The importance of inversion disorder in the visible light induced persistent luminescence in Cr 3+ doped AB 2 O 4 (A = Zn or Mg and B = Ga or Al)
Neelima Basavaraju, K. R. Priolkar, Didier Gourier, Suchinder K. Sharma, Aurelie Bessiere, Bruno Viana
aa r X i v : . [ c ond - m a t . m t r l - s c i ] D ec Importance of inversion disorder in visible light induced persistent lu-minescence in Cr + doped AB O (A = Zn or Mg and B = Ga or Al) Neelima Basavaraju, a Kaustubh R. Priolkar, ∗ a Didier Gourier, b Suchinder K. Sharma, b Aur´elieBessi`ere, b and Bruno Viana, b Received Xth XXXXXXXXXX 20XX, Accepted Xth XXXXXXXXX 20XXFirst published on the web Xth XXXXXXXXXX 200X
DOI: 10.1039/b000000x Cr + doped spinel compounds AB O with A=Zn, Mg and B=Ga, Al exhibit a long near infrared persistent luminescence whenexcited with UV or X-rays. In addition, persistent luminescence of ZnGa O and to a lesser extent MgGa O , can also be in-duced by visible light excitation via A → T transition of Cr + , which makes these compounds suitable as biomarkers for invivo optical imaging of small animals. We correlate this peculiar optical property with the presence of antisite defects, whichare present in ZnGa O and MgGa O . By using X-ray absorption fine structure (XAFS) spectroscopy, associated with electronparamagnetic resonance (EPR) and optical emission spectroscopy, it is shown that an increase in antisite defects concentrationresults in a decrease in the Cr-O bond length and the octahedral crystal field energy. A part of the defects are in the close envi-ronment of Cr + ions, as shown by the increasing strain broadening of EPR and XAFS peaks observed upon increasing antisitedisorder. It appears that ZnAl O , which exhibits the largest crystal field splitting of Cr + and the smallest antisite disorder,does not show considerable persistent luminescence upon visible light excitation as compared to ZnGa O and MgGa O . Theseresults highlight the importance of Cr + ions with neighboring antisite defects in the mechanism of persistent luminescenceexhibited by Cr + doped AB O spinel compounds. Absorption of incident radiation (visible, UV or higher energy)by materials, and delayed subsequent emission most often inthe visible range, is termed as long-lasting phosphorescence(LLP) or persistent luminescence. This phenomenon is causedby the trapping of charges (electrons and/or holes) by defectspresent in the material, preventing fast recombination of thecharges. Detrapping of these charges most often proceeds viathermal activation giving rise to a progressive radiative recom-bination prolonging the emission up to several hours. Thesematerials known as persistent phosphors, were demonstrated tobe used as probes for in vivo small animal optical imaging in2007, when emitting in the red or near infrared (NIR) range. This technique is advantageous over conventional fluorescencetechniques as it avoids autofluorescence of body tissues undercontinuous illumination and thus improves the signal to noiseratio. † Electronic Supplementary Information (ESI) available: [details of anysupplementary information available should be included here]. See DOI:10.1039/b000000x/0 a Department of Physics, Goa University, Taleigao plateau, Goa 403206, India.Tel: 0832 651 9084; E-mail: [email protected] b PSL Research University, Chimie ParisTech - CNRS, Institut de Recherche deChimie Paris, 75005, Paris, France.
The technique of in vivo imaging was first demon-strated using silicate nanoparticles with compositionCa . Zn . Mg . Si O :Eu + , Dy + , Mn + (CZMSO), with 2.5mol% of Mn + luminescent ion doping. A new compound,Cr + doped ZnGa O was reported by Bessi`ere et. al. in2011 to be a potential candidate for this imaging application,with enhanced LLP properties. It was demonstrated inthere that LLP results from Cr + distorted by a neighboringantisite defect, which is the defect resulting from a zinc ionexchanging site position with Ga ion or vice versa. Soon afterthat, 0.25 mol% Cr + doped ZnGa O compound, preparedby hydrothermal method was tested for in vivo imaging andwas shown to be a suitable biomarker. However for solidstate synthesis, 0.5 mol% Cr + doping was found to be theoptimum concentration to get highest LLP. Recently, manyother modified gallate spinels doped with Cr + ions have beendiscovered to show similar red/NIR LLP emission. Pan et.al. reported Cr + doped zinc gallogermanates which showeda very long NIR afterglow. An extensive study was doneby Allix et. al. on Ge and Sn substituted zinc gallates andthey concluded that, substitution of Ga + by Ge + or Sn + in octahedral sites probably increases inversion in the spinelstructure, although the substitution may play an extra role increating more local defects and increase the delayed emission. + doped magnesium gallate was also shown to be agood enough phosphor for in vivo imaging with an argumentthat structural inversion is an important factor governing theLLP property. However, ZnGa O host is a much preferredmaterial to study the LLP mechanism due to its comparativelysimpler structure and well resolved Cr + energy levels.ZnGa O is known to crystallize in normal spinel structure(cubic space group Fd3m ) with Zn + ions in tetrahedral coor-dination and Ga + ions in octahedral coordination. However,the existence of a few percent of inversion in site occupanciesof Zn and Ga is reported in literature. Among the spinelfamily AB O (with A=Zn, Mg and B=Ga, Al), MgGa O isby and large an inverse spinel compound with about 44% oc-tahedral sites occupied by Mg + ions. On the other hand,ZnAl O is reported to be a near normal spinel structure withless than 1% cation disorder. When these spinels are dopedwith Cr + in octahedral positions and excited by UV/X-rays,they emit in red/NIR region (around 700 nm). This emissionarises due to E( G) → A ( F) Cr + d-d transition. Very recently, the LLP mechanism was