1. Introduction

Phosphors with down-conversion emissions in the infrared have attracted attention due to their applications in lasing systems (wavelength, λ = 1000-3000 nm), optical amplifiers (λ = 1400-1600 nm), and photovoltaics (λ = 800-1100 nm) [1–4]. In addition to these applications, infrared emissions in the range of 1200-1700 nm are of interest for biomedical imaging applications [5]. The first (700-900 nm) and second (1200-1700 nm) windows are ranges of optical transparency in biological tissues due to reduced absorption and scattering [5–9], allowing contrast agents which operate at these wavelengths to penetrate biological systems deeper on the scale of millimeters to centimeters. Infrared emissions for imaging in the 1100-1600 nm range have been reported to improve imaging sensitivity by up to ~10 times [10–12].

Rare earth doped materials can be tailored to exhibit absorption and emission lines within the first and second windows. The absorption and emission properties of rare earth doped phosphors can be tailored by controlling the local environment, such as site symmetry, crystal field strength and electron-phonon interaction strength of the rare-earth dopants. Tunable emissions in the second window can be achieved under the same excitation by selecting different rare earth ions as emitting centers, which will allow for multispectral imaging. Multispectral imaging is the use of various surface targeting ligands on two or more phosphors with non-overlapping emissions to identify different diseased sites and states. Figure 1 shows how Yb³⁺ can serve as a sensitizer to enhance 980 nm absorption and energy transfer to a number of rare earth emitting ions (Ho³⁺, Pr³⁺, Tm³⁺_, Er³⁺) with infrared emissions within the second window. In phosphate and telluride glasses, it has been shown that Yb³⁺- Ln³⁺ (Ln³⁺ = Ho³⁺, Tm³⁺, and Pr³⁺) systems are capable of down-conversion of 980 nm excitation to 1200, 1450, and 1300 nm [13–16]. The 1550 nm emission characteristics and optical efficiency of another shortwave infrared emitting chemistry, NaYF₄:Yb³⁺-Er³⁺, was previously reported and will be used as a reference in this work [17]. Halide hosts (e.g. NaYF₄, YF₃, LaF₃, CaF₂) are favored for their low phonon energies that minimize non-radiative losses to enable bright SWIR emissions. The use of hexagonal NaYF₄ as a host material is advantageous due to its low phonon energy which promotes radiative relaxation, and favorable lattice sites for substitution with lanthanide ions [18]. However, the widespread use of NaYF₄ as a biomedical imaging marker is limited due to its potential cytotoxicity and lack of functionalized surface sites [9]. Recent work has shown that the use of functionalized albumin coatings can mitigate these factors and yield contrast agents comparable to quantum dots in terms of functionality [9], while displaying low systemic toxicity [19]. Besides the unique advantage of narrower emission bandwidths (<100 nm) and spectral tunability within both infrared windows, rare earth doped nanoparticles are also more stable than conventional organic dyes, showing almost no loss in emission intensities over months. In comparison, most organic dyes suffer from poor photostability, which results in reducing emission intensities after a day. Alternative infrared-emitting inorganic semiconductor substitutes comprise several well-known toxic elements (e.g. Hg, Cd and Pb), and are thus not favorable for biomedical applications, aside from their poor emission characteristics [5,11,12]. The detailed comparison and discussion of the rare earth doped materials, organic dyes and quantum dots is discussed elsewhere [20].

 

Fig. 1 Energy levels of Yb³⁺, Tm³⁺, Pr³⁺, Ho³⁺, and Er³⁺ showing excitation and emission pathways.

Download Full Size | PDF

To date, the vast body of work concerning NaYF₄: Yb³⁺, Ln³⁺ particle systems has focused on the up-conversion and down-conversion properties in the range of 300-1000 nm, with no evaluation of infrared down-conversion spectra in the range of 1000-1500 nm. In this work, we will examine the feasibility of using NaYF₄ as a host for multispectral shortwave infrared emissions in the second “biologically transparent” window. A complete study on the application of these materials for biomedical imaging is described elsewhere [20]. Hexagonal microcrystals of NaYF₄ at various dopant concentrations (i.e. Pr³⁺, Ho³⁺ and Tm³⁺) were synthesized to investigate the optimal doping concentrations for infrared emissions. Optical efficiencies were then measured to quantify their emissions brightness and ranked in comparison to a Kigre phosphate glass (QE-7S, Kigre Inc. Hilton Head, SC) and NaYF₄: Yb³⁺, Er³⁺ as internal references [17]. Optical efficiency, unlike quantum efficiency, provides a practical brightness metric to evaluate the performance of these fluorophores for biomedical imaging [17].

