can be seen from field-cooled M(T) dependences, Fig. 5, as well as from M(H) dependences at the temperatures above Tp, Fig. 6. At the temperatures of the nanoparticle system lower than Tp the magnetic hysteresis was recorded, Table 2. The higher coercivity value at 2 K in Co/Au1 contrary to Co/Au2 corresponds to the stronger interactions confirmed by previous ac analysis in the first sample. Once the system is in superparamagnetic state at higher temperatures, the data M(H) should obey the modified Langevin law which takes in account the non-negligible size distribution of magnetic moments. Than the magnetization of superparamagnetic nanoparticles is described as the weighted sum of Langevin functions [21].
M¼
Z∞ mp H f mp dmp ; L kB T
(6)
0
where f (mp) is the distribution of magnetic moments, L Langevin
function and mp magnetic moment of the particle. Assuming nanoparticles in superparamagnetic state and applying fit of the Eq. (6) to the experimental data of M(H) obtained at 280 K the average magnetic moment value mp ~ 86 mB and mean diameter of the particle dMAG ~ 5 nm were established for the Co/ Au1 and mp ~ 145 mB, dMAG ~ 7 nm for the Co/Au2. Since the calculations of the particle's diameter utilizing Langevin fit consider only cobalt core of the particle, not diamagnetic gold shell, the size of the particle determined by Langevin fit was lower than the size obtained from HRTEM images and Scherrer formula. Since the results of structural measurements introduced some doubts about the existence of partial oxidation in Co core, especially in the sample Co/Au1, we attempted to verify the presence of additional cobalt oxide interlayer performing the exchange bias effect measurements. Magnetization vs. field (isothermal magnetization curves) at 2 K (below blocking temperature) before and after cooling in external dc field of 30 000 Oe were recorded. No shift of hysteresis loop along the x axis after cooling in external dc field (FC ¼ 30 000 Oe) was observed what is the evidence of the exchange bias absence in the sample. This confirms that the exchange interaction between ferromagnetic/antiferromagnetic
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
phases is not present. In order to confirm superparamagnetic state presence in the samples we employed the method described in Ref. [31]. If the samples were truly superparamagnetic the isothermal magnetization curves measured at different temperatures above the blocking temperature should overlap when plotting M vs H/T [31]. This feature is indeed achieved as can be seen in Fig. 7, thus confirming the superparamagnetic behavior of our sample for temperatures above 60 K in both studied samples. 4. Conclusions In summary, we have investigated the static and dynamic magnetic behavior of Co/Au nanoparticles, prepared employing the microemulsion method. The magnetoestructural correlations show that the samples are composed predominantly from magnetic Co core with the size around 6e7 nm coated by diamagnetic Au shell of the thickness around 3 nm and 6 nm for two studied samples Co/Au1 and Co/Au2, respectively. Magnetic measurements confirm the evidence of magnetic relaxation processes and existence of novel superspin glass state in both studied systems. In fine Co/Au1 and CoAu/2 nanoparticles the very strong dipolar magnetic interactions between particles are present. These interactions lead to superspin freezing and individual particles exhibit the collective behavior below the glass transition temperature TSSG. In the sample Co/Au2 was observed that, the energy barrier distribution has two maxima corresponding to freezing and blocking process at different temperatures of TSSG ¼ 4.04 K and Tp ¼ 57.14 K, respectively. We demonstrated that the size of the diamagnetic Au shell capping the magnetic Co core can crucially influence the magnetic properties of the system. The thickness of the diamagnetic Au shell affects the center-to-center distance between magnetic Co nanoparticles what has significant influence on the strength of magnetic interaction. Our observation can open up the possibility of tailoring the SSG (superspin glass) state and its onset temperature by suitable selection of the particle size and can help in the application of very fine Co/Au bimetallic nanoparticles. Acknowledgment This work was supported by the Slovak Research and Development Agency under the contracts APVV-0132-11 and APVV-007314 and by the VEGA projects of Ministry of Education of the Slovak Republic (No. 1/0583/11, No. 1/0861/12) and by the ERDF EU grant under No. ITMS 26220120035. The authors (A.Z and V.Z) would like to thank DESY/HASYLAB project under No. I-20110282 EC and also
111
Dr. N. Murafa (AS CR, Rez, Czech Republic) for HRTEM measurements.
