Surface ferromagnetism in Pr0.5Ca0.5MnO3

Cite as Haruna et al., Chem Mater 30, 7138-7145 (2018)

Surface ferromagnetism in Pr0.5Ca0.5MnO3 nanoparticles as a consequence of local imbalance Mn3+:Mn4+ ratio. Haruna D Aliyu,1,2 Jose M Alonso,1,3 Patricia de la Presa,*,1,4 Walmir E Pottker, †1 Benedict Ita,ǂ2 Mar Garcia-Hernández,3 Antonio Hernando1,4 1

Instituto de Magnetismo Aplicado, UCM-ADIF-CSIC, 28260 Las Rozas, Spain

2

Department of Chemistry, University of Abuja, Nigeria

3

Instituto de Ciencia de Materiales, CSIC, 28049-Madrid, Spain

4

Dto Física de Materiales, Univ. Complutense de Madrid, 28040 Madrid, Spain

ABSTRACT Half-doped praseodymium manganites, Pr0.5Ca0.5MnO3, nanoparticles synthesized by sol-gel method produce very crystalline nanoparticles with no structural disorder at the surface. As the most of half-doped (Mn3+:Mn4+ = 1) antiferromagnetic (AFM) chargeordered manganites, ferromagnetism (FM) appears as particle size decreases. A possible origin of FM phase development is the lacking oxygen ligands at the surface that increases Mn3+ at expenses of Mn4+ in order to maintain the electronic neutrality. In this work, we show that Mn3+:Mn4+ = 1 ratio is preserved in the whole crystalline particle with the exception of the surface, where this ratio changes to Mn3+:Mn4+ >1. This Mn3+ excess makes double exchange Mn3+O2-Mn4+ prevails over the super-exchange giving place to the FM interactions. We show that a “shell” thickness of only one-unit cell, with acell = 0.38 nm, is enough to explain the onset of FM at the surface, whereas the volume remains AFM. Furthermore, the FM to AFM ratio fits to the increase of surface to volume ratio with decreasing particle size.

1


INTRODUCTION In the last years, after numerous publications on the influence of the particle size on the magnetic and transport properties of manganites nanoparticles, as the vanishing of charge ordering (CO) or the onset of ferromagnetism (FM) in an antiferromagnetic (AFM) phase, etc, the physical phenomena that provoke these changes are still under discussion. Some authors propose that, in the case of manganites nanoparticles with FM ground state as LaxSr1-xMnO3, a shell of 2-3 nm thickness -formed by structural defects as cationic vacancies, oxygen defect or excess, loss of stoichiometry, etc- can cause either this shell to be a non-magnetic dead layer that weakens the FM interaction of the core or superexchange prevails over double-exchange and nanoparticle surface becomes AFM.1-4 In the case of manganites with AFM ground state, as Ln0.5Ca0.5MnO3 with Ln = La, Pr, Nd, different explanations are under discussion. Some authors suggest that the particle size reduction gives place to an induced hydrostatic pressure that promotes the FM interactions and hinders CO.5-6 However, others propose that, following the theoretical model by Dong et al.,7 the increase of the surface charge density is the responsible of the FM interactions and CO frustration.8 On the other hand, the works from Jirak et al.9 and Shankar and Sing10 in 24 nm Pr0.5Ca0.5MnO3 make a different interpretation: Jirak et al.9 consider that the magnetic state of the samples is a mix of a small FM metallic fraction with another, majority, insulating phase which presents charge and orbital disorder with frozen spins. Shankar and Singh10 propose that the increase of the cell volume originated by the reduction of the particle size to the nanoescale gives place to a bandwidth increase that vanishes the CO and onsets the FM ordering.

