Introduction

Up-converting nanoparticles (UCNPs) have been intensively investigated in last years because of their unusual properties allowing, in general, for conversion of near-infrared light (NIR) to higher energetic, visible and even ultraviolet [1,2,3,4,5]. This phenomenon can be observed for materials containing lanthanide ions (Ln3+) because of the possibility of electronic transitions within their 4f subshell. Because of the properties of the 4f shell, sharp emission spectra, long luminescence lifetimes and massive Stokes shifts can be observed [6,7,8]. Nanomaterials containing Ln3+ ions are very attractive in many fields of science and industry. Their small size and the possibility of conversion of NIR light make them excellent for many applications, e.g. optical materials, displays, sensors, biological markers, drug delivery systems and many others [9,10,11,12].

Considering UCNPs for different applications, particularly in the biomedical field, the ability to control their morphology and spectroscopic properties is essential. These issues are still a challenge [13]. The simplest solution is to adjust the type and conditions of the synthesis route. So far a few main synthesis methods have been proposed, e.g. thermal decomposition, solvo(hydro)thermal synthesis or co-precipitation with polyols [14,15,16]. The most common and useful method is the thermal decomposition of precursors, which leads to highly uniform nanoparticles of specific shapes and sizes [15, 17,18,19]. However, in the hydrothermal synthesis, it is also possible to control the conditions so that to obtain particles of desired properties. The hydrothermal synthesis is usually conducted in water, under high pressure and temperature, in a special type of autoclave [15, 20, 21]. Moreover, the process, as well as equipment needed, is quite simple [22,23,24].

During hydrothermal synthesis, there are many variables like pressure, temperature, synthesis time, the volume of solution or stirring that can significantly influence the morphology and spectroscopic properties of UCNPs. Great importance in synthesis route has also the addition of hydrophilic compounds like sodium citrate, ethylenediaminetetraacetic acid (EDTA), cetyltrimethylammonium bromide (CTAB), as well as polymers, e.g. polyethylene glycol (PEG), polyethylenimine (PEI) and other [14, 16, 25]. These additives not only promote the formation of particles of certain size and shape but also improve the dispersion of UCNPs in water and stabilize the colloids. Additionally, surfactants may affect the spectroscopic properties of NPs.

Calcium fluoride, CaF2, is one of the best hosts for Ln3+ ions, thanks to its stability, low phonon energy, fluorite structure and very good compatibility with Ln3+ ions [26,27,28]. Moreover, CaF2 is a material with high rigidness, low refractive index and is optically transparent in a range from mid-infrared to UV [29]. As a host compound, CaF2 can be used in lasers [30], for bioimaging (high biocompatibility with living cell, non-toxic material) [31], biomedical sensors [32] and other applications. Also, this material can be easily obtained by a variety of methods such as sol–gel, solvo(hydro)thermal methods, polyol-mediated, thermal decomposition of precursors or colloidal techniques [33,34,35,36,37]. CaF2 is also easier to obtain by hydrothermal method as nanocrystals than NaREF4 materials [38,39,40]. Furthermore, our previous research indicated that using similar to presented here, hydrothermal method, sub-microspheres can be synthesized instead of nanoparticles and by incorporation of Mn2+ ions into CaF2 sub-microspheres the colour of up-conversion can be tuned [41].

In this work, CaF2 was used as a model, allowing to track the way in which dopant ions are incorporated, which allows for a better understanding of the UC process in similar materials. Moreover, the presented results of synthesis in different volumes or with, and without stirring give insight into problems of production of Ln3+-doped nanomaterials at larger scale.

Experimental section

Characterization

Powder diffractograms were recorded on a Bruker AXS D8 Advance diffractometer, with Cu Kα1 radiation λ = 1.5406 Å. The reference data were taken from the International Centre for Diffraction Data (ICDD). The composition of prepared materials was analysed by energy-dispersive X-ray spectroscopy (EDS), using Quanta 250 FEG, FEI, with voltage 30 kV. Transmission electron microscopy (TEM) images were recorded on a JEOL 1400 Transmission Electron Microscope, which used an accelerating voltage of 120 kV. Fourier transform infrared spectra (FT-IR) were recorded using a JASCO 4200 FT-IR spectrophotometer. Dynamic light scattering (DLS) and zeta potential measurements were performed by using a Malvern Zetasizer Nano ZS instrument, where the sample concentration was 0.25 mg/mL.