References [1] M. Yigit, A. Moore, Z. Medarova, Pharm. Res. 29 (2012) 1180. [2] T. Sadhukha, T.S. Wiedmann, J. Panyam, Biomaterials 34 (2013) 5163. [3] V.V. Mody, A. Cox, S. Shah, A. Singh, W. Bevins, H. Parihar, Appl. Nanosci. 4 (2014) 385. [4] J.H. Jung, J.H. Leea, S. Shinkai, Chem. Soc. Rev. 40 (2011) 4464. [5] S. Bedanta, W. Kleemann, J. Phys. D Appl. Phys. 42 (2009) 28. [6] O. Petracic, X. Chen, S. Bedanta, W. Kleemann, S. Sahoo, S. Cardoso, P.P. Freitas, J. Magnet. Magnetic Mater. 300 (2006) 192. [7] K. Nadeem, H. Krenn, T. Traussnig, R. Wrschum, D.V. Szabo, I. Letofsky-Papst, J. Magnet. Magnetic Mater. 323 (2011) 1998. [8] F. Jimenez-Villacorta, J. Sanchez-Marcos, E. Cespedes, M. Garcia-Hernandez, C. Prieto, Phys. Rev. B 82 (2010) 134413. [9] S. Thomas, M. Uhlig, U. Wiedwald, L. Han, P. Ziemann, M. Albrecht, Nanotechnology 24 (2013) 155703. [10] M. Perovic, V. Kusigerski, V. Spasojevic, A. Mrakovic, J. Blanusa, M. Zentkova, M. Mihalik, J. Phys. D Appl. Phys. 46 (2013) 165001. [11] J.A. De Toro, S.S. Lee, R. Mathieu, P.S. Normile, D. Salazar, J.L. Cheong, P. Muniz, J.M. Riveiro, M. Hillenkamp, A. Tamion, F. Tournus, P. Nordblad, J. Phys. Conf. Ser. 521 (2014) 012011. [12] J.A. De Toro, S.S. Lee, D. Salazar, J.L. Cheong, P. Muniz, J.M. Riveiro, M. Hillenkamp, A. Tamion, F. Tournus, P. Nordblad, Appl. Phys. Lett. 102 (2013) 183104. [13] M. Tadic, V. Kusigerski, D. Markoviic, M. Panjan, L. Miloseviic, V. Spasojeviic, J. Alloys Compd. 525 (2012) 28. [14] A. Ebbing, O. Hellwig, L. Agudo, G. Eggeler, O. Petracic, Phys. Rev. B 84 (2011) 012405. [15] Y. Song, J. Ding, Y. Wang, J. Phys. Chem. C 116 (2012) 11343. [16] Y.K. Su, C.M. Shen, T.Z. Yang, H.T. Yang, H.J. Gao, H.J. Li, Appl. Phys. A 81 (2005) 569. [17] J. Lin, W. Zhou, A. Kumbhar, J. Wiemann, J. Fang, E.E. Carpenter, C.J. O'Connor, J. Solid State Chem. 159 (2001) 26. [18] Y. Bao, H. Calderon, K.M. Krishnan, J. Phys. Chem. C 111 (2007) 1941. [19] Y. Bao, K.M. Krishnan, J. Magnet. Magnetic Mater. 293 (2005) 15. [20] G. Cheng, A.R. Hight Walker, J. Magnet. Magnetic Mater. 311 (2005) 31. [21] A. Zelenakova, V. Zelenak, J. Bednarcik, P. Hrubovcak, J. Kovac, J. Alloys Compd. 582 (2014) 483. [22] O. Petracic, Superlattices Microstruct. 47 (2010) 569. [23] S. Morup, M.F. Hansen, C. Frandsen, Beilstein J. Nanotechnol. 1 (2010) 182. [24] M. Tadic, S. Kralj, M. Jagodic, D. Hanzel, D. Makovec, Appl. Surf. Sci. 322 (2014) 255. [25] M. Tadic, V. Kusigerski, D. Markoviic, M. Panjan, L. Miloseviic, V. Spasojeviic, J. Alloys Compd. 509 (2011) 7134. [26] H. Sompolinsky, A. Zippelius, Phys. Rev. Lett. 47 (1981) 359. [27] T. Nishioka, Y. Tabata, T. Taniguchi, Y. Miyako, J. Phys. Soc. Jpn. 69 (2000) 1012. [28] J. J. Binney, N. J. Dowrick, A. J. Fisher, and M. E. J. Newman, Oxford University Press (1992). [29] M. Tadic, D. Markoviic, V. Kusigerski, L. Spasojeviic, L. Miloseviic, J. Magnet. Magnetic Mater. 321 (2009) 12. [30] M. Suzuki, S.I. Fullem, I.S. Suzuki, L. Wang, C.-J. Zhong, Phys. Rev. B 79 (024418) (2009). [31] M. Tadic, D. Markoviic, V. Spasojeviic, V. Kusigerski, M. Remskar, J. Pirnat, Z. Jaglicic, J. Alloys Compd. 441 (2007) 291.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
The study of magnetic properties and relaxation processes in Co/Au bimetallic nanoparticles k a, Adriana Zelen a kova a, *, Vladimir Zelen a k b, Jozef Kova Pavol Hrubov ca cc a rik University, Park Angelinum 9, Kosice, Slovakia Department of Condensed Matter Physics, P.J. Saf rik University, Moyzesova 11, Kosice, Slovakia Department of Inorganic Chemistry, P.J. Safa c Institute of Experimental Physics, SAS, Watsonova 41, Kosice, Slovakia a
b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 June 2015 Accepted 5 July 2015 Available online 15 July 2015
Co/Au bimetallic fine nanoparticles were prepared employing the method of microemulsion using reverse micelle as nanoreactor, controlling the particles size. Magnetic and structural properties of two different samples Co/Au1 and Co/Au2 with almost comparable size of Co core and different size of Au layer were studied. The investigation of magnetic relaxation processes present in the particles was carried out by means of ac and dc magnetization data obtained at different temperatures and magnitudes of magnetic field. We observed the existence of superspin glass state characterized by the strong interparticle interactions in the nanoparticle systems. In this paper, we discuss the attributes of novel superspin glass magnetic state reflected on various features (saturated FC magnetization at low temperatures, shift of the ColeeCole arc downwards) and calculated parameters (relaxation time, critical exponent zv ~ 10 and frequency dependent criterion p < 0.05). Comparison of the magnetic properties of two studied samples show that the thickness of diamagnetic Au shell significantly influences the magnetic interactions and change the relaxation dynamics. © 2015 Elsevier B.V. All rights reserved.