2


In order to understand the origin of FM order in half-doped manganite, we propose to study Pr0.5Ca0.5MnO3 because it is purely AFM in bulk, unlike La0.5Ca0.5MnO3 where a FM contribution is always present and it is difficult to distinguish the FM order of the materials itself from that coming from size effects. A systematic investigation of the physical properties of nanocrystals of half-doped Pr0.5Ca0.5MnO3 with particle sizes ranging from 16 to 50 nm is performed. The HRTEM shows the particles are very crystalline even close to the surface, no evidence of amorphous or disordered shell is observed. The magnetic results show a noticeable increase of FM like behavior for 15 nm nanoparticles that decreases with increasing particle size. The FM phase can be estimated, as proposed by Dong et al.7, by an extra electron (eg) per Mn3+ in the first cells at the surface that provokes Mn3+: Mn4+>1; this charge imbalance is compensated in the volume where the ratio Mn3+ to Mn4+ is 49.6% to 50.4%, i.e, Mn3+:Mn4+ ~1 and, consequently, the volume remains AFM. This is confirmed by the measuring FM to AFM fractions and comparing them with the surface to volume ratio with increasing particle size. Additionally, hysteresis cycles under field cooled procedure show an increase of the coercivity and a shift to negative fields, indicating that both magnetic phases are coupled and the applied field provokes FM shell increases from 1 to 3 unit cells. EXPERIMENTAL PROCEDURE Single phase nanoparticles are synthesized by sol-gel method by adding 6.25 mM Pr6O11 to 250 ml of acidified distilled water with 5 ml nitric acid, heated at 50o C, and vigorously stirred for 45 min to form a clear solution; thereafter, 10 g of citric acid is added. MnCO3 (25 mM) is added under vigorously stirring until a colourless solution is obtained. CaCO3 (12.5 mM) and 10 ml of ethylene glycol (Eg) are successively incorporated. The solution was evaporated to dryness, after several hours at 90oC, 3


forming a dark brown powder. Finally, the powder is milled in agate mortar and then heated at 600 ºC during 24 h to remove the organic materials. Four different samples are prepared by subsequent thermal treatments for 2 h at different annealing temperatures: 700 ºC (PC700), 800 ºC (PC800), 900 ºC (PC900) and 1000 ºC (PC1000). Structural characterization and average crystallite size are measured by X-ray diffraction (XRD) using Cu-Kα radiation in a PANalytical X'Pert Pro MPD diffractometer. The mean crystallite sizes are calculated from Scherrer's formula. The XRD patterns are analyzed by the Rietvelt method, using FullProf program. For samples PC700 and PC1000, thermodiffraction pattern are taken from 50 to 250 K (by step of 5 K), 275 K and 300 K, on a PANalytical X’Pert Pro MPD diffractometer equipped with the X’Celerator detector and an Oxford Phenix cryostat. Particles size and shape are determined by transmission electron microscopy (TEM) by using a JEOL JEM 2100F operated at 200 keV and JEOL 3000 FEG operated at 300 keV. In addition, an elemental analysis by energy dispersive spectroscopy (EDS) is performed in each sample by means of JEOL JSM 6400. The anionic composition is determined by means of thermogravimetric analysis in a Cahn D-200 electrobalance. The magnetic characterizations are carried out on powder samples by means of a Quantum Design SQUID magnetometer. Hysteresis loops, including virgin curves, are measured at 5 T at 2, 5, 25, 50, 100, 150, 200 and 300 K. Field-cooled (FC) hysteresis loops are measured at 5 T cooling field. Zero field-cooled and field-cooled curves (ZFC-FC) are obtained from 5 to 350 K at 50 and 1000 Oe applied magnetic field. A 3 T demagnetizing field is applied at room temperature before ZFC measurements in order to avoid any remnant in the superconducting magnets. RESULTS 4


EDS analyses in sample PC800 show an average composition Pr0.48(4)Ca0.48(4)Mn1.04(4)Oy that, taking into account the experimental errors, fits pretty well to the nominal one. Since all the samples are made starting from the same nanoparticles batch, they all have same stoichiometry and carrier concentration irrespectively of the size. The oxygen content, precisely determined by thermogravimetric analysis, indicates that the anionic sublattice

is

complete,

thus,

final

composition

of

the

samples

is

Pr0.48(4)Ca0.48(4)Mn1.04(4)O3.00(1). XRD patterns of Pr0.5Ca0.5MnO3 nanoparticles recorded at room temperature (fig. 1) show single phase well-defined pattern; by means of Rietveld powder diffraction profile it is determined that the four samples have orthorhombic structure with space group Pnma.11 The particles synthesized at the lowest temperature (PC700) show the broadest diffraction peaks and they become thinner as the annealing temperature increases. The crystallite size, calculated by the Scherrer formula by means of the FWHM at the (2 0 0) diffraction peak, smoothly increases from 16 to 50 nm with increasing annealing temperature.

