Defect controlled room temperature ferromagnetism in

0 downloads 0 Views 3MB Size Report
Dec 14, 2011 - This article has been downloaded from IOPscience. Please scroll .... the program package WinCSD [22]. For all samples ... Ba1−x Ti1−yCoy Oz a. Sample .... the nanocrystallites tend to return to the defect-free cubic structure with a ..... [17] Vanderah T A, Loezos J M and Roth R S 1996 J. Solid State. Chem.
Home

Search

Collections

Journals

About

Contact us

My IOPscience

Defect controlled room temperature ferromagnetism in Co-doped barium titanate nanocrystals

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2012 Nanotechnology 23 025702 (http://iopscience.iop.org/0957-4484/23/2/025702) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 143.129.150.51 The article was downloaded on 10/01/2012 at 08:33

Please note that terms and conditions apply.

IOP PUBLISHING

NANOTECHNOLOGY

Nanotechnology 23 (2012) 025702 (10pp)

doi:10.1088/0957-4484/23/2/025702

Defect controlled room temperature ferromagnetism in Co-doped barium titanate nanocrystals Sugata Ray1,2 , Yury V Kolen’ko3,4, Kirill A Kovnir5,6 , Oleg I Lebedev7,8, Stuart Turner7, Tanushree Chakraborty2, Rolf Erni7,9 , Tomoaki Watanabe3, Gustaaf Van Tendeloo7 , Masahiro Yoshimura3 and Mitsuru Itoh3 1 Department of Materials Science, Indian Association for the Cultivation of Science, Jadavpur, Kolkata 700 032, India 2 Center for Advanced Materials, Indian Association for the Cultivation of Science, Jadavpur, Kolkata 700 032, India 3 Materials and Structures Laboratory, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8503, Japan 4 International Iberian Nanotechnology Laboratory, Avenida Mestre Jos´e Veiga, s/n, P-4715-330 Braga, Portugal 5 Max Planck Institute for Chemical Physics of Solids, N¨othnitzer Straße 40, 01187, Dresden, Germany 6 Chemistry Department, University of California at Davis, Davis, CA 95616, USA 7 Electron Microscopy for Materials Science, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgium 8 CRISMAT UMR 6508, CNRS–ENSICAEN, Universit´e de Caen, Caen, France 9 Empa, Swiss Federal Laboratories for Materials Science and Technology, Electron Microscopy Center, Ueberlandstraße 129, CH-8600 Dubendorf, Switzerland

E-mail: [email protected]

Received 3 August 2011, in final form 27 October 2011 Published 14 December 2011 Online at stacks.iop.org/Nano/23/025702 Abstract Defect mediated high temperature ferromagnetism in oxide nanocrystallites is the central feature of this work. Here, we report the development of room temperature ferromagnetism in nanosized Co-doped barium titanate particles with a size of around 14 nm, synthesized by a solvothermal drying method. A combination of x-ray diffraction with state-of-the-art electron microscopy techniques confirms the intrinsic doping of Co into BaTiO3 . The development of the room temperature ferromagnetism was tracked down to the different donor defects, namely hydroxyl groups at the oxygen site (OH•(O) ) and oxygen vacancies (V•• (O) ), and their relative concentrations at the surface and the core of the nanocrystal, which could be controlled by post-synthesis drying and thermal treatments. (Some figures may appear in colour only in the online journal)

magneto-optic and magneto-electronic devices and can also be ideal model systems to study dopant impurity dependent optical and electronic properties, finite size effects on magnetic ordering, anisotropy etc. Therefore, fabrication of well characterized nano-DMS systems is becoming increasingly important from the viewpoint of technology as well as fundamental scientific interest [2–7]. However, since different synthesis

1. Introduction Dilute magnetic semiconductors (DMS) with Curie temperatures (Tc ) above 300 K are materials of high technological importance due to their potential use in spin based electronic devices operable at room temperature [1]. Moreover, DMS systems in nanostructured form can be useful components in 0957-4484/12/025702+10$33.00

