Multiferroic phase stability in non-stoichiometric MnWO
4H. W. Yu, X. Li, L. Li, M. F. Liu, Z. B. Yan, and J.-M. Liua)
Laboratory of Solid State Microstructures, Nanjing University, Nanjing 210093, China
(Presented 7 November 2013; received 14 September 2013; accepted 14 November 2013; published online 26 February 2014)
We investigate the multiferroic phase stability of MnWO4in response to the non-stoichiometry of Mn and W, given the Mn:W ratiog. It is observed that the non-stoichiometry does not affect remarkably the ferroelectric transition point (the AF3-AF2 transition point) and the AF2-AF1 transition point, but the non-polar AF1 phase is partially replaced by the ferroelectric AF2 phase. The measured electric polarization is slightly enhanced with increasing stoichiometric deviation jg1j. The possible underlying mechanism for these effects is discussed.VC 2014 AIP Publishing LLC.
[http://dx.doi.org/10.1063/1.4866087]
Multiferroic materials have inspired much attention in recent years. The coexistence of spontaneous polarization and magnetization in the same phase and the magnetoelectric coupling between them makes the mutual control of magne- tism and ferroelectricity possible.1–3These potentially useful phenomena are observed in a wide class of materials, includ- ing orthorhombic rare-earth manganites,4–6 triangular cuprates,7–9 Ni3V2O8,10 Ca3CoMnO6,11 and so on. While most of them show more than one magnetic phase, the magne- toelectric coupling is available only in certain phases that fea- ture noncollinear spin orders. The Dzyaloshinskii-Moriya (DM) interaction is believed to induce such couplings based on a magnetic structure that results in a macroscopic electric polarization,12–14 mainly in 3d magnetic transition metal oxides of noncollinear spin order.15,16
MnWO4(MWO) has been well studied in recent years.
MWO undergoes successive magnetic phase transitions upon decreasing temperatureT. The first antiferromagnetic transi- tion occurs atT¼TAF3¼13.5 K, where a sinusoidal incom- mensurate spin structure is developed (AF3 phase). The AF3 phase is replaced by the AF2 phase with a tilted elliptical spiral structure atT¼TAF2¼12.6 K. The collinear commen- surate AF1 phase with characteristic""##spin order appears atT¼TAF1¼7.8 K and below. It is noted that only the heli- cal AF2 phase is ferroelectric with polarization P aligned along theb-axis. Plenty of works have been done to modu- late the stability of these phases, based on the fact that the spin structure is sensitive to any small perturbations such as external fields17,18and chemical substitutions.19–22
The purpose of this work is to investigate the multiferroic phase stability of MWO against the Mn:W non-stoichiometry.
The spin structure is a compromise between multifold interactions’ competition and the effective interactions can be over the 11th-neighbor.23 The spin-lattice interaction is believed to play a substantial role. It was reported that substi- tution of Mn2þby Co, Zn, or Mg can seriously suppress the AF1 phase and slightly destabilize the AF2 phase.20,21,24,25
Since these substitutions all result in chemical disorder and lattice contraction, it is argued that the spin-lattice interactions
in addition to the multifold spin interactions are major ingre- dients for the magnetic transitions.
Here, for non-stoichiometric MWO, either Mn- or W-deficiency can be viewed as a kind of chemical disorder too. In this case, the relative stability of one magnetic phase over the other will be modulated. By synthesizing a series of MWO samples with different g, we investigate the stability of the AF1 and AF2 phases.
We define an atomic ratiog¼Mn/W and quantityjg1j measures the non-stoichiometry. Our experiment is on poly- crystalline samples prepared using standard solid-state reac- tions. The high-purity WO3and MnO powder was chosen as reagents and thoroughly mixed for 24 h. The dried mixture was ground for 1 h each and then annealed in air for 12 h at 600C.
After another intermediate grindings, the mixture was com- pressed into pellets and annealed at 950C for 20 h in air.
The sample crystallinity was checked using X-ray dif- fraction (XRD) with Cu Karadiation at room temperature.
