Journal of Alloys and Compounds 661 (2016) 331e344
Contents lists avai
Journal of Alloys and Compounds
journal homepage: http: / /www.elsevier .com/locate/ ja lcom
Mechanism of defect induced ferromagnetism in undoped and Crdoped TiO2 nanorods/nanoribbons
Batakrushna Santara a, Kenji Imakita b, Minoru Fujii b, P.K. Giri a, c, *
a Department of Physics, Indian Institute of Technology Guwahati, Guwahati 781039, Indiab Department of Electrical and Electronic Engineering, Graduate School of Engineering, Kobe University, Kobe 657-8501, Japanc Centre for Nanotechnology, Indian Institute of Technology Guwahati, Guwahati 781039, India
a r t i c l e i n f o
Article history:Received 3 August 2015Received in revised form23 September 2015Accepted 10 November 2015Available online 29 November 2015
We have studied the effect of doping concentrations, growth temperature and calcinations on thestructural, optical and magnetic properties of the undoped and Cr doped TiO2 nanorods/nanoribbons(NR/NRb), in order to develop an improved understanding on the mechanism of room temperature (RT)ferromagnetism (FM) in these nanostructures. Both undoped and doped TiO2 NR/NRb exhibit RT FM anda ~2.6 fold enhancement in magnetization is observed in 0.3% Cr doped TiO2 NR/NRb as compared to theundoped NR/NRb, and the magnetization increases considerably after vacuum annealing. However, nomeasureable FM is observed for the precursor TiO2 powder, despite the presence of high concentration ofoxygen vacancies in it. On the other hand, the magnetization decreases at higher doping concentration(0.7% Cr) as compared to 0.3% Cr doped sample. Thus, our studies revealed that a simple oxygen vacancyand/or Cr ions alone cannot yield the RT FM in TiO2 nanostructures. We argue that the oxygen vacancywith appropriate charge redistribution and their orbital overlapping, and exchange interaction with thenearby unpaired 3d electron of Ti3þ in undoped TiO2 and/or unpaired 3d electrons of Cr3þ dopant withproper charge distribution and occupation inside the host lattice in Cr doped TiO2 nanostructures playthe pivotal role for the ferromagnetic ordering and observed RTFM. The presence of oxygen vacancyrelated Fþ-center and Cr3þ are confirmed from the electron spin resonance and x-ray photoelectronspectroscopy measurements.
Over the past decades, there has been tremendous interest inunderstanding the origin of room temperature (RT) ferromagne-tism (FM) in a variety of transition metal (TM) doped oxide semi-conductors. These materials could be useful for future spintronicand magneto-optic devices [1e3], where both spin and chargedegrees of freedom can be manipulated and used to transport andstore charge, and process information in novel ways. This tech-nology may provide enhanced performance and new functional-ities in order to increase both speed and storage capacity intraditional microelectronic devices where only electronic chargesare considered. Therefore, intensive attention has been focused ondilute magnetic semiconductors (DMS), such as TM doped ZnO
dian Institute of Technology
[4,5], SnO2 [6,7], In2O3 [8,9] and TiO2 [10e13] due to realization ofRT FM in these nanostructured non-magnetic semiconductors. Inparticular, Pereira et al. [14] reported intrinsic ferromagnetism inCo doped TiO2 anatase nanopowders grown by a hydrothermalmethod and it was shown that oxygen vacancy (Ov) play animportant role in promoting the long-range ferromagnetic order inthe material, in addition to the transition metal doping. Later thesame group [15] reported a stable ferromagnetism in theTi1�xCoxO2�d nanopowders (with 0.03 � x � 0.10) which providedan evidence of preserved long-range ferromagnetic order in suchsystems.
Despite several reports on the TiO2-based DMS [11e13,16e19]since the prediction and discovery of RTFM in Co doped TiO2 sys-tem [3], there is no clear agreement about the nature and origin ofthe observed FM in the diluted magnetic oxide doped with a fewpercent of 3d cations. It still remained a subject of intense debatethat whether the FM is indeed intrinsic which is mediated by car-riers or defects inside the host semiconductor and/or related to thepurely extrinsic origin due to formation of ferromagnetic secondary
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344332
phases, metallic clusters or impurities[11e13,16e19]. A wealth ofinformation derived from both experimental and theoreticalstudies on the FM in Co and Fe doped TiO2 systems can be found inliterature [18e22]. The reported results are controversial regardingthe intrinsic and extrinsic nature of Fe and Co doped TiO2 DMSsystems, since these elements themselves are ferromagnetic innature. Unlike Fe, Co and Ni metals, Cr itself is antiferromagneticand its clusters/compounds (except nanocrystalline CrO2) do notcontribute to FM. Thus, it would not induce an extrinsic FM even ifCr clustering occurs in the Cr doped TiO2 (Cr:TiO2) nanostructures.Moreover, the pure phase ferromagnetic CrO2 is difficult to syn-thesize, since it is metastable at atmospheric pressure and gener-ally stable at high oxygen pressure depending upon the method ofpreparations, process temperatures and precursors [23e25]. Anintrinsic FM in annealed Cr:TiO2 nanorods was reported [12] and itwas suggested that ferromagnetic coupling between the Ov and theunpaired 3d electrons of Cr dopant forming bound magnetic po-larons (BMP) is the origin of FM signal. Choudhury et al. [13]observed diamagnetism in undoped TiO2 nanoparticles and FM inCr:TiO2 nanoparticles, and suggested that F-center mediated BMP isthe active mechanism for induced FM. More recently, Liao et al. [11]reported that Ov plays a critical role in the enhancement of FM in Crdoped anodic TiO2 nanotubes by comparing the FM behavior ofdifferent annealed Cr:TiO2 nanotubes. Similarly, Xing et al. [8] andWang et al. [9] demonstrated that Ov defects mediated BMP isresponsible for the FM in Cr doped In2O3 nanostructures and films,respectively. Recently, Da Pieve et al. [16] reported that F-centermediated exchangemechanism is not active for inducing FM in caseof Cr:TiO2 systems and they proposed a mechanism of super ex-change interaction of t2g electrons, which is favored by the presenceof structural defects and Ov. Further, the creation of grain boundarydefects with Ov at the nanocrystal fusion interfaces was proposed tobe the origin of FM in Cr3þ doped TiO2 nanostructures [17]. Thesecontradicting understandings on the RTFM in Cr:TiO2 materialssuggest that the specific mechanism and the physical origin of FMin Cr:TiO2 is still unclear and it needs further investigation.
Interestingly, RTFM has been observed in a wide range ofundoped oxides, such as TiO2 [26e28], HfO2 [29], In2O3 [27], SnO2[30] and ZnO [31]. Recently, Hoa and Huyan [32] reported enhancedFM in undoped TiO2 nanowires as compared to Ni doped TiO2nanowires. Some of the outstanding reports revealed no evidenceof ferromagnetic ordering in Fe doped TiO2 (Fe:TiO2) systems[20,22]. These reports help to address the controversies on the is-sues related to the role of defects in the ferromagnetic ordering. Infact, recent experimental evidences [26,28] and theoretical studies[33,34] suggest that intrinsic defects are responsible for the ferro-magnetic signal in undoped TiO2. However, unambiguous deter-mination of the nature of defects responsible for the observed FMremains a considerable challenge to researchers. It is unclearwhether Ov or Ti-vacancy defects contribute to the magnetic mo-ments, since both titanium and oxygen vacancies were proposed tobe responsible for the FM in undoped TiO2 [28,33e35]. Some re-ports based on first-principle calculations indicate that Ti-vacanciesand Ti-divacancies are ferromagnetically coupled [34], and oxygenvacancies do not produce any magnetic moment in undopedanatase [34,36] and rutile [36] TiO2. On the other hand, newertheoretical and experimental studies show that ferromagneticordering in undoped TiO2 is strongly related to Ov in TiO2(B) [28],rutile [28,35] and anatase [28,35] TiO2, and thus it was thought tobe the source of RTFM in undoped semiconducting oxides. Based onfirst-principles calculations, Wang et al. [33] reported a multi-defects induced FM mechanism, in which one kind of defect (Ti-vacancy) produces local moments while the electrons induced byanother kind of defect (Ov) mediate long-range ferromagneticordering through exchange interaction, and suggested that Ov alone
cannot induce RTFM in undoped rutile TiO2. These diversities ofresults and interpretations intensified the controversy related toissue whether Ov or Ti-vacancy defects contribute to the ferro-magnetic response in undoped TiO2. Therefore, it is important tounderstand the presence of specific defect types and their rolestowards the FM. In fact, this may provide the key for interpretingthe theoretical predictions and experimental data consistently.