2. Experimental methods

2.1 Hydrothermal synthesis of NaYF₄: Yb³⁺, Ln³⁺ down-conversion particles

Powders with dopants concentrations corresponding to the Y³⁺: Ln³⁺ atomic ratios of ~(0.80 – x): x = (0.001, 0.0025, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.3, 0.5) and constant Yb: (Y + Ln) atomic ratios at ~0.2: 0.80 were prepared. These powders were used to study the effects of varying the emitting ion concentrations on the emissions intensities. Using the optimized dopants concentration of each system (x_opt), powders with concentrations corresponding to Ln: Yb atomic rations of x_opt: (0.10, 0.15, 0.30, 0.35) were prepared to study the sensitizer concentration effects on emissions intensities.

Stoichiometric amounts of 99.8% yttrium (III) nitrate, 99.9% ytterbium (III) nitrate, 99.9 + % lanthanide (III) nitrate; where lanthanide = holmium, praseodymium, and thulium (Sigma Aldrich, St. Louis, Missouri), were mixed with 1.5 times excess 99.0% sodium fluoride (Sigma Aldrich), and 8.0 g of polyvinyl-pyrrolidone (average Mn ~40 000 g/mol from Sigma Aldrich) in ~70 mL water/ethanol mixture (80:20 v/v) for 30 min. The mixture was then transferred to a 125 mL Teflon^TM liner, loaded in a Parr pressure vessel (Parr Instrument Co., Moline, IL) and heated to ~240°C for 4 hours. The as-synthesized particles were washed three times in ethanol followed by three times in deionized water by centrifuging (Beckman Coulter Avanti J-26 XP, Fullerton, CA) and dried at 70°C in a mechanical convection oven (Thermo Scientific Thermolyne, Waltham, MA) for powder characterization.

2.2 Powder characterization

Scanning electron microscopy (SEM) images of powder samples were taken using a Carl Zeiss Σigma field emission scanning electron microscope (Carl Zeiss, Carl Zeiss SMT Inc., Peabody, MA) using secondary electron detector operating at an accelerating voltage of 5.0 kV with a working distance of ~8.0 mm. Particle sizes, aspect ratios (i.e. major to minor axis), and morphology were evaluated manually from the SEM micrographs. Approximately thirty random particles were selected and measured for every powder sample from approximately 3 to 4 micrographs to provide the distribution of the particle sizes, and major to minor axis measurements. Energy dispersive X-ray (EDX) spectroscopy area scans of the powder samples were also completed to determine their elemental compositions under an accelerating voltage of 25 kV and a working distance of 8.5 mm for an aperture of 30 µm. The EDX elemental composition was evaluated by averaging elemental composition from five area scans and comparing the relative peak intensities assuming their total intensities to be 100%.

Powder X-ray diffraction (XRD) patterns were collected with a resolution of 0.04°/step size, 2 s/step using a Siemens D500 (Bruker AXS Inc., Madison, WI) powder diffractometer (40 kV, 30 mA) with Cu K_α radiation (λ = 1.54 Å). Powder diffraction file (PDF) from International Centre for Diffraction Data (Newton Square, PA) for hexagonal NaYF₄ PDF #97-005-1917 was used as reference.

The powder samples were packed in demountable Spectrosil far UV quartz Type 20 cells (Starna Cells, Inc., Atascadero, CA) with 0.5 mm path lengths for emission collection. The emission spectra of the particles were collected using an Edinburgh Instruments FSP920 spectrometer (Edinburgh Instruments, Livingston, United Kingdom) equipped with a Hamamatsu R928P photomultiplier tube detector under an excitation of ~976 nm with an external 2.5W laser (BW976, B&W Tek, Newark, NJ).

~0.2 g of powder samples of the particles were dry pressed into pellets of 1 cm diameter and 2 mm thickness. These pellets were used to measure the optical efficiencies of the powders. A modification of the C9220-03 quantum yield measurement system (Hamamatsu, Bridgewater, NJ) was used to measure the optical efficiency. In brief, the measurement principle is based on direct illumination and indirect reflection. Light enters the integrating sphere through the sample port, goes through multiple reflections, and is scattered uniformly around the interior of the sphere. For our measurements, the integrating sphere was set up in reflectance mode to measure total integrated reflectance of a surface. The PD300-IR power detector (Ophir-Spiricon, Logan, UT) measured the power of emitted light was used in place of the photomultiplier tube that was originally on the C9220-03 quantum yield measurement system. It was positioned at the port on the side of the sphere were the emitted beam is independent of the angular properties of light at the sample port. A further assumption made during measurements is that all light emanating from the different samples is isotropic, since there is no preferential ordering of the powders during the consolidation to form pellets.

