Nd-doped NASICON-type nanophosphors for near-infrared excitation and emission

ABSTRACT Neodymium-doped phosphates, (Nd1-x Gd x )0.33Zr2(PO4)3 (0 ≤ x ≤ 1), were synthesized by co-precipitation. (Nd1-x Gd x )0.33Zr2(PO4)3 was obtained as a single-phase and was confirmed to be a NASICON-type structure consisting of a three-dimensional network of PO4 tetrahedra sharing corners with ZrO6 octahedra. The particle size of the (Nd1-x Gd x )0.33Zr2(PO4)3 samples was in the nanoscale, which is suitable for in vivo optical imaging. The (Nd1-x Gd x )0.33Zr2(PO4)3 samples showed characteristic luminescence corresponding to the f – f transitions of Nd3+. The highest emission intensity at 1072 nm with excitation at 824 nm was observed for (Nd0.75Gd0.25)0.33Zr2(PO4)3, which was 4.5 times higher than that of Nd0.33Zr2(PO4)3. The near-infrared (NIR) emission intensity of this nanophosphor was significantly higher than that of indocyanine green, which is actually used as an in vivo optical probe reagent.


Introduction
Bioimaging using optical probes is an essential technology for biomedical research and clinical diagnosis. In particular, optical probes that emit near-infrared (NIR) light with NIR excitation have attracted attention as a promising material to dramatically improve in vivo optical imaging. The NIR light in the wavelength range of 700-1400 nm is called the "optical transmission window" and provides deep tissue penetration, reduced photodamaging effects, and low autofluorescence and light scattering. Thus, safe and high signalto-noise ratio images can be obtained [1][2][3][4][5][6][7][8][9]. Indocyanine green (ICG) is well known as the only practical reagent capable of excitation and emission within the narrow wavelength region of the "optical transmission window" of tissues [10]. However, there are problems to be solved, such as low emission intensity, low photostability, shallow penetration depth, and low quantum yield (QY) [4,6]. In addition to organic fluorescent dyes, several materials have been investigated that are excited by NIR light and emit NIR: semiconductor quantum dots (QDs) [11][12][13][14][15], carbon-based materials such as single-walled carbon nanotubes (SWCNTs) [16][17][18][19][20], and up-conversion phosphors [21][22][23][24]. Unfortunately, however, there are concerns with these materials. The QDs often contain toxic elements such as cadmium, arsenic, lead, and mercury, making them unsuitable for in vivo use. A concern with less toxic NIR emitting QDs is their low QY. For example, Ag 2 S is a low toxicity QD, but has a QY of only 15% [24], and SWCNTs have low toxicity and good photostability, but low QY (about 10%) [18]. The up-conversion phosphorescent materials that convert NIR light into visible light exhibit critically low QY because of the inevitable multiphoton excitation in the emission mechanism. Therefore, new NIR luminescent materials for optical probes with high QY are needed, since low QY generates extra infrared (IR) emission and causes undesirable heat load during in vivo optical imaging.
Rare-earth doped ceramic phosphors that emit light based on f-f transitions usually exhibit low long-term cytotoxicity, low photobleaching, long luminescence lifetime, and thermal and chemical stability. They also generally have high QYs because the luminescence mechanism is classified as a downshift process. Phosphors containing trivalent neodymium ions (Nd 3+ ) show emissions around 880, 1060, and 1340 nm due to the f-f transition of Nd 3+ upon NIR excitation, which is suitable for in vivo optical imaging [25][26][27][28][29][30]. However, the excitation and emission efficiencies of Nd 3+ -doped phosphors are usually insufficient, because the f-f transitions of Nd 3+ are parity forbidden. Doping the host lattice with high concentrations of Nd 3+ solves this problem [31][32][33], but excess doping of Nd 3+ ions above an appropriate concentration usually quenches the luminescence. This is due to the migration of the excitation energy among activator ions [34] and is called "concentration quenching". One means of solving the problem is to select a host material with a large distance between the activator ions in the lattice to suppress concentration quenching.