The valence states of Mn and W ions and chemical composi- tion were examined by X-ray photoelectron spectroscopy (XPS; ULVAC-PHI PHI5000 VersaProbe using Al Karadia- tion). The magnetizationManddcmagnetic susceptibilityv were measured using the Quantum Design Superconducting Quantum Interference Device (SQUID) in the zero-field cooled (ZFC) mode and field-cooling (FC) mode, respec- tively. The cooling and measuring fields were both 1000 Oe.
The electric polarization as a function of T was measured using the pyroelectric current method by Keithley 6514 elec- trometer connected to the Physical Properties Measurement System.26 The samples were first poled in a static electric field of10 kV/cm, and then the pyroelectric currents meas- ured through warming the samples were integrated to obtain polarizationPas a function ofT.
The high-precision XRD data of several samples are plotted in Fig.1. All the spectra fit the standard database sat- isfactorily. Here, it is critical to exclude any manganese oxides or tungsten oxides. By focusing on the local (030) and (022) reflections in Fig. 1(b), one sees gradual shifting of the two peaks towards the high-angle with increasing jg1j, indicating clearly the lattice contraction. For clarify- ing the details, we perform the Rietveld refining of the XRD data using the GSAS program. For a reliable refining of the
a)Author to whom correspondence should be addressed. Electronic mail:
0021-8979/2014/115(17)/17D722/3/$30.00 115, 17D722-1 VC2014 AIP Publishing LLC
JOURNAL OF APPLIED PHYSICS115, 17D722 (2014)
data, one needs to know the variations of lattice occupation in these non-stoichiometric samples. First, we employ the XPS to probe the g value and Mn/W valence states. For examples, the XPS spectra for samplesg¼1.04,g¼1, and g¼0.94 are shown in Fig.1(d) for W and (e) for Mn. No identifiable shift of the peaks corresponding to the Mn- 2p1/2,3/2 and W-4f5/2,7/2 is observed, implying that the va- lence states of Mn and W ions remain to be Mn2þand W6þ, although the existence of tiny Mn and/or W ions with differ- ent valences cannot be excluded. Second, generation of ionic defects in theg6¼1 samples is needed to meet the electric neutrality condition. It is noted that Mn2þion is bigger than W6þion and there is big valence difference between them.
Forg>1, no possibility for occupation of the W-vacant site by Mn ion is expected and thus, the W-vacancies are assumed with oxygen vacancies surrounding the W vacan- cies. On the other hand, for g<1, two possible situations may appear. One is that the excess W6þ ions occupy the Mn2þ-vacant sites. The electrostatic energy due to the big valence difference between Mn2þand W6þcan be high. The other is the generation of Mn vacancies and surrounding ox- ygen vacancies, which are favored from the energy consideration.
Based on this site occupation model, high precision Rietveld refining is obtained. The evaluated lattice constants (a,b, andc) and lattice unit volumeVas a function ofgare plotted in Figs. 1(e)–1(h). These parameters decrease with increasingjg1j. This feature is similar to those cases with the Co, Zn, or Mg-substituted MWO.20,21,24,25It is then sug- gested that the AF1 phase may be destabilized, and the AF2 phase may be or not. The AF1 phase will most likely be par- tially replaced by the AF2 phase, to be confirmed below.
The v-T data in both the ZFC and FC modes for five samples are presented in Fig.2. Rather than any quantitative discussion, we list several qualitative features. First, the magnetization over the whole T range is suppressed with increasingjg1j. This phenomenon may be understandable for g<1 since the magnetic Mn species is deficient.
However, it is true for g>1 too. A reasonable argument is that the non-stoichiometry releases the spin frustration and stabilizes the spin structures. Second, a comparison of the ZFC and FC data shows that the separation between the two modes becomes weaker with increasing jg1j, and the two modes even merge together atg¼0.96. This also evidences that the spin phase stability against, e.g., thermal activations, is enhanced with increasingjg1j. Third, thev-Tcurves at the ZFC mode show quite similar shapes, indicating that the magnetic transition sequence does not change much for dif- ferent g. Unfortunately, the AF3-AF2 and AF2-AF1 transi- tion points cannot be clearly identified from the v-T data although several weak anomalies can be roughly seen, which are marked asTAF1,TAF2, andTAF3shown in Fig.2.
