Rapid and Energy-Saving Microwave-Assisted Solid-State Synthesis

Research Article www.acsami.org

Rapid and Energy-Saving Microwave-Assisted Solid-State Synthesis of Pr3+-, Eu3+-, or Tb3+-Doped Lu2O3 Persistent Luminescence Materials Cássio C. S. Pedroso,† José M. Carvalho,†,‡ Lucas C. V. Rodrigues,† Jorma Hölsa,̈ †,‡,§,⊥ and Hermi F. Brito*,† †

Institute of Chemistry, University of São Paulo, BR-05508-000 São Paulo, São Paulo, Brazil Department of Chemistry and §Turku University Centre for Materials and Surfaces, University of Turku, FI-20014 Turku, Finland ⊥ Department of Physics, University of the Free State, Bloemfontein ZA-9300, South Africa ‡

S Supporting Information *

ABSTRACT: Persistent luminescence materials Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) were successfully and rapidly (22 min) prepared by microwave-assisted solid-state synthesis (MASS) using a carbon microwave susceptor and H3BO3 as flux. Reaction times are reduced by up to 93% over previous synthetic methods, without special gases application and using a domestic microwave oven. All materials prepared with H3BO3 flux exhibit LuBO3 impurities that were quantified by Rietveld refinement from synchrotron radiation X-ray powder diffraction patterns. The flux does not considerably affect the crystalline structure of the C-Lu2O3, however. Scanning electron micrographs suggest low surface area when H3BO3 flux is used in the materials’ synthesis, decreasing the amount of surface hydroxyl groups in Lu2O3 and improving the luminescence intensity of the phosphors. The carbon used as the susceptor generates CO gas, leading to complete reduction of TbIV to Tb3+ and partial conversion of PrIV to Pr3+ present in the Tb4O7 and Pr6O11 precursors, as indicated by X-ray absorption near-edge structure data. Persistent luminescence spectra of the materials show the red/near-IR, reddish orange, and green emission colors assigned to the 4fn → 4fn transitions characteristics of Pr3+, Eu3+, and Tb3+ ions, respectively. Differences between the UV-excited and persistent luminescence spectra can be explained by the preferential persistent luminescence emission of R3+ ion in the S6 site rather than R3+ in the C2 site. In addition, inclusion of HfIV and Ca2+ codopants in the Lu2O3 host increases the emission intensity and duration of persistent luminescence due to generation of traps caused by charge compensation in the lattice. Photonic materials prepared by MASS with H3BO3 flux show higher persistent luminescence performance than those prepared by the ceramic method or MASS without flux. Color tuning of persistent luminescence in Lu2O3:R3+,M provides potential applications in bioimaging as well as in solar cell sensitizers. KEYWORDS: microwave-assisted solid-state synthesis, photonic material, persistent luminescence, lutetia, praseodymium, europium, terbium, color tuning

1. INTRODUCTION

according to present knowledge it is the reduction product of the mineral barite, barium sulfide (BaS). The luminescence was latter assigned to the electronic transition of Cu+ impurities present in BaS.1,4 Sulfide vacancies, formed as charge compensation due to Cu2+ reduction to Cu+, act as the traps. The discovery of new persistent luminescence materials, much more efficient than those used in the mid-1990s (e.g., SrAl2O4:Eu2+,Dy3+),5 created a new benchmark for the performance of persistent emission in the visible region. Since then, study of these new photonic materials has focused

Persistent luminescence is a phenomenon where the material emits radiation (usually in the UV−visible−near IR region) from seconds to several hours after cessation of irradiation, which can be light, UV radiation, electron beam, etc.1 Currently it is known that persistent luminescence, a special case of thermally stimulated luminescence, results from storage of the excitation energy in traps and its subsequent release induced by thermal energy available at the appropriate temperature.2 The phenomenon has been known for over 1000 years in natural materials, documented in, for example, ancient Chinese paintings.3 However, the first persistent luminescence material, prepared by Vincenzo Cascariolo, was scientifically documented in 1640.3 This material was called the Bologna stone, and © 2016 American Chemical Society

Received: April 19, 2016 Accepted: July 7, 2016 Published: July 7, 2016 19593

DOI: 10.1021/acsami.6b04683 ACS Appl. Mater. Interfaces 2016, 8, 19593−19604

Research Article

ACS Applied Materials & Interfaces

application areas), was studied and elaborated. Persistent luminescence materials Lu 2 O 3 :R 3+ ,M (Pr,Hf IV ; Eu; or Tb,Ca2+) were thus prepared by the MASS method. The materials’ properties, such as phase purity, crystallinity, particle morphology, and persistent luminescence, were investigated and compared to those of materials prepared by a ceramic method. These materials were characterized by different techniques: Fourier transform infrared (FTIR) and Raman spectroscopy, synchrotron radiation X-ray powder diffraction (SR-XPD), scanning electron microscopy (SEM), synchrotron radiation X-ray absorption spectroscopy (SR-XAS), and UV− visible luminescence. Eventually, the uncommon spectral features, including quenching of the putative emission, were both analyzed and the appropriate mechanisms elucidated.