The Na + ion super ionic conductor (NASICON) structure, which is thermally and chemically stable, nontoxic, and has long distances between the active agent ions, is one of the leading candidates for this study [35][36][37]. The NASICON-type structure in the space group of P-3c1 illustrated by the VESTA program is shown in Figure 1 [38]. This structure adopts a three-dimensional network of AO 6 (A = Ti, Zr, and Hf) octahedra sharing corners with PO 4 tetrahedra, and M ions (M = alkaline or alkaline earth metal), which can be substituted by rare earth ions, are present in the interstitials [39,40]. The interstitial sites are separated from each other by a distance of more than 5 Å [41]. Although several studies have been reported on Nd 3 + -containing oxides with NASICON-type structures such as Nd 0.33 Zr 2 (PO 4 ) 3 [35,40,42], to our knowledge their luminescence properties have not been investigated and their potential as in vivo optical imaging probes has not been explored. In this study, nano-sized (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x ≤ 1.0) phosphors were synthesized by a precipitation method and their fluorescence properties were investigated. To suppress the concentration quenching of Nd 0.33 Zr 2 (PO 4 ) 3 , Gd 3+ , which is chemically and thermally stable and whose ionic radius is close to that of Nd 3+ , was selected as a buffer ion.

Synthesis of (Nd 1-x Gd x ) 1/3 Zr 2 (PO 4 ) 3 (0 ≤ x ≤ 1) nanoparticles
(Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x ≤ 1.0) nanoparticles were synthesized by the precipitation method [41]. Nd 2 O 3 and Gd 2 O 3 weighed in stoichiometric ratios were dissolved in diluted nitric acid solution (20 cm 3 ), and then a stoichiometric amount of ZrCl 2 O solution (92.136 g L −1 ) was added. A stoichiometric amount of H 3 PO 4 solution (8.5%) was then added subsequently to the mixed solution under constant stirring, heated at 50°C for 1 h, and then evaporated to dryness in the air at 80°C for 12 h. The resulting powder was carefully ground and calcined in a muffle furnace (Isuzu, EPTR-26K) at 700 − 900°C for 5 − 20 h. To test sample dispersibility, the sample powders (0.01 g) were dispersed in deionized water (10 mL) using ultrasound for 5 min. The solutions were left to rest at room temperature for one hour to allow for sedimentation, after which the coarse particle residue was removed by decantation. The resulting solution is treated as sample solution.

Characterization
The sample composition was analyzed by X-ray fluorescence spectrometry (XRF; Rigaku, ZSX Primus). The crystal structure was identified by X-ray powder diffraction (XRD; Rigaku Ultima IV) using Cu-Kα radiation (40 kV and 40 mA). The data were collected by step scanning in the 2θ range from 20° to 80° with a step size of 0.02° and a scan rate of 6° min −1 . Rietveld refinement of the resulting XRD patterns obtained in the 2θ range from 10° to 120° by an X-ray diffractometer (Bruker D2 PHASER) with monochromatic CuKα radiation (10 mA and 30 kV) was performed by the RIETAN-FP software package [43]. The lattice parameters and lattice volumes were calculated from the XRD peak angles refined with α-Al 2 O 3 as a standard material using the CellCalc Ver. 2.20 software. Transmission electron microscope images were taken at an acceleration voltage of 300 kV (TEM; Hitachi H-9000NAR). The average particle size was estimated by measuring the maximum diameter of 200 particles in one direction on the TEM images. The NIR photoluminescence (PL) excitation and emission spectra of the sample powders were measured at room temperature using a fluorescence spectrometer (Horiba, Fluorolog-3), where the emission spectra were obtained for excitation at 824 nm and the excitation spectra were recorded for emission at 1072 nm. Figure 2 shows the XRD patterns of the Nd 0.33 Zr 2 (PO 4 ) 3 samples after heating at 700-900°C in air for 20 h. The sample synthesized at 700°C was obtained in the amorphous phase. On the other hand, the samples synthesized at 800°C and above yielded a trigonal NASICONtype structure in a space group of P-3c1 as the main phase, although Zr 2 O(PO 4 ) 2 and ZnO 2 are detected as impurities. This result indicates that the precursor obtained as an amorphous phase crystallized upon heating at 750°C and above. Since a single phase was  obtained at 750°C but the crystallinity was low, the optimum reaction temperature was 800°C, although a very small amount of ZrO 2 was observed.