Finally, we consult to the ferroelectricity for illustrating the effect of Mn/W non-stoichiometry. The measured I-T curves and evaluatedP-Tcurves for several samples are shown in Figs.3(a)–3(f). TheP-Tdata for theg¼1 sample is similar to earlier results.17Several clear features are shown. First, if defining the initiating point of the pyroelectric current Ifrom the high-Tside asTAF2, and the initiating point from the low-T side asTAF1, one sees that bothTAF2andTAF1do not shift with increasingjg1j, noting that the two points for theg¼1 sam- ple coincide well with earlier reports.17,21,22,24,25This is differ- ent from those Co, Mg, or Zn-substituted MWO where TAF1
disappears immediately andTAF2is slightly down-shifted. For the Fe-substituted case, however,TAF1is up-shifted andTAF2is down-shifted. Second, for bothg>1 andg<1, not only theP within TAF1<T<TAF2 is gradually enhanced but also non- zero polarization is identified below TAF1. Meanwhile, the polarization atT¼2 K,P(2 K), increases too. TheP(max) and P(2 K) data as functions ofgare plotted in Figs.3(g)and3(h), respectively. In those substituted MWO, however, thePof the AF2 phase is gradually damaged upon the increasing substitu- tion. Third, the above data show the incomplete AF2-AF1 phase transitions initiating atTAF1in theseg6¼1 samples, lead- ing to the AF1 and AF2 phase coexistence belowTAF1. This is also different from those substituted MWO.20–22,24
From the measured results, we can highlight the multi- ferroic phase diagram in the (g,T) plane atH¼0, as shown
FIG. 1. (a) Measuredh-2hXRD spectra of several samples, and (b) the amplified (030) and (022) reflections. The XPS spectra of Mn(2p1/2,3/2) and W(4f5/2,7/2) core levels forg¼1.04, 1.00, and 0.94 are shown in (c) and (d).
The evaluated lattice constants (a,b,c) and lattice unit volumeVare plotted in (e)-(h).
FIG. 2. Measuredv-Tcurves under the ZFC and FC modes for five samples g¼1.04, 1.02, 1.00, 0.98, and 0.96. The measuring field is 1000 Oe.
17D722-2 Yuet al. J. Appl. Phys.115, 17D722 (2014)
in Fig.4. In the low-T range, the phase diagram is divided into three regions: the AF1 phase close to g¼1 and the AF1þAF2 coexisting regions on the two sides.
Furthermore, from the ferroelectric data, the TAF1 at which the AF2-AF1 phase transition begins does not change. Here, we acknowledge that the proposed phase diagram is still more or less qualitative, and more data on the spin structure are needed for a quantitative phase diagram.
The obtained structural and ferroelectric data allows us to have a qualitative discussion on the underlying mecha- nism for the effects of the Mn/W non-stoichiometry.
First, the similarity between the present materials and those Co, Zn, and Mg-substituted MWO lies in the fact that chemical disorder is induced and the lattice contracts. The chemical disorder is believed to destabilize the magnetic phases via the spin-lattice interactions.24,25This effect seems more significant in suppressing the AF1 phase rather than the AF2 phase. The induced lattice contraction is equivalent to the effect of a chemical pressure, believed to destabilize the non-polar AF1 phase.27The chemical disorder induced here seems weak, and the lattice contraction at the maximal value here is only one-tenth of that for those Co, Zn, and Mg- substituted MWO. Therefore, it is reasonable to observe an incomplete taking-over of the AF1 phase by the AF2 phase, leading to the AF1þAF2 phase coexistence.
