Barium yttrium fluoride based upconversion nanoparticles as dual mode image contrast agents

Dual labeled contrast agents could provide better complementary information for bioimaging than available solely from a single modality. In this paper we investigate the suitability of Yb 3 + and Er upconversion nanoparticles (UCNPs) as both optical and X-ray micro computed tomography ( μ CT) contrast agents. Stable, aqueous UCNP dispersions were synthesised using a hydrothermal method with the addition of polyethyleneimine (PEI). UCNPs were single crystal and had a truncated cuboidal morphology, with average particle size of 47 ± 9 nm from transmission electron microscopy which was further used to characterize the structure and composition in detail. A zeta potential value of + 51 mV was measured for the aqueous nanoparticle dispersions which is beneficial for cell permeability. The outer hydrated PEI layer is also advantageous for the attachment of proteins for targeted delivery in biological systems. The prepared UCNPs were proven to be non- toxic to endothelial cells up to a concentration of 3.5 mg/mL, when assessed using an MTT assay. The particles showed intense green upconversion photoluminescence when excited at a wavelength of 976 nm using a diode laser. Quantitative X-ray μ CT contrast imaging confirmed the potential of these UCNPs as X-ray contrast agents and confirming their dual modality for bioimaging.


Introduction
Upconversion (UC) is a process that converts two or more lowerenergy photons in the near-infrared wavelengths into one higher energy photon in the visible or ultraviolet (UV) wavelengths [1]. Nanoparticles doped with lanthanide ions that can produce efficient upconversion are known as upconversion nanoparticles (UCNPs) and have been proven to be useful as fluorescent labels for biomedical imaging [1][2][3][4][5]. In a biological medium, UCNPs offer advantages such as higher penetration depth of infrared excitation radiation [2][3][4][5][6][7], excellent photostability and the absence of autofluorescence in comparison to fluorescent dyes, semiconductor quantum dots or down-conversion nanophosphors [8,9]. The long lifetime and ladder-like 4f n energy levels of trivalent lanthanide ions can produce higher energy anti-Stokes luminescence via incorporation into low phonon energy (<500 cm − 1 for fluorides) inorganic host nanocrystals. To be suitable for bioimaging or sensing, the nanocrystalline material for lanthanide ion doping should be transparent in the wavelength range of interest, should have a high optical damage threshold and be chemically stable. Fluoride-based host materials such as LaF 3 and NaYF 4 have been the most common materials of choice for effective upconversion as they possess all these characteristics [7][8][9][10][11][12][13]. There have been some studies on fluorides such as MF 2 and MF 2 -LnF 3 (where M represents a group II element) solid solutions [14][15][16][17]. The lanthanide ion-doped barium yttrium fluoride system has also been proposed as a host material due to the structural as well as optical properties [14,[18][19][20]; it is reported that bulk BaYF 5 doped with Er 3+ can present approximately eight times the UC emission than that of Er 3+ -doped LaF 3 nanocrystals [21]. Glass ceramics containing Er 3+doped BaYF 5 nanocrystals have shown 13 times brighter UC emission than correspondingly doped LaF 3 nanocrystals [22]. Syntheses of lanthanide-doped BaYF 5 nanocrystals have been carried out by codecomposition in organic solvents under inert gas atmosphere and high temperature [18][19][20][21][22][23]. Most of these methods produce hydrophobic nanoparticles which need further surface modification for biological applications. The synthesis of tetragonal BaYF 5 nanocrystals using EDTA as a chelating agent produced red emitting particles in multiple organic solvents with various size ranges depending on the pH as well as dopant concentrations, which were rendered soluble in water by a general PAA exchange procedure [24].
Apart from high UC emission, the barium fluoride system is also reported to show remarkable X-ray contrast [25] due to the high atomic number of Ba, as compared to Na, resulting in a K-edge energy of 37 keV. Hence, the BaYF 5 /Yb 3+ /Er 3+ system presents significant opportunities for dual modal imaging. Although larger particles will have higher photoluminescence, they are reported to be more susceptible to aggregation so limiting their penetration into cells and making them less suitable for biological imaging applications [26]. Moreover, spectroscopic investigation has revealed that upconversion luminescence intensity is dependent on particle morphology, size, and crystallinity [27]. Hence developing particles of the appropriate size for in vitro and in vivo imaging of cells with optimum photoluminescence efficiency is of significant interest.
Recent reviews have reported important advances in NIR lanthanidebased UC materials and their biomedical application [28][29][30]. However, there are few studies of the facile synthesis of aqueous-based UC systems, their complete structural characterisation and correlation with properties, as well as the assessment of their suitability for advanced application. Hence in the present work, we have modified the method to synthesise, and characterised the water dispersible BaYF 5 /Yb 3+ /Er 3+ nanoparticles coated with polyethyleneimine (PEI) for their use in bioimaging. The method reported here does not use organic solvents such as cyclohexane or toluene, and the direct dispersion of particles in water allows elimination of an additional hydrophilisation step. Detailed structural characterisation is another novelty of the work which is useful for advanced applications of these UCNPs in optical and X-ray imaging (dual mode) applications, confirmed via UC spectroscopy and μCT measurements. The non-toxic effects of these particles have been established via testing on endothelial cells, which demonstrates the suitability of the particles for use in biological applications.  2 (99.5%), were purchased from Sigma Aldrich. All chemicals were of analytical grade and were used without further purification. The final washing of the precipitate and dispersion of the nanoparticles were done using ultrapure water with a conductivity of 0.05 μS/cm, corresponding to a resistivity of 18 mega-ohm⋅cm (MΩ⋅cm).