At present, most of the reported FM in undoped TiO2 systemshas been for thin films [27,35] and nanoparticles (NPs) [37,38],while their undoped bulk counterparts are paramagnetic ordiamagnetic. This implies that the spatial dimensionality mightplay an important role in the ferromagnetic ordering. Compared tothin films and NPs, exploitation of TiO2 one dimensional (1D)nanostructures such as nanowires, nanorods (NRs) and nano-ribbons (NRbs) with high surface area make it easier to engineerhigh availability of defect sites for trapping electrons and may favorthe ferromagnetic ordering, thus making them an ideal candidatefor the realization of intrinsic enhanced RTFM. Moreover, 1Dnanostructures are favored over the 0D nanoparticles in terms ofelectron transport, storage and information processing that canenhance the performance of spintronic devices at the nanoscale forpractical applications. A survey of literature show that RTFM inundoped and doped 1D TiO2 nanostructures are relatively lessstudied [11,12,26]. In our earlier study [28], we reported RTFM inundoped TiO2 NRbs and a direct correlation between the Ov and FMwas established from various experimental tools. In another study[39], we examined the individual and combined effect of Fe dopantand Ov defects and clarified the controversial issues related to thelong-range ferromagnetic ordering in Fe:TiO2 systems. A cleardemonstration by correlating the RTFM with dopant concentrationand Ov defects in the host semiconducting oxide lattice is crucialand it needs a thorough investigation. In this work, we carried out asystematic study on the structural, optical and magnetic propertiesof undoped and Cr doped TiO2 1D nanostructures in order toinvestigate the mechanism of ferromagnetic ordering in thesenanostructures. We aim to provide an improved understanding ofthe mechanism of long-range ferromagnetic ordering in undopedand Cr doped TiO2 systems by discussing the interplay of specificdefects and their interaction in TiO2 nanostructures, which isessential to control and improve the magnetic properties forpractical device applications.
2. Experimental details
2.1. Growth and processing of undoped and doped TiO2 NRs/NRbs
Anatase TiO2 powder (particle size ~80 nm), chromium nitratenonahydrate (Cr(NO3)3,9H2O) were used as TiO2 and Cr precursors,respectively. Doubly distilled de-ionized (DI) water was used assolvent for the synthesis of Cr doped TiO2 nanostructures. Theappropriate quantity (0.3 at% and 0.7 at% of Cr) was mixed withTiO2 powder and ground it for 10 min in a ceramic mortar. Themixed powder was added in 10MNaOH (alkaline water) and stirredfor 30 min in a conical flask. Then the solution was transferred intothe Teflon-lined autoclave (Berghof, BR-100) for hydrothermal re-action at 170 and 190 �C for 24 h. The precipitates after HCl treat-ment and washing with DI water were allowed for calcinations in atube furnace at 500e900 �C temperatures. The undoped samplewas prepared under similar condition except adding the Cr pre-cursor. For the simplicity of discussion, the undoped sample grownat 170 �C and calcined at 500 �C is named as A500. The 0.3% Crdoped samples grown at 170 �C are named as “B” series and arenamed as B500, B700 and B900 after calcinations at 500, 700 and900 �C, respectively. The sample B500 after vacuum annealing at300 �C, 2 h under 1.2 � 10�2 mbar pressure is named as B500V. The
Fig. 1. Comparison of XRD patterns for different samples: (a) A500, B500, C500 andD500; (b) B500, B700 and B900.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344 333
0.3% and 0.7% Cr doped samples grown at 190 �C is termed as “C”and “D” series samples. The details of the growth/processing con-ditions and nomenclatures for different samples are presented inTable 1.
2.2. Characterization techniques
The crystal structures of the obtained samples are characterizedby x-ray diffraction (XRD) (Rigaku RINT 2500 TTRAX III, Cu Ka ra-diation) and micro-Raman spectroscopy (LabRam HR800, JobinYvon). Morphologies of the samples were studied by field emissionscanning electron microscopy (FESEM; Sigma, Zeiss). The energydispersive x-ray (EDX) spectra were collected with the help of an x-ray detector (Oxford Instruments, UK) attached with a scanningelectron microscope (SEM; LEO 1430VP). The high-magnificationsurface morphologies and structures of the samples were studiedby transmission electron microscopy (TEM), high-resolution TEM(HRTEM) and selected area electron diffraction (SAED) pattern(JEOL-JEM 2010 operated at 200 kV). The UVevisibleeNIR absorp-tion spectroscopy measurements were recorded using a commer-cial spectrophotometer (Perkin Elmer, LAMBDA-750). The steadystate photoluminescence (PL) spectrum was recorded at roomtemperature (RT) using a 405 nm diode laser (Coherent, Cube)excitation with the help of a spectrometer (focal length: 15 cm;blaze wavelength: 500 nm; groove density: 150 g mm�1) equippedwith a cooled charge coupled device (Princeton Instruments, PIX-IS100B) detector. The powder samples were put onto a conductivecarbon tape (black) for the PL measurements. Each spectrum wascorrected for the detector response as a function of wavelengthafter background subtraction. X-ray photoelectron spectroscopy(XPS) measurements were carried out with a PHI X-Tool automatedphotoelectron spectrometer (ULVAC-PHI, Inc.) using Al Ka x-raybeam (1486.6 eV) with a beam current of 20 mA. Carbon 1s spec-trum was used for the calibration of the XPS spectra recorded forvarious samples. Electron spin resonance (ESR) measurementswere done by a JEOL (JESFA200) instrument operating in the Xband. The magnetic properties of the samples (both field depen-dent magnetization at room temperature and temperaturedependent magnetization in the temperature range 30e300 K)were measured using a Lakeshore (Model no. 7410) vibratingsample magnetometer (VSM).
3. Results and discussion
3.1. Structural characterization
3.1.1. XRD studiesThe XRD patterns of undoped and Cr doped samples (B500,
C500 and D500) are shown in Fig. 1(a). All the peaks of differentsamples show the pure TiO2(B) phase, while the undoped sampleA500 shows the mixed phases of anatase TiO2 and TiO2(B). Thismay imply that Cr doping plays some role in the formation ofTiO2(B) phase. In the process of Cr doping in the TiO2 lattice, the
Table 1Details of the growth conditions, crystal structures and magnetization parameters for dSaturation magnetization (Ms), remanent magnetization (Mr), and Coercive field (Hc) we
Sample name Growth temperature, calcination, Cr doping % Crys
lattice distortion may be higher under higher pressure and tem-perature (reaction). As TiO2(B) phase is a high pressure form of TiO2obtained from the hydrothermal reaction followed by low tem-perature calcinations, the formation of pure TiO2(B) phase in thedoped sample indicates that the autogenous pressure inside theautoclave may be comparatively higher during the reaction process(precursors Cr(NO3)3,9H2O and TiO2 nanoparticles) in the presenceof alkaline water as compared to the undoped case. The 0.3% Crdoped samples grown at 170 �C after different calcination tem-peratures are shown in Fig. 1(b). All the peaks of samples B700correspond to pure anatase TiO2, while the sample B900 shows the
ifferent samples; Crystal structures were obtained from XRD and Raman studies;re determined from MeH loops.