1

© 2012 IOP Publishing Ltd Printed in the UK & the USA

Nanotechnology 23 (2012) 025702

S Ray et al

methods to produce one single DMS system are known to strongly modify the magnetic properties, this magnetism always remains clouded by experimental uncertainties, and the basic issue of whether the magnetism is intrinsic or not is never properly established [8]. Therefore, at present any proposal for a true DMS system must be accompanied by detailed characterizations which should first ensure the clean, intrinsic magnetic ion doping of the system. Curiously, almost simultaneously the importance of other defects (vacancies etc) in generating high temperature ferromagnetism, especially in nanocrystalline systems, has also been put forward very strongly [9–11]. Therefore, intrinsically doped systems with purposefully placed and manipulated defects seem to be the best possible option to achieve a true and useful DMS material. It is worth noting here that unlike the trends of more than a decade earlier, the discovery of high Tc ferromagnetism in Co-doped thin films of anatase TiO2 by Matsumoto et al [12] has pushed the focus toward oxides, which immediately opened up new benchmark. Specifically, in contrast to the more commonly studied III–V semiconductors [13], the advantage of perovskite oxide materials is the high equilibrium solubility of the dopant transition metal ions into its structure, which curbs down usual doubts about precipitation of secondary magnetic phases etc. These considerations prompted us to choose transition metal doped perovskite barium titanate (BaTiO3 ) [14, 15] as a potential nano-DMS system, since BaTiO3 is known to be doped easily with transition metal ions up to very large concentrations [16–18]. In this paper, we report room temperature ferromagnetism in Co-doped BaTiO3 nanoparticles, where the presence of vacancies and relative defect concentrations is found to regulate the occurrence/absence of ferromagnetism. Firstly, we characterize our nanocrystals in great detail, employing several different experimental tools. Then we show that even though the importance of acceptor and donor defects in modifying dilute magnetism has already been discussed [19], it is the relative defect concentration and also the location of the defects inside the nanocrystal that actually modify the character of dilute magnetism in these materials. Notably, high temperature ferromagnetism in undoped BaTiO3 nanocrystals has been reported earlier [9], and the origin was assumed to be surface defects. However, our results show that surface defects alone could not be sufficient for ferromagnetism because in many of our samples ferromagnetism was not observed even though intrinsic dopants as well as a considerable number of vacancy defects were present, which indicates that the relative concentration and location of dopant and vacancy defects must be the key factor.

was prepared by dissolution of 4.3 mmol of titanium(IV) tetraisopropoxide in 10 ml of absolute ethanol. This solution was added dropwise to the Ba/Co-containing one. Next, 6 ml of distilled water was added to the reaction mixture, and the PTFE vessel with the resultant white suspension was then capped by a PTFE cover (Hyper-Sheet Gore-Tex gasket sets between the vessel and cover) [21] and placed in a stainless steel autoclave. The autoclave was sealed and kept at 240 ◦ C for 4, 6 and 12 h to produce three samples with marginal differences in terms of synthesis duration. To remove the BaCO3 by-product, the derived nanopowders were washed by ultrasonication for 5 min in 0.5% aqueous acetic acid solution and then dried at 80 ◦ C for 12 h. The organic residues were consequently removed by thermal treatment of the samples at 500 ◦ C for 10 h in static air. Throughout this work, a set of symbols is used (table 1). The first letter B indicates the desired, as-produced BaTiO3 ; the numbers indicate three different drying times—4 h (1), 6 h (2) and 12 h (3); the last letters indicate the consequent treatments (if applicable): w is washing by dilute acetic acid and tt followed by thermal treatment at 500 ◦ C. 2.2. Characterization The composition and crystallographic parameters of the nanocrystalline products were studied by powder x-ray diffraction (XRD), using a Rigaku RINT 2000 diffractometer and Huber G670 Image Plate Camera, equipped with Cu Kα 1 ˚ The unit cell radiation, wavelength (λ) = 1.540 598 A. parameters were calculated from least-square fits using LaB6 ˚ as an internal standard utilizing (cubic, a = 4.156 92 A) the program package WinCSD [22]. For all samples the cubic symmetry (Sp. Gr. Pm 3¯ m ) was assumed. Chemical analysis for the Ba, Co and Ti content was performed by inductively coupled plasma-optical emission spectrometry (ICP-OES) using a Varian Vista RL spectrometer. All values are the average of at least three replicates. The phase purity and presence of impurity clusters or inhomogeneities were investigated using transmission electron microscopy (TEM), high resolution transmission electron microscopy (HRTEM) and electron diffraction (ED). TEM and HRTEM investigations were carried out on a crushed sample deposited on a holey carbon grid using a JEOL 4000EX microscope, operated at 400 kV. Energy filtered TEM (EFTEM), Z -contrast imaging, scanning TEM-electron energy loss spectroscopy (STEMEELS) experiments were performed using a JEOL 3000F TEM/STEM electron microscope, equipped with a GIF-2000 spectrometer. The STEM-EELS spectra were acquired at an approximate interspacing of 1 nm, using a convergence semiangle of 22 mrad, a collection semi-angle β of 29 mrad and nominal spot size of 0.5 nm and energy dispersion of 0.5 eV/pixel. The low-loss spectrum was acquired using EELS in diffraction mode, using a convergence semi-angle of 0.3 mrad, a collection semi-angle β of 0.7 mrad and an energy dispersion of 0.2 eV/pixel. Fourier transformed infrared spectroscopy (FT-IR) measurements were performed in a Shimadzu FT-IR 8400S instrument. The magnetic measurements were carried out in a Quantum Design SQUID magnetometer with a maximum field of 5 T.