Nanoparticle synthesis
A modified version of the experimental method adopted by Zhang et al. [31] was used to prepare BaYF 5 nanoparticles doped with Yb 3+ / Er 3+ using PEI and water as the solvent. The doping ratio was chosen to be BaY 0.78 Yb 0.2 Er 0.02 F 5 based on our previous studies of the NaYF 4 / Yb 3+ /Er 3+ system [32]. As the main objective of this current investigation was to test the suitability of these Ba-based UCNPs for dual mode imaging, we have not yet further optimised the doping ratio for this Babased system, which will be the subject of a future study.
In the first step, 1.6 g of PEI was dissolved in 15 mL of deionised water keeping the temperature at 50 • C. A second solution was prepared using 3.12 mmol of Y(NO 3 ) 3 ⋅6H 2 O, 0.8 mmol of Yb(NO 3 ) 3 ⋅5H 2 O, 0.08 mmol of Er(NO 3 ) 3 ⋅5H 2 O and 4 mmol of Ba(NO 3 ) 2 in 60 mL of deionised water. To this solution, the PEI solution was added dropwise followed by an ammonium fluoride solution (100 mmol of NH 4 F dissolved in 30 mL of deionised water) and this was stirred for 30 min to obtain a uniform homogenous mixture. The resultant solution yielded nanoparticles of the composition BaY 0.78 Yb 0.2 Er 0.02 F 5 . The pH of the solution was measured to be around 7.5. The solution was then transferred into a Teflon container and placed within a Parr pressure vessel, and heated at a temperature of 180 • C for 24 h in an oven. The resultant solution was then allowed to cool down to room temperature. The contents including the nanoparticulate precipitate in the cooled solution were then washed 3-4 times with ultrapure water by repeated ultracentrifugation at 80,000g for 30 min using a Beckman Avanti J20XP high speed centrifuge. The nanoparticle pellet thus obtained was re-dispersed in MilliQ water by sonication using an ultrasonic probe for a maximum of 2 min (Bandekin GM2070 with 100% power; cycle 0.7 s). The concentration of the suspension was determined to be 35 mg/mL. It was diluted 10 times and ultrasonicated before measuring the UC fluorescence and diluted as per the requirements for other measurements.