Results and discussion
In addition, the samples were synthesized at different reaction times to determine the optimum reaction time to obtain a single-phase sample. Figure 3 shows the XRD patterns of the Nd 0.33 Zr 2 (PO 4 ) 3 samples obtained by heating the precursor at 800°C for 5 to 20 hours in air. A single phase with NASICON-type structure was obtained by heating at 800°C for less than 10 h. From this result, the optimum reaction time to obtain a single phase while suppressing particle growth was determined to be 5 h.
The composition of each sample was confirmed to be stoichiometric by XRF analysis. Figure 4 shows the XRD patterns of the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x ≤ 1.0) samples. All samples were obtained in a single phase with a trigonal NASICON-type structure in a space group of P-3c1 (Inorganic Crystal Structure Database: ICSD No. 245204). Figure 5 shows the composition  dependence on lattice volume of the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 samples, estimated from the peak angles of X-ray diffraction. The lattice volume decreased monotonically with increasing Gd content, indicating that smaller Gd 3+ (ionic radius: 0.0938 nm for 6 coordination) [44] was substituted with larger Nd 3+ (ionic radius: 0.083 nm for 6 coordination) [44] to form solid solutions. As shown in Table 1, the a parameter increased whereas the c parameter decreased with increasing Gd content. Similar behavior has been observed in other solid solutions with NASICON-type structures, which can be explained by structural changes [45]: the NASICON-type M 0.33 Zr 2 (PO 4 ) 3 contains columns between two adjacent ZrO 6 octahedral faces along the c axis, and the decrease in the ionic radius of the M cation causes the c parameter decreases [35][36][37]. On the other hand, the a parameter increases with decreasing ionic radius of the M cation, due to the correlated rotation of the PO 4 tetrahedron connecting the columns parallel to the ZrO 6 octahedron. The introduction of Gd 3+ , which is smaller than Nd 3+ , into the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x ≤ 1.0) system resulted in an increase in the a parameter and a decrease in the c parameter. Figure 6 shows the photoluminescence emission spectra at room temperature of the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x < 1.0) samples excited at 824 nm. Three characteristic peaks due to the f-f transition of Nd 3+ were observed at 910 nm, 1072 nm, and 1309 nm in the near-infrared region for all samples, which were attributed to the 4 F 3/2 → 4 I 9/2 , 4 F 3/2 → 4 I 11/2 , and 4 F 3/ 2 → 4 I 13/2 transitions, respectively. Because the emission spectra between 800 and 850 nm should overlap with the excitation spectra peaked at 824 nm, an emission peak related to the 4 F 5/2 → 4 I 9/2 transition was not found. No emission due to the 4 F 5/2 → 4 I 9/2 transition was seen even under excitation with multiple wavelengths, most likely because the luminescence intensity was too low. Therefore, the emission due to the 4 F 5/2 → 4 I 9/2 transition was left out of the following discussion. The 4 F 3/2 → 4 I 11/2 transition showed the strongest emission line at 1072 nm, which was more than five times higher than the other emission lines in Figure 6. This characteristic can be attributed to the large crystal field splitting of the Nd 3+ 4 F 3/2 metastable manifold, as observed in other Nd 3+ -doped ceramic phosphors with low symmetry around Nd 3+ [46].