UV–Vis–NIR absorption spectra of powders were recorded with spectrophotometer JASCO V-770. The excitation and emission spectra of the prepared samples in the form of solid powders were measured on a Photon Technology International QuantaMaster™ 40 spectrofluorometer equipped with an Opolette 355LD UVDM tuneable laser, with a repetition rate of 20 Hz and a Hamamatsu R928 photomultiplier used as a detector. A continuous CNI multiwavelength (808, 975, 1208 and 1532 nm) 2 W CW diode laser was used as the excitation source, coupled to a 200 µm optical fibre and collimator for emission measurements and examination of relations between emission intensity and laser power. Laser beam size and power were measured by Ophir 10A-PPS sensor (CW laser) or by Coherent EnergyMax-USB J-10 MB-HE Energy Sensor (pulsed laser). As a detector, a Digital CCD Camera made by Princeton Instruments PIXIS:256E, equipped with an SP-2156 Imaging Spectrograph was applied, corrected for the instrumental response. Luminescence decay curves were recorded using a 200 MHz Tektronix MDO3022 oscilloscope, coupled to the R928 PMT and the QuantaMaster™ 40 spectrofluorometer. All spectra, i.e. excitation and emission were corrected for the instrumental response and OPO laser energy.

Synthesis of CaF 2 :20%Yb 3+ ,1%Er 3+ nanoparticles

In order to obtain 3.5 mmol of CaF2 doped with 20% of Yb3+ and 1% of Er3+, the aqueous solution of chlorides with concentration 1 M or 0.25 M was used. CaCl2 (2.77 mmol) and YbCl3 mixed with ErCl3 (0.7 mmol Yb3+ and 0.035 mmol Er3+) were added to 20 mL of 1 M aqueous solution of sodium citrate (NaCit) or 20 mL of 1 M aqueous solution of ammonium citrate (NH4Cit). Then, 5 mL of 2.10 M (1.5 excess to the stoichiometric amount, 1.5 × NH4F) or 5 mL of 4.2 M aqueous solution of NH4F (3 times excess to the stoichiometric amount, 3 × NH4F) as a source of fluoride ions was added to the solution containing CaCl2 and LnCl3 salts. The pH of the final solution was 7.5. The as-prepared transparent solution was transferred into 50 mL (35 mL of solution) or 100 mL (75 mL of solution) Teflon-lined vessel and hydrothermally treated for 12 h (200 °C, 15 bar), in an externally heated autoclave. Two different Berghof autoclaves were used: DAB-2 reactor for the smaller volume sample, without a stirrer, and BR-100 for the larger volume sample, with a stirrer. When the reaction was complete, the obtained white precipitate was purified by centrifugation and rinsed several times with water and ethanol. The final product was dried under ambient conditions. Additionally, CaF2:20%Yb3+, NaCit, 1.5 × NH4F in 35 mL of solution was prepared to investigate the influence of complexing agent on spectroscopic properties. Dopant concentrations were established based on literature data and earlier research, as well as synthesis conditions [25, 42].

Results and discussion

Structure and morphology

Cubic nanocrystals of CaF2 doped with lanthanide ions (Ln3+) were obtained by the hydrothermal synthesis in the presence of sodium citrate (NaCit) or ammonium citrate (NH4Cit) as a complexing agent. The prepared samples showed a single-phase structure with \(Fm\stackrel{-}{3}m\) space group, for both reactor volumes used (Fig. 1).

Figure 1
figure 1

source of citric ions

XRD patterns of the CaF2:Yb3+,Er3+ samples synthesized by the hydrothermal method: a without stirring, in 35 mL volume, b with stirring, 75 mL volume. The patterns are labelled according to the scheme: an excess of NH4F precipitating compound, a

Full size image

Additional information about the physical properties of prepared NPs was obtained from the cell parameter analysis (Table S1). An increase in the cell volume of all samples (163.26 to 166.65 Å3) in comparison with that of undoped CaF2 (163.04 Å3) was observed. The larger cell volumes were interpreted as a result of electronic repulsion between F− ions distributed in a cell in different positions due to local or nonlocal charge compensation as well as clusters formation [43, 44]. Additionally, differences in the cell size between the samples prepared with NaCit and NH4Cit are presented, which confirms the occurrence of two types of charge compensation processes (1st: 2 Ca2+  → Ln3+  + Na+, 2nd: Ca2+  → Ln3+  + F−). During the synthesis with NaCit, Na+ ions are incorporated into the structure replacing interstitial fluorine ions because their ionic radii are similar to those of calcium ions (rNa+ = 1.18, rCa2+ = 1.12, for coordination number CN = 8). In the process of synthesis using NH4Cit, F− ions act as a charge compensator, as the size of NH4+ cations is bigger (rNH4+ = 1.54 Å) [45,46,47,48]. Moreover, the change of cell size is also noticeable in the XRD patterns (Fig. 1) as a slight shift of the peaks towards lower angles for the samples prepared with NH4Cit and a shift towards higher angles for the sample with NaCit, obtained in the small volume, with 1.5 × NH4F excess.