mainly on divalent europium as the activator and on aluminate and silicate materials as stable host, in the front line being CaAl2O4:Eu2+,Nd3+,6 Sr4Al14O25:Eu2+,Dy3+,7 and Sr2MgSi2O7:Eu2+,Dy3+.8 Despite the dominance of the Eu2+doped persistent luminescence phosphors, other activators like the d-block metals (e.g., Ti3+,9 Cr3+,10 and Mn2+ 11) and trivalent rare earths (R3+; for example, Ce3+,12 Pr3+,13 Sm3+,14 Eu3+,15 Tb3+,16 and Dy3+ 17,18) have been introduced. Although scientific studies have lagged far behind, persistent luminescence phosphors have already been commercialized since the beginning of the 20th century and are widely used as nightvision materials in security and emergency route signs as well as in traffic signage.4 Other promising applications of persistent luminescence materials include radiation detection, sensors for structural damage, optical memory media, identification markers, medical diagnostics, optical probes for in vivo bioimaging, and solar cell sensitizers.18−22 Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials show moderate to strong persistent luminescence emission.23−25 In addition, the lutetia host lattice possesses an exceptionally high absorption coefficient for ionizing radiation. This high stopping power is due to the high density of the material (9.42 g·cm−3). These phosphors are suitable as scintillator materials and have useful applications, for example, in medical diagnosis imaging.25−27 As a drawback, in addition to high price, Lu2O3-based persistent luminescence materials seem to require preparation at high temperatures (∼1700 °C) in high vacuum or in H2−N2 by a solid-state method.23−25 A possible alternative to ceramic methods is to use rapid microwave-assisted solid-state synthesis (MASS). This method has attracted the attention of materials researchers since the 1980s, but it is poorly explored compared to organic and solution-phase microwave syntheses. Advantages of the MASS method include short processing time, selective dielectric heating, low energy consumption, and use of inexpensive equipment (domestic microwave oven), often affording highpurity and high-yield products.28,29 The MASS method enables the preparation of versatile ceramic materials such as oxides, chalcogenides, borides, carbides, oxynitrides, etc.28 However, most of the materials interact very weakly with the microwave radiation at room temperature (i.e., their dielectric loss tangents are small at room temperature).28,29 To compensate for this drawback, selected/specific compounds can act as microwave absorbers (also called susceptors), defined as substances that have the ability to absorb the microwave radiation and convert it to heat (e.g., SiC, CuO, or carbon).28,30−32 A susceptor can be the reagent itself, but if this is not the case, the susceptor must be in direct contact or close to the reagent(s). Although the magnitude of the dielectric loss tangent usually increases along with temperature, the penetration of microwaves and heat conductivity within the material play a role. Ideally, the material can be heated locally with low-temperature gradients.28,32 The aim of this work was to improve drastically the persistent luminescence properties of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials by using the novel and cost-effective MASS technique. H3BO3 was used as flux at high temperature with the prospect of obtaining Lu2O3 morphology with low porosity, to facilitate the introduction of dopants in the host and form a protective layer in order to decrease OH amounts on the surface. The luminescence of lutetia-based materials, especially the quenching of luminescence resulting in uncommon luminescence color (and thus opening new

2. EXPERIMENTAL SECTION 2.1. Preparation. Polycrystalline persistent luminescence materials containing Pr,HfIV (0.05 and 0−0.1 mol %), Eu (0.2 mol %), and Tb,Ca2+ (0.5 and 0−1.5 mol %) doped in the Lu2O3 host were prepared by use of thoroughly mixed and ground stoichiometric amounts of the starting materials: Lu2O3, Pr6O11, Eu2O3, and Tb4O7 (all 99.99%, CSTARM), CaO (>98%, Merck), and HfO2 (98%, Sigma−Aldrich). H3BO3 (>99.5%, Synth) at 2.5, 5.0, and 10.0 wt % was added as flux. The codopant concentrations and H3BO3 amount were optimized to obtain the highest intensity and longest duration of persistent luminescence. The modified microwave-assisted solid-state synthesis used is based on the work on yttria-stabilized zirconia.33 In the present synthesis, 15 g of granular activated carbon (diameter 1−2 mm, Synth) in an alumina crucible (50 cm−3) was used as the microwave susceptor and source of the CO reducing gas. The mixture of starting materials (1 g) was placed in a second alumina crucible (3 cm−3), which was partially surrounded by activated carbon for efficient heat transfer to the reagents. The larger crucible was covered with an alumina disk to avoid exposure to O2, which could jeopardize the reduction of Tb4O7 and Pr6O11 precursors. Both crucibles were placed inside a cavity cut into a block of aluminosilicate thermal insulation brick. The top and bottom of the block were closed with two flat lids of the same material. The schematic MASS device and crucible setup are depicted in Figure 1.

Figure 1. Microwave-assisted solid state synthesis (MASS) device (left) and crucible setup (right). Finally, the precursor materials were heated in a domestic microwave oven (Electrolux MEF41, 1000 W maximum power and 30 L volume) for 12 and 10 min at power levels of 1000 and 900 W, respectively. Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) persistent luminescence materials were prepared by a conventional ceramic method as well. The same precursors were heated in dense alumina crucibles in a static reducing atmosphere (5% H2/N2) at a constant heating rate of 5 °C· min−1 from room temperature to 1150 °C and remained at the maximum temperature for 12 h. To identify the impurities formed by reaction of R2O3 with H3BO3 flux, calcite- and vaterite-type LuBO3 were also synthesized by ceramic methods.34 2.2. Characterization. Infrared absorption spectra of the materials were measured by the KBr pellet technique on a Bomem MB102 FTIR 19594