The intensity of each transition depends on the branching ratio (β), defined as the relative emission probability for each transition. The branching ratio can be estimated from the relative contribution to  the total integrated luminescence of the emission spectrum in Figure 6 as follows: where I(λ) is the integrated emission intensity at a certain wavelength. In this case, the background intensity is removed. Note that the emission band due to the 4 F 3/2 → 4 I 15/2 transition is observed outside the detection range of our spectrometer, and the branching ratio corresponding to this transition is less than 1% [47][48][49][50]. Therefore, the contribution of this transition can be neglected in the following calculations and discussion without affecting the results. Figure 7 shows change in the branching ratios for the 4 F 3/2 → 4 I 9/2 , 4 F 3/2 → 4 I 11/2 and 4 F 3/2 → 4 I 13/2 transitions with increasing Gd content. The branching ratios at 910 nm and at 1309 nm increased from 12.8 to 18.8% and from 10.7 to 16.3%, while that at 1072 mm decreased from 76.5 to 64.8%. It is suggested that the branching ratio can be adjusted according to the application by changing the Gd content. The change in branching ratio implies a change in the spontaneous emission probability of the 4 F 3/2 → 4 I 9/2 , 4 I 11/2 and 4 I 13/2 transitions. The continuously enhanced 4 F 3/2 → 4 I 9/2 and 4 F 3/2 → 4 I 13/2 transitions suggest that the crystal field around Nd 3+ increased with increasing Gd content [51,52]. Since the development of detectors for in vivo optical imaging consisting of AsGaIn arrays, nanophosphors working in the so-called "second biological window" (1000-1400 nm) have attracted much attention. They can improve both image resolution and transmission depth. In such detectors, the relative detection sensitivities at 910, 1072, and 1309 nm are close to 0.4, 0.8, and 0.8 (arbitrary units), respectively [53]. The emission due to the 4 F 3/2 → 4 I 9/2 transition at 910 nm was observed in the region of low relative detection sensitivity. The low emission probability of the 4 F 3/2 → 4 I 13/2 transition at 1309 nm was not compensated by the high relative detection sensitivity in the corresponding spectral region. Moreover, the emission intensity of the 4 F 3/2 → 4 I 13/2 transition is significantly reduced in the solvent due to strong emission quenching by the hydroxyl group [54]. Therefore, monitoring the 4 F 3/2 → 4 I 11/2 emission intensity at 1072 nm will be suitable for in vivo optical imaging in the "second biological window" region using an AsGaIn array detector. The result that the branching ratio of the 4 F 3/2 -4 I 11/2 transition for the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x < 1.0) samples increased with decreasing Gd content indicates that the samples with lower Gd content are more suitable for in vivo optical imaging of the "second biological window" region. Figure 8 shows the Gd concentration dependence of the photoluminescence emission peak intensity at 1072 nm for the (Nd 1-x Gd x ) 0.33 Zr 2 Figure 7. Gd content dependence on the branching ratios for the 4 F 3/2 → 4 I 9/2 , 4 F 3/2 → 4 I 11/2 , and 4 F 3/2 → 4 I 13/2 transitions in the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x < 1.0) samples under excitation at 824 nm.