TEM images were used to determine the accurate size of CaF2 nanoparticles, and the results are listed in Table 1.

Table 1 Size of obtained NPs, calculated from TEM analysis
Full size table

NPs sizes were in the range of 17.4–46.5 nm when the small volume was used and 13.8–40.4 nm for the larger volume (Table 1 and Fig. 2). Additionally, some particle agglomerations can be observed in TEM pictures, which was also confirmed by DLS histograms (Fig. S1 in Electronic Supplementary Information, ESI). The tendency to agglomeration is visible mainly for the samples prepared without stirring (Fig. 2a–d). Moreover, the NPs, obtained with the use of NH4Cit for the synthesis, had irregular shapes and were of larger sizes. To investigate the incorporation of Ln3+ ions and the real structure of the obtained compound, EDS mapping was made and the results are presented in Table S2. The amount of Ln3+ dopants is lower than assumed, but similar in all samples, i.e. Yb3+ 11.03—14.50% and Er3+ 0.47—1.14%, with one exception, for the sample prepared in the presence of NH4Cit with 3 × NH4F and in 75 mL of solution. The EDS analysis confirmed the incorporation of Na+ ions into the structure at the sites occupied by Ca2+ and F− ions when the samples were prepared in the presence of NaCit. The amount of Na+ ions was estimated to be between 16.50 and 28.11%. Additionally, the presence of citrate groups on NPs surface was confirmed by FT-IR measurements (Fig. S2). Prepared samples exhibited negative charge on the surface, except for the samples prepared in the presence of NH4Cit with 3 × NH4F in 35 mL, and NH4Cit, 1.5 × NH4F in 75 mL (Table S3) which had a positive charge. What is more, the NPs showed different stability in water (zeta potentials varied between |20.3| and |7.8| mV) at physiological pH for 24 h.

Figure 2
figure 2

Nanocrystals size distribution of hydrothermally synthesized samples: a CaF2:Yb3+,Er3+, NaCit, 1.5 × NH4F, b CaF2:Yb3+,Er3+, NaCit, 3 × NH4F, c CaF2:Yb3+,Er3+, NH4Cit, 1.5 × NH4F, d CaF2:Yb3+,Er3+, NH4Cit, 3 × NH4F, e CaF2:Yb3+,Er3+, NaCit, 1.5 × NH4F, f CaF2:Yb3+,Er3+, NaCit, 3 × NH4F, g CaF2:Yb3+,Er3+, NH4Cit, 1.5 × NH4F, h CaF2:Yb3+,Er3+, NH4Cit, 3 × NH4F, where a–d 35 mL, without stirring, and e–h 75 mL, with stirring

Full size image

Summarizing, it is possible to control the CaF2:Yb3+,Er3+ NPs morphology by selecting the appropriate concentration of NH4F and the type of co-reagent. From the above-presented results, it is also seen that the synthesis in a reactor with stirring should be more favourable, resulting in lower agglomeration of NPs. It is possible to obtain small NPs of the size below to 20 nm even with a high content of NH4F, which is quite essential when the synthesis is conducted in water.

Spectroscopic properties

It is well known from the literature, that only for low concentrations (< 0.1%), the Ln3+ dopants form isolated centres. Yb3+ ions are sensitive to the site symmetry, and their absorption spectra reflect this feature [49]. The mentioned centres have trigonal, tetragonal or cubic symmetry, depending on the location of charge compensating F− ions. In the heavily doped materials, Yb3+ ions form clusters, mainly cubooctahedral hexamers with the six Yb3+ ions site in square antiprisms [44, 50]. Cluster formation is well confirmed on the basis of the broad Yb3+ excitation bands that do not allow drawing conclusions on the Yb3+ ions site symmetries as it overlaps any other possible signals. However, for the samples prepared with NaCit, in 35 mL of solution, the formation of centres with the cubic symmetry (Oh) is responsible for the presence of peaks at 966 nm (10,352 cm−1) and 920 nm (10,870 cm−1), which are intense and well separated (Fig. 3a). After Na+ ions introduction into the structure as a charge compensator, Yb3+–Na+ ion pairs are formed. At the same time, a decrease in the number of Yb3+–Yb3+ pairs and formation of Yb2+, which are responsible for quenching luminescence through cooperative energy transfer from Er3+ ions, are observed [47].