Research Article

ACS Applied Materials & Interfaces spectrometer in the range 400−4000 cm−1 with a spectral resolution of 4.0 cm−1. Raman spectra of the materials were obtained on a FTRaman Bruker RFS 100/S spectrometer using the 1064 nm line of a Nd:YAG laser, typically with 50 mW output power. The setup gave a spectral resolution of 1.0 cm−1. Room-temperature synchrotron radiation X-ray powder diffraction (SR-XPD) data were collected at beamline XPD at the Laboratório ́ Nacional de Luz Sincrotron (LNLS), Centro Nacional de Pesquisa em Energia e Materiais (CNPEM), Campinas, Brazil. The XPD patterns were recorded with θ−2θ geometry, an image-plate Dectris Mythen 1K detector providing a 0.001° resolution. The patterns were measured at a wavelength of 1.613 32 Å, 2θ range of 10−120°, stepscanning mode (step size 0.5°), and count time of 5 s·step−1. Rietveld refinements employing pseudo-Voigt reflection profile function were carried out by use of the general structure analysis system (GSAS)35,36 software suite. Crystal structures were illustrated with the VESTA software.37 Scanning electron microscopy (SEM) images and energy-dispersive spectroscopy (EDS) elemental mapping were obtained with a JEOL JSM-740 1F field-emission scanning electron microscope with acceleration voltages of 5 and 15 kV, respectively. For SEM images, the solid samples were deposited on double-sided graphite tape, attached to a graphite sample holder. In EDS mapping, the samples were prepared by dispersing the materials in deionized water. The suspension was submitted to ultrasound for 15 min and then placed on a graphite grid sample holder. The valences of praseodymium, europium, and terbium in Lu2O3 were studied by the X-ray absorption near-edge structure (XANES) method at room temperature by use of beamline XAFS1 (LNLS, CNPEM, Brazil). XANES data was collected on Pr, Eu and Tb LIII edges at 5964, 6977, and 7514 eV, respectively. The measurements were carried out by use of a Si(111) double-crystal monochromator (energy resolution Δλ/λ = 1.31 × 10−4)38 in the fluorescence mode with a Ge-15 solid-state detector. The energy range of the collected data was from −100 to 100 eV with respect to the absorption edge (E0) with a step of 0.2 eV. Freshly annealed Pr6O11, Eu2O3, and Tb4O7 as well as Pr(NO3)3·6H2O and TbF3 were used as references.13,39,40 UV-excited and persistent luminescence spectra were measured on a Horiba Jobin Yvon Fluorolog 3-11 spectrofluorometer equipped with a mono- and double-grating monochromator (focal lengths of 0.3 and 0.5 m, respectively) for excitation and emission (iHR320), respectively. A 450 W xenon lamp was used as the irradiation source, and a Synapse Horiba Jobin Yvon E2 V CCD30 (1024 × 256 pixels) was used as the detector. Persistent luminescence charging and decay times were measured on a SPEX Fluorolog-2 spectrofluorometer equipped with two 0.22 m SPEX1680 double-grating monochromators, a 450 W xenon lamp as the irradiation source, and photomultiplier detection. All data were collected at an angle of 22.5° (front face). The CIE (Commission Internationale de l’Eclairage) color coordinates were calculated from the emission spectra by use of SpectraLux v2.0 software.41 Measurement of vacuum UV-excited luminescence was performed at LNLS (CNPEM) in the Toroidal Grating Monochromator beamline (TGM). The emission spectrum was recorded at room temperature under excitation at 110 nm selected by a diffraction grating of Pt (75 lines·mm−1). Luminescence was recorded by use of Ocean Optics fiber optics (diameter 1 mm) connected to an Ocean Optics QE65000 spectrometer with resolution up to 1 nm (at 500 nm).

Figure 2. FTIR (top) and Raman (bottom) spectra of Lu2O3 materials prepared with 5.0 wt % H3BO3 flux and without flux, as well as spectra of LuBO3 (both vaterite and calcite forms).

of the vaterite form of LuBO3.43,44 The original cubic C-type structure of Lu2O3 is maintained after heat treatment, though a LuBO3 (vaterite) impurity can be detected in materials prepared with H3BO3 flux. The LuBO3 (calcite) vibrations were not observed in the FTIR spectrum of Lu2O3 prepared with 5.0 wt % H3BO3. FTIR spectra of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) prepared with 5.0 wt % H3BO3 are similar to the spectrum of nondoped Lu2O3 with 5.0 wt % H3BO3 (Figure S1). The similarities are due to the low dopant concentrations and not yet a proof of complete solid solubility of dopants in the Lu2O3 host. In the Raman spectrum of Lu2O3 with 5.0 wt % H3BO3, there are lines at 120, 350, 390, 455, 500, and 610 cm−1 that can be assigned to Lu−O stretching vibrations of Lu2O3 (Figure 2, bottom). A weak line at 265 cm−1 is due to the B−O−B bending of vaterite LuBO3. However, the strongest Raman line of LuBO3 (calcite) at 385 cm−1, assigned to BO3 bending, can overlap with the most intense line of Lu−O stretching (390 cm−1) of Lu2O3.45,46 It is thus not possible to rule out unambiguously the presence of LuBO3 (calcite) from the Raman spectra. FTIR and Raman data suggest that the Lu2O3 host maintains the original structure after heat treatment, though a LuBO3 (vaterite) impurity was observed in the materials prepared with H3BO3 flux. The use of H3BO3 flux is necessary, however, since it increases the persistent luminescence intensity and duration of the Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials, as will be discussed further. Phase purities of all materials were investigated by SR-XPD analyses as well. XPD patterns of Lu2O3 prepared with different amounts of H3BO3 flux (0, 2.5, 5.0, and 10.0 wt %; Figure 3) and of the Lu2O3 precursor (Figure S2) include Rietveld plots calculated with GSAS35,36 software and using the starting values reported.47−49 All fits converged well, and the figure of merit values (Rwp, Rp, Rexp, and χ2) as well as atom coordinates,

3. RESULTS AND DISCUSSION 3.1. Phase Purity and Structural Analysis. FTIR absorption spectra of the Lu2O3 host with and without H3BO3 flux (Figure 2, top) show absorption bands at 450− 650 cm−1 assigned to Lu−O stretching vibrations of the LuO6 polyhedral units.42 In addition, the spectrum of Lu2O3 with H3BO3 flux reveals absorption bands corresponding to the O− B−O bending of BO3 (720 cm−1) and the B−O stretching vibrations of tetrahedral BO4 (790−1250 cm−1), characteristic 19595


Research Article


concentrations (Figure S3). The MASS synthetic method induces a fast temperature rise and a temperature high enough to decompose H3BO3 to B2O3. The molten B2O3 acts as a liquid medium at high temperatures. Formation of the LuBO3 phases occurs due to reaction of the flux with Lu2O3 precursor, suggesting that either the flux amount should be reduced or reaction conditions should be modified to obtain impurity-free materials. However, for practical purposes, to obtain the materials with best luminescence properties, substantial flux amounts are needed. The LuBO3 (vaterite) impurity was quantified by Rietveld refinement (Figure 4, right) and the amount of impurity increases linearly with increasing flux amount. For the total amount of LuBO3 impurities, materials prepared with 2.5 and 10.0 wt % H3BO3 have practically the same amounts as the theoretical ones, indicating complete reaction of H3BO3 with Lu2O3. However, the material prepared with 5.0 wt % H3BO3 has a deficiency of 4 wt % LuBO3. The synthesis can reach a temperature of about 1350 °C (measured with a hand-held pyrometer), which is high enough to partially evaporate B2O3, suggesting a lower LuBO3 amount. The low LuBO3 amount may be due to formation of calcite-type LuBO3 not detected due to a difference between theoretical and experimental values of LuBO3 that is in the limit of XPD detection (Figure 3). As a conclusion, all B2O3 may not necessarily react with Lu2O3. Cell parameters, interatomic distances, and polyhedron volume were obtained from Rietveld refinements for the lutetia precursor and MASS-prepared materials with and without flux (Table 1 and Table S2). All materials have similar a values in comparison with the C-Lu2O3 precursor. It is thus not possible to detect changes in the C-Lu2O3 cell parameter after heat treatment as well as the presence of boron in the sesquioxide lattice. Besides, interatomic distances of the C-Lu2O3 structure in precursor and prepared materials are similar, except for d1Lu1−O1(C2) and dLu2−O1(S6) that have inverse behavior. The variation in dLu2−O1(S6) distance in the materials has a high impact on the volume of the S6 LuO6 polyhedron. However, the C2 symmetry site of C-Lu2O3 has several Lu−O distances, so d1Lu1−O1 value has less impact on the polyhedron volume. This is caused by lutetium ions that remain practically in the same position and the considerable displacement of the oxygen atoms (Table S1). This result is inevitable because the low