Fig. 2. Raman spectra for different samples: (a) A500 and B500; (b) B500, C500 andD500; (c) B700 and B900. The characteristic Raman modes of TiO2(B) are indicated ineach case, the peak positions marked with arrows in (a) show the anatase TiO2 phasesand that in (c) show the rutile TiO2 phases.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344334
mixed phases of anatase-rutile TiO2. No signature of additionalpeaks other than the TiO2 is observed in these Cr doped samples,indicating the Cr may be successfully doped into the TiO2 crystallattice. Due to the similar ionic radii of Ti4þ (75 Å) and Cr3þ (76 Å) in6-coordinate octahedral system, Cr3þ is more likely to occupy theregular lattice position and substituted for the Ti4þ cation in theTiO2 lattice.
3.1.2. Raman scattering studiesThe as-synthesized undoped and Cr doped, and vacuum
annealed samples are characterized by Raman spectroscopy tofurther confirm the structure and phase composition of the sam-ples. This study helps us to identify the impurity/secondary ferro-magnetic phases such as CrO2, if these are present in the Cr:TiO2samples. All the Raman peaks of the 0.3% Cr doped sample B500correspond to the pure TiO2(B) phase, while the Raman modes ofthe undoped sample A500 correspond to mixed TiO2(B)-anataseTiO2 phases as shown in Fig. 2(a). The Raman spectra of dopedsamples calcined at 500 �C are shown in Fig. 2(b). All the Ramanmodes correspond to pure TiO2(B) phase. Fig. 2(c) shows the Ramanspectra of B700 and B900. The Ramanmodes of sample B700 revealpure anatase TiO2, while B900 shows mixed anatase-rutile phases.Note that the Cr-oxides related Raman peaks are not observed inthese Cr doped samples [40,41], especially CrO2 (expected peakpositions are at 149, 458, 570, 682 cm�1 at room temperature) [40],which is the only ferromagnetic phase of Cr and these resultsconfirm that the Cr is successfully doped into the TiO2 crystal lat-tice. The Raman results are fully consistent with the XRD results.The Raman modes of vacuum annealed B500V correspond to pureTiO2(B) phase (not shown), indicating that the phase remains un-changed after vacuum annealing at 300 �C.
3.2. FESEM and TEM studies
3.2.1. FESEM imagingThe morphologies of the Cr doped TiO2 nanostructures studied
by FESEM imaging are shown in Fig. 3. The high magnificationimage of a single nanorod (NR) in sample B500 is shown in Fig. 3(a).Fig. 3(b) shows a single straight and thick nanoribbon (NRb) likestructure of the sample B900. The FESEM image of D500 shows theNRb like structures co-existing with small nanorods (Fig. 3(c)).These samples reveal the well-defined 1D NRs/NRbs withoutappearance of any additional impurity clusters, indicating that allthe precursor anatase TiO2 and Cr(NO3)3,9H2O are uniformlymixed and homogeneously distributed throughout the solvent, andit implies that chromium ions are incorporated into the lattice ofTiO2 after hydrothermal treatment forming doped 1D TiO2 NRs/NRbs. Fig. 3(d) shows the FESEM image of precursor TiO2 powder(PTiO2). The EDX spectrum of NRs/NRbs (D500) is shown inFig. 3(e), which shows only Ti, O and Cr elements, indicating that noother contaminant/element was introduced into the Cr:TiO2 duringthe sample preparation.
3.2.2. TEM imagingFig. 4(a) shows the TEM image of undoped sample A500, indi-
cating NR morphology with nanobricks/nanostones like structureson the surface. B500 shows the NRs/NRbs with nanobricks/nano-stones like structures on the surface making it a homogeneousnanoporous like structures (Fig. 4(b)). Similar growth of nano-bricks/nanostones/nanopits was reported by other groups [42,43].Zhao et al. [42] explained the evolution of nanopits like structureson the surface of TiO2(B) nanowires via an oriented attachmentgrowth. The inset in Fig. 4(b) shows the SAED patterns of the cor-responding NRs/NRbs, indicating single crystalline nature ofTiO2(B) phase. The lattice fringe of the NR of B500 is shown in
Fig. 4(c). The d-spacing of 3.12 Å corresponds to (002) planes ofTiO2(B) phase. The TEM image of C500 in Fig. 4(d) shows the NRs/NRbs morphology with nanobricks/nanostones like structures onthe surface. A single NRb of D500 is shown in Fig. 4(e). Nanobricks/
Fig. 3. ESEM images of the morphology of TiO2 NRs/NRbs: (a) B500, (b) B900, (c) D500, (d) precursor TiO2 powder. (e) EDX spectrum of D500 indicating Ti, O and Cr elements.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344 335
nanostones like morphologies are formed on the surface of NRb.The SAED pattern of the NRb is shown in the inset of Fig. 4(e)showing the single crystalline nature of TiO2(B) phase. The mono-clinic structure of TiO2(B) is clearly observed in the well spottedSAED pattern. The lattice fringe of the corresponding NRb of D500is shown in Fig. 4(f). The d-spacing of 3.12 Å corresponds to (002)plane of TiO2(B) phase. We observed that Cr doping does not alterthe morphology of the TiO2 nanostructures. No cluster relatedstructures is observed in HRTEM images. Further, clear SAEDpattern of TiO2(B) phase dictates that Cr is well incorporated intothe TiO2 crystal lattice. The formation of nanobricks/nanostones onthe surface of the NRs/NRbs may give rise to structural defects atthe interface of NRs/NRbs and nanobricks/nanostones.
3.3. Optical absorption and PL studies
3.3.1. Optical absorption studiesThe UVevisibleeNIR absorption spectra of B500, B700 and B900
along with precursor TiO2 are shown in Fig. 5(a). The absorptionedges of Cr doped samples are red-shifted when compared withthat of the precursor PTiO2. The red-shift in doped samples can beassigned to the charge transfer band Cr3þ/ Ti4þ [44] or may be dueto the Ov [28]. A large red shift is observed for the sample B900,which is attributed to presence of Ti interstitial defects [45]. TheB900 sample shows strong Ti interstitial defect related NIR PL,discussed in the next paragraph. The indirect band gaps of the Crdoped samples are confirmed from the Tauc plot as shown in
Fig. 4. High magnification TEM image of (a) a single NR in A500; (b) NR/NRb in B500, inset is the corresponding SAED pattern; (c) HRTEM lattice fringe of B500; (d) nanoporousNRs/NRbs in C500; (e) single nanoporous NRb in D500, inset is the corresponding SAED pattern; (f) HRTEM lattice fringe of D500.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344336
Fig. 5(b). It is found that the band gaps of the doped samples arereduced as compared to the PTiO2 (3.22 eV). The band gap iscalculated from the linear fits to the linear portion of the (ahn)1/2
versus hn plot (Tauc plot). The calculated band gaps are 2.73, 2.73and 2.38 eV for the samples B500, B700 and B900, respectively,which are very important for the visible light photocatalyticapplications.
3.3.2. Photoluminescence studiesPhotoluminescence is a very useful technique to observe the
presence of defect related trap states in the materials and is thusemployed to confirm the presence of Ov defects in the as-grown
TiO2 NRs/NRbs. Fig. 6(a) shows the room temperature PL spectraof 0.3% Cr doped samples grown at 170 �C after different calcinationtemperatures. The broad visible PL is commonly attributed to the Ovrelated traps [28,45]. The NIR PL is related to the Ti interstitialdefect related traps [45]. With the increase in calcination temper-atures, the Ov related visible PL decreases, while Ti-interstitialrelated NIR PL increases and these results are consistent with theprevious studies on undoped TiO2 NRbs [45]. Similar response of PLis observed for the 0.7% Cr doped samples grown at 190 �C withdifferent calcination temperatures (not shown). These resultsindicate that the visible PL decreased due to reduction of oxygenvacancies during the calcinations in air at higher temperatures.