X-ray diffraction (XRD)
X-ray diffraction (XRD) measurements were made with a PANalytical X'Pert Pro Multipurpose Diffractometer at a scanning rate of 4 • /min in the 2θ range from 20 • to 80 • , with Cu Kα radiation (λ = 1.5406 A • ). An accelerating voltage and current of 40 kV and 40 mA respectively were used for measurement.

Transmission electron microscopy (TEM)
TEM samples were prepared by placing a diluted drop of UCNPs onto a holey carbon-coated copper TEM grid. TEM measurements were performed using a FEI Titan Themis 300 G2 FEG S/TEM operated at 300 kV and equipped with a Gatan OneView 16 Megapixel CMOS digital camera and Super-X EDX system with a windowless 4-detector design. Selected area electron diffraction (SAED) and lattice spacings in the images were analysed to confirm the crystallinity of the UCNPs. EDX analysis was used to obtain the approximate atomic ratios of the elements present in the UCNPs. The size-distribution histogram of nanoparticles was determined by image analysis (ImageJ software) of low magnification TEM images. High angle annular dark field (HAADF) scanning TEM (STEM) imaging and EDX mapping were carried out with a 1000 nA probe current and 23 μs dwell time with a specimen pixel size of 4 nm. High signal to noise ratio atomic resolution STEM HAADF images were recorded with a HAADF inner detector angle of 43.5 mrad and obtained by stacking of up to 93 frames; the accumulated electron dose being 8 × 10 4 electrons/Å 2 .

Dynamic light scattering (DLS)
A Malvern Zetasizer Nano ZS system (Malvern, UK) was used to determine the average hydrodynamic diameter and zeta potential of the nanoparticle suspension at a concentration of 3.5 mg/mL.

Luminescence measurements
The measurement of luminescence spectra and the corresponding decay curves of the nanoparticle suspensions were undertaken in a Spectrofluorimeter (Edinburgh Instruments, UK). The samples were excited using a focused 976 nm continuous wave (CW) laser with an approximate power density at focus of 162 kW/m 2 . The UC photoluminescence spectrum was measured in the wavelength range 400-800 nm. The decay of the UC photoluminescence were measured by modulating the laser at 100 Hz and the decay time obtained by linear fitting of the logarithmic intensity data.

Fluorescence microscopy
Fluorescence images of the particles were captured using a Zeiss Imager Z1 AX10 Apotome Microscope attached with a 976 nm CW diode laser (BL976-SA300, Thorlabs).

In vitro cytotoxicity evaluation
In vitro cell toxicity studies of UCNPs was performed on human umbilical vein endothelial cells (HUVECs) and assessed using the 3-(4,5dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma Aldrich) assay. Briefly, 10 4 HUVECs were grown overnight in a 96-well plate. The growth medium was prepared using Promocell Endothelial Cell Growth Medium supplemented with Promocell Endothelial cell supplements and 10% (v/v) foetal calf serum. Doses of nanoparticles, ranging from 0.3 μg/mL to 3.5 mg/mL in growth media, were added and incubated for a period of 24 h. The MTT assay was carried out the next day by adding 50 μL of 5 μg/mL MTT to each well. After the addition of MTT, the plate was incubated for a period of 3 h, until the purple product was formed. The control corresponds to cells without particles. These purple crystals were dissolved using 300 μL/well of isopropanol and the plate was read using 576 nm illumination using a Bio-Rad (Model-680) microplate reader.

Micro computed tomography (μCT) contrast
X-ray micro computed tomography (μCT) images of the samples were taken in a SkyScan 1176 μCT Scanner, using a source voltage of 65 kV, a source current of 380 (μA), and a 1 mm Al filter. The image reconstructions were performed using NRecon software with a smoothing parameter of 5, a ring artefact correction of 11 and a beam hardening correction of 30%. Inside each sample, equally sized round volume of interests (VOIs) were drawn and reconstructed. Image analysis was executed with CTAn, with values calibrated to water.