Keywords: Magnetic nanoparticles Superparamagnetism Relaxation process Magnetic susceptibility
1. Introduction Magnetic nanoparticles have been investigated from different aspects for more than three decades. Its unique properties are suitable for numerous applications. Their introduction in medicine (cancer diagnosing [1], cancer treating [2], drug delivery [3]) or environmental science [4] posed new trends and challenges in these fields for the incoming 21st century. Therefore, the understanding of processes and phenomena present in the magnetic nanoparticle systems is crucial. Unique properties of monodomain magnetic nanoparticles stem in existence of giant magnetic moment, superspin (several orders larger than the magnetic moment of single atom), in individual nanoparticle. The superspin can freely fluctuate along the magnetocrystalline axis [5]. If the relaxation time of superspin reversal is short, superparamagnetic state is recognized [5e9]. However, this behavior can be suppressed by the influence of temperature, external magnetic field as well as by the effect of magnetic interparticle interactions. Proper combination of these conditions slows the magnetic moment relaxation and may cause superspin blocking, superspin freezing or even superspin long range ordering * Corresponding author. a kov E-mail address: [email protected] (A. Zelen a). http://dx.doi.org/10.1016/j.jallcom.2015.07.044 0925-8388/© 2015 Elsevier B.V. All rights reserved.
in magnetic nanoparticles [5,6]. Several works concerning the influence of the strength of inter-particle magnetic interactions on magnetization relaxation have been reported [7e11]. These studies show that with increasing concentration of the particles in the assembly the dipoleedipole magnetic interactions between superspins became predominant and collective behavior appears. It is considered to be the origin of superspin glass state observed and discussed in the last years [9,10]. The concept of superspin glass state was introduced for the magnetic state of the nanoparticle assembly which shows magnetic characteristics similar to the atomic spin glasses. However, the origin of the interactions controlling the magnetic ordering is completely different in both cases. De Toro et al. [11,12] observed the ideal superspin glass behavior in the ensemble of uniform 8 nm maghemite nanoparticles after they had been pressed into the disc. On the other hand, Tadic et al. [13] presented non-compressed maghemite nanoparticles (4 nm) which exhibited superparamagnetic properties. These results are in accordance with the finding of Jimenez-Villacorta et al. [8] who were investigating the concentration effect on magnetic characteristics of the fine Fe nanoparticles embedded in non-magnetic matrix. They concluded the higher concentration of the particles affect the strength of inter-particle magnetic interaction and leads to the change of the superparamagnetic to superspin or even superferromagnetic state. Ebbing et al. [14] reported the influence
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
105
of platinum capping on fine cobalt nanoparticles and documented the enhanced magnetic anisotropy and superspin glass behavior for the particles with thicker Pt layer. Song et al. [15] observed slight decrease of blocking temperatures with increasing coating layer in core@shell Co@Au nanoparticle system. They assumed the effect was induced by enhancing the inter-particle distances what led to the decay of the interaction strength. Corresponding with the theory of superparamagnetism, Su et al. [16] found the decrease of the blocking temperature with diminishing diameter of the Co nanoparticles. Also the values of coercivity of the Co nanoparticles measured below the blocking temperature turned out to be proportional to the volume of the particles. Obviously, superspin glass state can be induced by various factors. Concerning the above mentioned facts we attempted to produce systems of cobalt/gold bimetallic nanoparticles with the aim to encounter the superspin glass state and to study the processes leading to its formation. In the presented paper we have studied magnetic state and relaxation process in cobalt nanoparticles coated by protective gold layer, prepared in two different samples with different size of gold shell. We have shown that the nanoparticles collectively behave like a superspin glass and exhibit different magnetic responses when the size of diamagnetic shell differs. The existence of superspin glass state in bimetallic Co/Au nanoparticles has not yet been published.
Cetyltrimethylammonium bromide (CTAB, 99%), sodium borohydride (NaBH4; 98%), were purchased from Aldrich and Sigma. Nanoparticles were prepared using microemulsion method in reverse micelles [17]. All reverse micelle solutions were prepared using CTAB as the surfactant with octane as the oil phase. 1butanol was used as co-surfactant, increasing the polarity of the surfactant and helping to stabilize the micelle solutions [17]. Aqueous reactants of CoCl2, NaBH4, and HAuCl4 were used to form the reverse micelle. The metal particles were formed inside the reverse micelle by reduction of a metal salt using sodium borohydride. At first the pure Co core were produced using a magnetic stirrer and under flowing argon. Co based nanoparticles were subsequently coated by Au shell also by using of reverse micelle solutions. After nanoparticle preparation, the micelles in the reaction mixture were disrupted using aceton causing nanoparticles to precipitate. Repeated washing using a 1:1 mixture of chloroform/methanol removed the surfactant. The size of prepared nanoparticles was controlled by the water to surfactant molar ratio w ([H2O]/[CTAB]). Two different samples of Co/Au1 and Co/Au2 were prepared, where the same molar ratio for preparation of Co core (wCo ¼ 8) and different ratio for Au shell (wAu1 ¼ 5 and wAu2 ¼ 10) were used.