2. Experimental methods 2.1. Synthesis Nanosized BaTiO3 was synthesized via a solvothermal drying route [20, 21]. The synthesis proceeds by dissolution of 4.73 mmol (10 mol% excess) of Ba(OH)2 ·8H2 O and 0.215 mmol of Co(NO3 )2 ·6H2 O in a 40 ml poly(tetrafluoro ethylene) (PTFE) vessel backfilled with 20 ml of 2methoxyethanol. In parallel, the Ti-containing solution 2

Nanotechnology 23 (2012) 025702

S Ray et al

Table 1. Synthesis, structural and compositional details of all the samples Ba1−x Ti1−y Co y Oz a

Sample B1 B1w B1wtt B2 B2w B2wtt B3 B3w B3wtt

Synthesis and post-synthesis treatments

T = 240 ◦ C, t = 4 h B1 washed by dilute acetic acid B1w calcined at 500 ◦ C for 10 h in air T = 240 ◦ C, t = 6 h B2 washed by dilute acetic acid B2w calcined at 500 ◦ C for 10 h in air T = 240 ◦ C, t = 12 h B3 washed by dilute acetic acid B3w calcined at 500 ◦ C for 10 h in air

Admixture

˚ cubic a (A)

Ba (wt%)

Co (wt%)

Ti (wt%)

1−x

y

z

3 − xb

BaCO3 —

4.0400(4) 4.0421(8)

55.3(2) 43.7(2)

1.05(1) 0.94(2)

16.4(1) 23.1(1)

1.12 0.64

0.05 0.03

— 4.0

— 2.64



4.0273(5)

48.4(1)

1.00(1)

25.9(1)

0.63

0.03

2.7

2.63

BaCO3 —

4.0395(5) 4.0426(4)

55.2(4) 44.1(2)

1.08(1) 0.87(1)

15.8(1) 23.4(1)

1.16 0.64

0.05 0.03

— 3.9

— 2.64



4.0265(4)

49.0(4)

0.93(2)

24.6(7)

0.67

0.03

3.0

2.67

BaCO3 —

4.0331(5) 4.0386(5)

56.9(5) 45.8(2)

1.02(1) 0.95(1)

17.9(2) 23.0(2)

1.06 0.67

0.04 0.03

— 3.8

— 2.67



4.0288(5)

47.2(1)

1.05(2)

24.5(1)

0.65

0.03

3.2

2.65

a

The oxygen stoichiometry was calculated by subtracting the measured mass contributions of Ba, Co and Ti ions from the total sample mass, assuming the presence of only Ba, Ti, Co and O in the samples. The oxygen content from the as-prepared samples has not been included due to the unaccounted for presence of BaCO3 admixture. b The values 3 − x have been calculated assuming a Co4+ oxidation state maintaining charge neutrality and this value matches well with the experimental value only in case of B1wtt sample.

description is briefly given below, while our results turned out to be in complete agreement with these earlier observations. It was shown earlier that even in the absence of any excess barium reactant, BaCO3 is formed as a by-product during the hydrothermal/solvothermal syntheses of BaTiO3 in the presence of water. This is due to the incorporation of water molecules into the oxygen sublattice in form of hydroxyl groups [23]. It was predicted that replacement of O2− ions by OH− groups is a highly favorable process in this case, which can be compensated by vacancies in the Ba and/or Ti sublattices [23]. The post-synthesis drying/calcination process can drive a few more reactions in which the displaced Ba2+ ions form BaCO3 using the CO2 from the synthesis medium, while the displaced Ti4+ ions can form TiO2 [23, 24], although in the present case only BaCO3 impurity was found, most likely due to the excess of the Ba precursor (see section 2). The chemical composition of the products was studied by ICP-OES, and the results are presented in table 1. The asprepared samples possess greater than stoichiometric amounts of Ba, consistent with the fact that the initial precursor solution contained an excess of Ba salts. Interestingly, the measurements also showed that after washing the samples with an acidic solution (BaCO3 is almost completely soluble in water under acidic pH [25]), a substantial amount of Ba nonstoichiometry (∼35%) appears in BaTiO3 itself which also remains unchanged after the post-washing thermal treatment. Now, in agreement with earlier reports [24, 26], a chemical process leading to effective removal of Ba2+ ions from the BaTiO3 nanocrystals can be described as follows: BaTiO3 (s) + 0.8H+ (aq.) = 0.4Ba2+ (aq.) (1) + Ba0.6 TiO2.2 (OH)0.8 . Finally, the calcination of the washed samples (Bnw; n=1–3) at 500 ◦ C in air (table 1) for the removal of organic admixtures

3. Results and discussions In figure 1(a), the powder x-ray diffraction (XRD) patterns from all three B1 series of samples are shown, which closely resemble single phase cubic BaTiO3 (Sp. Gr. Pm 3¯ m ). The B2 and B3 series of samples exhibit very similar XRD patterns (not shown). For all the solvothermally derived products Bn (n = 1, 2 or 3) the presence of a barium carbonate admixture was detected while for the washed and annealed samples only phase-pure BaTiO3 was confirmed. In the left inset to figure 1(a) an expanded view of the region with the strongest BaCO3 peaks is presented for the B1 series of samples. This shows the effective removal of BaCO3 by washing with acetic acid. The sizes of the nanocrystallites were estimated by means of the Debye–Scherrer equation from the average broadening of the (100), (110) and (111) XRD reflections, yielding an average diameter of about 14 nm for all the samples. The lattice parameters are extracted from the XRD patterns and are listed in table 1. Although the lattice parameters appear to be expanded slightly by going from Bn to Bnw (n = 1, 2 or 3), a distinct cell volume contraction is observed in Bnwtt compared with Bnw. Clear evidence for this lattice contraction is found from XRD patterns collected together with a LaB6 internal standard, as is shown for the B1 series of samples in figure 1(b). This is further supported by electron diffraction (ED) experiments, shown for samples from the B1 series in the right inset to figure 1(a). Importantly, no other crystalline phases were detected from the XRD studies of the washed and post-washed products although strong line broadening effects are not conducive for locating the presence of minor impurity phases. The synthesis and the post-synthesis treatments of BaTiO3 nanoparticles, followed here, were studied earlier in detail. A 3