X-ray diffraction measurements
The X-ray diffraction pattern of the UCNP sample prepared in this research is presented in Fig. 1. The XRD data was found to agree well with ICSD Reference: 169849, which corresponds to sample with a cubic structure [33]. The results are also comparable with the studies of Liu et al. for a similar nanoparticle system where the authors compared the pattern to cubic BaCeF 5 (JCPDS Reference: space group-Fm3m 43-0394) [19,25]. Residual peaks at 2θ equal to 43 • and 49 • in Fig. 1 may result from traces of BaF 2 in the compound phase [34].
BaF 2 is matched with the ICSD Reference: 41649. It is reported that cubic BaF 2 acts as a structural director for the formation of cubic BaYF 5 nanocrystals [19]. The presence of these additional XRD peaks may also possibly correspond to very mild traces of the tetragonal BaYF 5 phase [33]. However, this latter explanation was ruled out for the nanoparticles reported here as these XRD peaks do not correspond to JCPDS Reference: 46-0039 corresponding to the tetragonal phase and were also not observed in the TEM analysis presented below. More detailed XRD patterns showing the fitted calculated curve, via the Rietveld method and the stick models corresponding to cubic BaYF 5 , cubic BaCeF 5 , BaF 2 and tetragonal BaYF 5 compared to the original measured XRD of the sample is presented in Fig. S1. The refinement shows a weighted profile R-factor (R wp ) value of 0.0049, consisting of 98.70 wt% BaYF 5 and 1.30 wt% BaF 2 . The calculated lattice parameter of the UCNP sample is 0.5962 ± 0.0016 nm. From a Williamson Hall analysis (Fig. S2) using the {111}, {200}, {220}, {311} XRD peaks, the average crystallite size was calculated to be 36.5 ± 6.4 nm with a lattice strain 0.27 ± 0.07%.

High resolution transmission electron microscopy
High resolution, phase contrast conventional TEM (CTEM) images and selected area electron diffraction (SAED) patterns of the synthesised nanoparticles are shown in Fig. 2. Fig. 2a is a CTEM image of particles recorded at 45,000× magnification suggesting the particles have a cuboidal, truncated cuboidal and truncated octahedral morphology and, owing to the presence of diffraction contrast, are crystalline.
The particle size distribution, obtained from a larger number of particles (ca. 100) is shown in Fig. 2b. Particles were distributed over the size range of 20-75 nm with a mean size of 47 ± 9 nm in good agreement with the crystallite size derived using XRD. The interplanar spacings (d hkl ) obtained from the selected area electron diffraction (SAED) pattern (Fig. 2c) were: d 111 = 0.347 ± 0.014 nm, d 200 = 0.304 ± 0.012 nm, d 220 = 0.214 ± 0.008 nm and d 311 = 0.182 ± 007 nm corresponding to a cubic lattice [19] with a lattice constant, a calculated to be 0.604 ± 0.021 nm. A comparison of the d spacings and lattice parameter calculated from TEM and XRD is presented in Supporting Information as Table S1. The lattice parameters derived from XRD (a = 0.5962 ± 0.0016 nm) and TEM (a = 0.604 ± 0.021 nm) agree within experimental error. Fig. 2d shows a higher magnification image (570,000×) of a single particle showing a truncated cuboidal shape while Fig. 2e corresponds to the fast Fourier transformed (FFT) image of the area marked in Fig. 2d which confirmed that the crystal lattice is face centered cubic [35]. The spots in the FFT are labeled with their {hkl} notation and it is clear from a comparison of Fig. 2d and e that the surface normals and hence large surface facets are predominantly {111} type with additional {100} type facets indicating strictly a truncated octahedral morphology. Fig. 3a is a high-resolution CTEM image revealing lattice fringes recorded from the particle shown in Fig. 2d; lattice fringes extend across the whole particle indicating that it is single crystal in nature. A Fourier filtered image of the lattice is shown in Fig. 3b and the intensity profile in Fig. 3c indicates an average d spacing of 0.345 nm, close to the {111} lattice spacing.
EDX spectra of the nanoparticles (Fig. S3) showed the presence of the main elements in atomic ratios, very comparable to the expected theoretical composition derived from the synthesis procedure. The average atomic% of the elements determined from EDX spectra from four different areas, each containing a cluster of nanoparticles, together with  Table 1. The experimental values indicate that the sample shows only relatively minor variations in homogeneity and are close to the desired composition.
In order to assess further the variation in elemental composition between particles, high angle annular dark field (HAADF) STEM imaging and STEM/EDX mapping of the various elements were recorded. Fig. 4a shows a STEM HAADF image of a group of nanoparticles and  STEM/EDX elemental mapping (Fig. 4b-f) qualitatively indicates the presence of all major elements evenly distributed amongst the particles. Fig. 5a is a magnified HAADF STEM image showing relatively even Zcontrast, indicative of compositional homogeneity within and between particles (except where particle overlap is apparent). A STEM/EDX line scan across an aggregate of three particles is presented in Fig. 5b (HAADF image and line scan position) and 5c (EDX intensity profiles). Again the results show an absence of intraparticle segregation and a reasonably homogeneous distribution of elements. Note any variation in the fluorine signal (and overall composition in Table 1) may be due to its beam sensitivity relative to the other elements present.