Fig. 5. UVeViseNIR absorption spectra: (a) B500, B700, B900 and precursor TiO2; (b)(ahn)1/2 vs hn plot indicating the indirect band gap for the corresponding absorptionspectrum. The band gap energy is calculated from extrapolated line (dashed) fitted torespective linear portions in PTiO2 and B500.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344 337
Interestingly, the center of NIR PL is shifted to higher energy withincreasing calcination temperatures. The NIR PL at lower energyside can be attributed to the Ti3þ interstitial defects, while the NIRPL at 1.45 eV is attributed to Ti4þ interstitial defect states within thebandgap of Cr:TiO2 nanostructures [45]. The inset in Fig. 6(a) showsa magnified view of the visible PL of the corresponding samplesalong with broad visible PL of D500 (0.7% Cr doped) sample. Thecomparison of visible PL intensities of B500 and D500 (inset inFig. 6(a)) indicates that larger concentrations of Ov defects arepresent in B500 as compared to D500. The observed intensity ofNIR PL for the sample D500 is about 6 times higher than that ofB500 (not shown); this may be due to the higher growth temper-ature of D500. It appears that more interstitial defects are evolvedat higher growth temperature, which is consistent with our earlierreport [45]. The comparison of PL intensities for B500, B500V andPTiO2 is shown in Fig. 6(b). Note that the spectra shown in Fig. 6(b)are obtained using a visible PMT detector. After vacuum annealing,the PL intensity is considerably enhanced for B500V as compared toas-synthesized B500, indicating large concentrations of Ov defects
are introduced during the vacuum annealing and this is responsiblefor the observed enhancement of FM in B500V. The strong visiblePL of PTiO2 reveals that large concentrations of Ov defects arepresent in PTiO2. However, no measureable FM is observed inPTiO2, which is very interesting and is an indication that simple Ovdefects are not responsible for the ferromagnetic ordering in TiO2system.
In order to understand the origin of broad visible PL, we fit thespectrum of B500 with Gaussian peaks (Fig. 6(c)). The spectrum iswell fitted with three Gaussian peaks. The peak 1 is attributed tothe emission from self trapped excitons, while peak 2 and peak 3are assigned to the emission from Ov related Fþ and Ti3þ trap states[28,45,46]. Upon the loss of an O atom in TiO2 lattice, the electronpair that remains trapped in the vacancy cavity leaves behind a pairof electrons that give rise to an F-center. The basic assumption isthat one of the electrons in F-center tends to occupy the neigh-boring Ti4þ ion and yield Ti3þ-center and Fþ-center [28]. Further,we monitored the PL emission at 525 nm and measured the PLdecay rate (Fig. 6(d)). The PL decay is well fitted with a doubleexponential decay curve with time constants t1 ¼ 0.3 ns andt2 ¼ 2.5 ns. Thus, the emission at 525 nm is contributed by twodefect states, i.e., peak 2 and 3, as shown by the dotted vertical linein Fig. 6(c). Thus the steady state PL and time-resolved PL results arefully consistent with each other and it provides an unambiguousproof regarding the strong presence of Ov defects in these samples.
3.4. XPS and ESR studies
3.4.1. XPS studiesThe XPS measurements were carried out to obtain information
about the electronic valence states of the elements in the Cr:TiO2samples. The Ti 2p3/2 and Ti 2p1/2 core level peak positions ofsamples B500, B700 and D500 are at 458.0 eV and 463.7 eV,respectively, which indicates that Ti is in 4 þ oxidation state (seeFig. 7(a)). Fig. 7(b) shows the O 1s core level spectra of B500, B700and D500. The intense peak at 529.5 eV is attributed to the lattice Oassociated with Ti-cation, while the asymmetric broad shoulderingat higher binding energy may be due to the O attached to surfaceatmospheric carbon (OC) contamination and hydroxyl group (OH)[47,48]. The Cr 2p core level spectrum of B500 is shown in Fig. 7(c).The asymmetric Cr 2p3/2 peak indicates overlapping of multiplepeaks. So, the Cr 2p3/2 core level peak is deconvoluted by twoGaussian peaks centered at 577.0 and 579.7 eV. The peak at 577.0 eVis assigned to Cr3þ 2p3/2 state, while the peak at 579.7 eV is likely tobe related to Cr6þ 2p3/2 state. Further, the Cr 2p1/2 core level peak isobserved at ~586.0 eV, which is attributed to the Cr3þ state. Thecore level binding energy observed for the Cr 2p3/2 is different from~574.0 eV of Cr metal or 576.3 eV of CrO2 [49]. It strongly suggeststhat there is no metallic Cr or clusters of CrO2 phase on the surfaceof the nanostructure. The surface compositions of the Ti, O and Crare found to be 8.2%, 34.8% and 0.7% for B500, 5.2%, 35.5% and 0.8%for B700, 10.7%, 38.8% and 1.0% for D500, respectively (remainingpercentage is C% in each sample due to the carbon tape used for themeasurement). From the compositional analyses, it is clear thatwith calcination at higher temperature the O concentration isincreased, while the Ti concentration is decreased systematically.Further, the intensity of the OC/OH peak in the O1s spectrum isincreased dramatically in B700, indicating higher concentration ofO attached to the surface during calcination. Note that severalparamagnetic species related to surface adsorbed oxygen areobserved in sample B700 in the ESR spectrum, which is consistentwith XPS results. This indicates that B700 has less concentration ofoxygen vacancy as compared to B500 and thus the magnetism inB700 is less as compared to B500, as observed in VSM measure-ments, supporting the mechanism of oxygen vacancy induced
Fig. 6. Room temperature PL spectra for: (a) B500, B700 and B900, inset is the magnified view of the visible region of B500, B700, B900 and D500; (b) comparison of the visible PL ofB500, B500V and PTiO2; (c) visible PL of B500 with Gaussian peak fitting; (d) PL decay of B500 with bi-exponential decay fit. The PL peaks corresponding to the oxygen vacancy andTi interstitial defects are labelled as Ov and Tii, respectively.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344338
ferromagnetism. The surface Cr% is increased in the orderB500 < B700 < D500. However, the FM is decreased in the orderB500 > B700 > D500. This supports our assertion that the FM in Crdoped samples is due to the interaction between Ov and Cr dopants.
3.4.2. ESR studiesESR is an exceptionally powerful tool in detecting the spin
polarized charge states in defective TiO2 nanostructures. The roomtemperature ESR spectra of undoped and Cr doped samples alongwith PTiO2 are shown in Fig. 8. Fig. 8(a) shows the ESR spectra ofA500, B500 and C500 with g-values 2.00 and 1.97. The signalhaving g-value at 2.00 is assigned to singly ionized Ov (Fþ) [45,50],while the weak signal with g-value at 1.97 is assigned to regularlattice bulk Ti3þ [45,51] in undoped sample (A500) and the strongsignal at g ¼ 1.97 in Cr doped samples (B500, C500) can beassigned to the presence of both Cr3þ [52] and Ti3þ [45,51] species.The comparison of ESR spectra of B500 and B700 is shown inFig. 8(b). Besides the signal of Fþ, Cr3þ and Ti3þ, two additionalfeatures with g-values at 2.04, 2.09 are observed in B700. Thesesignals can be assigned to superoxide ion (O�
2 ) on the anatase TiO2surface due to surface adsorbed oxygen [51,53]. It has been re-ported that there are two specific sites for O2 adsorption inreduced TiO2, i.e., a five coordinate Ti3þ site and an Ov site [53].Since the sample B700 is calcined at 700 �C in air, the atmosphericO2 may be adsorbed on the surface of TiO2 and there is a possi-bility of forming two kinds of O�
2 as follows:
(i) Adsorbed O2 at five co-ordinate site may interact with Ti3þ
and one electron transfer from Ti3þ to O2, and generate a pairof Ti4þ and O�
2 species [51].