5


Figure 1: XRD patterns and Rietveld fits of the samples PC700, PC800, PC900 and PC1000. The red points are the experimental data, the black lines the fits, and the blue lines are the differences between fits and experimental data.

The results of the Rietveld refinement for all samples are summarized in table 1. The low values of the reliability factors represent the goodness fits. Pr/Ca atoms are located at 4c (x, ¼, z), Mn atoms at 4a (0, 0, 0) and oxygen atoms occupy two different sites, namely O(1) at 4c (x, y, ¼) and O(2) at 8d (x,y,z) positions. Table 1: Refined structural parameters obtained from the Rietveld refinement of room temperature of Pr0.5Ca0.5MnO3 annealed at 700 °C, 800 °C, 900 °C and 1000 °C.

Parameters

700 °C

800 °C

900 °C

1000 °C 6


a (Å)

5.428(3)

5.421(1)

5.408(7)

5.404(7)

b (Å)

7.616(2)

7.602(1)

7.604(6)

7.607(6)

c (Å)

5.415(3)

5.386(9)

5.389(9)

5.395(6)

221.959(1)

221.609(4)

221.779(4)

V (Å3)

223.854(2)

Mn-O(1) (Å)

1.942(6)

1.939(3)

1.949(4)

1.941(1)

Mn-O(2) (Å)

1.877(3)

1.849(3)

1.792(1)

1.850(2)

Mn-O(2)’(Å)

2.014(3)

2.034(3)

2.079(1)

2.032(2)

Mn-O(1)-Mn (°)

157.192(3)

151.592(1)

154.500(2)

157.062(3)

Mn-O(2)-Mn (°)

160.389(4)

163.581(1)

161.068(1)

159.377(3)



1.28

1.26

1.58

1.59

Fig. 2 shows the variation of the cell parameter and volume as a function of the annealing temperature. The PC700 sample presents the highest cell parameters and, consequently, the largest volume. The cell parameters a, b and c show a different behavior with the thermal annealing. Whereas b and c take minima values at 800 ºC and then increase up to 1000 ºC, a decreases monotonously up to reaching the bulk value at 1000 ºC. As a result, the cell volume takes a minimum at 900 ºC and reach the bulk value at 1000 ºC. This behavior is similar to that observer by Shankar and Kumar10, but opposite to that observed by Zhang et al.8 for similar particle size but smaller cell volume.

7


Figure 2: Cell parameters and volume obtained by Rietveld for nanoparticles annealed at different temperatures and the bulk material.

As previously reported, the thermal dependence of the crystal parameters in Ln0.5Ca0.5MnO3 in bulk is governed by the charge localization in Mn3+/Mn4+ ions, origin of the strong lattice-orbital coupling that occurs in the long-range ordered pattern, known as CO.12-13 The onset of CO is accompanied by a change in magnetic interactions from FM to AFM. Pr0.5Ca0.5MnO3 is a paramagnetic insulator that, upon cooling, suffers a CO transition at TCO=225 K followed by an AFM transition at TN=155 K.9 It is well-known that the thermal dependence of the crystal parameters varies significantly between these two temperatures.9, 14-16 The evolution of the lattice parameters of PC700 and PC1000 samples was determined by means of XRD in the 50 K ≤ T ≤ 300 K range (Fig. 3). The thermal behavior of a, b and c in PC1000 resembles the behavior of the crystal parameters in bulk,14 thus, a significant change takes place around the transition from CO to AFM phase. On the other hand, the thermal dependence of the crystal parameters for PC700 has an almost lineal behavior in the same thermal range. This is in concordance with the loss of the CO for the smallest particles observed by magnetic characterization, as it will be shown later.

8


Figure 3: Thermal dependence of the cell parameters for samples PC800 (top) and PC1000 (bottom).

The mean particle size measured by TEM increases from 20 to 100 nm (see Fig. 4). The TEM images show the nanoparticles are often interconnected by narrow bridges. The discrepancies at large crystallite size from XRD and TEM characterizations lay in the Scherrer formulation: the indetermination of crystallite size increases with the increasing size. It is worth noting that even when the distributions have maximum at 20 and 103 nm for PC800 and PC1000, respectively, the most frequent particle sizes are 23 and 85 nm, respectively.