Figure 3
figure 3

Excitation spectra of CaF2:20%Yb3+,1% Er3+ samples (900–1050 nm): a small volume, without stirring, b large volume, with stirring, excited by pulsed laser as excitation source (at 25 mJ·cm−2)

Full size image

As a result, the emission of particles with incorporated Na+ ions is more intense, which is observed in Figs. 4 and S3. What is interesting, the domination of Oh symmetry is only visible for the sample prepared with NaCit, 3 × NH4F in 35 mL solution, where more effective incorporation of Na+ ions replacing interstitial F− ions occurred. Furthermore, in the analogous sample but prepared in larger volume (75 mL) with stirring, a lower amount of Yb3+ sites with the Oh symmetry are present, despite the fact that the determined concentration of Na+ ions was higher (28%) than for the sample described above. It is worth noting that Na+ ions can be also present on the surface of NPs due to the bonding to citrate groups, which are incorporated into CaF2 NPs (Fig. S2). Interestingly, the excitation peak near 966 nm can be detected in the excitation spectra of all of the synthesized samples (Fig. 3), hence confirming the presence of the Oh symmetry of Yb3+ ions in the prepared NPs, regardless of the used surfactant, which has been also observed by another research group [44]. Absorption spectra of samples prepared are presented in Fig. S11 showing different characteristics of the Yb3+ absorption peaks, hence revealing also different efficiency of energy transfer between various types of Yb3+ ions and Er3+ ions.

Figure 4
figure 4

Luminescence (450–860 nm) spectra and emission colour of CaF2:20%Yb3+,1%Er3+ samples: a small volume, without stirring, b large volume, with stirring, where: (i) NaCit, 1.5 × NH4F, (ii) NaCit, 3 × NH4F, (iii) NH4Cit, 1.5 × NH4F, (iv) NH4Cit, 3 × NH4F, excited by laser under continuous excitation source (at 25 W·cm−2)

Full size image

For prepared samples, the emission spectra were measured under continuous diode laser with λex = 975 nm wavelength and are presented in Fig. 4. The brightest luminescence was recorded for the samples synthesized in the presence of NaCit and with 1.5 × NH4F in both volumes used, 35 mL, without stirring and 75 mL with stirring. The observed luminescence intensities confirm the influence of sodium ions on the effectiveness of emission. Additionally, samples prepared without stirring, with NaCit in 35 mL, exhibited twice time stronger emission than the best samples obtained in 75 mL solution with stirring. High luminescence can be connected with a decreased number of formed Yb3+–Yb3+ clusters which reduced the non-radiative energy losses. Changes in the cooperative energy transfer between Yb3+ ions and shorter distances between Yb3+ and Er3+ ions (see Table S1) can improve energy transfer.

The synthesis procedure also influences the samples emission colour, which is illustrated in the photographs, the calculated ratios between two of the strongest bands and the chromaticity diagrams (Figs. 4, S4, S5). Interestingly, for the samples prepared without stirring and in the small volume, the domination of red band over green is strongly visible, in comparison with the luminescence of the samples obtained with stirring and in larger volume, for which the 4S3/2 → 4I15/2 and 4F9/2 → 4I15/2 intensities are similar. As a result, the samples obtained in larger volume, with stirring, showed the yellow-green colour of emission. The most significant shift to the green region was recorded for the sample prepared in 75 mL with stirring, NaCit as co-reagent and 3 × NH4F, which is also noticeable in Figs. 4b (ii) and S5. Comparing the luminescence of the obtained samples prepared in the same way, it is possible to receive different emission intensity and colour, just by changing the volume and application of stirring.