Figure 3. Synchrotron radiation X-ray powder diffraction (SR-XPD) patterns (λ = 1.613 32 Å) including Rietveld fits for Lu2O3 without and with H3BO3 flux (2.5, 5.0, and 10.0 wt %). Reference patterns for Lu2O3 as well as vaterite and calcite LuBO3 were obtained from refs 47−49.

occupations, and isotropic temperature factors are rather good, when one takes into account the problems with modeling the baseline and extra impurity from absorption of CO2 generated by carbon combustion (Table 1 and Table S1). It is evident that materials prepared by the MASS technique maintain the cubic C-type Lu2O3 structure [powder diffraction file (PDF), International Centre for Diffraction Data (ICSD), entry 431021; Figure 4 left].47 However, H3BO3 causes a hexagonal LuBO3 (vaterite) impurity (PDF, ICSD, entry 74-1938)48 with all flux amounts, and with 10.0 wt % flux the trigonal LuBO3 (calcite) impurity (PDF, ICSD, entry 72-1053)49 is present as well (Figure 3). Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials prepared with 5.0 wt % H3BO3 have similar XPD patterns as the nondoped material due to the low dopant

Table 1. Parameters for Lu2O3 Precursor and Lu2O3 Heated without and with Different H3BO3 Flux Amountsa precursor

crystal system space group Z crystalline fraction, wt % lattice parameter a,b Å lattice parameter c, Å cell volume,f Å3 fwhm of C-Lu2O3 refln at 31.2°, deg χ2

without flux

with 2.5 wt % H3BO3 flux

C-Lu2O3

C-Lu2O3

C-Lu2O3

cubic Ia3̅ 16

cubic Ia3̅ 16

10.3909

10.3911

cubic Ia3̅ 16 91 10.3902

1121.9 0.1982 2.407g

1122.0 0.0873 1.556h

1121.7 0.0849 2.035i

LuBO3 vaterite hexagonal P63/mmc 2 9 3.7282 8.7220c 105.0 2.035i

with 5.0 wt % H3BO3 flux C-Lu2O3 cubic Ia3̅ 16 85 10.3912 1122.0 0.0832 1.490j

LuBO3 vaterite hexagonal P63/mmc 2 15 3.7284 8.7288d 105.1 1.490j

with 10.0 wt % H3BO3 flux C-Lu2O3 cubic Ia3̅ 16 66 10.3912 1122.0 0.1047 1.610k

LuBO3 vaterite

LuBO3 calcite

hexagonal P63/mmc 2 29 3.7272 8.7219b 104.9

trigonal R3c̅ 6 5 4.9140 16.2112e 339.0

1.610k

1.610k

a Obtained from Rietveld refinements of synchrotron radiation powder X-ray diffraction data. bCalculated (from Rietveld refinements) estimated standard deviation (esd) value 1 × 10−4. cCalcd esd value 7 × 10−4. dCalcd esd value 4 × 10−4. eCalcd esd value 3 × 10−4. fCalcd esd value 1 × 10−1. g Figure of merit (Rwp, Rp, Rexp) values: 9.87%, 6.17%, 6.36%. hFigure of merit (Rwp, Rp, Rexp) values: 9.19%, 6.70%, 7.37%. iFigure of merit (Rwp, Rp, Rexp) values: 9.80%, 7.40%, 6.87%. jFigure of merit (Rwp, Rp, Rexp) values: 9.83%, 7.07%, 8.05%. kFigure of merit (Rwp, Rp, Rexp) values: 9.08%, 6.72%, 7.16%.

19596


Research Article


Figure 4. (Left) VESTA37 representation of C-type Lu2O3 unit cell, showing C2 and S6 sites. (Right) Relationship between H3BO3 flux and LuBO3 (vaterite and total) weight fractions, the latter from Rietveld refinement.

oxygen electron density increases uncertainty in its position. Therefore, it was not possible to elucidate the influence of boron on the C−Lu2O3 structure by Rietveld refinements. The Scherrer formula for calculating nanocrystal sizes is not reliably applied when the sizes are on the order of several hundred nanometers. The full width at half-maximum (fwhm) value of the most intense XPD reflection (222) at 31.2° (Table 1) was used instead, according to the method reported earlier.32 The fwhm value of the Lu2O3 precursor is higher than the values observed for lutetia in prepared materials. These results indicate that heat treatment increases the crystallite size. However, the fwhm values for Lu2O3 prepared with and without H3BO3 flux are similar, suggesting that the addition of flux has no effect on crystallite size (Table 1). 3.2. Morphology and Dopant Distribution. SEM images (Figure 5, top row) exhibit particles of the materials synthesized without and with H3BO3 flux. Particles of Lu2O3 prepared without flux are smaller than those prepared with H3BO3. The SEM image of the material without flux shows agglomerates with well-defined grain boundaries, while the materials prepared with H3BO3 flux are sintered ceramics with smooth surfaces and low porosity. The MASS method can hinder morphological control of the material formed during the synthesis due to fast sintering. H3BO3 flux was thus used to minimize this problem since the flux works well as a liquid medium at high temperature, promoting efficient particle aggregation. The low surface area of Lu2O3 particles due to H3BO3 flux decreases the amount of hydroxyl groups at the surface that can decrease the nonradiative luminescent decay rate. This effect can thus improve the luminescence intensity of the phosphors.50 EDS mapping of the Pr-, Eu-, or Tb-doped and Ca-codoped Lu2O3 prepared with H3BO3 shows apparent homogeneous distribution of the dopants (Figure 5). Homogeneous distribution of the dopants in a host is very important because the spatial distribution directly influences the luminescence intensity. Locally concentrated dopants may lead to luminescence quenching, for example, by cross-relaxation processes.51 For the Hf codopant, EDS mapping was not recorded since its M and L edges overlap with lutetium M and L edges, respectively. 3.3. Valence of Pr, Eu, and Tb Dopants. In order to probe the oxidation state of Pr, Eu, and Tb dopants in Lu2O3, X-ray absorption spectroscopy in the XANES region was carried out (Figure 6). XANES spectra of Pr-doped Lu2O3 indicate the presence of both trivalent and tetravalent forms of praseodymium (Figure 6, left). Similar spectra were obtained