Nanotechnology 23 (2012) 025702

S Ray et al

Figure 1. (a) Powder x-ray powder diffraction patterns of B1, B1w and B1wtt samples. The positions of the BaTiO3 diffraction peaks are shown by ∗. The left inset shows the angular region from 20◦ < 2θ < 27.0◦ where the most intense peaks (indicated by dotted lines) of the BaCO3 by-product are expected to be. In the right inset, the ED patterns of B1 and B1wtt samples are compared, indicating a smaller unit cell parameter for B1wtt. (b) The powder XRD patterns of B1, B1w and B1wtt samples measured with LaB6 as an internal standard (the peaks of the standard are denoted by +). Dotted lines are drawn at the positions of the diffraction maxima of sample B1 to indicate the peak shifts.

Figure 2. (a) FT-IR spectra from the as-prepared B1, B2 and B3 samples. Panel (b) shows the FT-IR spectra, normalized at parent BaTiO3 bands from the 12 h dried (B3w) and 4 h (B1w) dried samples. The inset shows the background correct IR bands from the surface hydroxyl groups from these two samples.

the nanocrystallites tend to return to the defect-free cubic structure with a smaller lattice volume. In order to check the truth of the above description and also to follow the relative changes of different species in the sample as a function of various chemical and thermal treatments, we have carried out detailed FT-IR experiments on these samples. The most informative results are summarized in figure 2. The raw FT-IR spectra from the three solvothermally derived samples B1, B2 and B3 are shown in panel (a). All the spectra are similar in nature and are consistent with previous reports [27–31]. The sharp feature at 2362 cm−1 indicates the band for asymmetric stretching of CO2 , which is an extrinsic feature and will be ignored in the discussion. A comprehensive assignment of IR bands is presented in table 2. As expected, the IR bands corresponding to the bending and asymmetric stretching vibrations of the CO23− group of BaCO3 impurity dominate the spectra and can be easily identified primarily at 1438, 1384, 857 and 670 cm−1 . The weak bands corresponding to the C–H stretching vibrations of the accompanying organic alkyl groups are also observed at 2855

is reported to drive the following additional reaction [23]: 2OH•(O) = O(O) + V•• (O) + H2 O

(2)

resulting in a significant number of oxygen vacancies in the lattice caused by evaporation of water. Here, OH•(O) indicates a hydroxyl group at the oxygen site, which acts as a single electron donor because the rest of the system must accept one electron from this site in order to maintain charge neutrality. Similarly V•• (O) represents the double electron donor oxygen vacancy. Reaction (2) is followed by partial annihilation of the point defects due to the migration of the cation vacancy [V◦◦ (Ba) ] and the anion vacancy [V•• (O) ] to the same point, generating pore-like structures [23]. The significant reduction of the lattice parameters in the Bnwtt samples (figure 1 and table 1) supports this phenomenon of overall defect annihilation where 4

Nanotechnology 23 (2012) 025702

S Ray et al

Table 2. Positions of vibrational bands (cm−1 ) detected in BaTiO3 .