More information about spectroscopic properties of synthesized CaF2 NPs was obtained from the luminescence decays of Er3+ ions, measured under λex = 976 and 966 nm wavelength with pulsed laser as the excitation source (for experimental data, see ESI, Figs. S6 and S7). Moreover, luminescence rise times as well as decays of Yb3+ under λex = 966/977 nm pulsed laser excitation were also recorded to investigate the energy transfer between Yb3+ ions in different sites and the influence of complexing agent on their lifetimes (Fig. S8). On the basis of these measurements, luminescence lifetimes were calculated for transitions of Er3+ and Yb3+ ions (collected in Tables 2, S4 and S5).

Table 2 Emission lifetimes calculated from the measured luminescence decay of CaF2:Yb3+,Er3+ NPs under 976 nm laser excitation (for decays, see Fig. S6, err < 1.6 μs). The number of photons involved in the up-conversion mechanism, determined from the dependencies of luminescence intensity on laser energy for CaF2:Yb3+,Er3+ NPs (for experimental results, see Fig. S9, err < 0.06)
Full size table

The longest luminescence decay of Er3+ ions was recorded for the sample prepared in the presence of NaCit and with 3 × NH4F, without stirring in 35 mL solution as well as NPs with NaCit, 1.5 × NH4F, in 75 mL solution. From all Er3+ transitions, the decay of 4F9/2 → 4I15/2 was recorded as the longest one for all samples (16.4 μs to 206.9 μs), which may be a result of strong emission of this band and the mechanism responsible for 4F9/2 energy level excitation. Importantly, for the samples with a large number of Yb3+ ions at Oh symmetry sites (NaCit, 1.5 × NH4F and 3 × NH4F in 35 mL of solution), longer lifetimes for Er3+ were calculated under 966 nm excitation than under 976 nm.

The Yb3+ decay time measurements did not reveal significant changes in the lifetimes of Yb3+ ions upon excitation with 966 or 976 nm wavelengths, even for the samples with a large number of Yb3+ ions at the sites of Oh symmetry. Taking into account a significant impact of site symmetry on Ln3+ emission lifetimes, it is expected to find longer lifetimes for the structure with sites of higher symmetry [51]. The explanation of the difference between the literature-based expectation and observations can be the domination of Yb3+–Yb3+ clusters in the sample’s structure or at least the presence of a mixture of sites of different symmetries, which makes it impossible to excite Yb3+ ions at the sites of a single symmetry. However, for the samples prepared in the presence of NaCit, relatively long rise times were observed, especially in comparison with those of the samples obtained with NH4Cit. What is more, the difference was also detectable for the same samples obtained by the two synthesis routes (NaCit, 3 × NH4F, 35 mL and NaCit, 3 × NH4F, 35 mL, Table S5, Fig. 5). The reason for this observation is related to the energy transfer from Oh centres of Yb3+ of higher energy (10 352 cm − 1) to Yb3+ cluster centres with lower energy (10 246 cm−1, 5 → 5 transitions, Fig. 6a). On the basis of these measurements and calculations, it can be concluded that the values of Yb3+ lifetimes are independent of the dominant symmetry (the presence or absence of Na+ ions) as well as of the excitation wavelength. At the same time, the luminescence rise times seem to be sensitive to the site symmetry of Yb3+ ions. This result is an additional confirmation of Yb3+ ions multisite positions in NPs; there is a fraction at sites of Oh symmetry and a fraction of those in clusters. According to the Hraiech et al., when a high number of Na+ ions are present in the sample, the band characteristic of Yb3+ ions with Oh symmetry appears in the NIR emission spectra with a maximum near 1028 nm. However, when a small number of Na+ ions were added, or for the samples without sodium ions, only the broad electronic transitions 5 → 3 and 5 → 4 of Stark’s level with a band maximum near 1030 and 1050 nm appeared in the spectra [51]. For the all obtained samples, the lifetime, as well as rise time, was measured for the emission at 1050 nm, which proves the presence of Yb3+ ions at the sites of Oh symmetry and clusters in all of obtained NPs.