Figure 5. (Top row) SEM images of Lu2O3 without flux (left) and with 5.0 wt % H3BO3 flux (right). (Middle row) EDS mapping of Pr (L edge; left) in Lu2O3:Pr3+,HfIV (1.0 and 0.1 mol %) and Eu (L edge; right) in Lu2O3:Eu3+ (2.0 mol %). (Bottom row) EDS mapping of Tb (L edge; left) and Ca (K edge; right) in Lu2O3:Tb3+,Ca2+ (1.0 and 1.5 mol %).

for materials prepared with and without HfIV codopant. In contrast, the Lu2O3:Eu material exhibits only the trivalent state of europium (Figure 6, middle). Though the europium ion can commonly be present in either divalent or trivalent forms, Eu2+ is not stable in the Lu2O3 host due to its ionic radius being much larger than for Eu3+. The charge compensation due to introduction of tetravalent rare earths (or, in general, tetravalent species) can be carried out easily by introduction of interstitial oxygen in the oxygen lattice sites of the cubic RO2 vacant in C-Lu2O3. Quite surprisingly, in Lu2O3:Tb,Ca2+, terbium is only in the trivalent state in the materials with and without Ca2+ codopant (Figure 6, right) despite the fact that Ca2+ codoping is expected to promote the presence of tetravalent terbium. 19597


Research Article


the defect clusters formed by aliovalent doping can affect the redox process in oxide materials. However, further studies are needed to verify the formation of dopant clusters in Lu2O3 materials. Since persistent luminescence is obtained by charging with different forms of radiation, including the X-rays used in XANES, important information was gathered from the XANES measurements. The presence of only the R3+ forms indicates that the presumed photoionization of Eu3+ and Tb3+ during the persistent luminescence charging process creates Eu3+−e− and Tb3+−h+ pairs instead of the Eu2+ and TbIV species. However, Pr-doped Lu2O3 may create the Pr3+−h+ and/or PrIV species due to the presence of both trivalent and tetravalent oxidation states. 3.4. UV−Visible Excitation of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) Materials. In order to obtain the optimal energy to excite Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials, their UV−vis excitation spectra (Figure 7) were recorded at room temperature. The excitation spectrum of Lu2O3:Pr3+,HfIV (0.05 and 0.1 mol %), monitoring the intensity of the 1D2 → 3H4 transition at 634 nm, consists of a high-intensity broad band with a maximum at 267 nm, assigned to the Pr3+ 4f2 → 4f15d1 transition (Figure 7, left). In addition, there are 4f2 → 4f2 transitions as low-intensity groups of narrow lines from the 3H4 ground state to the excited levels, which are assigned to 3P0 (ca. 20 000 cm−1); 1I6, 3P1 (ca. 20 850 cm−1); and 3P2 (ca. 21 950 cm−1).53 The 4n → 4fn−15d1 transitions are broad bands with high intensity because they involve parity-allowed electronic transitions to the outer 5d orbitals. The broad nature of the band is due to vibronic fine structure not resolved at room temperature. The 4fn → 4fn transitions have low intensity due to their parity-forbidden nature. Their sharp line appearance is due to weak electron−phonon interaction, that is, interaction between 4f orbitals and the lanthanide environment, because of effective shielding of the 4f electrons by the external filled 5s and 5p subshells.51,54−56 When the materials with and without HfIV codopant are compared, the Lu2O3:Pr3+ excitation spectrum shows the 4fn → 4fn−15d1 band as less broad than that for the HfIV-codoped material (Figure 7, left). This result can be explained by defects generated in the aliovalent HfIVcodoped material changing the crystal field around the Pr3+ ion. The Lu2O3:Eu3+ (0.2 mol %) excitation spectrum, monitoring the 5D0 → 7F2 transition intensity at 613 nm, exhibits groups of narrow lines due to the 4f6 → 4f6 transitions from the 7 F0,1 levels to excited levels of Eu3+ (Figure 7, middle). The

Figure 6. Synchrotron radiation Pr (left), Eu (middle), and Tb (right) LIII edge XANES spectra of Lu2O3:R,M (Pr,HfIV; Eu; or Tb,Ca2+, respectively) and reference materials [Pr(NO3)3·6H2O, Pr6O11, Eu2O3, TbF3, and Tb4O7].

Both praseodymium and terbium are stable as tetravalent species (PrIV and TbIV) as well in the trivalent state. For Lu2O3:Tb(,Ca2+) materials, TbIV present also in the Tb4O7 precursor and presumably in the two Lu3+ sites, together with charge compensation, is completely reduced by the reducing CO gas sphere (Figure 6, right). In contrast, XANES data for the Lu2O3:Pr(,HfIV) materials (Figure 6, left) show a line assigned to the PrIV form (at 5964 eV) though with much lower intensity than for Pr6O11, indicating only partial reduction of PrIV to Pr3+, both of which are present in the Pr6O11 precursor. Leaving aside its higher tendency for the tetravalent state (Pr6O11 vs Tb4O7), which may explain the existence of PrIV in Lu2O3, the behavior in Lu2O3 needs further consideration. In spite of the same coordination number (CN), six, of the two R3+ sites in C-R2O3 (Figure 4, left), the ionic sizes of the ions differ: Lu3+ has an ionic radius of about 0.86 Å.52 Due to lanthanide contraction, the Tb3+ radius (0.92 Å) is closer to that of Lu3+ than that of Pr3+ (0.99 Å). The small ionic radius of TbIV (0.76 Å), the charge mismatch, and the reducing atmosphere during synthesis make TbIV unstable in the Lu3+ sites. On the other hand, the larger size of PrIV (0.85 Å) can stabilize PrIV even in the aliovalent Lu3+ sites due to the same ionic radii (Figure 6, left). There is evidence in literature40 that