3. Results and discussion

The XRD profiles in Fig. 2 show that pure hexagonal phase NaYF₄: Yb³⁺, Ln³⁺ (Ln³⁺ = Ho³⁺, Tm³⁺, Pr³⁺) particles with different rare earth emitting ions (Ln³⁺) concentrations and different sensitizer (Yb³⁺) concentrations were synthesized and optimized using the hydrothermal method. Evaluation of the (100) diffraction peak located at 2θ = 17.2° using Scherrer’s equation [21] showed average crystallite sizes of ~41 ± 7, 38 ± 5, and 33 ± 3 nm for Ho³⁺, Tm³⁺, and Pr³⁺ doped powders, respectively. The substitution of similarly sized rare earth dopants and the variation in concentration did not have an effect on observed XRD crystallite sizes. SEM images show that irregular, elongated micron-sized NaYF₄: Yb³⁺, Ln³⁺ particles were prepared regardless of dopant chosen (Fig. 3). The major axis and minor axis of particle sizes were determined from the SEM micrographs. Since the crystallite sizes were significantly smaller than the particle sizes, it was concluded that polycrystalline phosphors were synthesized. No statistically significant differences in both particle and crystallite sizes at a 95% confidence level were observed between the different doping levels. All the particles synthesized had a broad size distribution. The major axes lengths ranged from ~0.3 to 4 µm while the minor axes lengths ranged from ~0.1 to 0.9 µm. The differences in crystallite sizes and particle sizes indicate that polycrystalline micron-sized particles of NaYF₄:Yb³⁺, Ln³⁺ (Ln³⁺ = Ho³⁺,Tm³⁺,Pr³⁺) were synthesized.

 

Fig. 2 XRD profile of NaYF₄:Yb³⁺:RE³⁺ for RE³⁺ = Ho³⁺, Tm³⁺ and Pr³⁺. All diffraction peaks were attributed to hexagonal NaYF₄.

Download Full Size | PDF

 

Fig. 3 SEM micro-graphs showing synthesized NaYF₄: Yb_0.2, Ln_0.01, where Ln = (a) Ho³⁺, (b) Tm³⁺, and (c) Pr³⁺ doped NaYF₄ particles. Distributions of major axis were 1.31 ± 0.49, 1.21 ± 0.51, and 1.25 ± 0.47 µm, respectively. Distributions of minor axis were 0.39 ± 0.10, 0.36 ± 0.11, 0.40 ± 0.12 µm, respectively.

Download Full Size | PDF

The elemental composition of the powders synthesized with different dopant concentrations were measured using EDX (Fig. 4). This was done to ensure the theoretical concentrations of rare earth dopants (Ho³⁺, Tm³⁺, Pr³⁺) were obtained in the as-synthesized particles, particularly since rare earth concentrations strongly affects emission intensities. Figure 4 shows the EDX elemental composition of NaYF₄:Yb³⁺, Ho³⁺ particles synthesized with different Ho level concentrations. Since the theoretical values are within the 95% confidence interval (i.e. within 2σ), there is no statistically significant difference for the measured fractions of both Y³⁺ and Yb³⁺:Ho³⁺ when compared with that of the theoretical values. In addition, the trends show an increasing measured dopant concentration in accordance to the magnitude expected based on theoretical estimates. Any further citation in this work of the dopant concentration refers to the atomic fractions of the respective species in the precursor chemicals.

 

Fig. 4 EDX elemental composition of NaYF₄: Yb³⁺, Ho³⁺ particles synthesized with different Ho³⁺ concentrations. The blue and red horizontal lines represent the theoretical atomic fractions for Y³⁺ and Yb³⁺ + Ho³⁺, respectively. The error bars represent the range of ± 1σ, where σ is the standard deviation. For a statistical difference at a 95% confidence level, the values should be bounded within ± 2σ.