and 2925 cm−1 . Most importantly, a broad ‘fingerprint’ band from the O–H stretching vibration at around 3420 cm−1 is observed, which confirms the presence of hydroxyl groups in the lattice, as predicted earlier. A certain amount of physisorbed water may also contribute to this, but the existence of this feature even in calcined Bnwtt samples confirms chemical bonding of hydroxyl groups to the lattice. It is also important to note here that the relative intensity of the O– H band generally increases in the Bnw samples compared to their raw counterparts Bn, which indicates further replacement of O(O) by OH•(O) groups during acid washing. The overall charge neutrality consideration can be a driving force toward this change. At this juncture it is important to search for the differences that the three series of samples (n = 1–3) might possess as a result of the variations in their respective synthesis drying time. For proper comparison, the spectra have to be normalized at a specific band, which should remain invariant through all the chemical and thermal treatments. Accordingly, we have chosen the broad band around 530–600 cm−1 , which corresponds to lattice modes of inorganic BaTiO3 itself [27–29] and which with increasing drying time during synthesis is not expected to change transmittance, while all the C–H, CO23− , and O–H bands may change. Two normalized spectra from the lowest and highest dried samples, after acid washing, are shown in figure 2(b). Firstly, the effect of acid washing becomes evident from this pair of spectra as all the characteristic bands corresponding to the BaCO3 impurity are found to vanish. However, the most intriguing observation is the significantly decreasing intensity of the O–H band in the B3w sample with respect to the B1w sample. This observation indicates that prolonged drying at 240 ◦ C during synthesis causes modifications in the concentration of the bound O–H defects. A closer look at these spectra reveal another interesting feature, which helps us to understand the reason behind such behavior. The inset to figure 2(b) shows an expanded part of the spectra after background correction, which exhibits two weak but clear bands centered at 3730 and 3697 cm−1 . These two bands can be attributed to surface hydroxyl groups, since it is well known that the free hydroxyl groups at the surface absorb at higher wavenumbers than the lattice bound hydroxyl groups [27]. Thus, prolonged drying of the nanocrystals leads to their structural relaxation where strains and other defects are released by pushing out the ionic impurities to the surface. This results in a rather defectfree nanocrystalline core. Recently, similar behavior has been observed in many nanostructured materials [32, 33], where the core of the nanocrystals prefers to push the impurities to the surface, in order to relax structurally. In the present case, this ‘self-purification’ mechanism [32, 33] operates for the H-ions, which propagates to the surface oxygen atoms with prolonged heating as if the hydroxyl ions themselves migrate to the surface. Thus, the FT-IR analysis reveals that with increasing drying time during synthesis (B3 > B2 > B1), the doped nanocrystals tend to possess a more crystalline defect-free interior, while the hydroxyl impurities migrate to the surface. In other words, the B1w crystals are expected to be more defective with large numbers of hydroxyl ion impurities residing at the core in closer proximity, while the

Positions (cm−1 )

Assignments

538, 597 670, 857 1058 1384, 1438 1625, 1750 2341, 2362 2855, 2925 3420 3697, 3730

Lattice modes of BaTiO3 In plane and out plane bending of CO23− Symmetric C–O stretch of CO23− Asymmetric C–O stretch of CO23− C–H bending and BaCO3 CO2 impurity C–H stretching Bound O–H stretching Free O–H stretching

B3w crystals should be crystallographically more perfect with a substantial number of O–H defects residing at the surface of the nanocrystals. This is further confirmed by electron microscopy, as described below. Next, the controversial issue of intrinsic Co doping in these BaTiO3 nanocrystals was probed using different electron microscopy techniques. HRTEM investigation of single nanoparticles indicates the presence of Ba vacancies in varying concentrations, in accordance with our previous understanding. Figures 3(a)–(c) show the HRTEM and ED images from a B1w nanoparticle along the [111]C zone axis. There is a clear variation of the lattice fringe contrast inside one particle (figure 3(b)). This variation is manifested by a reduced background contrast within 1–2 nm sized areas. Such local reduction of contrast in HRTEM micrographs might be attributed to clusters of Ba vacancies. Exceeding a certain vacancy concentration leads to internal stress that can be reduced by the formation of core dislocations which are observed in sample B1w (figures 3(d)–(e)). The existence of dislocations is particularly surprising in the case of nanoparticles where the energy stability of a dislocation is not a priori warranted. The closure failure of the Burgers circuit in the HRTEM image (figure 3(e)) determines that the projection √ ¯ ]C , where a is the of Burgers vectors b1 and b2 is a 2[110 lattice parameter of cubic BaTiO3 . In figures 4(a)–(c), three representative HRTEM images of Bnwtt (n = 1, 2 or 3) are shown, respectively. Detailed TEM and ED studies confirm that the samples are single phase Co-doped BaTiO3 , devoid of other impurity phases or Co metal clusters. The [111]C and [001]C HRTEM images and the corresponding Fourier transform (FT) patterns of individual nanoparticles show a defect-free crystal structure for all Bnwtt samples. They are faceted along the main crystallographic directions ([110]C and [010]C ) and free from any possible inclusions. However, close inspection of individual particles reveals the presence of similar brighter regions within the particles mentioned above for sample B1w (figure 3(b)). In order to clarify the origin of these contrast variations, EFTEM and Z -contrast imaging in STEM mode have been employed. This is demonstrated in figures 5(a)–(f) for the case of sample B1w. The EFTEM Ti map (figure 5(a)) confirms a narrow Ti distribution inside the B1w nanocrystals while the Z -contrast imaging (figure 5(b)) indicates that the bright spots observed in HRTEM correspond to pore-like structures in the nanocrystallites, and are not related to any chemical inhomogeneity or metal clustering effects. Figure 5(c) shows a HRTEM micrograph of the two 5

Nanotechnology 23 (2012) 025702

S Ray et al

Figure 3. (a) [111]C HRTEM image of a Co-doped BaTiO3 nanoparticle in sample B1w. (b) Filtered HRTEM image from selected region marked in (a) by a white frame. Notice the variation of the contrast within the nanoparticle. (c) FT pattern from the B1w nanoparticles in (a). (d) [111]C HRTEM image of a defective Co-doped BaTiO3 particle of B1w sample and corresponding FT pattern given as an inset. (e) Corresponding filtered HRTEM image showing the presence of core dislocations. The associated Burgers circuits are indicated.