9


Figure 4: Mean particle size for samples PC800 (top, left) and PC1000 (bottom, left) and TEM image for PC800 (top right) and PC1000 (bottom right). The mean particle size has been fitted with a Gaussian distribution.

Figure 5 shows HRTEM image of a single PC800 nanoparticle. As can be observed, the particle is crystalline up to the surface; there is no evidence of a core-shell structure or an amorphous layer at the surface. The interplanar space measured at PC800 is 2.71 Å close to the value 2.73 Å corresponding to the (0 0 2) plane. More HRTEM images at different augments are shown in supporting information for PC700, PC800 and PC1000.

10


Figure 5: HRTEM image for PC800. It illustrates the high crystallinity of the sample; no disordered shell is observed at the surface.

The dependence with the annealing temperature of the crystallite size calculated by Scherrer and the cell volume is shown on Fig. 6, it is observed the particle size smoothly increases from 16 to 50 nm with increasing annealing temperature, whereas the cell volume decreases.

Figure 6: Mean crystallite size calculated by Scherrer´s equation (black points, left axis) and cell volume as a function of annealing temperature (red points, right axis).

11


The thermal dependence of the magnetization with annealing temperature is shown in Fig. 7 for 50 Oe applied field. The first result to highlight is the character of the irreversibility of the ZFC-FC curves on the low temperature region (T < 150 K), suggesting a highly non-uniform magnetic state. For PC1000 sample, the maximum observed at T250 K can be assigned to the onset of CO (see insets in Fig. 7D). As can be seen from insets Fig. 7, the intensity of this maximum decreases with the particle size, to finally disappear for the smallest particles PC700. Furthermore, the CO onset Temperature, TCO, shifts to lower temperatures as particle size decreases, from 255 K in bulk to 242 K for PC800.

Figure 7: ZFC (full circles) and FC (open circles) curves under 50 Oe for samples annealed at 700 (A), 800 (B), 900 (C) and 1000 ºC (D). The insets show the ZFC results.

At low temperature, the maximum of the ZFC curves at 50 K suggests a superparamagnetic or glassy behavior of the nanoparticles which is confirmed by

12


measuring ZFC-FC curves at different applied fields. The ZFC curves of PC1000 sample at 50 and 1000 Oe show that the maximum shifts to lower temperature as field increases (Fig. 8), as expected for a blocking temperature, whereas the Tco is not affected by the applied field. The peak at 40 K in the ZFC curves is observed only in PC1000 and it is also independent of applied field. This behavior has been ascribed to the reentrant spin glass and the temperature as the spin freezing temperature Tf.8, 17 The onset of reentrant spin glass suggests that the spins frozen can take place only for particle size larger than 40 nm.

Figure 8: Thermal dependence of the magnetization under ZFC for PC1000 by applying 50 Oe (red points, right axis) and 1000 Oe (blue points, left axis). The vertical line points out the reentrant spin glass temperature.

Figure 9 shows the M vs H curves at 5 K under ZFC and FC procedures. Two different behaviors can be clearly observed. At low magnetic field, a sharp hysteresis loop, characteristic of FM materials, appears and this contribution decreases with increasing particles size. The magnetization at 5 T for the PC1000 reaches a value close to the bulk but still shows a weak FM contribution on the low field region; the TC of this FM contribution, determined by the thermal dependence of M under ZFC (Fig. 7), varies from 106 K for PC700 to 113 K for PC1000. The hysteresis loops do not reach

13


saturation, on the contrary, the magnetization changes linearly with the applied field in the high-field region suggesting an AFM contribution in addition to the FM one. It`s worth noting that the loss of CO for small particles, as confirmed by ZFC-FC and DRX results, suggests the loss of the AFM CE type; however, as proposed by other authors, 5, 9, 16

the onset of another type of AFM order (A, C, etc.) is observed for particles smaller

than 100 nm. By measuring the hysteresis loops under 5 T cooling field, the magnetization at 5 T increases significantly for all samples excepting the bulk one (Fig. 9). The FM contribution of PC700, PC800, PC900 y PC1000 can be obtained by subtracting the linear behaviour of the magnetisation at high-field range (20 kOe  H  50 kOe) to the loops. Table 2 shows the saturation magnetization of the FM contribution for all the samples obtained by ZFC and FC procedures.

Figure 9: Hysteresis loops nanoparticles annealed at different temperature and bulk samples. Left: hysteresis cycles under ZFC procedure. Right: hysteresis cycles under 5 T FC procedures.