Particle size distribution and zeta potential from dynamic light scattering (DLS)
The particle size distribution of UCNPs in aqueous solution and their zeta potential was measured using DLS and the results are reported in Fig. 6. From Fig. 6a we can see that relatively monodispersed particles with mean hydrodynamic particle diameter of around 90 nm dominate the particle size distribution. This is higher than the average size obtained from TEM (Fig. 2b) and is consistent with hydrodynamic diameters generally obtained in DLS measurement, whereby the hydrodynamic diameter includes the contribution of the surfactant coating [36]. PEI-UCNP particles showed long-term stability in an aqueous environment without any noticeable aggregation. The maximum zeta potential value of +51 mV for the particles obtained from the zetasizer analysis measurement is presented in Fig. 6b. It suggests that PEI-UCNPs had cationic surfaces attributed to the protonated amino groups of the PEI at neutral pH which act as a hydrophilic head. This repulsion between cationic particles results in a highly stable dispersion which showed excellent long-term stability which was assessed by DLS measurement of the aqueous dispersion after 6 months and revealed no noticeable change in the particle hydrodynamic size distribution and no observable sedimentation of the particles. Aggregation does occur at higher pH when the nanoparticles are deprotonated. The pH of our synthesised aqueous suspension is found to be 7.5, where the particles are well dispersed. The stability of the particles in hydrophilic conditions makes these particles attractive for biological applications. The positive zeta potential is also advantageous for better cell membrane permeability, relative to negatively charged particles [37]. Thus, overall the high positive zeta potential value of the UCNPs is beneficial for tissue and cellular imaging.