a

c

b

Figure 4. HRTEM images along the [111]C , [001]C and [111]C for the B1wtt (a), B2wtt (b) and B3wtt (c) Co-doped BaTiO3 nanocrystals, respectively. The FT patterns are shown as an inset.

probable conclusion is that physical voids are present within the particles. This is consistent with earlier reports [21, 23] where the appearance of such porous structures of BaTiO3 nanoparticles and their origin have been described. In order to further investigate the chemical composition of the nanoparticles, detailed EELS measurements were also

particles from sample B1w, whereas figure 5(d) shows the corresponding plasmon-loss image and figures 5(e) and (f) are elemental maps of Co and Ti, respectively. The elemental maps and the plasmon-loss image reveal that the observed spots with brighter contrast in the HRTEM image do not correlate with increased concentrations of Ti or Co. Therefore, the most 6

Nanotechnology 23 (2012) 025702

S Ray et al

Figure 5. (a) EFTEM Ti mapping and (b) low and high (inset) magnification Z -contrast images of the same area in the STEM mode from sample B1w. (c) Zero-loss TEM image of two separated nanoparticles from sample B1w exhibiting structural inhomogeneities and corresponding EFTEM images at the plasmon loss (25 eV energy loss) (d) Co M2,3 edge (62 eV energy loss) (e) and Ti M2,3 edge (45 eV energy loss) (f).

performed. Unfortunately, the presence of Co dopants cannot be derived directly from the core-loss spectra as the most intense Co L2,3 -edge at 779 eV is known to appear at the same energy loss as the Ba L4,5 -edge (figure 6(c)). Therefore, to detect the doped Co atoms in the material matrix a lowloss EELS spectrum was acquired from several particles simultaneously using EELS in diffraction mode. This lowloss spectrum (figure 6(a)) shows the Co M2,3 -edge at 62 eV, confirming the presence of Co atoms inside the nanoparticles. However, owing to the overlap of the Co L2,3 edge (2p core level) and the Ba M4,5 edge (3d core level) and the low Co concentration in the particles, it is probable that no spectroscopy technique like EELS, x-ray photoelectron spectroscopy (XPS) or x-ray absorption spectroscopy (XAS)

will be effective in determining the oxidation state of Co within this system. To rule out any error in the compositional analysis due to possible diffraction contrast in the Z -contrast or EFTEM images, detailed EELS measurements were also performed in scanning mode (STEM-EELS). These measurements should confirm the chemical composition of the dark regions in the Z -contrast images (bright regions in the HRTEM images). To this end an EELS line scan was performed on a typical particle showing a dark region in Z -contrast (inset to figure 6(b)). EELS spectra were acquired at an interspacing of 1 nm, over a scan length of 20 nm (indicated by the dashed line). Three core-loss spectra acquired at representative positions A, B and C of the scan are shown in figure 6(c). All the spectra show 7

Nanotechnology 23 (2012) 025702

S Ray et al

Along with the peak strengths, the Z -contrast image intensity was also plotted. The peak strengths all show an intensity drop at the position of the dark contrast in the Z -contrast image. No notable shift in elemental composition can be determined. The drop in intensity and peak strength at the position of the dark image contrast can therefore be attributed to the presence of less material at this position, confirming earlier suggestions that the dark contrast in the Z -contrast images represents physical voids and not metallic clusters or other chemical inhomogeneities. Thus, our comprehensive electron microscopy characterization of the samples confirms intrinsic doping of Co ions inside the BaTiO3 nanocrystals and also reveals the formation and annihilation of different defect components as a result of various post-synthesis treatments, described above. Finally, detailed magnetic studies were carried out on all samples using a superconducting quantum interference device (SQUID) magnetometer. The room temperature magnetization versus magnetic field ( M(H )) dependence is the most important feature for checking the high temperature ferromagnetism in a dilute magnetic system, and those from our nanocrystals are shown in figure 7. In the upper three panels, results from samples B1wtt, B2wtt and B3wtt are shown. Evidently, room temperature ferromagnetism is found to develop regularly by moving from B3wtt toward B1wtt. The corresponding low field region data are shown as insets, exhibiting the development of weak, but finite, ferromagnetic hysteresis loops for B1wtt and B2wtt (the magnetic coercivities being 90 Oe and 45 Oe for B1wtt and B2wtt, respectively). In contrast, B3wtt behaves as a perfect paramagnet (also confirmed by the susceptibility data, not shown here) while decreasing the initial drying time during synthesis (see table 1) seems to support ferromagnetic interaction in these samples along with a persistent paramagnetic-like background. It should be emphasized that no other extrinsic, experimental parameters can be responsible for this systematic growth of ferromagnetism from one sample to the other since all the samples described here are synthesized from the same starting materials and were also handled in identical manner during synthesis and the magnetic measurement process. It is also to be noted that surface oxygen vacancy driven ferromagnetism, discussed in the context of pure BaTiO3 nanoparticles [9, 10], does not appear to be operative here because in such a case all the samples would have shown room temperature ferromagnetism. The surface nonstoichiometry discussed in [9] originates from high temperature annealing of the bare particles, while no such high temperature treatments have been carried out in our sample. While our samples do contain oxygen vacancies at the core this vacancy creation occurs via a completely different route, where a couple of hydroxyl ions migrate to each other during annealing and create a molecule of water and an anion vacancy. Further, a significant number of these anion vacancies get annihilated by collaborating with available cation vacancies, and as a result physical voids are created in the samples. Therefore, the vacancy distribution and the intrinsic mechanism of ferromagnetism is quite different in the two cases. In the present case, connecting