14


SAMPLE

Ms (emu/g)

Hc (Oe)

Ms (emu/g)

HC (Oe)

HE (Oe)

ZFC

ZFC

FC

FC

FC

PC700

13.2

1220.5

27.9

1687

144.5

PC800

9.1

1119

28.6

1648

96

PC900

3.4

1080

13.3

1588

80.5

PC1000

1.2

1125.5

5.8

1569

120.5

Bulk

0.05

-

0.05

-

Table 2: Saturation magnetization (Ms) and coercive field (Hc) of the FM phases and exchange bias field (HE) for the nanoparticle and bulk samples measured under ZFC and FC under 5 T.

Considering that the magnetic moments of bulk Pr0.5Ca0.5MnO3 are fully ferromagnetically aligned at fields as high as 30 T, with a saturation magnetization MFM = 110 emu/g.18-20, it is possible to calculate the FM fraction in the different samples. These results are shown in Fig. 10.

15


Figure 10: Relationship of the FM contribution Ms in the NPs to the fully ferromagnetically aligned moment of the bulk material, MFM = 110 emu/g.

Hysteresis loops under FC procedure below the Neel temperature (TN) of the AFM phase (assuming TN > TC) determine if AFM and FM phase are coupled or not.21 The interaction between both magnetic fractions is clearly assessed by measuring exchange bias field (HE), which is characteristic of AFM regions in intimate contact with FM ones. As observed in Fig. 11 and Table 2, these samples show the fingerprint of an exchange bias: the coercivity increases and shifts to negative fields. Therefore, we can discard that the system is a mix of a small fraction of FM nanoparticles with a majority fraction of AFM ones. Our results suggest that both AFM and FM phases not only coexist but are also coupled.

16


Figure 11: Hysteresis loops for sample PC700 under ZFC and FC procedures

Finally, virgin magnetization curves were measured in PC800 at 2, 5, 25, 50, 100, 150, 200 and 300 K, as shown in Fig. 12. Virgin curves allow observing the onset of FM transitions as well as the coupling between FM and AFM phases. From the hysteresis cycles with virgin magnetization curves it observed that, at low temperatures (T = 2 and 5 K), only the moments at the surface are aligned FM with the field. As the temperature increases (T = 25 and 50 K), the particle surface orders FM for H < 30 kOe, but for higher fields (H > 30 kOe) there appears a metamagnetic transition due to the FM alignment of the AFM core, and this transition is irreversible. For T > 100 K, wherein the FM ordering of the surface vanishes as shown in ZFC-FC curves (Fig. 7), the metamagnetic transition also disappears.

17


Figure 12: Hysteresis loops and virgin magnetization curves for PC800 at 2, 5, 25, 50, 100, 150, 200 and 300 K.

Discussion It is well known that magnetic behavior of FM Ln2/3A1/3MnO3 (or similar) or AFM Ln0.5A0.5MnO3 change dramatically as particle size decreases down to few nanometers. In order to explain these changes, a core-shell model has been proposed by different authors according to whom the properties of the core is similar to the bulk but the shell behaves quite different.1-3,

6, 22

Usually, a structural disorder of 2-3 nm thick is

associated to the shell inducing a different magnetic behavior regarding to the core.1-3 We show in this work that the presence of broken links at the surface, as a consequence lacking symmetry and lower coordination, is enough to explain qualitative and quantitatively the singular magnetic behavior of Pr0.5Ca0.5MnO3 nanoparticles.

18


Let us assume a half doped manganese perovskite Ln0.5A0.5MnO3 with pseudocubic cell parameter acell  0.39 nm and cell volume Vcell  0.0593 nm3. The volume of a spherical particle with rp = 10 nm is Vp = 4188.8 nm3 and the number of unit cells per particle is Vp/Vcell  70637. Assuming a core-shell model with a shell of a cell unit thick, ts  acell , the radio of the core is rc  rp  ts  9.61 nm and the core volume Vc  3717.6 nm3 (see Scheme 1). The number of unit cells in the core is given by Nc  Vc / Vcell  62691 and the number of unit cells at the surface is N s  N p  Nc  7946 . Therefore, 11.25% of the unit cells of the particle are at the surface.

Scheme 1: Nanoparticle of radius rp, core radius rc and shell thickness ts.