investigated in de-tail by thermally stimulated luminescence (TSL) studies doneon chromium doped zinc gallate demonstrating that, direct d-dexcitation of Cr + ions by visible light gives a persistent lumi-nescence dominated by Cr + ions possessing an antisite defectin its first cationic neighbour (referred to as Cr N ion). Theessential feature of the proposed mechanism is that, charge sep-aration and trapping occur in assistance with the local electricfield created by the presence of a pair of complementary an-tisite defects around Cr + ion. Thus the energy is stored inthe form of an electron-hole pair and does not involve valencestate change of Cr + ion. The model efficiently explains howLLP can be excited even with lower energy visible radiation. This hypothesis was further augmented by electron paramag-netic resonance (EPR) studies carried out on chromium dopedzinc gallate compounds with varying Zn/(Ga+Cr) nominal ra-tio. The defects around chromium were identified by corre-lating photoluminescence (PL) and EPR spectroscopy. Hereagain it was shown that, Cr N ion plays a key role in both LLPexcitation and emission. To further substantiate the mechanism deduced from op-tical and EPR studies, X-ray absorption fine structure spec-troscopy (XAFS) measurements were carried out in Cr + doped ZnGa O to understand the local structures aroundall the cations. Moreover, the validity of the mecha-nism is extended by studying local environment of Cr + inMgGa O :Cr + and ZnAl O :Cr + . Based on interference ofa photoelectron wave emitted from an atom due to absorptionof a X-ray photon and a back scattered wave resulting due toits scattering from neighboring ions, XAFS provides informa-tion about immediate surroundings of the central absorbing ion.Therefore, XAFS can be used to probe structural defects arounda metal ion and is an ideal technique to understand the local structure of an atom in the lattice. It works virtually for allelements and even in cases where the concentration of absorb-ing ion is very low (few ppm). This makes it a versatile toolto study persistent luminescent materials since the dopant con-centration will usually be very low. In this paper, we havecarried out XAFS measurements at Cr K edges in chromiumdoped ZnGa O , MgGa O and ZnAl O compounds, charac-terized by optical and EPR spectroscopy. Further, local struc-tures around Zn, Ga and Cr have been studied in Cr + dopedZnGa O compounds with varying Zn/Ga nominal ratio. The samples were synthesized by solid state method with theirrespective metal oxides ZnO (Sigma Aldrich 99.99% pure),Ga O (Sigma Aldrich 99.99% pure), MgO (Sigma Aldrich99.995% pure), Al O (SRL 99.75% pure) and CrO (SRL99% pure) as precursors. Weighed powders along with propan-2-ol were thoroughly mixed in an agate mortar and the driedmixture was pelletized under 4 tons pressure in a hydraulicpress. All the pellets were annealed in air at 1300 ◦ C for 6hours except ZnAl O which was annealed at 1400 ◦ C, andlater crushed to get fine powders for further characterizations.ZnGa O (noted ZGO) compounds were prepared in three dif-ferent molar ratios - Zn/Ga = 0.5 (noted s-ZGO for stoichiomet-ric ZGO), Zn/Ga = 0.495 (d-ZGO for 1 mol% Zn deficiency)and Zn/Ga = 0.505 (e-ZGO for 1 mol% Zn excess) with 0.5mol% Cr introduced relative to Ga. It has to be noted thatvarying the nominal Zn/Ga ratio in the reactants mix does notmean that Zn or Ga atoms are in excess/deficiency in the ob-tained ZGO compounds. It was indeed shown that tiny quanti-ties of ZnO/Ga O could either be present as very minor impu-rities or evaporate during high temperature synthesis. How-ever, varying the nominal Zn/Ga ratio had a very definite effecton the number of point defects in d-ZGO, s-ZGO and e-ZGOcompounds.
Therefore letters d, s and e only reflect defi-ciency, stoichiometry and excess in the reactants mix withoutpre-supposing the stoichiometry of the actually formed ZGOcompounds. To compare the effect of Cr + doping on LLPproperties in different host lattices, MgGa O (noted MGO)with 1 mol% Mg deficiency (noted d-MGO) and 0.5 mol% Crdoping relative to Ga, and ZnAl O (noted ZAO) with 1 mol%Zn deficiency (noted d-ZAO) and 0.5 mol% Cr doping relativeto Al, were synthesized.Room temperature (RT) photoluminescence (PL) excitationspectra were recorded on Varian Cary Eclipse spectrofluorime-ter in the range 190 nm-650 nm with xenon lamp as excitationsource. Pulsed laser excited PL was run on 8 mm-diameterpellets silver glued on the cold finger of a cryogenic systemmaintained at 20 K. The emitted light was collected by an opti-cal fiber and transmitted to a Scientific Pixis 100i CCD cameracooled at -20 ◦ C and coupled to a monochromator with 12002roves/mm grating. The pellets were excited at 230 nm by anoptical parametric oscillator (OPO) EKSPLA NT342B. The PLemission spectra were measured with 10 ms gate width and 26ns gate delay. LLP measurements were carried out at RT on 180mg samples filled into a 1 cm diameter circular sample holder.The samples were illuminated for 15 minutes with X-rays (Mo-tube, 20 mA-50 kV) or for 30 minutes in the A → T Cr + band with OPO. After this excitation, emission was collectedusing a Scientific Pixis 100 CCD camera via an optical fiberlinked to an Acton SpectraPro 2150i spectrometer for spectralanalysis. LLP emission spectra were recorded during the exci-tation and 5 s after the end of excitation. All the samples werebleached at 250 ◦ C for 30 minutes and were kept in the darkprior to LLP measurements.X-band ( ∼ a radiation.The spectra were recorded in 2 q range 20 ◦ -80 ◦ with 0.02 ◦ stepand 2 ◦ /min scan speed. Rietveld refinement on the XRD pat-terns was carried out using FullProf software. XAFS at RTwere measured on the samples in fluorescence mode for CrK edge and in transmission mode for Zn K and Ga K edges,at SAMBA beamline in Soleil synchrotron facility, France. Si(111) crystal