2. Experimental
The high-energy powder X-ray diffraction (HE-PXRD) experiments were carried out at BW5 wiggler beamline of DORIS positron storage ring in DESY (Hamburg, Germany) using monochromatic synchrotron radiation with beam energy of 100 keV (l ¼ 0:12398 Å). The experiments were carried out at room temperature in the transmission mode. LaB6 standard was used to
2.1. Preparation All chemicals, namely: Chloroauric acid (HAuCl4, 99.9%), Cobalt chloride (CoCl2, 97%) octane, 1-butanol (99.8%),
2.2. Characterization
Fig. 1. High resolution transmission electron images of samples Co/Au1 (a) and Co/Au2 (c) samples and its size distributions (b), (d), respectively.
106
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
calibrate the sample-to-detector distance. The background intensity was subtracted directly from XRD patterns and the result was integrated using the software package FIT2D. The HRTEM micrographs were taken with a JEOL 2100F microscope. Copper grid coated with a holey carbon support lm was used to prepare samples for the TEM observation. The bright-field TEM image was obtained at 200 kV. The magnetic measurements were performed on a commercial SQUID-based magnetometer (Quantum Design MPMS 5XL) over a wide range of temperatures (2e300 K) and applied magnetic fields (up to 5 T). Zero field cooling (ZFC) and field-cooling (FC) measurements M(T) were carried out in dc mode. The samples were encapsulated into a plastic sample holder. The diamagnetic contribution of plastic capsule and plastic sample holder were measured and extracted and experimental data were corrected. The same instrument was used for the experimental study of dynamical properties by ac magnetic susceptibility. The complex ac magnetic susceptibility c0 (f)- i c00 (f), where c0 represents in-phase ac susceptibility (real) component and c00 out-of-phase (imaginary) component, was recorded in the temperature interval 2e300 K and in the frequency interval of 0.1e1000 Hz. 3. Results and discussion 3.1. Structural characterization Nearly spherical shape and crystallinity of individual nanoparticles were observed by TEM in both samples, Fig. 1. Sample Co/ Au1 is more homogenous and the particles are smaller in comparison with the Co/Au2 sample. Size distributions obtained from the TEM images with high number of particles (min. 100 particles) are displayed on Fig. 1 (b), (d) and yield the average particle diameter dTEM ~ 7 nm and dTEM ~ 11 nm for the Co/Au1 and Co/Au2, respectively. HRTEM micrographs of gold-coated cobalt nanoparticles show, that the gold shell can be identified in the majority of the particles and the cobalt core appears as dark black area in the center of the nanoparticles. The images of the nanoparticles appear different even at the same defocus because of unhomogeneous shapes and different orientations of the nanoparticles. The phase analysis of the composition carried out by X-ray diffraction is displayed on Fig. 2. The broadness of diffraction peaks refers to nanocrystalline character of the samples, but on the other hand, hampers the fine phase analysis. The FCC metallic gold structure is clearly visible on both diffraction patterns, but the cobalt signal is difficult to recognize. Numerous authors dealing with Co@Au core@shell nanoparticles reported the lack of cobalt signal in the XRD pattern [18e20]. Some of them presume this is the consequence of overlapping FCC gold and 3-cobalt diffraction peaks [19]. In our samples, we suppose the existence of magnetic Co core and its diffraction pattern is hidden under the pattern of gold due to the overlapping of their diffraction peaks. The dominant peak in diffraction pattern (111) was used for calculating the size of the nanoparticles employing the Scherrer formula [21]. The approximate values of dXRD ~ 10 nm and dXRD ~ 13 nm were obtained for the Co/Au1 and Co/Au2, respectively. In the sample Co/Au1, the cobalt oxide diffraction peaks can be also identified as a consequence of surface oxidation of the cobalt cores. Since the thickness of Au shell in Co/Au1 is smaller than in Co/Au2, there is a possibility of the surface oxidation of cobalt core in Co/Au1. 3.2. Ac magnetic susceptibility We have performed the ac susceptibility measurements as function of temperature (2e300 K) and frequencies (0.1e1000 Hz) of external alternating low magnetic field to study the dynamic
Fig. 2. The XRD pattern of Co/Au1 (a) and Co/Au2 (b).
behavior and relaxation processes in prepared samples. Both inphase (real) and out-of-phase (imaginary) components of ac susceptibility were recorded. As it is evident from Fig. 3 (a), (b), one sharp and narrow maximum corresponding to each attempt frequency was recognized for the Co/Au1 at the temperature around Tmax ~ 7 K. Above this temperature all the susceptibility vs. temperature curves are linear and they merge. Sample Co/Au2 displayed different ac susceptibility dependence, see Fig. 3 (c), (d). The sharp and narrow maximum at low temperature of about Tmax1 ~ 5 K was recognized, similar to the maximum of sample Co/Au1. Moreover the second very broad maximum at temperature interval Tmax2 ~ 50e90 K was observed. The distinct behavior of two maxima in the Co/Au2 predicts on different magnetization processes. The common features of ac susceptibility maxima in both samples is so called frequency shift. Increasing the frequency the decrease of maximum value of inphase susceptibility and its slight shift towards higher temperature was observed. Out-of-phase susceptibility maximum also shifts towards higher temperatures but its value rises with frequency enhancement, see Fig. 3. According to numerous authors [5,6,13] such behavior of the ac susceptibility components is typical of systems in which blocking or freezing of magnetic moments occurs. Ac susceptibility measurements are useful tool for quantitative analysis of relaxation processes. Several theoretical and empirical models for ac susceptibility data evaluation were introduced and their results provide the information on character of magnetic
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
107
relaxation as well as on the strength of inter-particle interactions [5,6,12,13,22]. The assembly of magnetic nanoparticles can exhibit different behavior depending on the strength of possible interactions between particles. If mutual interactions are negligible (e.g. superparamagnetic state), the relaxation time t of single parl-Arrhenius law [23]. ticle moment follows the Nee EA
t ¼ t0 ekB T ;
(1)
where t0 denotes the pre-relaxation constant, EA ¼ KV the activation energy, K effective magnetocrystaline anisotropy constant, V the volume of the magnetic particle, kB Boltzman constant and T temperature. With increasing interaction strength the relaxation l-Arrhenius law, Eq. (1) and some time does not follow the Nee authors [12,13,24] employ the Vogel-Fulcher law for data evaluation. According to this model the relaxation time is determined by the equation [22]. E*
t ¼ t0 ekB ðTT0 Þ ;
(2)
where E* is the energy barrier (activation energy) modified by an effective contribution of the inter-particle interactions and T0 is the parameter corresponding to the strength of the interactions. For even stronger interactions the collective behavior of superspins predominate and so called superspin glass state below the critical temperature for superspin ordering TSSG can be observed [25]. In this case the relaxation time is given by equation [22].