Figure 7. UV−vis excitation spectra of Lu2O3:Pr3+(,HfIV) (left), Lu2O3:Eu3+ (middle), and Lu2O3:Tb3+(,Ca2+) (right) materials. (Right, inset) First derivative of UV−vis excitation spectrum of the Lu2O3:Tb3+(,Ca2+) material, indicating splitting of the 5d levels. 19598


Research Article


Figure 8. UV-excited luminescence spectra of Lu2O3:Pr3+(,HfIV) (left), Lu2O3:Eu3+ (middle), and Lu2O3:Tb3+(,Ca2+) (right) materials.

transitions can be assigned according to the final level as follows, though level labeling is uncertain due to mixing of levels’ wave functions: 5D0 (ca. 17 250 cm−1); 5D1 (ca. 18 800 cm−1); 5D2 (ca. 21 500 cm−1); 5D3 (ca. 24 100 cm−1); 5L6 (ca. 25 400 cm−1); 5G2−6 (ca. 25 800 cm−1); 5L7 (ca. 26 600 cm−1); 5 D4 (ca. 27 600 cm−1); and 5HJ, 5FJ, 5IJ, and 3P0 (ca. 29 600− 35 100 cm−1).56 In addition, a broad band centered at ca. 253 nm is assigned to the O2−(2p) → Eu3+ ligand-to-metal charge transfer (LMCT) transition. This transition is intimately related to the low reduction potential of Eu3+ (substantially lower than those of Tb3+ or Pr3+),57,58 the easy oxidation of O2−, and the position of Eu3+/2+ energy levels in the Lu2O3 host band structure.59 The excitation spectrum of Lu2O3:Tb3+,Ca2+ (0.5 and 1.5 mol %), monitoring the 5D4 → 7F5 transition at 544 nm, presents ostensibly at least three broad bands arising from the 4f8 → 4f75d1 transitions of Tb3+ (Figure 7, right). In reality, there should be five bands, each with fine structure, which are smeared away by the effect of high temperature and structural disorder caused by the Ca2+ codopant. In addition to the highintensity bands centered between ca. 260 and 300 nm, due to parity-allowed 4f8 → 4f75d1 transitions, the much less intense bands may be due to parity-forbidden 4f8 → 4f8 transitions (or high-spin 4f 8 → 4f7 5d 1 transitions). The intense Tb 3+ interconfigurational transitions and the ostensibly complete absence of 4f8 → 4f8 transitions may indicate that the 4f8 → 4f75d1 transitions are not within the conduction band in Lu2O3, as reported earlier for Tb3+-doped CdSiO3.16 However, there are well-documented Ce3+ cases (e.g., YAG:Ce) where excitation to the 5d1 levels, at least partially in the conduction band, is strong and the bands are intense. This matter needs further study, however. Comparison of the excitation spectra of Lu2O3:Tb3+ and that of the material with Ca2+ codopant (Figure 7, right) shows a small red shift for the Ca2+-codoped material. Defects generated by the aliovalent Ca2+ codopant can change the crystal field around the Tb3+ dopant and affect the Tb3+ 4f8 → 4f75d1 transitions in a manner similar to Pr3+ (as described). 3.5. UV Excited Luminescence of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) Materials. UV-excited luminescence spectra of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials (Figure 8) were recorded at room temperature, under excitation, via the 4f2 → 4f15d1 (Pr3+), O2−(2p) → Eu3+ LMCT, and 4f8 → 4f75d1 (Tb3+) transitions, respectively. The Lu2O3:Pr3+,HfIV (0.05 and 0.1 mol %) emission spectrum (Figure 8, left) shows somewhat overlapping groups of narrow emission lines of the intraconfigurational transitions of Pr3+: 3P0 → 3H5, 3F3,4, 1G4 and 1 D2 → 3H4−6, 3F2. Similar Lu2O3:Pr3+(,HfIV) UV-excited

luminescence spectra (Figure 8, left) indicate that aliovalent codoping with HfIV has no visible effect on the 4f2 → 4f2 transitions. The red and near-IR emissions observed from Pr3+(,HfIV) doped Lu2O3 materials are due to coupling of 3P0 → 1D2 relaxation with 3H4 → 3H6 excitation via a crossrelaxation process (Figure 9, top left). Coupling of 3P0 → 3H6

Figure 9. (Top row) Cross-relaxation processes between two Pr3+ ions via coupling of 3P0 → 1D2 relaxation with 3H4 → 3H6 excitation (left) or 3P0 → 3H6 relaxation with 3H4 → 1D2 excitation (right). (Bottom row) Cross-relaxation processes between two Tb3+ ions via coupling of 5 D3 → 5D4 relaxation with 7F6 → 7F0 excitation (left) or 5D3 → 7F0 relaxation with 7F6 → 5D4 excitation (right).

relaxation with 3H4 → 1D2 absorption involving much higher doses of energy cannot be ruled out (Figure 9, top right), however. This Pr3+ coupling causes strong quenching of 3P0−2 → 3H4 transitions (blue-green emission) and their replacement with mainly 1D2 → 3H4 transition (red emission). Formation of X X PrLu −OXO−PrLu clusters is a possible explanation for this 19599


Research Article


Figure 10. Persistent luminescence spectra of Lu2O3:R3+,M materials (R = Pr and M = HfIV, left; R = Eu, middle; or R = Tb and M = Ca2+, right) prepared by MASS (with or without H3BO3 flux) and ceramic methods.