At the surface, the Mn can be not completely coordinated by oxygen leading to a change in the Mn oxidation state in order to maintain the electronic balance. Consequently, the ratio Mn3+/Mn4+=1 is not energetically favorable at the surface and makes Mn3+ increases at expense of Mn4+, thus leading to Mn3+/Mn4+>1. Recently, N´Goc et al. reported a variation of O2- content and oxidation state of Mn at the surface of La2/3Sr1/3MnO3 nanoparticles.23 Assuming that, as proposed by Dong et al.7, there exists an extra 0.25 electron (eg) per Mn3+ position at the surface, then the surface has a 25% more Mn3+ than Mn4+ favoring the double-exchange to the super-exchange and allowing the onset of FM order at the shell. Consequently, the nanoparticles are conformed by an AFM core and FM shell of 19


one-unit cell thickness. However, although this excess is always present, the surface is negligible for bulk materials or single crystals and the behavior of the bulk prevails over the surface phenomena. Therefore, this effect is only observable as the particle size decreases. The FM fraction can be calculated from the particle size and cell volume. Fig. 13 shows the experimental and calculated FM fraction. For the sample PC1000, the particle size is calculated from TEM results. Qualitatively, the calculation and experimental results follow the same behavior. Quantitatively, the values do not match exactly to each other, which does not surprise considering that the calculation has been done with a mean particle size without examining the size distribution; however, the FM fraction fits pretty well to the Vs/Vp ratio.

Indeed, let us consider PC800 with d = 22 nm (rp =11 nm) as an example. This material has 8.6% of FM fraction for a total volume of 5575 nm3, thus, the FM volume is VFM = 5575 x 0.086 ≈ 479 nm3. For the core-shell model of Scheme 1, the core has a volume of 5575 - 479 ≈ 5096 nm3 with a core radius of rc ≈ 10.7 nm. The shell thickness is thus calculated as ts = rp – rc ≈ 0.3 nm, which is slightly smaller than the cell parameter for this material acell ≈ 0.38 nm. Similar calculations can be made for the rest of the samples, all with similar results: the FM to AFM ratio fits to the surface to volume ratio when the surface thickness is a single unit cell with Mn3+:Mn4+ >1.

20

Cite as Haruna et al., Chem Mater 30, 7138-7145 (2018) Figure 13. Ferromagnetic fraction (red points, left axis) and surface volume (Vs) to particle volume (Vp) ratio (blue, right axis). The inset shows the dependence of Vs/Vp as a function of particle size, the dotted vertical lines indicate the region of our results.

The saturation magnetization obtained by extrapolating the FC hysteresis loops at high fields of the sample PC800 (see Fig. 10) is Ms = 29.5 emu/g, that corresponds to 27% of the FM fraction. Following the previous calculations, the surface thickness of the FM fraction under FC is ts = 1.1 nm, which corresponds to a “shell” of 3 unit cells (0.38 x 3 = 1.14 nm). This confirms that the part of the AFM core that orders FM is the closest one to the surface and the ferromagnetism of the last one stabilizes the metamagnetic transition, as observed in the virgin curves of Fig. 12. This also explains that the exchange bias observed for the sample PC700 (see Fig. 11) is a consequence of the coupling of this FM shell of 3 unit cells thickness with the AFM bulk. The spin glassy state observed below 150 K for all the samples (see fig. 7) is also in concordance with a high crystalline AFM particle with a FM surface thickness of single unit cell since a glassy state can be originated by magnetic frustration of the FM order. Therefore, these results show that the onset of FM in AFM half doped manganite nanoparticles is always present even when the nanoparticles are very crystalline, as in our case, and it is consequence of the lacking oxygen ligands at the surface. The presence of a structural disordered shell of 2-3 nm thickness is not a requirement for the observation of the FM order in the Ln0.5Ca0.5MnO3 nanoparticles.