Second, the ferroelectric polarization in the AF2 phase is enhanced in the g6¼1 MWO, while it is suppressed for those Co, Zn, and Mg-substituted MWO. This P-enhance- ment may be related to the overall high stability of the mag- netic phases (both AF2 and AF1) in the g6¼1 MWO with
respect to theg¼1 MWO (Fig.2). This higher stability and thus the ferroelectric domains in the AF2 phase against inter- nal fluctuations reasonably allows larger polarization.
Third and surely, both magnetic and non-magnetic substitutions in the g6¼1 samples and those Co, Zn, and Mg-substituted MWO definitely impose impact on the spin interactions. It was claimed that the effects of non-magnetic substitutions seem independent of the nature of substituting species, and the three-dimensional nature of magnetic inter- actions and the spin frustration remain less affected.25 The Mn-deficient MWO is equivalent to a kind of non-magnetic substitution, while the W-deficient MWO does not include additional magnetic species either since the substituted sites are W-vacancies. Therefore, a reasonable argument is that, sim- ilar to the Zn/Mg-substituted cases, the non-stoichiometry here at lowjg1jimposes less influence on the magnetic interactions, and the major contributions are from the chemical disorder and modulated spin-lattice interactions.
This work was supported by the National 973 Projects of China (Grant No. 2011CB922101), the Natural Science Foundation of China (Grant Nos. 11234005 and 51332006), and the Priority Academic Program Development of Jiangsu Higher Education Institutions, China.
1Y. Tokura,Science312, 1481 (2006).
2S. W. Cheong and M. Mostovoy,Nat. Mater6, 13 (2007).
3K. F. Wang, J.-M. Liu, and Z. F. Ren,Adv. Phys.58, 321 (2009).
4T. Kimuraet al.,Nature (London)426, 55 (2003).
5T. Gotoet al.,Phys. Rev. Lett.92, 257201 (2004).
6N. Huret al.,Nature (London)429, 392 (2004).
7S. Parket al.,Phys. Rev. Lett.98, 057601 (2007).
8T. Kimuraet al.,Phys. Rev. B73, 220401(R) (2006).
9K. Kimuraet al.,Phys. Rev. Lett.103, 107201 (2009).
10G. Laweset al.,Phys. Rev. Lett.95, 087205 (2005).
11L. Linet al.,J. Appl. Phys.111, 07D901 (2012).
12H. Katsuraet al.,Phys. Rev. Lett.95, 057205 (2005).
13M. Mostovoy,Phys. Rev. Lett.96, 067601 (2006).
14I. A. Sergienko and E. D. Dagotto,Phys. Rev. B73, 094434 (2006).
15M. Kenzelmannet al.,Phys. Rev. Lett.95, 087206 (2005).
16S. G. Condranet al.,J. Phys.: Condens. Matter22, 162201 (2010).
17K. Taniguchiet al.,Phys. Rev. Lett.97, 097203 (2006).
18K. Taniguchiet al.,Phys. Rev. Lett.101, 207205 (2008).
19F. Yeet al.,Phys. Rev. B78, 193101 (2008).
20Y. S. Songet al.,Phys. Rev. B79, 224415 (2009).
21R. P. Chaudhuryet al.,Phys. Rev. B83, 014401 (2011).
22H. W. Yuet al.,Phys. Rev. B87, 104404 (2013).
23H. Ehrenberget al.,J. Phys.: Condens. Matter11, 2649 (1999).
24F. Yeet al.,Phys. Rev. B86, 094429 (2012).
25L. Meddaret al.,Chem. Mater.21, 5203 (2009).
26G. Q. Zhanget al.,Phys. Rev. B84, 174413 (2011).
27R. P. Chaudhuryet al.,Physica B403, 1428 (2008).
FIG. 3. (a)-(f) Measured pyroelectric currentI(red solid curves) and eval- uated electric polarizationP(blue dashed curves) as a function of T. The arrows indicate TAF2 and TAF1, respectively. (g) and (h) The evaluated P(2 K) andP(max) as a function ofg.
FIG. 4. Evaluated multiferroic phase diagram in thegTplane atH¼0.
17D722-3 Yuet al. J. Appl. Phys.115, 17D722 (2014)