Ti3þ þ O2/Ti4þ þ O�2 (1)
(ii) Adsorption of O2 at oxygen vacancy site may interact with F,Fþ, Fþþ-centers and one electron transfers from F, Fþ-centerto adsorbed O2 and forms O�
2 as
F þ O2/Fþ þ O�2 (2)
Fþ þ O2/Fþþ þ O�2 (3)
Another possibility of formation of O�2 is due to the adsorbed O2
at the Fþþ center as shown below [54].
Ti3þe,eTi3þ þ O2/Ti3þeO�2eTi
4þ (4)
i.e., one electron from Ti3þ is transferred to O2, as a result Ti4þ
and O�2 species are formed. Where, the symbol ‘,’ represents Fþþ
with no electron. Note that in case of Fþþ-center, two of the elec-trons interact with nearby two Ti4þ and creates two Ti3þ speciesand no more electrons are left in the Fþþ-center. The signal atg ¼ 1.94 can be attributed to surface/interstitial Ti3þ present in theTiO2 NRs/NRbs samples [45,51,54]. The ESR signal of PTiO2 is shownin Fig. 8(c). The ESR spectrum of PTiO2 is similar to the spectrum ofB700, both having pure anatase phase, except for g ¼ 1.97, which isassigned to Cr3þ and/or bulk Ti3þ in B700. This reveals the presenceof super oxide ion (O�
2 ) along with the Fþ and surface/interstitialTi3þ species in anatase PTiO2 and indicates that Ov defects arepresent in the PTiO2, which is consistent with the PL results(Fig. 6(b)). Fig. 8(d) shows the comparison of ESR spectra for B500
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344 339
(0.3% Cr doped) and D500 (0.7% Cr doped) samples. We observedthat the ESR signal with g-values at 1.97 is enhanced for higher Cr3þ
dopant, i.e., for 0.7% doped sample as compared to 0.3% dopedsample, indicating that large concentration of Cr3þ species arepresent in D500 sample. Note that no trace of Ti3þ is observed inXPS Ti 2p spectrum of B500, but it is found in ESR spectrum, indi-cating the Ti3þ species are mostly present inside the bulk ratherthan the surface of NRs/NRbs in B500. This plays a very crucial rolein enhancing the stability of ferromagnetic properties in the TiO2system.
3.5. Magnetization studies
Magnetic measurements of the undoped and Cr doped TiO2nanostructures were carried out using a VSM. The field dependentmagnetizations (MeH) at room temperature of the undoped anddoped samples are shown in Fig. 9. All the as-synthesized samplesexhibit clear ferromagnetic hysteresis behavior with a saturationmagnetization (Ms). Fig. 9(a) shows the hysteresis loop of undopedA500 alongwith 0.3% Cr doped B500 and B500V. There is a ~2.6 foldenhancement in Ms in doped sample B500 as compared to theundoped A500. Interestingly, the magnetization further increasedafter moderate vacuum annealing (B500V). This indicates that Crdopant and Ov both contribute significantly in enhancing theferromagnetic ordering. Note that the precursor PTiO2 does notshow any measurable magnetization, which is shown in the inset(left top panel) of Fig. 9(a). Fig. 9(b) shows the comparison of MeHloop for samples B500, B700 and B900, which were calcined atthree different temperatures in air. The magnetic moment clearlydecreases with increase in calcination temperatures, which isattributed to the lower concentration of Ov as confirmed from thePL spectra (Fig. 6(a)). The comparison of magnetization of B500 andC500, which were grown at two different temperatures is shown inFig. 9(c). The saturation magnetization of C500 is less compared toB500. This may be due to different concentration of Ov in B500 andC500. It has been found that the samples grown at higher reactiontemperature have comparatively lower concentration of Ov [45].This may be a possible reason for the less magnetization in C500 ascompared to B500. The MeH loop of C500 (0.3% Cr doped sample)and D500 (0.7% Cr doped sample) are shown in Fig. 9(c). It isobserved that the sample D500 with higher doping concentration(0.7% Cr) has less magnetization as compared to C500 (0.3% Crdoped) and this strongly indicates that the observed FM in dopedsamples is not of extrinsic origin, i.e., not due to the CrO2 phase. Thesaturation magnetization (Ms), coercive field (Hc) and remanentmagnetization (Mr) values of the samples are summarized inTable 1. Clearly, theMs is the highest in the vacuum annealed B500Vsample, that shows highest concentration of Ov.
In order to clarify the origin of observed FM, representativetemperature dependent magnetic behavior, i.e., zero field cooled(ZFC) and field cooled (FC) magnetizations were measured in anapplied field of 500 Oe for the sample B500. Prior to the mea-surements, the sample was cooled from 300 to 30 K without a field(ZFC) and with a field of 500 Oe (FC). The ZFC and FC magnetiza-tions were recorded with increasing temperature from 30 to 300 K.The combined plot of both ZFC and FC curves [Fig. 10] shows thatmagnetization decreases with increasing temperature throughoutthe measured temperature range in both cases. The ZFC and FCcurves exhibit a clear deviation (i.e., branching between ZFC and FCcurves) up to 265 K [Fig. 10], which indicates the presence of
Fig. 7. XPS spectra: (a) Ti 2p core level spectra for B500, B700 and D500; (b) O 1s corelevel spectra for B500, B700 and D500; (c) Cr 2p core level spectra for B500, inset is theGaussian peak fitting of Cr 2p3/2, indicating contribution of Cr3þ and Cr6þ 2p3/2 peaks.
Fig. 8. Room temperature ESR spectra showing comparison of: (a) A500, B500 and C500; (b) B500 and B700; (c) B700 and PTiO2; (d) B500 and D500. The vertical dotted line in (a)indicates Fþ signal position.
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344340
ferromagnetic ordering in the sample [55,56]. No magnetic block-ing phenomenon (i.e., no hump or peak signal) has been observedin the ZFC and FC curves of the sample B500 [Fig. 10], indicating theabsence of any magnetic impurities and magnetic phases in thesample [57].
3.6. Mechanism of defect induced FM
Despite decades of studies on the oxide based DMS, there is alack of consensus on the origin of the FM inwide range of materialsobtained from a variety of synthesis methods, and the actualphysical mechanism remains an open question. Experimental re-sults reported by various groups are often contradictory. Inparticular, the underlying origin of RTFM in TiO2 based DMSs isintensely debated. The possible reasons for the inconsistent resultsmay have arisen due to the creation and distribution of differentconcentration of defects and/or formation of secondary ferromag-netic phases/metal clusters that are dependent on the growthmethods and processing conditions during sample preparations.However, due to the fact that the RTFM is observed in undoped TiO2
system, it helps to resolve the controversies partly about the issuesrelated to the crucial role of defects in the ferromagnetic ordering inTiO2 nanostructures [27,32].