Download Full Size | PDF

The down-conversion emission spectra for dry powders of Ho³⁺, Tm³⁺, and Pr³⁺ doped NaYF₄:Yb³⁺ were collected. Bright and distinct emission peaks at 1185, 1475, and 1310 nm of Ho³⁺, Tm³⁺, and Pr³⁺ doped samples were observed under 980 nm excitation, respectively. Based on previous reports, these emission peaks can be attributed to the ⁵I₆→⁵I₈, ³H₄→³F₄, and ¹G₄→³H₅ transitions, respectively [13–15]. Figure 5 shows the normalized down-conversion emission spectra of the particles doped with the various emitting ions plotted on the same graph. The emission peaks intensities of Ho³⁺, Tm³⁺ or Pr³⁺ doped particles were normalized to their respective maximum peak emission to show clearly the relative position of the emission peaks of the three systems in comparison to NaYF₄:Yb³⁺, Er³⁺ [17]. The ranking for maximum emission peak intensities was NaYF₄:Yb³⁺, Er³⁺ > NaYF₄:Yb³⁺, Ho³⁺ > NaYF₄:Yb³⁺, Tm³⁺ > NaYF₄:Yb³⁺, Pr³⁺, where the ratio of the maximum peak intensities was 53:11:3:1, respectively. In addition, the emission peak for Pr³⁺-doped particles was two times broader than the peaks of either Ho³⁺-or Tm³⁺-doped particles. The difference in the emission intensities and peak widths was attributed to higher density of energy levels for Pr³⁺ as compared to that for Ho³⁺, Tm³⁺ or Er³⁺. The higher density of energy levels offers an increase in the number of possible transitions that can diminish emission brightness [22]. A more detailed investigation into the radiative and non-radiative mechanisms to determine the emission quenching mechanisms is required in the future.

 

Fig. 5 Emission spectrum of NaYF₄:Yb³⁺:Ln³⁺ for Ln³⁺ = Ho³⁺, Tm³⁺, Pr³⁺ and Er³⁺ particles upon excitation at 980 nm, showing tunable emission peaks at 1185, 1475, 1310, and 1530 nm, respectively. The labels indicate the ratio of the maximum emission peak intensities for the non-normalized emission spectra.

Download Full Size | PDF

The effect of Ho³⁺, Tm³⁺, and Pr³⁺ emitting ion concentrations on down-conversion emission brightness were investigated while keeping the sensitizer (Yb³⁺) concentration constant at 20 mol%. Distinct differences in the emission intensities of NaYF₄: Yb³⁺, Ln³⁺ (Ln³⁺ = Ho³⁺, Tm³⁺, Pr³⁺) were observed. The integrated emission intensities for each system at different emitting ion concentrations are shown in Fig. 6 on a semi-logarithmic scale. Figure 6(a) shows the non-linear behavior of emission intensities with respect to the concentration of rare earth emitting ions. The profiles for the various emitting ions show similar parabolic shapes on the semi-log scale. As the concentration of Ho³⁺, Tm³⁺, or Pr³⁺ in the host lattice initially increased, the emission intensities increased to an optimal point (see Fig. 6(a)). With further concentration increases, the emission intensity for Tm³⁺ or Pr³⁺ doped particles declined, while Ho³⁺ doped particles showed a pronounced plateau in their emission intensities before declining too. The optimal concentration levels were found to be 1 mol% for Ho³⁺ and Tm³⁺ and 0.1 mol% for Pr³⁺. The parabolic behavior in emission intensities results from two competing factors as the Ln³⁺emitting ion concentration is varied. Below the optimal concentration, an increase in Ln³⁺ concentration increased the number of emitting ions, leading to increased emission intensities with the increase in concentration up until the optimal concentration. The further increase in concentration of Ln³⁺ emitting ion leads to concentration quenching [23,24]. Concentration quenching is determined mainly by the dipole-dipole interaction between rare earth ions. The quenching effects vary according to R⁻⁶, where R is the interionic distance between emitting ions. The luminescence is completely quenched for ions separated at a distance shorter than R, whereas ions separated by a distance longer R is not subjected to complete quenching. The typical critical interionic distance where concentration quenching occurs is approximately between 0.5 to 2 nm for a range of different host systems [10,23,24]. The rare earth interionic distance of 5, 5 and 11 nm was estimated based on the optimum concentrations of Ho³⁺, Tm³⁺ and Pr³⁺ (i.e. 1.0, 1.0 and 0.1 mol%, respectively) observed at the maximum emission intensities in Fig. 6(a). These calculated rare earth interionic distances were found to be beyond the range where concentration quenching is typically significant [10,23,24].

 

Fig. 6 Integrated emission intensity as a function of (a) Ho³⁺, Tm³⁺, or Pr³⁺ concentration in NaYF₄, and (b) Yb³⁺ sensitizer concentration upon excitation at ~980 nm.