t ¼ t*0
T TSSG TSSG
zv
;
(3)
with the critical exponent zv, where v is the critical exponent of the correlation length
xf
T TSSG v TSSG
(4)
and the exponent z relates the relaxation time with the correlation length via tfxz . The Eq. (3) is the modification of the equation derived from the dynamic scale hypothesis [26]. The denomination superspin glass comes from the similarity of the system behavior to the behavior of canonical spin glasses [27,28]. On the other hand there is fundamental difference in the nature of the ordering interactions. While the collective behavior of spins in atomic spin glass is induced mainly by exchange interaction and RKKY (RudermaneKitteleKasuyaeYosida) interaction the superspin glass state existence is caused predominately by the magnetic dipoleedipole interaction [23]. We have analyzed our experimental data in terms of above mentioned theory and the results we obtained are listed in Table 1. The interpretation of various parameters is introduced in following discussion. el-Arrhenius The fit of the data according to the Eq. (1) and Ne model [23] designed for mutually non-interacting particles provided physically non-reasonable small value of pre-relaxation constant t0 ¼ 3.77 1049 s for Co/Au1 maximum. According to several authors [5,6,22,29] typical values of this parameter are in the interval 1012 e 109 s. Due to this, we expected stronger inter-particle interactions in the systems and we employed Vogel-Fulcher model for the data analysis. The value t0 ¼ 1.17 1019 s established from
Fig. 3. Ac susceptibility of studied samples: (a), (c) in-phase and (b), (d) out-of-phase magnetic susceptibility dependence on temperature recorded at different frequencies of alternating magnetic field.
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
108
Table 1 Magnetic characteristics calculated from ac susceptibility data. Sample
Co/Au1 Co/Au2- 1. RP Co/Au2- 2. RP a b c
N-A theorya
VeF theoryb
SSG theoryc
p
t0 (s)
EA/kB (K)
t0 (s)
E*/kB (K)
T0 (K)
t0 (s)
TSSG (K)
zy
3.77 1049 4.47 1023 1.39 1012
794 264 1871
1.17 1019 1.03 1014 2.08 1012
115 97 1821
4.47 2.12 e
3.25 1012 9.68 107 e
6.57 4.04 e
10.7 10.2 e
0.022 0.046 0.092
el-Arrhenius theory using Eq. (1). Parameters were calculated in accordance with Ne Parameters were calculated in accordance with Vogel-Fulcher theory using Eq. (2). Parameters were calculated in accordance with dynamic scaling hypothesis using Eq. (3).
the fit of the data to the dependence ln(t) on Tmax derived from Eq. (2) matched better the expecting value, but still was out of acceptable interval. The best results t0 ¼ 3.25 1012 s and TSSG ¼ 6.57 K provided the data fit according to the dynamic scale hypothesis, Eqs. (3) and (4). Moreover, the dynamic critical exponent zn ¼ 10.7 is in good accordance with the values reported for superspins' freezing process [5]. All these facts are indicating the presence of very strong magnetic dipoleedipole interactions between nanoparticles leading to freezing of the superspins into the superspin glass state. Co/Au2 susceptibility data suggest on the presence of two distinct magnetization processes, therefore we evaluated both of the maxima separately. The character of the first maximum at low temperature Tmax1 ~ 5 K recommends the maximum recognized in Co/Au1. Similarly, we established physically non-reasonable values of preel-Arrhenius, but relaxation constant employing the model of Ne the Vogel-Fulcher model released t0 ¼ 1.03 1014 s. Dynamic scale hypothesis applied on the data established the values zn ¼ 10.2 and TSSG ¼ 4.04 K what is in good accordance with expectations. Only the pre-relaxation constant t0 ¼ 9.68 107 s was out of usually reported range. Assuming gained results, we infer the freezing of superspins is dominant process in the temperature region close to 5 K, however, the rate of collective behavior is lower in comparison with the Co/Au1. The second process (high temperature) in Co/Au2 at Tmax2 ~ 50e90 K is completely different. The pre-relaxation conel-Arrhenius model is stant t0 ¼ 1.39 1012 s obtained utilizing Ne on the edge of the mentioned acceptable interval and refers to blocking of superspins. Since the Vogel-Fulcher law, Eq. (2), provided negligible value of T0 parameter and the dynamic scale hypothesis, Eqs. (3) and (4), absolutely failed, we can conclude that the superspins blocking influenced by weak interactions dominates. Several authors [29] utilize frequency dependent criterion p to distinguish between superspin blocking and freezing process.