First, we discuss the mechanism behind the FM in undoped TiO2
nanostructures. The role of Ov defects in inducing the FM inundoped TiO2 NRbs was demonstrated in our earlier report [28].Here, we explore a better understanding on the specific role ofoxygen vacancy related defects towards the observed FM bycomparing the ESR, PL and VSM results of both as-prepared
undoped TiO2 NRs/NRbs and PTiO2. If defects are the source ofFM, then what are the roles of defects in inducing ferromagneticordering in undoped TiO2? To answer the question, we made acomparative study of the as-synthesized undoped TiO2 NRs/NRbsand PTiO2. A large concentration of Ov defects is observed in PTiO2from the PL spectrum (Fig. 6(b)). However, no measureablemagnetism is observed in PTiO2 (inset of Fig. 9(a)), while theundoped A500 show a clear FM behavior (Fig. 9(a)). This indicatesthat a simple Ov model is not adequate to explain the FM orderingin undoped TiO2, though there must be some specific roles playedby the Ov defects in inducing the FM ordering. Due to absence of anO atom, the two electrons left behind in reduced TiO2 may formthree kinds of Ov related color centers, such as Ov with doublyoccupied electrons (F-center, with two electrons), singly ionized Ov(Fþ-center, with one electron) and doubly ionized Ov (Fþþ-center,with no electron at all). Another kind of defect species is possibledue to missing of an O atom from TiO2 lattice as: the extra twoelectrons may interact with neighboring Ti4þ ions and create twoTi3þ ions or one Ti3þ and one Fþ species. Coey et al. [58] reportedthat Ov with two electrons (F center) form 1s2 state, which mediateweak antiferromagnetic exchange. Da Pieve et al. [16] reported thatthe Ov with a Fþþ center and the associated absence of electronswould not favor FM ordering in Cr doped TiO2 systems. Pandey andChoudhary [59] proposed that an Ov leads to electron doping in theTiO2 system, but does not induce an appreciable magnetic moment.Yang et al. [36] reported antiferromagnetism in undoped TiO2which arises due to the antiferromagnetic ordering of two Ti3þ ionsclose to the Ov sites. They interpreted that the two electronsassociated with Ov will convert neighboring Ti4þ ions into two Ti3þ
Fig. 9. Magnetic field dependent magnetization (MeH) loop at room temperatureshowing comparison of hysteresis in samples: (a) A500, B500 and B500V; (b) B500,B700, B900; (c) B500, C500 and D500. The insets in each case show the MeH loops atlow field near origin in magnified scale. The inset in (a) (left top panel) shows the MeHbehavior of PTiO2, indicating no measurable magnetism.
Fig. 10. Temperature dependent magnetization (ZFCeFC) curves in the temperaturerange from 30 to 300 K under an applied field of 500 Oe for the sample B500.
Fig. 11. Schematic atomic presentation of the formation of BMPs, illustrating ferro-magnetic, antiferromagnetic ordering and isolated paramagnetic species. The yellowarrows on the Ti3þ and Cr3þ atoms represent unpaired 3d-electrons spins while the redarrow on Fþ is the spin of single electron in isolated Fþ-center. The shaded circlesrepresent the BMPs showing ferromagnetic coupling, and the shaded squares show theantiferromagnetic coupling. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of this article.)
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344 341
ions, which induce an equal local magnetic moment of 1mB at thesesites and results in net antiferromagnetism.37 Kim et al. [35] pre-dicted a weak ferromagnetic moment in TiO2 due to Ov defects.Moreover, isolated Ti3þ cations, interstitial Ti3þ and isolated Fþ areparamagnetic species, and do not contribute to FM without anyexchange interactions [see Fig. 11]. Therefore, creation of a partic-ular charge configuration and distribution of Ov, and its interactionwith neighboring Ti 3d electron is necessary to induce
ferromagnetic ordering. From the ESR spectrum, we observed Fþ
centre, surface/interstitial Ti3þ and O�2 species in PTiO2, while Fþ
centre, regular lattice bulk Ti3þ in undoped TiO2 NRs/NRbs. Thesurface Ti3þ may be isolated Ti3þ species and contribute to para-magnetism. The O�
2 species is also a paramagnetic species. Most ofthe electron spins may be randomly oriented in PTiO2 and thus wedo not observe any measureable magnetism. Moreover, the Fþ andsurface Ti3þ species may be formed due to the adsorbed O2, asdiscussed in Eqs. (2) and (4), respectively. So, we expect absence ofstrong exchange interaction between 1s1 electron spin of Fþ-centerand 3d1 electron spin of Ti3þ-center nearby Ov. For the sampleA500, the absence of O�
2 species in ESR spectrum revealed that theFþ and Ti3þ species are not due to adsorbed O2. One of the electrons
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344342
in F-center may interact with neighboring Ti4þ and generate Fþ andTi3þ trap centers. The electrons in Fþ-center localize and may formbound magnetic polarons (BMPs) by ordering the Ti3þ (3d1) elec-tron spin neighboring the oxygen vacancies, thereby gaining ex-change energy. The sed exchange interaction between the 1s1
electron spin in the Fþ center, which is localized in the vicinity of3d1 electron spin of Ti3þ ions within an orbit around oxygen va-cancies favors long-range FM [Fig. 11, shaded circle]. Thus the for-mation of BMP, which includes electrons locally trapped by oxygenvacancies, with the trapped electron in the Fþ center occupying anorbital overlapping with the unpaired electron (3d1) of Ti3þ isbelieved to explain the observed FM in our undoped TiO2 NRs/NRbs. In the literature, the following scenarios aremainly discussedto explain FM in oxide semiconductors: (i) oxygen vacancy-mediated (short range) exchange coupling, (ii) chemical in-homogeneity such as spinodal decomposition [60], and (3) carrier-mediated ferromagnetism [61,62].
Our data favors the oxygen vacancy-mediated (short range)exchange coupling model, while the contribution of carrier medi-ated FM cannot be fully ruled out, since the presence of Ti3þ and OVmay induce extra carriers. However, in absence of relevant data, itscontribution cannot be assessed properly.
Now, we discuss the mechanism of FM in Cr doped TiO2 NRs/NRbs. Unlike many other TMs, Cr itself is antiferromagnetic and sothe observed FM cannot have any extrinsic origin even if Cr clus-tering occurs. Further, trivalent Cr3þ ions exhibit 3d3 high-spinconfiguration, which may induce long range ferromagneticordering by exchange interaction with nearby oxygen vacancy inthe defective host semiconductors. The mechanism of FM in TMdoped TiO2 is rather complicated and several mechanisms havebeen proposed, such as double exchange and super exchangecoupling [16,21], carrier mediated RudermaneKitteleKasuyaeYashida-like interaction [63] and coupling of BMP which is formedwith a localized electron (usually induced by Ov defects) and thesurrounding TM cations [11e13], etc. However, the dominantmechanism is still unclear, though one common feature in all thesemodels is the vital role played by the defects and carriers and/ortheir coupling. The low concentration of Cr doping (0.3 at%) in ourCr:TiO2 systems does not favor the so-called double exchange andsuper exchange mechanism of FM. Since we observed FM inundoped system and a lower magnetic moment in 0.7% Cr dopedsample as compared to 0.3% Cr doped sample (Fig. 9), it stronglysuggests that a simple carrier mediated mechanism may not beapplicable in Cr:TiO2 NRs/NRbs. We observed that with increase incalcination temperatures, the FM decreases considerably due to thedecrease in Ov defects (see Fig. 6(a)). Further the FM is enhanced inthe vacuum annealed sample due to the introduction of largeconcentration of Ov. So, the strong correlation between Ov and FMsuggests that Ov is the key ingredient in the FM ordering. From theMeH measurements, it is found that the saturation magnetizationof 0.3% Cr doped sample is ~2.6 times higher than the undopedsample (Fig. 9(a)). This is quite reasonable, since the ionic charge ofCr3þ is different from that of Ti4þ; substitution of Cr3þ for Ti4þ
generate oxygen vacancies nearby Cr3þ in the lattice of TiO2 tomaintain charge neutrality and the additional Ov enhanced the FMin 0.3% Cr doped sample. Therefore, it is believed that the Ov playsthe pivotal and complimentary role in ferromagnetic ordering of3d3 spins of Cr3þ. Note that the presence of Cr3þ is confirmed fromour ESR and XPS analyses. However, the oxygen vacancies alonewithout any proper charge distribution are not adequate to estab-lish a robust ferromagnetism in the Cr:TiO2 nanostructures.