Download Full Size | PDF

Yb³⁺ (sensitizer)-concentration dependence was investigated in each system by holding the emitting ion concentration constant at levels that correspond to optimized emissions (1 mol% of Ho³⁺ and Tm³⁺, and 0.1 mol% of Pr³⁺). The integrated emission intensities of the particles at different sensitizer ion concentrations are shown in Fig. 6(b). The profiles show that emissions increase with increasing Yb³⁺ ions until it peaks at an optimal concentration before dropping off in a parabolic behavior. Results indicate that 20 mol% Yb³⁺ is the optimal concentration level regardless of emitting ion dopant in system. The behavior in emission intensities with variations in the sensitizer concentration is attributed to two competing factors that arise when the Yb³⁺-Ln³⁺ distances are varied. An increase in sensitizer concentration increases the number of Yb³⁺ ions and reduces the Yb³⁺-Ln³⁺ distances. Below the optimal concentration, this leads to enhanced energy transfer from Yb³⁺ to Ln³⁺ as the absorption of the 980 nm excitation was increased and transferred to the emitting ions. Successful energy transfer from sensitizer to emitting ion increases the likelihood of radiative transitions. At concentrations of Yb³⁺ greater than the optimal, additional Yb³⁺ ions reduced Yb³⁺ to Ln³⁺ transfer in favor of Yb³⁺ to Yb³⁺ transfer and relaxation via non-radiative pathways [16,25].

To quantify and compare the performance of these down-conversion infrared-emitting particles, the optical efficiencies of their respective emissions were measured under the same excitation wavelength of ~980 nm. Table 1 shows the the optical efficiencies of the Ln³⁺ doped powders in dry pressed pellets form. Kigre Erbium phosphate glass QE-7S and NaYF₄:Yb³⁺, Er³⁺ microparticles are included as an internal reference for the measurement of optical efficiencies [17]. The ranking of the down-conversion performance of these particles was NaYF₄:Yb³⁺, Er³⁺ (1.17%) > Kigre (0.45%) > NaYF₄:Yb³⁺, Ho³⁺ (0.021%) > NaYF₄:Yb³⁺,Tm³⁺ (0.0080%) > NaYF₄:Yb³⁺,Pr³⁺ (0.0017%). This ranking is consistent with the ranking of emission intensities discussed earlier (Fig. 5), where the ratios of the relative optical efficiencies are 688:265:12:5:1, respectively (Table 1). For the optical efficiency measurements, the emission intensities of the samples from all angles are collected using the integrating sphere. In contrast for measurements collected within the spectrometer, only the emission at one specific angle (i.e. 90° to incident excitation source) was measured. Therefore, the difference in the values for the ranking ratios obtained from Fig. 5 and Table 1 was due to the difference in accounting for sample emissions, where the optical efficiency reflects a more representative performance value as discussed in our other work [17]. The lower optical efficiency measured for Pr³⁺-, Ho³⁺- or Tm³⁺-doped systems compared with Er³⁺- doped systems (Fig. 1) can be attributed to several possibilities: density of energy levels, oscillatory strength, branching ratios, energy transfer rates and energy transfer or cross-relaxation pathways [22]. Further discussion regarding these differences is warranted when more detailed spectroscopy studies are performed in the future.

Table 1. Optical Efficiencies of Various Samples upon Excitation at ~980 nm

View Table

4. Conclusions

Polycrystalline hexagonal NaYF₄:Yb³⁺,Ln³⁺ (Ln³⁺ = Ho³⁺, Tm³⁺, Pr³⁺) microparticles with down-conversion properties were synthesized and optimized using a facile hydrothermal process. Distinct emissions in the second infrared window were observed when excited at 980 nm so as to render them potentially useful in multispectral imaging. Dopants concentration levels were optimized to give the maximum possible emissions from these rare earth doped systems. The optimum concentration level of Pr³⁺ dopant was found to be 10 times less than those of Ho³⁺ and Tm³⁺ dopant concentrations. Efficiencies were ranked as Er³⁺>Ho³⁺>Tm³⁺>Pr³⁺. The Pr³⁺ system was found to be ~12 times less efficient than Ho³⁺ and ~5 times lower than Tm³⁺. For the actual application of these rare earth doped systems in biomedical imaging, further research to improve the optical efficiency of these particles to levels comparable to the NaYF₄:Yb³⁺, Er³⁺ is warranted.