plane was used as the monochromator. For fluo-rescence measurements, absorbers were prepared by mixing 50mg compound with 100 mg boron nitride and pressing each ofthem into 10 mm pellets while for transmission, the appropriateamount of finely ground powder was deposited on a membrane.Fluorescence yield was collected via Canberra 35 pixels SSDdetector and transmitted photons were counted using ionizationchamber with appropriate gases. Extended X-ray absorptionfine structure spectroscopy (EXAFS) fitting was carried out us-ing Ifeffit software with Athena and Artemis programs. EX-AFS data in the k range of 2 to 11 ˚A − for Zn, 2 to 14 ˚A − forGa and 3 to 10 ˚A − for Cr K edge were Fourier transformed,and the fitting was performed in the R range of 1 to 3.6 ˚A toobtain reasonable fits. Theoretical amplitude and phase infor-mation for various scattering paths were obtained using FEFF6.01 and the Rietveld refined parameters. Photoluminescence excitation spectra measured at RT for s-ZGO, e-ZGO, d-ZGO, d-MGO and d-ZAO are presented inFigure 1 (a). The spectra consist of host band gap excitationpeaking at about 245 nm for ZGO compounds, 230 nm for d-MGO and 205 nm for d-ZAO, and three broad absorption bandsaround 230-270 nm, 410 nm and 560 nm belonging to A ( F)
200 300 400 500 600
684 687 690 693 (a)
Wavelength (nm) d-ZGOs-ZGOe-ZGO d-MGOd-ZAO P L i n t e n s it y ( a . u . ) (b) N2N1R1R2*15 d-ZGO d-MGO d-ZAO s-ZGO e-ZGO
Fig. 1 (a) PL excitation spectra of Cr + doped ZGO compounds withvarying Zn/Ga nominal ratio, d-MGO and d-ZAO compounds,measured at room temperature. (b) Corresponding Zero-phonon PLemission spectra at 20 K and excited at 230 nm. → T ( P), A ( F) → T ( F) and A ( F) → T ( F) Cr + d-d transitions respectively. It was previously shown that, crys-tal field around Cr + ion in d-MGO is weaker than that of ZGOcompounds and hence a red shift is seen in Cr + absorptionbands of d-MGO compared to ZGO. On the contrary in d-ZAO, Cr + absorption bands are shifted to shorter wavelengthscompared to ZGO compounds indicating that the crystal fieldaround Cr + ion is stronger than that of ZGO. Within the ZGOcompounds, the Cr + absorption bands are seen to be graduallymoving towards longer wavelengths with the increasing Zn/Ganominal ratio (deficiency to stoichiometric to excess) revealinga decrease in the crystal field around Cr + ion.The influence of defects around Cr + ions among differentsamples could also be seen in PL emission at 20 K under 230nm excitation (Figure 1 (b)). The emission lines correspond totypical E → A transitions of Cr + with either unperturbedenvironment and/or perturbation from the nearby crystal de-fects. Only zero phonon lines (ZPL) are shown in Figure 1 (b).3ulti phonon side bands (PSB) which occur at shorter wave-lengths, are not shown here. For the sake of comparison, thespectra except for d-MGO sample are normalized to R1 line.The d-MGO spectrum is amplified by a factor of 15 with re-spect to d-ZGO, for better visibility and comparison. The emis-sion spectra can be categorized into three main ZPL regions,namely, (i) R1 and R2 region; (ii) N1 region and (iii) N2 re-gion, separated by dotted lines in Figure 1 (b). R1 and R2 linescorrespond to the Cr + ions with an unperturbed ideal environ-ment (referred to as Cr R centers). The distinct R1 and R2 linesare ascribed to the splitting of E excited state of Cr + due totrigonal distortion into two levels separated by ∼
40 cm − inZGO and ∼ − in ZAO. Similarly, the N1 line (re-ferred to as Cr N ) may correspond to either a Cr + - V Zn pair( V Zn is a Zn vacancy) or a Cr + - Zn i pair ( Zn i is an intersti-tial Zn) or a Cr + close to an antisite defect . The N2line is assigned to the presence of antisite defects (presumably Zn Ga ′ ) close to Cr + ion as first cationic neighbour (referred toas Cr N ). The emission wavelengths of all these linesare reported in Table 1. As N lines are due to Cr + perturbedby neighboring antisite defects, their intensity (compared to Rlines) increases with increasing inversion disorder. This showsthat d-ZAO and d-ZGO are the less disordered materials, fol-lowed by s-ZGO and e-ZGO in increasing order. The disorderis so high in d-MGO that emission features are difficult to in-terpret and the main emission line is situated at 707 nm withtotal of six ZPLs revealing up to six different environments forCr + ions. Table 1
Characteristics of zero-phonon emission lines (ZPL) ofCr + -doped ZGO, ZAO and MGO compounds. Sample ZPL emission line (682-695nm)R1 (nm) R2 (nm) N1 (nm) N2 (nm)d-ZAO 686.0 684.2 688.5 692.2d-ZGO 686.4 684.6 689.3 692.9s-ZGO 686.9 685.1 689.7 693.4e-ZGO 687.2 685.4 689.9 693.6d-MGO N. V. N. V. N. V. N. V.N. V. - Not visibleThe persistent luminescence decay curves obtained with X-ray and laser (wavelength at the maximum of A ( F) → T ( F) Cr + d-d transition) excitations are presented in Fig-ure 2. d-MGO is known to show less LLP intensity comparedto ZGO with both X-ray and laser excitations as reported ear-lier. This is presumably due to excess cationic inversion whichpartially quenches LLP by introducing a direct recombinationpathway between abundant defects. The d-ZAO compoundwhich is a near normal spinel, shows almost same or slightly Laser LL P i n t e n s it y ( a . u . ) Time (s) d-ZGO d-MGO d-ZAO
X-ray
Fig. 2