(PO 4 ) 3 (0 < x ≤ 1.0) samples. Among the samples synthesized in this study, the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample exhibited the highest emission, which was found to be 4.5 times higher than that of the Nd 0.33 Zr 2 (PO 4 ) 3 sample without Gd. In rareearth phosphors, concentration quenching generally occurs at luminescent ion concentrations below 5 mol %. However, higher critical concentrations were obtained in the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 system. This is because Nd 3+ , which acts as the activating ion, is located at the M cation sites more than 5 Å away. The Rietveld analysis of the XRD pattern of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample revealed that the distance between the nearest neighbor Nd 3+ ions was 8.74894(0)Å; the Rietveld refinement profile of the sample is shown in Figure  S1, and the crystallographic data and structural refinement parameters are summarized in Tables S1 and S2, respectively. It was experimentally confirmed that the Nd 3+ ions are separated from each other by more than 5Å. Such long-range separation suppresses concentration quenching [36,37]. The concentration quenching mechanism of phosphors was investigated by calculating the critical distance (R c ) between Nd 3+ ions. The critical distance for energy transfer can be approximately estimated using the following relation proposed by Blasse [49]: where X c is the critical concentration, N is the number of cation sites in the unit cell, and V is the volume of the unit cell. For the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 system, X c = 0.75, N = 2, and V = 1521.8 Å 3 , which are obtained in the Rietveld refinement (Table S1), resulting in a calculated R c of 12.5 Å. Two general concentration quenching mechanisms have been proposed: nonradiative energy transfer between ions (exchange interaction) and electric multipolar interaction. In the exchange interaction model, the R c between the sensitizer and the activator has to be shorter than 5 Å [55] and the emission and excitation spectra have to overlap for radiation reabsorption. However, the critical distance between the Nd 3+ ions is longer than that required for the exchange interaction and thus is not involved in the energy transfer. Therefore, the multipolar interactions between Nd 3+ ions are considered to be involved in the energy transfer mechanism of the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x < 1.0) system. The (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample exhibited the highest emission intensity and high branching ratio (72.2%) from 4 F 3/2 to 4 I 11/2 among the (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x < 1.0) powders. From these results, it can be concluded that the sample with x = 0.25 is most suitable for in vivo optical imaging. Figure 9 shows TEM photographs of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample. Although agglomerated particles were observed, the average primary particle size was less than 100 nm. Since the particle size suitable for in vivo optical imaging is 10-200 nm, the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 samples is expected to be a candidate as an in vivo optical imaging material. Figure 10 shows photographs of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample dispersed in deionized water under daylight and NIR light (980 nm). The solution is transparent (see Figure 10(a)). The Tyndall effect was detected in the solution, as shown in Figure 10(b), showing that the sample particles are disseminated as colloidal particles in deionized water. Despite the TEM image in Figure 9 showing particle agglomeration, this finding suggested that the sample particles may be disseminated steadily in deionized water without sedimentation. The colloidal suspension exhibits NIR luminescence under NIR light excitation (980 nm) as shown in Figure 10(c); the photograph of deionized water without the sample is shown in Figure 10(d), as a comparison.
The emission intensity of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample was compared with that of indocyanine green (ICG), a green organic dye that is in practical use as a bioimaging reagent. Figure 11 shows the photoluminescence emission spectra of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 sample and ICG at room temperature. No photoluminescence emission peak of ICG was observed under the same measurement conditions. The emission spectrum of ICG was multiplied by a factor of 200 for better comparison. The emission intensity of ICG at 910 nm was at least 200 times lower than that of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2  (PO 4 ) 3 sample. Furthermore, the emission intensity of ICG at 1072 nm, which will be suitable for in vivo optical imaging in the "second biological window" region using an AsGaIn array detector, is too low. Therefore, it is noteworthy that the emission intensity of (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 was significantly higher than that of ICG.

Conclusions
(Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x ≤ 1.0) nanophosphors were successfully synthesized by a precipitation method. The nanoparticles had a particle size of less than 100 nm and ranged from 10 to 200 nm, making them suitable for in vivo optical imaging materials. The (Nd 1-x Gd x ) 0.33 Zr 2 (PO 4 ) 3 (0 ≤ x < 1.0) nanophosphors exhibited NIR emission at 910 nm, 1072 nm, and 1309 nm due to the f -f transitions of Nd 3+ under NIR excitation at 824 nm. The highest emission intensity was obtained in the sample containing 75 mol% Nd 3+ . This indicates that the NASICON-type structure, in which the Nd 3+ ions are separated by more than 5 Å, can be highly doped with Nd 3+ while suppressing concentration quenching. Furthermore, the luminescence intensity of the (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 nanophosphor was found to be significantly higher than that of ICG. These results indicate that (Nd 0.75 Gd 0.25 ) 0.33 Zr 2 (PO 4 ) 3 nanophosphor is a promising optical probe for in vivo optical imaging.