Cytotoxicity analysis
Cytotoxicity is an important concern where the application of nanoparticles in biological systems is considered. Hence, the toxicity of the PEI-coated nanoparticles towards endothelial cells was evaluated using the MTT assay. Fig. 7 shows the cell viability data, which is a measure of the cytotoxicity of the sample. The data indicated no toxicity on endothelial cells with UCNP concentration ranging from 0.3 μg/mL to 3.5 mg/mL, compared to control (cells without particles). Exposing the cells to the UCNPs for 24 h resulted in an average cell viability of 90-100%. Fig. 8a shows a photograph of the as prepared colloidal suspension of the nanoparticles at a concentration of 35 mg/mL while Fig. 8b shows their green UC photoluminescence under diode laser illumination at a wavelength of 976 nm. The luminescence emission spectrum of the UCNP suspension is illustrated in Fig. 8c and shows an intense doublet for green emission as compared to red. The emission bands in the range 510 nm to 530 nm and between 530 and 570 nm corresponding to the 2 H 11/2 → 4 I 15/2 and 4 S 3/2 → 4 I 15/2 transitions of Er 3+ ions contribute to the green emission intensity. The relatively less intense doublet peak in the red wavelength region (635-694 nm) is due to the 4 F 9/2 → 4 I 15/2 transition [11]. The lineshape of the observed UC spectrum differ from previously reported research in these type of nanoparticles [38,39], and in this research we also report the lifetimes of both green and red emissions. The combination of both red and green emission under 976 nm NIR wavelength excitation appeared more greenish to the eye/ camera as shown in Fig. 8b. Owing to their higher absorption cross- Fig. 6. a) DLS hydrodynamic particle size distribution and b) zeta potential values of the BaYF 5 /Yb 3+ /Er 3+ UCNP dispersion (concentration of 3.5 mg/mL). to the control: a) 0.3 μg/mL b) 3 μg/mL c) 30 μg/mL d) 300 μg/mL e) 500 μg/ mL f) 700 μg/mL g) 1 mg/mL h) 3 mg/mL and i) 3.5 mg/mL. section, Yb 3+ ions absorb more photons from the laser exciting them from the 2 F 7/2 to 2 F 5/2 level and subsequently transferring the energy to Er 3+ to excite them from 4 I 15/2 level to the 4 I 11/2 [1].