LLP decay curves of Cr + -doped compounds after 15 minuteX-ray irradiation and after 30 minute laser excitation of wavelength560 nm for d-ZGO, 585 nm for d-MGO and 540 nm for d-ZAOsamples. These wavelengths correspond to the maximum of the A ( F) → T ( F) absorption band of Cr + . better LLP intensity as compared to d-ZGO compound withX-ray excitation. However, d-ZAO shows very feeble LLP in-tensity with 540 nm laser excitation, although this excitationcorresponds to the maximum of the A ( F) → T ( F) ab-sorption band of Cr + . On the other hand, d-ZGO shows con-sistently better LLP with both excitations as seen in Figure 2.Also, this distinctive property of d-ZGO to show LLP with laserexcitation renders the possibility for re-excitation of the ZGOnanoparticles inside the animal body by visible light illumina-tion, making it a favourable candidate for the application of invivo imaging. Comparison of emission spectra recorded for d-ZGO and d-ZAO compounds during the excitation and during LLP emis-sion (after the end of excitation) are shown in Figure 3. Emis-sion spectra measured during X-ray excitation (Figure 3 (a))show intense R-line for both the compounds, and correspondsto the E → A ZPL of Cr + . This ZPL is flanked by multiphonon sidebands. N2 line which arises due to emission fromCr + ions with an antisite defect in its first cationic neighbouris clearly seen in d-ZGO spectrum. It is difficult to identify4
00 750 600 750
RRN2
X-ray ONd-ZGO P L i n t e n s it y ( a . u . ) d-ZAO (a) (b) RRN2
X-ray OFF (c)
RN2R
Laser ON (d)
Laser OFF
Wavelength (nm) N2 Fig. 3
Emission spectra measured for d-ZGO (black curves) andd-ZAO (blue curves) (a) during X-ray irradiation; (b) 5 s after the endof 15 minute X-ray illumination; (c) during optical laser excitation at540 nm for d-ZAO and 560 nm for d-ZGO; (d) 5 s after the end of 30minute laser excitation. Position of R line is indicated by an arrow. the presence of N1 and N2 in d-ZAO because of the intensemulti PSB in the same range. However low temperature emis-sion spectra (Figure 1 (b)) without PSB clearly shows the pres-ence of N1 and N2 lines. The X-ray excited LLP emissionspectrum for d-ZAO shows an intense R-line (which is weaklysplit) whereas d-ZGO exhibits a prominent N2 line (Figure 3(b)). This indicates that for X-ray excitation, the charge recom-bination is taking place mainly at undistorted Cr + ions (Cr R )in d-ZAO, while it occurs through Cr N in d-ZGO. With laserexcitation, the emission spectra recorded while the laser is on(Figure 3 (c)) show R and N2 lines for both compounds. Whenthe laser is switched off (LLP emission), d-ZAO shows a broadweak emission (Figure 3 (d)), whereas d-ZGO spectrum againexhibits a prominent N2 line. This indicates the crucial role ofCr N ions and antisite defects in LLP emission when excited inlow energy absorption bands of Cr + .Experimental and simulated EPR spectra of Cr + in d-ZAO,d-ZGO and d-MGO are reported in Figure 4. For the directspinels d-ZGO and d-ZAO, which exhibit a weak inversion dis-order, the spectrum is composed of a strong line around 175mT, a weak line around 330-350 mT and other lines at ∼ ∼ + ion in a weakly ax- ially distorted octahedral site, in agreement with the C3 sym-metry of Ga and Al sites in these compounds. The effect ofantisite defects is a strain broadening of these EPR lines, whichincreases with magnetic field strength so that EPR lines at highmagnetic field are no longer observed in s-ZGO and e-ZGO. A zoom of transition at the low magnetic field (175 mT) showsthat the line is slightly broader in d-ZGO than in d-ZAO, point-ing to a slightly higher disorder in d-ZGO than in d-ZAO. Theinverse spinel d-MGO gives a very different EPR spectrum. Westill recognize broad lines at ∼ ∼
900 mT attributable to Cr + in octahe-dral sites affected by important disorder due to abundant anti-site defects. The spectrum also exhibits a broad and symmet-rical line in the field range ∼
300 mT. This line was also ob-served in disordered s-ZGO and e-ZGO materials. Based onthe temperature dependence of its intensity, this line was ten-tatively attributed to clusters of antiferromagnetically coupledCr + ions. The decomposition of the simulated spectrum ofCr + in d-MGO is shown in Supplementary Figure 1, and thesimulation parameters of all compounds are reported in Sup-plementary Table 1. The parameter which controls the shape ofthe powder EPR spectrum of Cr + in these compounds is thezero field splitting (ZFS) parameter ‘D’. It represents the split-ting 2D of the A ground state by the combined effect of thespin-orbit coupling and the trigonal distortion of the octahedralcrystal field. d-ZAO exhibits the largest splitting (2D = 1.864cm − ) compared to d-MGO (2D = 1.270 cm − ) and d-ZGO(2D = 1.050 cm − ). This corresponds to a decreasing trigonaldistortion in the sequence d-ZAO > > d-MGO > d-ZGO, whichcan be compared to octahedral crystal field splitting decreasingin the sequence d-ZAO > d-ZGO > d-MGO (see Figure 1 (a)).Thus the aluminate d-ZAO has more crystal field splitting D,more trigonal distortion and less antisite disorder than gallatesd-ZGO and d-MGO.To complement optical and EPR studies, and to investigate inmore details the environment around Cr + ion in these lattices,structural studies including XRD and EXAFS measurementshave been carried out. XRD patterns of all the compoundss,d,e-ZGO, d-MGO and d-ZAO are shown in SupplementaryFigure 2 and Rietveld refinement parameters are given in Sup-plementary Table 2. XRD indicates the formation of pure cu-bic spinel compounds. As previously reported, a minor ∼ A clear shift in the peak positionsis observed in Cr + doped host lattices spectra, correspondingto the variation in lattice parameters which is mostly the resultof difference in ionic sizes of cations. The lattice parameterdecreases in the sequence ZGO > d-MGO > d-ZAO. Not muchvariation was visibly seen among ZGO compounds however,Rietveld refinement indicated lowering lattice constant valueswith increasing Zn/Ga nominal ratio. Rietveld refinement wasdone on all the XRD patterns with A site cations in tetrahedral5
200 400 600 800 1000 1200
120 160 200 d-ZAO d-ZGO E P R i n t e n s it y ( a . u . ) d-MGO Sim.Sim.Sim.Exp.Exp.Exp. Magnetic Field (mT) d-MGOd-ZGOd-ZAO
Fig. 4
X-band EPR spectra of d-ZAO, d-ZGO and d-MGO recordedat room temperature. Microwave power 2 mW; Modulation depth 1mT at 100 kHz modulation frequency. The insert shows a zoom ofthe low magnetic field part of the spectra.