Conclusions

21


Pr0.5Ca0.5MnO3 nanoparticles have been synthesized by sol-gel method and annealed at different temperatures to produce nanoparticles with sizes ranging from 16 to 50 nm. HRTEM shows highly crystalline particles with no evidence of surface disorder or shell. Rietveld powder diffraction profile determines that the four samples have orthorhombic structure with space group Pnma and cell volume decreases with increasing particle size. The thermodiffraction pattern, taken for the 16 and 102 nm NPs from 50 to 300 K, show that the strong lattice-orbital coupling, known as charge ordering, that occurs in the long-range ordered pattern is suppressed for the smallest particles, whereas the largest one still resembles the thermal dependence of the cell parameters in the bulk. This suppression is also confirmed in the ZFC-FC curves, where CO is observed only in the largest nanoparticles. The ZFC-FC curves also show the onset of FM interactions at low temperature, the maximum of the ZFC curves at 50 K suggests a superparamagnetic or glassy behavior, which is confirmed by the shift of the maximum at different applied fields. Hysteresis loops under FC procedure measured below the TN show a coercivity increase and a shift to negative field, these results confirm that both AFM and FM phases not only coexist but are also coupled. It is also shown that a shell of a one-unit cell thickness with Mn3+:Mn4+ >1 can explain the onset of FM interactions in the AFM charge-ordered CE manganites when particle size is reduced. Although this imbalance is always present, its effect is negligible for bulk materials; however, it becomes relevant as the surface to volume ratio increases as occurs at the nanoscale range. The FM to AFM ratio fits to the surface to volume ratio when considering that at the most external shell the Mn3+:M4+ >1. It is worth noting that this FM order is always present despite of the high crystallinity degree of the samples.

22


ASSOCIATED CONTENT Supporting Information: HRTEM images at different augments for samples PC700, PC800 and PC1000. This material is available free of charge via the Internet at http://pubs.acs.org. AUTHOR INFORMATION Corresponding author *P de la Presa: [email protected], to whom correspondence should be addressed ORCID: Patricia de la Presa: 0000-0002-9456-8320 Mar García Hernández: 0000-0001-8170-0794 Walmir E Pottker: 0000-0002-4126-1181 Benedict Ita: 0000-0001-8170-0794 Permanent Addresses †

W.E. Pottker: Universidade Tecnológica Federal do Paraná, Brazil

ǂ

B Ita: University of Calabar, Nigeria

Acknowledgment This work was supported by grants from the Spanish Ministry of Science and Innovation MAT2015- 67557-C2-1-P. Walmir E Pottker acknowledges the financial support from CNPq, Brazil, for one-year postdoctoral fellowship at the Instituto de Magnetismo Aplicado, UCM, Spain.