bottom left) or 5D3 → 7F0 relaxation with 7F6 → 5D4 absorption (Figure 9, bottom right) due to the defect cluster TbXLu−OXO−TbXLu. Differences between the ionic radii of these dopants and Lu3+, difficult diffusion of dopants in the Lu2O3 host, or insufficient homogenization of the precursor material can be plausible causes of formation of terbium clusters. Another explanation for the absence of emission from the 5D3 level is the position of this energy level, which can be inside (or close to) the conduction band, as observed earlier for Tb3+doped CdSiO3.16 3.6. Persistent Luminescence Phenomena of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) Materials. Persistent luminescence spectra of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials (Figure 10) were recorded at room temperature with 3 min of delay after cessation of irradiation at 267, 255, and 261 nm, respectively. The broadness of the initial lines of the 4fn → 4fn transitions is due the use of wide slits required by the lower emission intensity of persistent luminescence than UV irradiation. The presence of several persistent emission species due to lattice defects (excitation energy traps) cannot be excluded, either, as a source of broad lines. Persistent luminescence emission of Lu2O3:Pr3+,HfIV (0.05 and 0.1 mol %) is in the red and near-IR region (Figure 10, left). The spectra are composed of emission bands assigned to 1 D2 → 3F2, 3H4−6 and 3P0 → 1G4, 3F3,4, 3H5 transitions of Pr3+. In general, the emission spectra are similar to the UV-excited luminescence ones (Figure 8, left) although the 1D2 → 3H4 transition has small differences in crystal field fine structure and the intensities of 1D2 → 3F2, 3H5,6 and 3P0 → 1G4, 3F3,4 transitions are higher for persistent luminescence than under UV excitation. The Lu2O3:Eu3+ (0.2 mol %) persistent luminescence spectra show emission assigned to the 5D0 → 7F0−4 transitions of Eu3+ and are similar to those obtained with UV excitation (Figure 8, middle). Indeed, the major difference is the low intensity of the 5 D0 → 7F2 transition and the absence of emission assigned to the 5D0 → 7F1 transition of Eu3+ in C2. This result has been interpreted by more efficient energy transfer from the energy stored in traps to Eu3+ in the S6 site than to Eu3+ in the C2 site.24,62 Since the 5D0 energy of the S6 site is expected to be higher than that of the C2 site, the observation suggests that the traps feed mainly the S6 emission, for example, situating spatially close to this site and making note of the weakness of energy transfer from S6 site to C2. Materials with persistent luminescence emission in the red to near-IR range, such as Lu2O3:Pr3+,HfIV (and Lu2O3:Eu3+), are not common and they

photoluminescence phenomenon.40 However, more detailed structural studies, for example, using the extended X-ray absorption fine structure (EXAFS) method, should be done to solve this kind of statement. Also theoretical (e.g., transition probabilities) and experimental (e.g., decay times) spectroscopic investigations may throw more light on this matter. The UV-excited luminescence spectrum of Lu2O3:Eu3+ (0.2 mol %) exhibits 5D0 → 7FJ transitions (Figure 8, middle), with the barycenters at 7F4 (ca. 14 050 cm−1); 7F3 (ca. 15 300 cm−1); 7 F2 (ca. 16 300 cm−1); 7F1 (ca. 16 800 cm−1); and 7F0 (ca. 17 170 cm−1). Because the magnetic dipole 5D0 → 7F1 transition is allowed for Eu3+ in both the C2 and S6 sites of Lu2O3, there are emission lines at 16 600, 16 800, and 17 000 cm−1 as well as at 16 710 and 17 110 cm−1 originating from the C 2 and S6 sites, respectively, in agreement with the literature.56,60,61 Since the Eu3+ ion in the Lu2O3 host also occupies a site with a center of inversion (the S6 site; Figure 4, left), the 5D0 → 7F2 hypersensitive transition is rigorously forbidden by the electric dipole mechanism (Laporte’s rule). The high-intensity line at 16 300 cm−1, so characteristic of Eu3+ emission in the cubic rare earth sesquioxides, is assigned to the 5 D0 → 7F2 transition arising from Eu3+ exclusively in the C2 site (transition allowed as a forced electric dipole).56 However, despite most electronic transitions being forbidden for Eu3+ in the S6 site, very weak lines due to vibronic coupling can be found as well. The emission spectrum of the Lu2O3:Tb3+,Ca2+ (0.5 and 1.5 mol %) material (Figure 8, right) shows narrow emission lines from Tb3+ assigned to the 5D4 → 7FJ transitions: 7F0 (ca. 14 500 cm−1); 7F1 (ca. 15 000 cm−1); 7F2 (ca. 15 300 cm−1); 7 F3 (ca. 16 000 cm−1); 7F4 (ca. 16 900 cm−1); 7F5 (ca. 18 200 cm−1); and 7F6 (ca. 23 300 cm−1). The emission spectrum of Tb3+,Ca2+-doped material excited with vacuum UV radiation at 110 nm (Figure S4) is similar to the one excited in Tb3+ 4f8 → 4f75d1 transitions. This result indicates that despite exciting Lu2O3:Tb3+,Ca2+ with an energy much higher than the Lu2O3 band gap (220 nm), which is able to generate electron−hole pairs, the emission process remains the same. Besides, the material without Ca2+ codopant shows similar UV-excited luminescence to Lu2O3:Tb3+,Ca2+ that indicates no apparent visible influence of Ca2+ aliovalent doping on 4f7 → 4f7 transitions (Figure 8, right). The absence of 5D3 → 7FJ transitions in the blue (and near-UV) range is rather serendipitous when one takes into account the low Tb3+ dopant concentration (0.5 mol %).16 This may be interpreted by nonradiative decay via a cross-relaxation process coupling 5 D3 → 5D4 relaxation with 7F6 → 7F0 absorption (Figure 9, 19600


Research Article


Figure 11. Persistent luminescence decay curves for Lu2O3:R3+,M materials (R = Pr and M = HfIV, left; R = Eu, middle; or R = Tb and M = Ca2+, right) prepared by MASS (with or without H3BO3 flux) and ceramic methods. (Right, inset) Persistent luminescence decay curve for Lu2O3:Tb3+,Ca2+ prepared by MASS with H3BO3 flux during the first 8 h.