8a positions (site symmetry: T d ) and B site cations in octa-hedral 16d positions (site symmetry: D d ). Parameters likelattice constant, cationic site occupancy along with scale fac-tor, background and instrumental parameters were varied in fit-ting. The fits to the experimental patterns along with residuesare shown in Supplementary Figure 2. The cationic occupancycould not be varied for ZGO compounds since Zn + and Ga + ions are isoelectronic having similar X-ray scattering powersfor the Cu-K a radiations. As reported earlier, d-MGO com-pound shows 45.2(3)% cationic site inversion confirming thenear inverse spinel structure. Site occupancy was also variedfor d-ZAO compound and yielded no inversion in the latticehinting towards its normal spinel crystal structure.EXAFS measurements were carried out on Cr + dopedZGO, d-MGO and d-ZAO compounds, along with ZnCr O (noted ZCO) compound which was used as a reference for octa-hedral Cr + environment. Magnitude of Fourier transform (FT)of Cr K and Ga K edge EXAFS in d-ZGO are presented in Fig-ure 5 (a) in comparison with Cr K edge spectrum of ZCO. TheCr K EXAFS spectra in all three ZGO compounds are largelysimilar and hence the individual spectra are not presented here.It can be seen that the first peak corresponding to Cr-O corre-lation in d-ZGO Cr edge appears at a lower R value comparedto the first peak in both d-ZGO Ga edge and ZCO Cr edge,which implies that the average Cr-O distance in d-ZGO is lessthan the Ga-O distance in the same compound or the Cr-O dis-tance in ZCO. This anomaly gains more importance becauseunlike the first peak, the second intense peak observed around2.7 ˚A in d-ZGO Cr edge spectrum is almost at the same posi-tion as that in d-ZGO Ga edge spectrum. Also, an asymmetri-cal peak broadening of the first peak is observed in d-ZGO Credge spectrum revealing a distribution of Cr-O bond distances,especially towards the shorter distances. These observations in-dicate that the local octahedral environment around Cr + ion islargely distorted compared to that of Ga + environment in ZGOor the ideal octahedral Cr + environment in ZCO. FT magni-tude of k weighted EXAFS spectra of Cr + doped host latticesare presented in Figure 5 (b). A comparison of the FT mag-nitudes in the host lattice compounds with that of ZCO showsCr-O distances in d-ZGO and d-MGO to be shorter than that inZCO. Again the peaks are asymmetrically broadened indicatinga distribution in Cr-O bond distances. However in d-ZAO, Cr-O distance is larger than that in ZCO and no broadening is seenin the peak indicating a smaller distribution of Cr-O distances.Experimentally obtained EXAFS spectra were fit using Ifef-fit software for all the compounds to obtain bond distances (R)and mean-square disorders ( s ) for each path. The raw spectrawere background corrected and energy calibrated using Athenaand then fitted using Artemis. Atomic coordinates and latticeparameters obtained for each compound from Rietveld refine-ment were used as inputs to generate a FEFF input file, withCr or Zn or Ga as core for the corresponding Cr or Zn or Ga6
15 0 1 2 3015
ZCOCr edged-ZGOGa edge | ( R ) | ( ¯ - ) (a) d-ZGOCr edge ZCOd-ZGOd-ZAOd-MGO (b) R (¯)
Fig. 5
Fourier transform magnitude of EXAFS pattern in the k range3 to 10 ˚A − (a) of Cr K edge and Ga K edge for d-ZGO and ZCO; (b)of Cr K edge for d-ZGO, d-MGO, d-ZAO and ZCO. k (¯ -1 ) R e [( q )] ( ¯ - ) ExperimentalFit
Fig. 6
Experimentally obtained Cr K edge EXAFS pattern ford-ZGO along with fit in q space in the k range 3 to 10 ˚A − . K edge spectra. The photoelectron scattering paths were thencalculated and the experimental data was fitted up to 3.6 ˚A in R -space to obtain R and s for each scattering path. Good-ness of the fit, expressed by R factor , was less than 0.03 in allthe fits. The Cr K EXAFS in d-ZGO compound could only befitted using the third cumulant parameter ( C ) which indicatesan asymmetrical distribution of Cr-O bond distances. All otherattempts to fit the spectra resulted in negative (unphysical) val-ues of s . This deviation away from a normal distribution ofCr-O bond distances can be explained to be due to the disorderaround the Cr + ion, resulting in an asymmetric distributionof bond distances. An example of fit to EXAFS data in backtransformed k space is presented in Figure 6 for d-ZGO. Otherfits for Zn K edge and Ga K edge EXAFS patterns of d-ZGOare shown in Supplementary Figure 3. All the fitting parame-ters are given in Supplementary Tables 3 for Cr K edge EXAFSand Table 4 for Zn K and Ga K EXAFS.Unlike ZGO compounds, Cr K EXAFS in d-MGO couldbe fitted starting with two structural models based on normalspinel structure with Mg occupying all tetrahedral sites and Gaoccupying octahedral sites, and inverse spinel structure whereinall the Mg ions occupied octahedral sites and half of the Ga ionsoccupied the tetrahedral sites. Phase fraction of each modelwas taken as a fitting parameter. The inversion around Cr + for d-MGO was found to be 44(5)% which is consistent withthe value 45.2(3)% obtained from Rietveld refinement. In caseof d-ZAO, best fit was obtained with the normal spinel model,indicating presence of very little or no inversion around Cr + ion.7 Discussion