23


References 1. Curiale, J.; Granada, M.; Troiani, H. E.; Sánchez, R. D.; Leyva, A. G.; Levy, P.; Samwer, K., Magnetic dead layer in ferromagnetic manganite nanoparticles. Appl. Phys. Lett. 2009, 95, 043106. 2. López-Quintela, M. A.; Hueso, L. E.; Rivas, J.; Rivadulla, F., Intergranular magnetoresistance in nanomanganites. Nanotechnology 2003, 14, 212. 3. Zhang, T.; Wang, X. P.; Fang, Q. F.; Li, X. G., Magnetic and charge ordering in nanosized manganites. Appl. Phys. Rev. 2014, 1, 031302. 4. Žvátora, P.; Veverka, M.; Veverka, P.; Knížek, K.; Závěta, K.; Pollert, E.; Král, V.; Goglio, G.; Duguet, E.; Kaman, O., Influence of surface and finite size effects on the structural and magnetic properties of nanocrystalline lanthanum strontium perovskite manganites. J. Solid State Chem. 2013, 204, 373-379. 5. Sarkar, T.; Ghosh, B.; Raychaudhuri, A. K.; Chatterji, T., Crystal structure and physical properties of half-doped manganite nanocrystals of less than 100-nm size. Phys. Rev. B 2008, 77, 235112. 6. Iniama, G.; de la Presa, P.; Alonso, J. M.; Multigner, M.; Ita, B. I.; Cortés-Gil, R.; RuizGonzález, M. L.; Hernando, A.; Gonzalez-Calbet, J. M., Unexpected ferromagnetic ordering enhancement with crystallite size growth observed in La0.5Ca0.5MnO3 nanoparticles. J. Appl. Phys. 2014, 116, 113901. 7. Dong, S.; Yu, R.; Yunoki, S.; Liu, J. M.; Dagotto, E., Ferromagnetic tendency at the surface of CE-type charge-ordered manganites. Phys. Rev. B 2008, 78, 064414. 8. Zhang, T.; Dressel, M., Grain-size effects on the charge ordering and exchange bias in Pr0.5Ca0.5MnO3: The role of spin configuration. Phys. Rev. B 2009, 80, 014435. 9. Jirák, Z.; Hadová, E.; Kaman, O.; Knížek, K.; Maryško, M.; Pollert, E.; Dlouhá, M.; Vratislav, S., Ferromagnetism versus charge ordering in the Pr0.5Ca0.5MnO3 and La0.5Ca0.5MnO3 nanocrystals. Phys. Rev. B 2010, 81, 024403. 10. Shankar, U.; Singh, A. K., Origin of suppression of charge ordering transition in nanocrystalline Ln0.5Ca0.5MnO3 (Ln = La, Nd, Pr) ceramics. J. Phys. Chem. C 2015, 119, 2862028630. 11. Woodward, P. M.; Vogt, T.; Cox, D. E.; Arulraj, A.; Rao, C. N. R.; Karen, P.; Cheetham, A. K., Influence of cation size on the structural features of Ln1/2A1/2MnO3 perovskites at room temperature. Chem. Mater. 1998, 10, 3652-3665. 12. Millange, F.; Brion, S. d.; Chouteau, G., Charge, orbital, and magnetic order in Nd0.5Ca0.5MnO3. Phys. Rev. B 2000, 62, 5619-5626. 13. Jirák, Z.; Damay, F.; Hervieu, M.; Martin, C.; Raveau, B.; André, G.; Bourée, F., Magnetism and charge ordering in Pr0.5CaxSr(0.5-x)MnO3(x=0.09 and 0.5). Phys. Rev. B 2000, 61, 1181-1188. 14. Radaelli, P. G.; Cox, D. E.; Marezio, M.; Cheong, S. W., Charge, orbital, and magnetic ordering in La0.5Ca0.5MnO3. Phys. Rev. B 1997, 55, 3015-3023. 15. Tomioka, Y.; Asamitsu, A.; Kuwahara, H.; Moritomo, Y.; Tokura, Y., Magnetic-fieldinduced metal-insulator phenomena in Pr(1-x)Ca(x)MnO3 with controlled charge-ordering instability. Phys. Rev. B 1996, 53, R1689-R1692. 16. Levy, P.; Parisi, F.; Polla, G.; Vega, D.; Leyva, G.; Lanza, H.; Freitas, R. S.; Ghivelder, L., Controlled phase separation in La0.5Ca0.5MnO3. Phys. Rev. B 2000, 62, 6437-6441. 17. Cao, G.; Zhang, J.; Wang, S.; Yu, J.; Jing, C.; Cao, S.; Shen, X., Reentrant spin glass behavior in CE-type AFM Pr0.5Ca0.5MnO3 manganite. J. Magn. Magn. Mater 2006, 301, 147154. 18. Xiao, G.; McNiff, E. J.; Gong, G. Q.; Gupta, A.; Canedy, C. L.; Sun, J. Z., Magnetic-fieldinduced multiple electronic states in La0.5Ca0.5MnO3+\delta. Phys. Rev. B 1996, 54, 60736076. 24

Cite as Haruna et al., Chem Mater 30, 7138-7145 (2018) 19. Dash, S.; Kumar, K.; Banerjee, A.; Chaddah, P., Effect of simultaneous application of magnetic field and pressure on magnetic transitions in La0.5Ca0.5MnO3. Phys. Rev. B 2010, 82, 172412. 20. Tokunaga, M.; Miura, N.; Tomioka, Y.; Tokura, Y., High-magnetic-field study of the phase transitions of R{1-x}Ca{x}MnO3 (R=Pr, Nd). Phys. Rev. B 1998, 57, 5259-5264. 21. Nogués, J.; Sort, J.; Langlais, V.; Skumryev, V.; Suriñach, S.; Muñoz, J. S.; Baró, M. D., Exchange bias in nanostructures. Phys. Rep. 2005, 422, 65-117. 22. Dong, S.; Gao, F.; Wang, Z. Q.; Liu, J.-M.; Ren, Z. F., Surface phase separation in nanosized charge-ordered manganites. Appl. Phys. Lett. 2007, 90, 082508. 23. N'Goc, H. L. T., et al., Surface-driven magnetotransport in perovskite nanocrystals. Adv. Mater. 2017, 29, 10.

25