p¼
DTmax ; Tmax Dlog n
(5)
where Tmax denotes the average value of c0 (T) maximum temperatures in the range of experimental frequencies, while DTmax denotes the difference between the maximum and minimum value of Tmax. The calculated values p ¼ 0.022 and p ¼ 0.046 for the low temperature maxima in both samples Co/Au1, Co/Au2, respectively, were from the interval 0.005e0.05. Various superspin glass systems with p parameter from this interval were reported [12], thus the observed properties can be regarded as collective freezing of superspins. On the other hand, the value calculated for the second maximum in Co/ Au2 p ¼ 0.092 approximated the interval typical of pure superspin blocking process 0.1e0.13 [29]. Concerning this fact and previous el-Arrhenius and the Vogel-Fulcher models we results from the Ne assume the blocking process of superspins is affected by weak dipoleedipole inter-particle interactions in this case. Thus, the frequency dependent criterion independently confirmed that the particles' dynamics critically slows down at TSSG ¼ 6.57 K in Co/Au1. In opposite, the distribution of energy
barriers in the sample Co/Au2 is found to have two maxima corresponding to freezing and blocking process at TSSG ¼ 4.04 K and Tp ¼ 57.14 K, respectively. Additionally, we used ColeeCole plots (Argand diagrams) [5,8] (the dependence of out-of-phase susceptibility on in-phase susceptibility components) at temperatures close below the temperature of susceptibility maximum for the determination of magnetic interaction strength in the investigated systems. The shape of the plots reflects the distribution of relaxation times of superspins [5]. If the single relaxation time is characteristic for the process, the superparamagnetic relaxation is confirmed and the shape of ColeeCole plot is perfect semicircle [5]. One of the possible indications of collective behavior presence in the system is the wide range of relaxation times [8]. This reflects the shift of the ColeeCole plot arc downwards. Fig. 4 shows the ColeeCole plots constructed at the temperature of susceptibilities maxima in Co/Au1 Fig. 4(a) and Co/Au2 Figs. (b), (c). Since the ColeeCole plots constructed for low temperature region (a), (b) of both samples exhibited significant translation under the horizontal axis, we infer the presence of stronger magnetic interactions. Moreover, the arc for the Co/Au1 (a) exhibited higher flatness than the arc for low temperature relaxation process at Tmax1 ~ 5 K in Co/Au2, thus the interactions in Co/Au1 should be stronger. This result is in line with the conclusions we discussed previously and confirms the freezing of superspins. On the other hand, the ColeeCole arcs are tending to have a shape of semicircle while approaching the temperature to the Tmax2 ~ 50e90 K what is evident from Fig. 4 (c). This suggests the second maximum in the distribution of energy bariers in sample Co/Au2, which is connected to the blocking process with low influence of magnetic interactions. 3.3. Dc magnetization Fig. 5 presents the dependence of magnetization of the samples on temperature recorded in both ZFC (zero-field-cooling) and in 100 Oe and 500 Oe FC (field-cooling) protocols. According to Tadic [29] the maximum of ZFC curve signifies the presence of relaxation process and the peak temperature Tp represents the characteristic temperature at which the superspins of the particles' majority become blocked or frozen. We determined the Tp ~ 6 K for the Co/ Au1 and Tp1 ~ 4 K, Tp2 ~ 57 K (see Table 2) for the first and the second maximum in Co/Au2, respectively. The character of FC magnetization vs. temperature curve below the Tp is the reliable feature commonly used for distinguishing between superparamagnetic and superspin glass state [30]. The FC magnetization monotonically increases with decreasing T for superparamagnet, while it tends to saturate to a constant value or even tends to decrease with temperature decay for superspin glass. The FC magnetization shows the tendency to saturate at very low temperatures what suggests the superspin glass state presence at this temperature region in the sample Co/Au1. Such effect cannot be documented responsibly in the Co/Au2 sample because the transition temperature is too low. According to [31], the temperature of irreversibility Tirr
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
109
Fig. 5. ZFC and FC magnetization of the samples Co/Au1 (a) and Co/Au2 (b). Inset the determination of blocking temperatures for the system Co/Au1 (a) and Co/Au2 (b).
Fig. 4. ColeeCole plots for relaxation process in Co/Au1 (a) and for the first (b) and the second (c) relaxation process in Co/Au2.
(temperature, at which the ZFC and FC curves begin to separate) can be associated with the blocking temperature of the largest nanoparticles in the assembly. The difference between Tirr and Tp is related to the width of activation energy distribution, thus, to the width of the size distribution of the particles [31]. The Tirr ~ 10 K close to Tp ~ 7 K was found for Co/Au1 confirming quite narrow size distribution. The existence of two ZFC M(T) maxima in sample Co/ Au2 we attributed to the existence of two maxima in energy barrier distribution corresponding to freezing and blocking processes at different temperature. One of the most relevant parameter typical of superparamagnetic system is blocking temperature TB. For T < TB the thermal fluctuations are insufficient for the superspin reversal over the activation energy barrier and the superspins will appear blocked or frozen, while for T > TB they will appear free comparable to a paramagnetic system. The blocking temperature is not an intrinsic temperature of the system, but depends strongly on the probing time scale of a measurement. It is defined as the temperature at which the relaxation time of the superspin equals the probing time scale of a measurement. This is the reason why some authors [22] establish its value from the relaxation time t vs. 1/T plot. The typical time scale of a standard SQUID magnetometry experiment is 10 s. It represents the dashed line in Fig. 5 (a) inset and intersects the curve t vs. 1/T at TB* ~ 7 K thus defining the blocking temperature of the Co/Au1 system. Similarly, we determined the TB1* ~ 5 K, TB2* ~ 60 K for the first and the second process in Co/Au2, Fig. 5 (b) inset.