This work contributes to our effort to explain an unexpectedproperty of Cr + -doped ZGO, which is the possibility to acti-vate LLP by visible light. Owing to the leading role of Cr + ions in this property, we examined the perturbation of Cr + en-vironment by structural defects, mainly antisite defects due toinversion disorder. By combining PL, TSL and EPR analy-ses, we recently proposed a mechanism whereby LLP excita-tion and emission are likely due to a particular type of [Cr + -defect] cluster (Cr N ), namely a Cr + ion with two neighboringantisite defects of opposite charge (a Zn Ga ′ defect at short dis-tance and a Ga Zn ◦ defect at longer distance). In this mech-anism, it was proposed that the electric field created by the twoneighboring defects of opposite charge triggers the formationof electron-hole pairs from the excited T chromium state (orother states of higher energy), the hole and the electron beingnext trapped at Zn Ga ′ site and Ga Zn ◦ site of the lattice, respec-tively. The present structural study provides a deeper insightinto this still speculative model. The importance of antisite de-fects is now clearly supported by the correlation between theamount of inversion disorder and the excitation of LLP by vis-ible light. ZGO is characterized by a few percent of inversion( ∼ which appears a good compromise for this opticalproperty. On the contrary d-ZAO exhibits no measurable inver-sion and no strain broadening of EPR, PL and XAFS spectra,and clearly shows no considerable visible light excited LLP, al-though it still exhibits a strong X-ray excited LLP (Figure 2).Indeed TSL studies on this compound clearly show the lack ofTSL peak arising due to antisite defects (occurring at 370 K). d-MGO shows a high level of inversion disorder ( ∼ + environment, is shown in Figure 7which correlates interatomic distances, lattice parameters andcrystal field energies. It appears that Cr-Zn (or Cr-Mg) dis-tances (open squares in Figure 7 (a)) and Cr-Ga (or Cr-Al) dis-tances (full squares in Figure 7 (a)) are entirely determined bylattice parameter (see also Supplementary Tables 3 and 4). Thepresence of defects does not affect this correlation as shown bythe fact that d-ZGO, s-ZGO and e-ZGO have almost the samelattice parameters and the same Cr-Ga distances of 2.95 ˚A.Contrary to Cr-cation distances, there is no correlation be-tween Cr-O bond lengths and lattice parameters (see Supple-mentary Table 3). However a linear correlation is clearly ob-served in Figure 7 (b) between Cr-O bond length and the crystalfield energy D c f deduced from the A → T absorption band s-ZGOe-ZGO d-MGO d-ZAOd-ZGO C F ( c m - ) Cr-O bond lengths (¯) (b) d-ZAO d-MGO L a tti ce p a r a m e t e r ( ¯ ) Cr-cation distance (¯)ZGO (a)
M=GaM=Zn e-ZGO s-ZGO d-ZGO M - O bond l e ng t h s ( ¯ ) Cr-O bond lengths (¯) (c)
Fig. 7 (a) Rietveld refined lattice parameter values plotted versusCr-first cationic neighbour distances obtained by EXAFS fitting.Filled black squares represent Cr-trivalent ion distances and open redsquares represent Cr-divalent ion distances. The three ZGOcompounds (d-ZGO, s-ZGO and e-ZGO) have almost same valuesand they are seen to be overlapping. (b) Crystal field energy D c f calculated from A → T Cr + absorption band in RT PL spectraversus Cr-O bond lengths obtained from EXAFS fitting. (c)Metal-oxygen bond lengths (with M=Ga, Zn) plotted versus Cr-Obond lengths obtained from EXAFS fitting for ZGO compounds.Filled squares represent Zn-O bond lengths and filled circlesrepresent Ga-O bond lengths. ig. 8 Energy level scheme representing the effect of s and p contributions to Cr-O bond. of Cr + , which occurs at energy h n = D c f . This correlation andthe lack of correlation with lattice parameter show that Cr + imposes its first shell environment. This can be understood byconsidering the relation between the strength of the Cr-O bondand the crystal field splitting energy as shown in Figure 8. In apurely ionic model and ignoring for the moment the electronrepulsion, the electronic configuration of Cr + in octahedralenvironment is t e with a crystal field splitting D c f betweent and e levels. Oxygen 2p orbitals participate to the top ofthe valence band (L in Figure 8). The t and e metal orbitalsare separated from the valence band by energies of the order of D E p and D E s , respectively. The p and s covalent characters ofthe Cr-O bond are described by non-zero transfer integrals b p = h t | H | L( p ) i and b s = h e | H | L( s ) i between metal 3d and oxygen2p orbitals, where H is the Hamiltonian operator and L( p ) andL( s ) are the symmetry adapted linear combinations of oxygenorbitals with p and s character, respectively. These transfer in-tegrals are proportional to metal-ligand overlap. A strengthen-ing of the covalent bonding (decrease of the Cr-O bond length)will manifest itself by a shift ∼ d to low energy for L orbitals(bonding character) and a shift ∼ d to high energy for metalorbitals (antibonding character). An increase of the p characterof the Cr-O bond will thus decrease the crystal field splitting D c f by an amount d p ≈ b p / d E p . Alternatively, an increaseof the s character of the bond will increase D c f by an amount d s ≈ b s / d E s (Figure 8). Within this scheme, the decrease ofCr-O bond length observed in the sequence d Cr − O (d-ZAO) > d Cr − O (d-ZGO) > d Cr − O (d-MGO) (Figure 7 (b)) clearly corre-sponds to a decrease of D c f , which points to an increase of the p contribution to the Cr-O bond along this series. Alternatively,an increase of the s contribution to the Cr-O bond would pro-duce the opposite correlation, i.e. an increase of D c f associatedto a decrease of Cr-O bond length. The same type of correlation, with a smaller slope, was alsoobserved by varying the defect concentration in ZGO (Figure7 (b)). As reported before, the concentration of defects aroundCr + is minimum in d-ZGO and increases in the sequence d-ZGO < s-ZGO < e-ZGO. The fact that the slope is smallerupon varying the defect concentration (full triangles in Figure7 (b)) than upon varying the A + and B + cations (full squaresin Figure 7 (b)) can be explained by symmetry effects. In thespinel series AB O (A=Mg, Zn and B=Al, Ga), the symme-try of the unperturbed Cr + site is trigonal, which introducesno mixing between t ( p ) orbitals and e( s ) orbitals. An antisitedefect in first