Photoluminescence and in vitro micro computed tomography (μCT)
The cross-relaxation energy transfer process leads to an increased probability of photon absorption by the excited Er 3+ : from the 4 I 11/2 level to the 4 F 7/2 level. As a result of non-radiative relaxation of Er 3+ ions from 4 F 7/2 to 2 H 11/2 and 4 S 3/2 and the subsequent radiative relaxation from these levels to I 15/2 results in green emission. Alternatively, the excited ion can relax non-radiatively from 2 H 11/2 and 4 S 3/2 levels and populate the 4 F 9/2 level leading to red emission due to the 4 F 9/2 → 4 I 15/2 radiative transition. Regarding the red emission, the energy level Er 3+ : 4 F 9/2 can also be populated through two other processes, the efficiency of these are dependent on the host material and doping concentrations. The first one is excited state absorption from the 4 I 13/2 level, which can be populated through non-radiative relaxation from the 4 I 11/2 level. The second process is relaxation of 4 F 7/2 → 4 F 9/2 and 4 I 11/2 → 4 F 9/2 between two nearby Er 3+ ions [40]. The green to red emission intensity ratio (G/R) was found to be 1.4 by consideration of the peak intensities at 539 nm and 667 nm for the excitation laser fluence used in this study. The photoluminescence decay lifetimes of the 541 nm and 667 nm emission of the UCNPs were obtained by linear fitting of the intensity in the logarithmic scale (Fig. 8d). From Fig. 8d, the BaYF 5 /Yb 3+ /Er 3+ particles studied in this paper have a photoluminescence lifetime values of τ = 324 μs for 541 nm emission peak and τ = 655 μs for 667 nm emission peak.
Photoluminescence imaging of the particles alone (i.e. following oven drying of a drop of UCNP suspension on a glass slide) showed green emission in accordance with the spectral results for the suspension (Fig. 9). During the process of drying the particles became aggregated and were monitored using the fluorescence microscope using 976 nm laser excitation.
In vitro X-ray μCT images of the UCNP suspensions at various nanoparticle concentrations are presented in Fig. 10 and these are quantified in terms of Hounsfield Units (HU) in Fig. 11. Hounsfield Units, which is a relative quantitative measurement of radio density used by radiologists in the interpretation of computed tomography are commonly used for the quantification of contrast in both in vitro and in vivo X-ray μCT images [41,42]. The enhanced contrast associated with an increasing concentration of nanoparticles is evident in Fig. 10. However, a clear contrast was not observed for lower concentrations of particles (images of all the concentrations are not shown in Fig. 10), despite the HU value increasing for concentrations in the range 0.5 to 3.5 mg/mL (Fig. 11). However, at higher concentrations, the contrast is visibly prominent in the images shown in Fig. 10.
In comparison, the commercial X-ray imaging agent (iopromide injection solution) shows a value of <40 HU in the 0-3 mg/mL concentration range. The higher atomic number and density of Yb (70 and 6.8 g/cm 3 ) compared to iodine (53 and 4.9 g/cm 3 ) and also the increased atomic number and density of Ba contribute to the overall CT contrast of the UCNPs. We report an HU value of 123 for the current BaYF 5 /Yb 3+ / Er 3+ particles at a concentration level of 3.5 mg/mL, which is much higher when compared to various other rare earth-based systems previously studied, where a value of 140 is achieved only at a concentration of around 6.5 mg/mL [43]. The quantification of the contrast using HU values have not been undertaken in many studies even though certain in vivo imaging results have been previously reported using materials system similar to this material [25]. Thus results of the present study demonstrate good upconversion PL and X-ray μCT properties evident from the reported PL spectrum and Hounsfield Unit values for BaYF 5 / Yb 3+ /Er 3+ UCNPs. The sharpness of the red and green photoluminescence peaks will be advantageous for biosensing applications as well. The synergistic combination of upconversion PL properties and Xray μCT contrast properties in a single nanoparticle system allows its application for dual mode imaging. As is well established, the phenomena of upconversion luminescence exhibited by the rare earth materials systems such as BaYF 5 /Yb 3+ /Er 3+ nanoparticles reported in this study produce minimal cell and tissue damage due to the 976 nm near infrared (NIR) excitation adopted. This is highly advantageous compared to the fluorescence exhibited by other nanoparticles, such as quantum dots and Au nanoparticles, which require UV excitation, resulting in limited tissue penetration and the possibility for phototoxicity, with damage to cells and tissues.
Future possible bioimaging applications of the material include improving the UCPL intensity by varying sensitizers-activator concentrations and combining with other modalities. Trimodal imaging combinations such as UCPL/CT/PET, by doping with PET isotopes such as 18 F, or UCPL/CT/MRI by doping with magnetic nuclei such as Gd 3+ or Fe 3 O 4 could be suitable candidates for more sensitive diagnostic imaging.

Conclusions
Yb 3+ /Er 3+ -doped BaYF 5 upconversion nanoparticles were prepared with PEI using a hydrothermal route. Particles were found to be single crystalline with a cuboidal morphology and of a homogeneous chemical composition. TEM imaging revealed an average particle size of 47 ± 9 nm, whereas DLS showed a hydrodynamic size distribution peaked around 90 nm due to the presence of a hydrated PEI layer which provided good stability and dispersability in aqueous media, as well as better cell membrane permeability, as indicated by the high zeta potential value of +51 mV. The particles were non-toxic towards endothelial cells (90-100% cell viability) across a wide concentration range as established by an MTT assay. Excitation of a colloidal suspension of the particles at 976 nm wavelength indicated intense green photoluminescence with a less intense red peak. Fluorescence microscopy of the dried nanoparticles indicated high green emission, under excitation by a 976 nm laser. Quantification of X-ray μCT contrast images strongly confirmed the suitability of these materials as CT contrast agents. The synergistic combination of upconversion photoluminescence and X-ray μCT contrast properties in a single nanoparticle system allows dual mode imaging application. Future possibilities of these dual mode upconversion particles include trimodal imaging combinations for sensitive diagnostic imaging, incorporating magnetic nuclei and PET isotopes. Further investigations and detailed surface modifications with appropriate bioconjugation procedures on these particles could be adopted for using these particles in biosensing of disease markers and bioimaging applications.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.