Recently, Da Pieve et al. [16] reported that the F-center exchangecannot be active in Cr-doped TiO2. Further, the Ov with two trappedelectrons (i.e., F-center) having 1s2 configuration exhibit antifer-romagnetic ordering [58]. Raebiger et al. [64] showed that in Cr
doped In2O3, the ferromagnetic CreCr interaction can be tuned andeven switched via electron doping. The extrinsic additional elec-trons doping, such as the large amount of donors induced by Ovcould provide the needed carriers to stabilize the long-rangeferromagnetic ordering, since Cr itself may not produce any freeelectrons [64]. The coupling between the two Cr3þ dopant spinsthrough Ov with trapped electron (Cr3þeOveCr3þ) may lead to longrange ferromagnetic ordering. However, if one of the Cr3þ is faraway from the Ov, then another possible source of antiferromag-netic interaction can be the existence of Ov with two trappedelectrons having 1s2 configuration [58]. Therefore, we believe thatthe partially filled Cr 3d states have to interact with the electron inFþ-center and the exchange interaction between 3d3 spins of Cr3þ
and 1s1 spin of Fþ-center is most likely responsible for the observedenhanced FM in Cr:TiO2 system along with the exchange interac-tion between 3d1 electron spin of Ti3þ and 1s1 spin of Fþ like inundoped systems [see Fig. 11, shaded circle]. Interestingly, thepresence of Fþ-center, Cr3þ and/or Ti3þ species in doped sampleswas confirmed from the ESR study. Note that O�
2 and surface Ti3þ
paramagnetic species created due to adsorbed O2 in B700 maypartly be responsible for the less magnetic moment in B700 ascompared to that of B500. The concentration of Ov is considerablylower in B900 as compared to that in B500, as confirmed from thePL analysis. Hence, the reduction of FM in the B900 is consistentwith the above model. Another possible mechanism for thereduction of FM with high temperature calcinations in 0.3% Cr:TiO2sample is the migration of Ti3þ defects towards the surface tointeract with atmospheric oxygen and convert to Ti4þ interstitialdefects during the course of calcinations and this may result in thereduction of BMPs. This is consistent with the PL analysis, since theinterstitial defects related NIR PL was strong in B900 (Fig. 6(a)).
Decrease in FM with increase in Cr concentration in TiO2 wasreported by Choudhury et al. [13] As we increase the doping of Cr3þ,some of the Cr3þ may reside in the interstitial site in the host TiO2or at the surface. All of the added Cr3þ may not be associated withan Ov and these Cr3þ ions may bind with oxygen ions byRef. Cr3þ�O2��Cr3þ bonds or may be associated with another Cr3þ
ion [13]. The decrease of magnetization may partly be due to theCr3þeCr3þ super exchange interaction in the absence of oxygenvacancy, which leads to antiparallel alignment of the spins, asshown schematically in Fig. 11 (shaded square). Chen et al. [21]suggested that the reduction of FM with high doping is possiblydue to the antiferromagnetic ordering between two nearby Fe3þ
ions in the absence of Ov by super exchange interaction. Forobserving FM, there should be exchange interaction of the 3d3 spinsof Cr3þ ions mediated by oxygen vacancies and for that purpose thedopant ions and defects should be situated close by. If the Cr3þ ionsare isolated or formed coupled pairs without any oxygen vacancies,antiferromagnetic interaction will take place and ultimately resultin the effective reduction of magnetization, which is quite plausiblefor higher concentration Cr:TiO2 nanostructures as in our case.Similar results were also observed in our Fe:TiO2 system with highconcentration of Fe [39].
Our results point to the fact that a simple Ov defect model is notadequate for the ferromagnetic ordering in TiO2, but Ov relatedspecific defects, dopant and their distribution, and the orbitaloverlapping/exchange interaction between 1s1 electron in Fþ and3d3 electron in Cr3þ or 3d1 electron in Ti3þ in the as-grown TiO2
nanostructures constitute the main ingredients to the observed FMin undoped and Cr doped TiO2 systems. The structural defectsenvironment associated with different morphologies may help inthe formation of BMPs, as the simple oxygen vacancies alonecannot decide ferromagnetic properties, as evidenced from theabsence of FM in the PTiO2 anatase powder. The structural andinterfacial defects may be responsible for the local atomic
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344 343
arrangement and orbital overlapping of 1s1 electron in Fþ and 3d3
electron in Cr3þ or 3d1 electron in Ti3þ and hence the formationBMPs and ferromagnetic ordering in TiO2 NRs/NRbs. Our resultssuggest that nature and structure of the defects in different mor-phologies and surface features of TiO2 nanostructures, such as NPs,NRs and NRbs may be different and gives rise to different concen-trations of BMPs. Our results indicate that the nanobricks/nano-stones like structures on the surface of TiO2 NRs/NRbs may behelpful in exhibiting strong ferromagnetic interaction. Our presentresults, however, show that the optical and magnetic properties ofthe TiO2 NR/NRb are dictatedmainly by the post-growth processingconditions. In particular, calcination in air shows reduced density ofoxygen vacancies and higher density of Ti interstitials mainly nearthe surface and these changes the respective PL and FM behaviordrastically. We find that higher temperature calcination showsmore nanoribbons than nanorods of TiO2 and these NRb exhibitreduced FM mainly due to reduced oxygen vacancy concentration,though nanoribbons possess higher surface area. Thus, it appearsthat the magnetic behavior of the TiO2 nanoribbons and nanorodsare not much different, if the post-growth processing temperaturesare identical for both. Further studies including theoretical calcu-lations are needed to understand the structure-property correla-tions in the context of BMP model.
4. Conclusions
Our studies revealed a strong correlation between Ov and RTFMin undoped and Cr doped TiO2 nanostructures grown by a hydro-thermal method. The enhanced FM after vacuum annealing of TiO2indicates direct involvement of Ov in the observed FM. Bycomparing the Ov related visible PL emission and saturationmagnetization from the MeH loops, we conclude that Ov is themost critical ingredient in dictating the magnitude of FM indifferent samples. Extrinsic origin of FM in TiO2 was ruled out fromour study. In contradiction to earlier reports, our results demon-strate that a simple Ov model is not adequate for the ferromagneticordering. Our results point to the fact that Ov related specific de-fects, optimum concentration of dopant and their distribution, andthe orbital overlapping/exchange interaction between 1s1 electronin Fþ and 3d3 electron in Cr3þ or 3d1 electron in Ti3þ in the as-grown TiO2 nanostructures are most critical ingredients to theobserved RTFM in the undoped and Cr doped TiO2 systems. Theenhanced saturation magnetization in Cr doped TiO2 NR/NRb ascompared to undoped samples indicates that Cr3þ play an impor-tant role in ferromagnetic ordering in the presence of Ov in Cr:TiO2systems. The decrease in FM at higher Cr concentration is attributedto the antiferromagnetic ordering caused by the super exchangeinteraction between two neighboring Cr3þ ions in the absence ofOv. The observed RTFM is explained on the basis of BMPs whoseformation is due to the sed exchange interaction between Ti3þ-Fþ
and Cr3þeFþ in the vicinity of Ov and the overlapping of more BMPsresult in enhanced FM. An optimal concentration of Cr doped intothe substitutional site of Ti4þ and the creation of nearby Ov are theprimary requirement for the enhanced FM in Cr:TiO2 nano-structures. These findings not only help to gain a better insight intothe defect engineering of RTFM in undoped and Cr doped TiO2-systems, but also constitute an important step for the developmentof practical nanospintronic devices, which can be operated at roomtemperature.
References
[1] H. Ohno, Science 281 (1998) 951.[2] S.A. Wolf, D.D. Awschalom, R.A. Buhrman, J.M. Daughton, S. von Moln�ar,
[3] Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura,M. Kawasaki, P. Ahmet, T. Chikyow, S.-y. Koshihara, H. Koinuma, Science 291(2001) 854.