neighbour position of Cr + , and located outsidethe C3 axis, decreases the site symmetry which becomes nonaxial. This introduces a mixing between t and e orbitals, sothat t orbitals take a small s character while e orbitals takea small p character. As s bonds are stronger than p bonds,this increasing s - p mixing upon increasing the defect concen-tration shortens the Cr-O bond length. However, both t and eorbital sets are now shifted to high energy, so that the increaseof D c f is smaller than for a pure axial symmetry, explaining thesmaller slope in the curve D c f = f ( d Cr − O ) in the ZGO series. In-troducing the electron repulsion between the three t electronsdoes not modify this scheme. The E state (emitting level) andthe A state (ground state level) both correspond to the con-figuration t with total spin S=1/2 and S=3/2, respectively. Theeffect of the crystal field variation is much smaller for the E- A splitting than for the T - A splitting, but varies with thesame trend. This effect thus explains why lower the symme-try of the Cr + site, smaller the Cr-O bond length, and lowerthe energy of the E- A emission of Cr + . This confirms thecorrelation previously proposed between the decreasing sym-metry of the [Cr + -defect] clusters determined by EPR (Cr R > Cr N > Cr N ) and the decreasing photon energy of the E- A emission line. Clusters with lowest symmetry give N2emission line at the lowest photon energy, while clusters witha weaker non axial distortion give N1 line at higher photon en-ergy, which are still at a lower photon energy than the R1 lineof unperturbed Cr + ions in axial symmetry.From this EXAFS study, it is also possible to gain infor-mation about the nature of the dominant defects in the ZGOseries. It has been previously shown that the less disorderedZGO material is d-ZGO, where the zinc deficiency in the start-ing composition compensates the Ga loss during high tempera-ture synthesis. Increasing the Zn content in the synthesis in-creases the defect concentration (e-ZGO > s-ZGO > d-ZGO)as shown by the increasing intensity of N1 and N2 emissionlines, presumably in the form of an excess of antisite defects Zn Ga ′ . This hypothesis is now reinforced by the correlationbetween M-O bond length (M=Ga, Zn) and Cr-O bond length(Figure 7 (c)). It is found that increasing the defect concentra-tion results in an increase in Zn-O bond length and a decreasein Ga-O bond length (see Supplementary Table 4). Zn-O bond9ength is nearly equal to Ga-O bond length at the highest defectconcentration (e-ZGO). This behaviour can be explained by thefact that increasing the number of Zn + in Ga + sites ( Zn Ga an-tisite defects) has two effects: (i) as shown before, it lowers thesymmetry of Cr + site, which reduces the Cr-O bond length andshifts the E- A emission to lower energy, and (ii) it introducesin the lattice more Zn-O bonds with the same length as Ga-Obonds, which shifts the average Zn-O bond length towards thatof Ga-O as observed in Figure 7 (c). The compensation forthe excess of negative charge induced by Zn Ga ′ antisite defectsshould be insured by the other antisite defect Ga Zn ◦ and/or byoxygen vacancies V O ◦ . In the case of compensation by Ga Zn ◦ antisite defect, we expect a shortening of the Ga-O bond uponincreasing defect concentration. As there are two Ga for oneZn in the lattice, the ratio Ga Zn ◦ / Ga Ga × should be two timessmaller than the ratio Zn Ga ′ / Zn Zn × , where Ga Ga × and Zn Zn × represent Ga and Zn in their normal site. We thus expect a ratioR = -2 between the slopes of the Zn-O and Ga-O variations,which is close to the experimental value R ≈ -2.17 deducedfrom Figure 7 (c). Thus the variations of Cr-O bond lengthwith Zn-O and Ga-O bond lengths can be accounted for by Zn Ga and Ga Zn antisite defects present in the lattice. EPR spec-troscopy previously indicated that, N2 emission line of LLP inZGO might be due to a Cr + ion with a Zn Ga ′ antisite defectin its first neighbour position (at 0.295 nm) and a Ga Zn ◦ an-tisite defect at slightly larger distances. This model is nowsupported by the present EXAFS study. It can also explain theorigin of the considerable improvement of LLP in the galloger-manate series Zn + x Ga − x Ge x O . In these compounds, twoGa + ions are replaced by one Ge + ion and one Zn + ion.Thus, increasing x should increase the number of Cr + ionsclose to a negatively charged Zn Ga ′ and a positively charged Ge Ga ◦ defects. This defect configuration is determinant in ourmodel for the visible light induced LLP mechanism. The role of antisite defects on the persistent luminescence in-duced by visible light excitation in Cr + -doped AB O spinels(A=Zn, Mg; B=Ga, Al) was investigated by a combined opti-cal, EPR, XRD and EXAFS study. The main conclusions arethe following. (i) Visible light excitation of persistent lumines-cence necessitates a small degree of inversion disorder, withthe optimal level corresponding to that in d-ZGO. (ii) Increas-ing the defect concentration decreases the Cr-O bond lengthand the crystal field energy, attributed to an increasing p bondcontribution to the Cr-O interaction. (iii) Defects responsiblefor this Cr-O bond variation in ZGO are likely to be Zn Ga ′ and Ga Zn ◦ antisite defects. It can be concluded that a doping strat-egy which can control the amount of antisite defects should al-low the optimization of the intensity and length of the persistentluminescence in AB O :Cr + spinels. Authors acknowledge SAMBA beamline, Soleil synchrotronfacility, for giving the beamtime. Ms. St´ephanie Belin,beamline scientist, is gratefully thanked for the experimentalsupport. Financial support from Indo-French Centre for thePromotion of Advanced Research (IFCPAR)/ CEntre Franco-Indien Pour la Recherche Avanc´ee (CEFIPRA) is acknowl-edged.