Figure 6. (a) Low-loss EELS spectrum indicating the Co doping by the Co M2,3 -edge at 62 eV. (b) Z -contrast image intensity, Ti, O and Ba-peak strengths as a function of probe position. Inset: Z -contrast image of scanned particle, the line scan is indicated by the dashed line. (c) Core-loss EELS spectra from probe positions A, B and C indicated in (b). All peak intensities drop in the dark region B, indicating a physical void. The peak positions of Ba (781 eV) and Co (779 eV) overlap, making the determination of the Co valency using the Co L2,3 -edge impossible.

typical core-loss edges for Ti (L2,3 -edge, 455 eV), O (Kedge, 532 eV) and Ba (M4,5 -edge, 781 eV). Using modelbased EELS data analysis [34], the normalized Ti, O and Ba peak strengths were plotted over the scan (figure 6(b)). 8

Nanotechnology 23 (2012) 025702

S Ray et al

Figure 7. (a)–(c) Magnetization ( M ) versus field ( H ) data from B1wtt, B2wtt, B3wtt samples at 300 K. The inset shows expanded views around zero field. Panels (d)–(f) show M versus H data from B1, B1w, and B1wtt samples at 300 K.

develops only after the post-washing calcination. This observation further reveals the importance of reaction (2) in developing ferromagnetism and establishes the oxygen vacancy defects as an essential ingredient of this unusual magnetic order. A similar trend is observed for the B2 series of samples, albeit to a lesser extent, while the B3 series of samples remained paramagnetic throughout. It is also important to mention here that the FC and ZFC M(T ) curves from the Bnwtt sample at different fields do not diverge at any temperature down to 2 K (not shown here) and susceptibility curves appear paramagnetic like (commonly observed in DMS systems [35, 36]) although the reverse susceptibility shows that

the SQUID and the FT-IR data, it can be concluded that the development of ferromagnetism is intricately connected with the crystalline defect content and donor defect distribution within the nanocrystallites. Another interesting development is observed in general for all these ferromagnetic samples and is described representatively in the bottom panels of figure 7 for the B1 series of samples, since the B1wtt sample is found to be the most strongly magnetic one. Room temperature M(H ) curves from samples B1 (figure 7(d)), B1w (figure 7(e)) and B1wtt (figure 7(f)) indicate that the as-synthesized sample does not exhibit any ferromagnetism, while distinct ferromagnetism 9

Nanotechnology 23 (2012) 025702

S Ray et al

the system does not exhibit any Curie–Weiss-like behavior. It was shown earlier that the internal defect content and, most importantly, the O–H defect locations are quite different in the B1, B2 and B3 series of samples. Therefore, it may happen that the reaction (2) at elevated temperature works most effectively for the B1wtt samples with a high density of O–H defects in the crystallographically distorted core and the resultant V•• (O) density in the interior is also higher for this sample. The oxygen contents calculated from ICP-OES experiments support such a description (table 1). Finally, the dopant Co ions start interacting ferromagnetically via the closely placed V•• (O) defects, an interaction which gradually weakens in crystallites with fewer defects and fewer anion vacancies at the core. Therefore, our results conclusively show that Co-doped BaTiO3 nanocrystals themselves do not exhibit high temperature ferromagnetism, while close proximity of V•• (O) donor defects is necessary for such ferromagnetism to develop.