[4] B. Pal, P.K. Giri, J. Appl. Phys. 108 (084322) (2010).[5] B. Ghosh, M. Sardar, S. Banerjee, J. Phys. D. Appl. Phys. 46 (2013) 135001.[6] X. Liu, J. Iqbal, Z. Wu, B. He, R. Yu, J. Phys. Chem. C 114 (2010) 4790.[7] K. Dodge, J. Chess, J. Eixenberger, G. Alanko, C.B. Hanna, A. Punnoose, J. Appl.
Phys. 113 (2013) 17B504.[8] G.Z. Xing, J.B. Yi, D.D. Wang, L. Liao, T. Yu, Z.X. Shen, C.H.A. Huan, T.C. Sum,
J. Ding, T. Wu, Phys. Rev. B 79 (2009) 174406.[9] S. Wang, Y. An, D. Feng, Z. Wu, J. Liu, J. Appl. Phys. 113 (2013) 153901.
[10] B. Santara, B. Pal, P.K. Giri, J. Appl. Phys. 110 (2011) 114322.[11] Y. Liao, H. Zhang, J. Li, G. Yu, Z. Zhong, F. Bai, L. Jia, S. Zhang, P. Zhong, J. Appl.
Phys. 115 (2014) 17C304.[12] S.K.S. Patel, N.S. Gajbhiye, J. Magn. Magn. Mater. 330 (2013) 21.[13] B. Choudhury, A. Choudhury, Mater. Sci. Eng. B 178 (2013) 794.[14] L.C.J. Pereira, M.R. Nunes, O.C. Monteiro, A.J. Silvestre, Appl. Phys. Lett. 93
M. Weissmann, S.J. Stewart, J. Phys. Conden. Matter 20 (2008) 135210.[23] L. Ranno, A. Barry, J.M.D. Coey, J. Appl. Phys. 81 (1997) 5774.[24] R.C. DeVries, Mater. Res. Bull. 1 (1966) 83.[25] B. Kubota, J. Am, Ceram. Soc. 44 (1961) 239.[26] A. Yahya, G. Ted, C. Cigdem, W. Iwamoto, P.G. Pagliuso, M. Sabee, Nanotech-
nology 24 (2013) 275704.[27] N.H. Hong, J. Sakai, N. Poirot, V. Briz�e, Phys. Rev. B 73 (2006) 132404.[28] B. Santara, P.K. Giri, K. Imakita, M. Fujii, Nanoscale 5 (2013) 5476.[29] M. Venkatesan, C.B. Fitzgerald, J.M.D. Coey, Nature 430 (2004) 630.[30] N.H. Hong, N. Poirot, J. Sakai, Phys. Rev. B 77 (2008) 033205.[31] X. Xu, C. Xu, J. Dai, J. Hu, F. Li, S. Zhang, J. Phys. Chem. C 116 (2012) 8813.[32] N. Hoa, D. Huyen, J. Mater. Sci. Mater. Electron 24 (2013) 793.[33] H. Wang, Z. Zong, Y. Yan, J. Appl. Phys. 115 (2014) 233909.[34] H. Peng, J. Li, S.-S. Li, J.-B. Xia, Phys. Rev. B 79 (092411) (2009).[35] D. Kim, J. Hong, Y.R. Park, K.J. Kim, J. Phys. Conden. Matter 21 (2009) 195405.[36] K. Yang, Y. Dai, B. Huang, Y.P. Feng, Phys. Rev. B 81 (2010) 033202.[37] A. Sundaresan, C.N.R. Rao, Nano Today 4 (2009) 96.[38] S. Ghosh, G.G. Khan, K. Mandal, A. Samanta, P.M.G. Nambissan, J. Phys. Chem.
C 117 (2013) 8458.[39] B. Santara, P.K. Giri, S. Dhara, K. Imakita, M. Fujii, J. Phys. D. Appl. Phys. 47
B 60 (1999) 33.[41] O. Monnereau, L. Tortet, C.E.A. Grigorescu, D. Savastru, C.R. Iordanescu,
F. Guinneton, R. Notonier, A. Tonetto, T. Zhang, I.N. Mihailescu, D. Stanoi,H.J. Trodahl, J. Optoelectron. Adv. Mater 12 (2010) 1752.
[42] B. Zhao, F. Chen, W. Qu, J. Zhang, J. Solid State Chem. 182 (2009) 2225.[43] T. Jiang, C. Jia, L. Zhang, S. He, Y. Sang, H. Li, Y. Li, X. Xu, H. Liu, Nanoscale 7
(2015) 209.[44] J. Zhu, Z. Deng, F. Chen, J. Zhang, H. Chen, M. Anpo, J. Huang, L. Zhang, Appl.
Catal. B 62 (2006) 329.[45] B. Santara, P.K. Giri, K. Imakita, M. Fujii, J. Phys. Chem. C 117 (2013) 23402.[46] J. Shi, J. Chen, Z. Feng, T. Chen, Y. Lian, X. Wang, C. Li, J. Phys. Chem. C 111
(2007) 693.[47] J. Zhuang, S. Weng, W. Dai, P. Liu, Q. Liu, J. Phys. Chem. C 116 (2012) 25354.[48] J.C. Yu, J. Yu, H.Y. Tang, L. Zhang, J. Mater. Chem. 12 (2002) 81.[49] I. Ikemoto, K. Ishii, S. Kinoshita, H. Kuroda, M.A. Alario Franco, J.M. Thomas,
J. Solid State Chem. 17 (1976) 425.[50] S.V. Chong, K. Kadowaki, J. Xia, H. Idriss, Appl. Phys. Lett. 92 (2008) 232502.[51] E. Carter, A.F. Carley, D.M. Murphy, J. Phys. Chem. C 111 (2007) 10630.[52] I.S. Pentegov, E.A. Konstantinova, Phys. Status Solidi C. 8 (2011) 1954.[53] K. Komaguchi, T. Maruoka, H. Nakano, I. Imae, Y. Ooyama, Y. Harima, J. Phys.
Chem. C 114 (2010) 1240.[54] J. Strunk, W.C. Vining, A.T. Bell, J. Phys. Chem. C 114 (2010) 16937.[55] B. Choudhury, R. Verma, A. Choudhury, RSC Adv. 4 (2014) 29314.[56] H.J. Meng, D.L. Hou, L.Y. Jia, X.J. Ye, H.J. Zhou, X.L. Li, J. Appl. Phys. 102 (2007)
073905.[57] R.K. Zheng, H. Liu, Y. Wang, X.X. Zhang, Appl. Phys. Lett. 84 (2004) 702.[58] J.M.D. Coey, M. Venkatesan, C.B. Fitzgerald, Nat. Mater 4 (2005) 173.[59] S.K. Pandey, R.J. Choudhary, J. Phys. Conden. Matter 23 (2011) 276005.[60] K. Sato, L. Bergqvist, J. Kudrnovský, P.H. Dederichs, O. Eriksson, I. Turek,
B. Sanyal, G. Bouzerar, H. Katayama-Yoshida, V.A. Dinh, T. Fukushima,
B. Santara et al. / Journal of Alloys and Compounds 661 (2016) 331e344344
H. Kizaki, R. Zeller, Rev. Mod. Phys. 82 (2010) 1633.[61] J. Philip, A. Punnoose, B.I. Kim, K.M. Reddy, S. Layne, J.O. Holmes, B. Satpati,
P.R. LeClair, T.S. Santos, J.S. Moodera, Nat. Mater 5 (2006) 298.[62] Y. Yamada, K. Ueno, T. Fukumura, H.T. Yuan, H. Shimotani, Y. Iwasa, L. Gu,
S. Tsukimoto, Y. Ikuhara, M. Kawasaki, Science 332 (6033) (2011) 1065e1067.[63] Z. Wang, W. Wang, J. Tang, L.D. Tung, L. Spinu, W. Zhou, Appl. Phys. Lett. 83
(2003) 518.[64] H. Raebiger, S. Lany, A. Zunger, Phys. Rev. Lett. 101 (2008) 027203.