aliovalent codopants that can generate charge-compensating defects. Lu2O3:Tb3+,Ca2+ (0.5 and 1.5 mol %) prepared by MASS with H3BO3 flux yielded the highest intensity and longest duration of persistent luminescence (Figure 11, right). The persistent luminescence of this material can achieve 8 h, although qualitatively the persistent luminescence was detected until 30 h by a charge-coupled device (CCD) camera (Figure 11, right inset). The long persistent luminescence lifetime for Tb3+-doped materials is due to a high density of traps from which the trapped energy (i.e., charge carriers) can be released at room temperature. Alternatively, the shallow traps emptied already at temperatures lower than room temperature are suppressed. This alternative is clearly true since most of the energy stored is released very fast; that is, there is a high density of shallow traps (cf. Figure 11). For the persistent phosphors based on electron trapping, such as the Pr3+-, Eu2+-, and Tb3+-doped materials, the shallow traps are electron traps just below the lower edge of the conduction band, for example, oxygen vacancies. Elimination of these shallow traps should result in good persistent emission. Borate flux is used frequently in the manufacture of aluminate persistent phosphors, not only with Eu2+,Dy3+codoped SrAl2O4 but also with probably the best aluminate phosphor, Eu2+,Dy3+-codoped Sr4Si14O25. The structures of these aluminates are rather rigid and the borates can offer a convenient way to modify the trap structure of these aluminates, since the borates can be formed with either tetrahedral BO4, triangular planar BO3 or more complicated polymeric units. The similar flexibility of the borate can help to remove the shallow traps detrimental to room-temperature persistent luminescence. Detection of these forms of borates is, however, not an easy task. The effect of the borate flux is lowest for the Eu3+-doped Lu2O3 material. The persistent luminescence of these materials is based on hole trapping, which, by definition, means low efficiency because of the low mobility of holes in the valence band. Very little is known presently on these phosphors, so the effect of the borate flux is impossible to elucidate at present. In order to investigate the influence of codopants on the persistent luminescence phenomenon of Lu2O3:Pr3+,HfIV and Lu2O3:Tb3+,Ca2+ materials, their emission spectra and decay curves (Figures S5 and S6), as well as those of the respective phosphors prepared without codopants, were recorded. The spectra of codoped and non-codoped materials show the same features, but the codopants significantly increase the intensity (by at least an order of magnitude) and duration of persistent

can be potential candidates for bioimaging probes and solar cell sensitizers. Lu2O3:Tb3+,Ca2+ (0.5 and 1.5 mol %) persistent luminescence spectra (Figure 10, right) also exhibit narrow emission lines assigned to 5D4 → 7F6−0 transitions of the Tb3+ ion, similarly to the UV-excited luminescence spectrum (Figure 8, right). The small differences, if any, observed between persistent luminescence and UV-excited spectra may be explained by the preferential persistent luminescence emission from Tb3+ in the S6 site than in the C2 site. The more complex crystal field energy level structure for the 5D4 level of Tb3+ than for the 5D0 level of Eu3+, together with the corresponding transition selection rules, makes difficult the confirmation of this hypothesis with Tb3+ spectra, however. Persistent luminescence spectra of Lu2O3:R3+,M (Pr,HfIV; Eu; or Tb,Ca2+) materials prepared by MASS with or without H3BO3 flux and by the ceramic method with H3BO3 flux show the same 4fn → 4fn transitions (Figure 10). More important for practical applications, the phosphors prepared by the MASS method with H3BO3 flux have the highest persistent intensity when all materials are recorded under the same experimental conditions. Persistent luminescence decay curves of the Pr3+-, Eu3+-, and Tb3+-doped Lu2O3 materials (Figure 11) were obtained by monitoring emission at 634, 613, and 544 nm, after cessation of excitation at 267, 255, and 261 nm, respectively. For Lu2O3:Pr3+,HfIV (0.05 and 0.1 mol %) material prepared by MASS with H3BO3 flux, the persistent luminescence decay curve (Figure 11, left; topmost line) shows a fast initial decay (ca. 40 s) before stabilizing to a slow linear decay with high intensity even until 1200 s. This indicates that the short and long red persistent luminescence emission of Lu2O3:Pr3+,HfIV materials is from the release of energy trapped in shallow and deep traps at room temperature, respectively. Lu2O3:Pr3+,HfIV has been reported to possess deep traps and can thus be applied as an energy storage material even at high temperatures.23 However, thermoluminescence studies are needed to confirm this claim in Lu2O3:Pr3+,HfIV as well as to obtain the trap densities and depths. The persistent luminescence decay curve of Lu2O3:Eu3+ (0.2 mol %) prepared by MASS with H3BO3 flux (Figure 11, middle) shows persistent luminescence emission even until 1100 s after UV irradiation ceases. Since all materials were recorded in similar experimental conditions, the lower intensity and shorter duration of Lu2O3:Eu3+ persistent luminescence compared to those of Lu2O3:Pr3+,HfIV and Lu2O3:Tb3+,Ca2+ (Figure 11, left and right) may be due to the absence of 19601


Research Article


lower contribution of the 5D0 → 7F2 transition (from Eu3+ in the C2 site) in the persistent luminescence spectrum compared with that under UV excitation. The Pr3+-, Eu3+-, and Tb3+doped Lu2O3 materials show red, reddish orange, and green persistent luminescence emission colors, respectively, indicating that trivalent rare earth dopants can tune the persistent luminescence emission color. Finally, these persistent luminescence emitter materials have potential applications, such as tracer particles for in vivo medical imaging and sensitizers to improve the efficiency of solar cells.

luminescence. These data indicate that HfIV and Ca2+ dopants act as appropriate trap generators or/and modifiers, considerably improving the persistent luminescence intensity and duration. The persistent luminescence decay curves of all materials prepared by MASS with H3BO3 flux (Figure 11) exhibit higher persistent luminescence intensity and duration than the phosphors prepared by other methods (ceramic method and MASS without flux), when analyzed under the same experimental conditions. However, more experimental analyses (e.g., thermoluminescence) are necessary to probe trap formation by H3BO3 flux. Besides, the melting of H3BO3 and the internal heating by MASS can improve the dopant distribution in the material, decreasing luminescence quenching due to R3+−R3+ clustering. When a persistent luminescence phosphor is excited, the luminescence intensity does not immediately reach a constant value. This phenomenon (so-called persistent luminescence charging) is related to the trap density.63 Since the curves of Lu2O3:R3+,M (Pr,HfIV; Eu; Tb3+,Ca2+) were recorded under similar experimental conditions (Figure S7), the Pr3+- and Eu3+doped Lu2O3 materials exhibit faster charging (

Rapid and Energy-Saving Microwave-Assisted Solid-State Synthesis

Recommend Documents