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
110
Table 2 Structural and magnetic characteristics of Co/Au1 and Co/Au2 samples. Sample
Tp (K)
TB* (K)
mp (mB)
MS (2 K) (emu/g)
HC (2 K) (Oe)
dTEM (nm)
dXRD (nm)
dMAG (nm)
Co/Au1 Co/Au2
6.71 4.37
7.2 4.77
86.33 144.94
40.62 16.93
1183 543
7.03 11.22
10.12 13.29
5.36 6.90
57.14
62.46
Fig. 7. Magnetization M of Co/Au1 (a) and Co/Au2 (b) expressed as a function of H/T. Fig. 6. Magnetization on magnetic field dependence measured at various temperatures for Co/Au1 (a) and Co/Au2 (b). Inset the Langevin fit of magnetization vs. field at 280 K.
Typical superparamagnetic behavior of the particles can be seen from field-cooled M(T) dependences, Fig. 5, as well as from M(H) dependences at the temperatures above Tp, Fig. 6. At the temperatures of the nanoparticle system lower than Tp the magnetic hysteresis was recorded, Table 2. The higher coercivity value at 2 K in Co/Au1 contrary to Co/Au2 corresponds to the stronger interactions confirmed by previous ac analysis in the first sample. Once the system is in superparamagnetic state at higher temperatures, the data M(H) should obey the modified Langevin law which takes in account the non-negligible size distribution of magnetic moments. Than the magnetization of superparamagnetic nanoparticles is described as the weighted sum of Langevin functions [21].
M¼
Z∞ mp H f mp dmp ; L kB T
(6)
0
where f (mp) is the distribution of magnetic moments, L Langevin
function and mp magnetic moment of the particle. Assuming nanoparticles in superparamagnetic state and applying fit of the Eq. (6) to the experimental data of M(H) obtained at 280 K the average magnetic moment value mp ~ 86 mB and mean diameter of the particle dMAG ~ 5 nm were established for the Co/ Au1 and mp ~ 145 mB, dMAG ~ 7 nm for the Co/Au2. Since the calculations of the particle's diameter utilizing Langevin fit consider only cobalt core of the particle, not diamagnetic gold shell, the size of the particle determined by Langevin fit was lower than the size obtained from HRTEM images and Scherrer formula. Since the results of structural measurements introduced some doubts about the existence of partial oxidation in Co core, especially in the sample Co/Au1, we attempted to verify the presence of additional cobalt oxide interlayer performing the exchange bias effect measurements. Magnetization vs. field (isothermal magnetization curves) at 2 K (below blocking temperature) before and after cooling in external dc field of 30 000 Oe were recorded. No shift of hysteresis loop along the x axis after cooling in external dc field (FC ¼ 30 000 Oe) was observed what is the evidence of the exchange bias absence in the sample. This confirms that the exchange interaction between ferromagnetic/antiferromagnetic
k et al. / Journal of Alloys and Compounds 649 (2015) 104e111 P. Hrubovca
phases is not present. In order to confirm superparamagnetic state presence in the samples we employed the method described in Ref. [31]. If the samples were truly superparamagnetic the isothermal magnetization curves measured at different temperatures above the blocking temperature should overlap when plotting M vs H/T [31]. This feature is indeed achieved as can be seen in Fig. 7, thus confirming the superparamagnetic behavior of our sample for temperatures above 60 K in both studied samples. 4. Conclusions In summary, we have investigated the static and dynamic magnetic behavior of Co/Au nanoparticles, prepared employing the microemulsion method. The magnetoestructural correlations show that the samples are composed predominantly from magnetic Co core with the size around 6e7 nm coated by diamagnetic Au shell of the thickness around 3 nm and 6 nm for two studied samples Co/Au1 and Co/Au2, respectively. Magnetic measurements confirm the evidence of magnetic relaxation processes and existence of novel superspin glass state in both studied systems. In fine Co/Au1 and CoAu/2 nanoparticles the very strong dipolar magnetic interactions between particles are present. These interactions lead to superspin freezing and individual particles exhibit the collective behavior below the glass transition temperature TSSG. In the sample Co/Au2 was observed that, the energy barrier distribution has two maxima corresponding to freezing and blocking process at different temperatures of TSSG ¼ 4.04 K and Tp ¼ 57.14 K, respectively. We demonstrated that the size of the diamagnetic Au shell capping the magnetic Co core can crucially influence the magnetic properties of the system. The thickness of the diamagnetic Au shell affects the center-to-center distance between magnetic Co nanoparticles what has significant influence on the strength of magnetic interaction. Our observation can open up the possibility of tailoring the SSG (superspin glass) state and its onset temperature by suitable selection of the particle size and can help in the application of very fine Co/Au bimetallic nanoparticles. Acknowledgment This work was supported by the Slovak Research and Development Agency under the contracts APVV-0132-11 and APVV-007314 and by the VEGA projects of Ministry of Education of the Slovak Republic (No. 1/0583/11, No. 1/0861/12) and by the ERDF EU grant under No. ITMS 26220120035. The authors (A.Z and V.Z) would like to thank DESY/HASYLAB project under No. I-20110282 EC and also
111
Dr. N. Murafa (AS CR, Rez, Czech Republic) for HRTEM measurements.