[3] Liao Y-F, Huang T-W, Lin M-Z, Yu K-L, Hsu H-S, Lee T-H, Lee C-H and Huang J C A 2007 J. Magn. Magn. Mater. 310 e818–20 [4] Wang J, Cai J, Lin Y-H and Nan C-W 2005 Appl. Phys. Lett. 87 202501 [5] Schwartz D A and Gamelin D R 2004 Adv. Mater. 16 2115–9 [6] Kittilstved K R, Norberg N S and Gamelin D R 2005 Phys. Rev. Lett. 94 147209 [7] Poddar P, Sahoo Y, Srikanth H and Prasad P N 2005 Appl. Phys. Lett. 87 062506 [8] Janisch R, Gopal P and Spaldin N A 2005 J. Phys.: Condens. Matter 17 R657–89 [9] Mangalam R V K, Ray N, Waghmare U V, Sundaresan A and Rao C N R 2009 Solid State Commun. 149 1–5 [10] Sundaresan A and Rao C N R 2009 Solid State Commun. 149 1197–200 [11] Coey J M D and Chambers S A 2008 MRS Bull. 33 1053–8 [12] Matsumoto Y, Murakami M, Shono T, Hasegawa T, Fukumura T, Kawasaki M, Ahmet P, Chikyow T, Koshihara S and Koinuma H 2001 Science 291 854–6 [13] Seipel B, Erni R, Gupta A, Li C, Owens F J, Rao K V, Browning N D and Moeck P 2007 J. Mater. Res. 22 1396–405 [14] Ray S, Mahadevan P, Mandal S, Krishnakumar S R, Kuroda C S, Sasaki T, Taniyama T and Itoh M 2008 Phys. Rev. B 77 104416 [15] Chakraborty T, Ray S and Itoh M 2011 Phys. Rev. B 83 144407 [16] Keith G M, Sarma K, Alford N M and Sinclair D C 2004 J. Electroceram. 13 305–9 [17] Vanderah T A, Loezos J M and Roth R S 1996 J. Solid State Chem. 121 38–50 [18] Grey I E, Christina L, Cranswick L M D, Roth R S and Vanderah T A 1998 J. Solid State Chem. 135 312–21 [19] Ohno H 1998 Science 281 951–6 [20] Ray S, Kolen’ko Yu V, Fu D, Gallage R, Sakamoto N, Watanabe T, Yoshimura M and Itoh M 2006 Small 2 1427–31 [21] Kolen’ko Yu V, Kovnir K A, Neira I S, Taniguchi T, Ishigaki T, Watanabe T, Sakamoto N and Yoshimura M 2007 J. Phys. Chem. C 111 7306–18 [22] Akselrud L G, Zavalij P Y, Grin Y N, Pecharsky V K, Baumgarthner B and W¨olfel E 1993 Mater. Sci. Forum 133–136 335–42 [23] Hennings D F K, Metzmatcher C and Schreinematcher B S 2001 J. Am. Ceram. Soc. 84 179–82 [24] Blanco-Lopez M C, Rand B and Riley F L 1997 J. Eur. Ceram. Soc. 17 281–7 [25] Lide R D 2009 Handbook of Chemistry and Physics (Florida: CRC Press) pp 4–49 [26] Blanco-Lopez M C, Fourlaris G and Riley F L 1998 J. Eur. Ceram. Soc. 18 2183–92 [27] Busca G, Buscaglia V, Leoni M and Nanni P 1994 Chem. Mater. 6 955–61 [28] Hennings D, Rosenstein G and Schreinemacher H 1991 J. Eur. Ceram. Soc. 8 107–15 [29] Hennings D and Schreinemacher S 1992 J. Eur. Ceram. Soc. 9 41–6 [30] Chen L, Shen Y, Xie A, Zhu J, Wu Z and Yang L 2007 Cryst. Res. Technol. 42 886–9 [31] Sharma P and Virk H S 2009 Open Surf. Sci. J. 1 34–9 [32] Erwin S C, Zu L, Haftel M I, Efros A L, Kennedy T A and Norris D J 2005 Nature 436 91–4 [33] Dalpian G M and Chelikowsky J R 2006 Phys. Rev. Lett. 96 226802 [34] Verbeeck J and Van Aert S 2004 Ultramicroscopy 101 207–24 [35] Kassaian A and Berciu M 2005 Phys. Rev. B 71 125203 [36] Das Sarma S, Hwang E W and Kaminski A 2003 Phys. Rev. B 67 155201

4. Conclusions In summary, we have synthesized a series of Co-doped BaTiO3 nanocrystalline samples using a novel solvothermal drying route, with varying solvothermal drying times. Detailed characterization has proved the intrinsic doping of Co ions in the nanocrystals and also the nature of various chemical processes, and changes of the fine microstructures have been probed. It is observed that the as-grown nanocrystals are not ferromagnetic in nature, while the presence of larger numbers of bound oxygen vacancy defects at the core, as a result of post-synthesis treatments, helps the dopant ions to interact ferromagnetically, and in some extreme cases room temperature ferromagnetism develops. Detailed theoretical analysis is needed to understand the critical interaction parameters and their effects in realizing ferromagnetic interaction in such systems.

Acknowledgments SR thanks DST Fast Track and DST-RFBR schemes for financial support. We thank Dr G Auffermann for chemical analysis and MPI CPfS Kompetenzgruppe Struktur for help in x-ray data collection. We also thank Dr P Mahadevan for useful discussions. YVK thanks the JSPS Postdoctoral Fellowships for Foreign Researchers. We acknowledge support from the European Union under the Framework 6 program under a contract from an Integrated Infrastructure Initiative (reference 026019 ESTEEM). ST gratefully acknowledges financial support from the Fund for Scientific Research Flanders (FWO).

References [1] Wolf S A, Awschalom D D, Buhrman R A, von Moln´ar S, Roukes M L, Chtchelkanova A Y and Treger D M 2001 Science 294 1488–95 [2] Schwartz D A, Norberg N S, Nguyen Q P, Parker J M and Gamelin D R 2003 J. Am. Chem. Soc. 125 13205–18

10