Figure 5
figure 5

Excitation spectra of CaF2:Yb3+,Er3+a, b and decay time of Yb3+ 2F5/2 → 2F7/2 transition c, d under 966/977 nm pulsed excitation source (at 15 mJ⋅cm−2), observed at 1050 nm, where a, c sample with NaCit, 3 × NH4F, 35 mL solution, b, d sample with NH4Cit, 3 × NH4F, 35 mL solution

Full size image
Figure 6
figure 6

Scheme of the up-conversion mechanism for CaF2:Yb3+,Er3+ systems, where a proposed mechanism of energy transfer between Yb3+ ions with different symmetries, b up-conversion energy transfer in Yb3+–Er3+ system, under NIR excitation (λex = 975 nm)

Full size image

To establish the up-conversion mechanism of the prepared NPs and investigate the effects of the synthesis methodology on it, the dependencies of the luminescence intensity on the laser power were measured. The results of the slope calculations are collected in Table 2 (measurement results, Fig. S9). The slope coefficients calculated for the samples studied took values from the range 1 to 2, which is lower than the theoretical value for energy transitions in Er3+ ions [25]. The highest slope coefficient was calculated for the CaF2: Yb3+,Er3+ samples prepared with NaCit and 1.5 × NH4F in the small volume (35 mL), without stirring and with NaCit and 3 × NH4F with magnetic stirring and in the large volume (75 mL). Such a result can be connected with effective emission, long luminescent lifetimes and high crystallinity of these two samples (appropriate distance between Yb3+ and Er3+ ions can minimize quenching effects). The highest slope was determined for the 4F9/2 → 4I15/2 transition. This result can be explained by the relaxation from 2H11/2 or 4S3/2 level and the highest emission of a band corresponding to the described transition for all samples. For the few prepared nanomaterials, the slope coefficient is close to one; however, the up-conversion emission can be treated as a two-photon process [52]. There are many factors, which can influence the experimental slope like saturation effect, heating of samples or cross-relaxation process between dopants [53, 54]. Furthermore, quite often the saturation effect which is attributed to the competition between linear decay and up-conversion depletion of the intermediate state when excitation density is high takes place [19, 55, 56].

On the basis of the number of photons established for a population of each excited level, a UC mechanism-energy transfer up-conversion (ETU) can be proposed (Figs. 6 and S10). In the first step, Yb3+ ion absorbs a photon and excitation of 2F5/2 from 2F7/2 is observed. For the Yb3+ ions at Oh symmetry sites present in the structure, absorption from the ground state to 5, 6 and 7 Stark’s sublevel is observed, from which the energy can be transferred to Yb3+–Yb3+ clusters or directly to Er3+ ions (4I15/2 → 4I11/2 transition). For the hexameric clusters, the energy is absorbed from ground state mostly to 5 Stark’s sublevel and transferred to activator ions. These two possible ways of excitation of Yb3+ ions can occur simultaneously in one sample with mixed symmetry. The next step of the mechanism is the energy transfer from the excited Yb3+:2F5/2 to Er3+:4I11/2 and the absorption of the second photon, to populate 4F7/2, from which relaxation to 2H11/2, 4S3/2 occurs. Two-photons emission is also observed from 4F9/2 and 4I9/2 to 4I15/2.

Conclusions

Up-converting nanoparticles based on CaF2 matrix doped with lanthanide ions (Yb3+ and Er3+) were synthesized by the hydrothermal method. The influence of such factors as the type of co-reagent, excess of fluoride ions, volume and stirring on the morphology and spectroscopic properties of the nanoparticles was investigated. The results provided the evidence illustrating the importance of the synthesis procedure because of its effect on emission intensity, colour and excitation mechanism.

The main factor influencing NPs morphology was the excess of NH4F; with the higher concentration of F− ions in the solution, the obtained NPs were bigger. It can be related to a more effective precipitation process. The effect of size of NPs as a result of NH4F excess used in the synthesis on the spectroscopic properties was also investigated. Moreover, in both synthesis route, the samples prepared with NaCit and 1.5 × NH4F were characterized by the most intense emission. Additionally, the presence of Na+ ions changes the symmetry of Yb3+ ions, which was visible for the products prepared in 35 mL of the solution without stirring whose luminescence was almost twice higher than that of the other NPs. Moreover, luminescence lifetimes of Er3+ and the rise times of Yb3+ ions depended on the surfactant used for the synthesis and show that NaCit is more favourable.

Summarizing, we have established the ideal hydrothermal conditions to obtain small NPs with bright up-conversion luminescence under 975 nm (see comparison with NaYF4:Yb3+, Er3+ in Fig. S12), which are: synthesis in the presence of NaCit as an anti-agglomeration agent, suppression of NPs growth and 1.5 × NH4F precipitating agent and 12 h time of synthesis. Avoiding stirring during the reaction and small reaction volume, 35 mL, resulted in the highest intensity of luminescence from all prepared samples. However, increasing reaction volume to 75 mL and the introduction of stirring during synthesis also brought products with satisfactory luminescence intensity and NPs sizes.