References [1] M. Yigit, A. Moore, Z. Medarova, Pharm. Res. 29 (2012) 1180. [2] T. Sadhukha, T.S. Wiedmann, J. Panyam, Biomaterials 34 (2013) 5163. [3] V.V. Mody, A. Cox, S. Shah, A. Singh, W. Bevins, H. Parihar, Appl. Nanosci. 4 (2014) 385. [4] J.H. Jung, J.H. Leea, S. Shinkai, Chem. Soc. Rev. 40 (2011) 4464. [5] S. Bedanta, W. Kleemann, J. Phys. D Appl. Phys. 42 (2009) 28. [6] O. Petracic, X. Chen, S. Bedanta, W. Kleemann, S. Sahoo, S. Cardoso, P.P. Freitas, J. Magnet. Magnetic Mater. 300 (2006) 192. [7] K. Nadeem, H. Krenn, T. Traussnig, R. Wrschum, D.V. Szabo, I. Letofsky-Papst, J. Magnet. Magnetic Mater. 323 (2011) 1998. [8] F. Jimenez-Villacorta, J. Sanchez-Marcos, E. Cespedes, M. Garcia-Hernandez, C. Prieto, Phys. Rev. B 82 (2010) 134413. [9] S. Thomas, M. Uhlig, U. Wiedwald, L. Han, P. Ziemann, M. Albrecht, Nanotechnology 24 (2013) 155703. [10] M. Perovic, V. Kusigerski, V. Spasojevic, A. Mrakovic, J. Blanusa, M. Zentkova, M. Mihalik, J. Phys. D Appl. Phys. 46 (2013) 165001. [11] J.A. De Toro, S.S. Lee, R. Mathieu, P.S. Normile, D. Salazar, J.L. Cheong, P. Muniz, J.M. Riveiro, M. Hillenkamp, A. Tamion, F. Tournus, P. Nordblad, J. Phys. Conf. Ser. 521 (2014) 012011. [12] J.A. De Toro, S.S. Lee, D. Salazar, J.L. Cheong, P. Muniz, J.M. Riveiro, M. Hillenkamp, A. Tamion, F. Tournus, P. Nordblad, Appl. Phys. Lett. 102 (2013) 183104. [13] M. Tadic, V. Kusigerski, D. Markoviic, M. Panjan, L. Miloseviic, V. Spasojeviic, J. Alloys Compd. 525 (2012) 28. [14] A. Ebbing, O. Hellwig, L. Agudo, G. Eggeler, O. Petracic, Phys. Rev. B 84 (2011) 012405. [15] Y. Song, J. Ding, Y. Wang, J. Phys. Chem. C 116 (2012) 11343. [16] Y.K. Su, C.M. Shen, T.Z. Yang, H.T. Yang, H.J. Gao, H.J. Li, Appl. Phys. A 81 (2005) 569. [17] J. Lin, W. Zhou, A. Kumbhar, J. Wiemann, J. Fang, E.E. Carpenter, C.J. O'Connor, J. Solid State Chem. 159 (2001) 26. [18] Y. Bao, H. Calderon, K.M. Krishnan, J. Phys. Chem. C 111 (2007) 1941. [19] Y. Bao, K.M. Krishnan, J. Magnet. Magnetic Mater. 293 (2005) 15. [20] G. Cheng, A.R. Hight Walker, J. Magnet. Magnetic Mater. 311 (2005) 31. [21] A. Zelenakova, V. Zelenak, J. Bednarcik, P. Hrubovcak, J. Kovac, J. Alloys Compd. 582 (2014) 483. [22] O. Petracic, Superlattices Microstruct. 47 (2010) 569. [23] S. Morup, M.F. Hansen, C. Frandsen, Beilstein J. Nanotechnol. 1 (2010) 182. [24] M. Tadic, S. Kralj, M. Jagodic, D. Hanzel, D. Makovec, Appl. Surf. Sci. 322 (2014) 255. [25] M. Tadic, V. Kusigerski, D. Markoviic, M. Panjan, L. Miloseviic, V. Spasojeviic, J. Alloys Compd. 509 (2011) 7134. [26] H. Sompolinsky, A. Zippelius, Phys. Rev. Lett. 47 (1981) 359. [27] T. Nishioka, Y. Tabata, T. Taniguchi, Y. Miyako, J. Phys. Soc. Jpn. 69 (2000) 1012. [28] J. J. Binney, N. J. Dowrick, A. J. Fisher, and M. E. J. Newman, Oxford University Press (1992). [29] M. Tadic, D. Markoviic, V. Kusigerski, L. Spasojeviic, L. Miloseviic, J. Magnet. Magnetic Mater. 321 (2009) 12. [30] M. Suzuki, S.I. Fullem, I.S. Suzuki, L. Wang, C.-J. Zhong, Phys. Rev. B 79 (024418) (2009). [31] M. Tadic, D. Markoviic, V. Spasojeviic, V. Kusigerski, M. Remskar, J. Pirnat, Z. Jaglicic, J. Alloys Compd. 441 (2007) 291.