References
Thermoluminescence of Solids ; Cam-bridge Solid State Science Series; Cambridge UniversityPress, Cambridge, U.K., 1985.2 Q. le Masne de Chermont, C. Chan´eac, J. Seguin, F. Pell´e,S. Maˆıtrejean, J.-P. Jolivet, D. Gourier, M. Bessodes andD. Scherman,
Proc. Natl. Acad. Sci. U.S.A. , 2007, , 9266-9271.3 A. Bessi`ere, S. Jacquart, K. Priolkar, A. Lecointre, B.Viana and D. Gourier,
Opt. Express , 2011, , 10131-10137.4 T. Maldiney, C. Richard, D. Scherman, D. Gourier, B.Viana and A. Bessi`ere, Nanoparticules `a luminescencepersistante excitable in situ pour limagerie optique invivo, limagerie multimodale optique-IRM in vivo et lath´eranostique. Patent PCT/EP2013/051727, 30/01/2013.5 T. Maldiney, A. Bessi`ere, J. Seguin, E. Teston, S. K.Sharma, B. Viana, A. J. J. Bos, P. Dorenbos, M. Bessodes,D. Gourier, D. Scherman and C. Richard, Nat. Mater. ,2014, , 418-426.6 S. K. Sharma, A. Bessi`ere, N. Basavaraju, K. R. Priolkar,L. Binet, B. Viana and D. Gourier, J. Lum. , 2014, ,251-256.7 Z. Pan, Y.-Y. Lu and F. Liu,
Nat. Mater. , 2011, , 58-63.8 M. Allix, S. Chenu,E. Ve, T. Poumeyrol, E. A. Kouadri-boudjelthia, S. Alahrache, F. Porcher and D. Massiot, Chem. Mater. , 2013, , 1600-1606.9 N. Basavaraju, S. Sharma, A. Bessi`ere, B. Viana, D.Gourier and K. R. Priolkar, J. Phys. D: Appl. Phys. , 2013, , 375401.10 I.-K. Jeong, H. L. Park and S.-i. Mho, Solid State Com-mun. , 1998, , 179-183.11 G. Anoop, K. M. Krishna and M. K. Jayaraj,
Appl. Phys.A. , 2008, , 711-715.102 L. Chen, Y. Liu, Z. Lu and K. Huang, Mater. Chem. Phys. ,2006, , 247-251.13 R. L. Millard, R. C. Peterson and I. P. Swainson, Phys.Chem. Minerals , 2000, , 179-193.14 N. Kashii, H. Maekawa and Y. Hinatsu, J. Am. Ceram.Soc. , 1999,
82 (7) , 1844-1848.15 W. Nie, F. M. Michel-Calendini, C. Linar`es, G. Boulonand C. Daul,
J. Lumin. , 1990, , 177-190.16 A. Bessi`ere, S. K. Sharma, N. Basavaraju, K. R. Priolkar,L. Binet, B. Viana, A. J. J. Bos, T. Maldiney, C. Richard,D. Scherman and D. Gourier, Chem. Mater. , 2014, ,1365-1373.17 D. Gourier, A. Bessi`ere, S. K. Sharma, L. Binet, B. Viana,N. Basavaraju and K. R. Priolkar, J. Phys. Chem. Solids ,2014, , 826-837.18 A. Bessi`ere, A. Lecointre, K. R. Priolkar and D. Gourier, J. Mater. Chem. , 2012, , 19039-19046.19 A. Lecointre, A. Bessi`ere, K. R. Priolkar, D. Gourier, G.Wallez and B. Viana, Mater. Res. Bull. , 2013, , 1898-1905.20 T. Aitasalo, J. Hassinen, J. H¨ols¨a, T. Laamanen, M. Las-tusaari, M. Malkam¨aki, J. Niittykoski and P. Nov´ak, J.Rare Earths , 2009, , 529-538.21 H. Okura, T. Murakawa, Y. Miyamoto, T. Honma and K.Ohmi, Jap. J. Appl. Phys. , 2012, , 062602.22 Z. Qi, C. Shi, W. Zhang, W. Zhang and T. Hu, Appl. Phys.Lett. , 2002, , 2857-2859.23 L. C. V. Rodrigues, H. F. Brito, J. H¨ols¨a, R. Stefani, M.C. F. C. Felinto, M. Lastusaari, T. Laamanen and L. A. O.Nunes, J. Phys. Chem. C , 2012, , 11232-11240.24 K.-S. Sohn, B. Cho, H. D. Park, Y. G. Choi and K. H. Kim,
J. Eur. Ceram. Soc. , 2000, , 1043-1051.25 T. Takeda, R.-J. Xie, N. Hirosaki, Y. Matsushita and T.Honma, J. Solid State Chem. , 2012, , 71-75.26 K. Korthout, K. Van den Eeckhout, J. Botterman, S.Nikitenko, D. Poelman and P. F. Smet,
Phys. Rev. B , 2011, , 085140.27 J. H¨ols¨a, T. Laamanen, M. Lastusaari, M. Malkam¨aki, E.Welter and D. A. Zajac, Spectrochimica Acta Part B , 2010, , 301-305.28 P. Ghigna, S. Pin, C. Ronda, A. Speghini, F. Piccinelli andM. Bettinelli, Opt. Mater. , 2011, , 19-22. 29 J. Rodriguez-Carvajal, Physics B , 1993, , 55-69.30 B. Ravel and M. Newville,
J. Synchrotron. Rad. , 2005, ,537-541.31 S. I. Zabinsky, J. J. Rehr, A. Ankudinov, R. C. Albers andM. J. Eller, Phys. Rev. B , 1995, , 2995-3009.32 W. Mikenda and A. Preisinger, J. Lum. , 1981, , 53-66.33 W. Mikenda, J. Lumin. , 1981, , 85-98.34 W. Zhang, J. Zhang, Z. Chen, T. Wang and S. Zheng, J.Lum. , 2010, , 1738-1743.35 W. Mikenda and A. Preisinger,
J. Lum. , 1981, , 67-83.36 J. Derkosh and W. Mikenda, J. Lumin. , 1983, , 431-441.37 Th Hahn, (ed) International Tables for CrystallographicVolume A: Space-group symmetry , Springer, The Nether-lands, 2005.38 S. K. Sharma, D. Gourier, B. Viana, T. Maldiney, E.Teston, D. Scherman and C. Richard,
Opt. Mater. , 2014,http://dx.doi.org/10.1016/j.optmat.2014.06.020.39 Y. Tanabe and S. Sugano,
J. Phys. Soc. Jap. , 1954, , 766-779.11
00 600 900 1200
Exp.
Sim1 (D= 0.635 0.001)
Sim. (Sim1 X 3*Sim2)